Research phase for Create test suite

* feat: add code variables capture

* feat: bump version and add changelog

.env.example

@@ -0,0 +1,11 @@
+# PostHog API Configuration
+# Copy this file to .env and update with your actual values
+
+# Your project API key (found on the /setup page in PostHog)
+POSTHOG_PROJECT_API_KEY=phc_your_project_api_key_here
+
+# Your personal API key (for local evaluation and other advanced features)
+POSTHOG_PERSONAL_API_KEY=phx_your_personal_api_key_here
+
+# PostHog host URL (remove this line if using posthog.com)
+POSTHOG_HOST=http://localhost:8000

.github/dependabot.yml

@@ -0,0 +1,36 @@
+version: 2
+updates:
+  - package-ecosystem: "pip"
+    directory: "/"
+    schedule:
+      interval: "daily"
+      time: "10:00"
+      timezone: "UTC"
+    groups:
+      ai-providers:
+        patterns:
+          - "openai"
+          - "anthropic"
+          - "google-genai"
+          - "langchain-core"
+          - "langchain-community"
+          - "langchain-openai"
+          - "langchain-anthropic"
+          - "langgraph"
+    allow:
+      - dependency-name: "openai"
+      - dependency-name: "anthropic"
+      - dependency-name: "google-genai"
+      - dependency-name: "langchain-core"
+      - dependency-name: "langchain-community"
+      - dependency-name: "langchain-openai"
+      - dependency-name: "langchain-anthropic"
+      - dependency-name: "langgraph"
+    open-pull-requests-limit: 1
+    reviewers:
+      - "PostHog/team-llm-analytics"
+    # Uncomment below to enable auto-merge for minor updates when CI passes
+    # pull-request-branch-name:
+    #   separator: "/"
+    # assignees:
+    #   - "PostHog/ai-team"

.github/workflows/call-flags-project-board.yml

@@ -0,0 +1,17 @@
+# This workflow is used to call the flags-project-board workflow when a pull request is opened, ready for review, review requested, synchronized, converted to draft, or reopened.
+# It is used to update the feature flags project board with the pull request information.
+
+name: Call Feature Flags Project Workflow
+
+on:
+    pull_request:
+        types: [opened, ready_for_review, review_requested, synchronize, converted_to_draft, reopened]
+
+jobs:
+    call-flags-project:
+        uses: PostHog/.github/.github/workflows/flags-project-board.yml@main
+        with:
+            pr_number: ${{ github.event.pull_request.number }}
+            pr_node_id: ${{ github.event.pull_request.node_id }}
+            is_draft: ${{ github.event.pull_request.draft }}
+        secrets: inherit

.github/workflows/ci.yml

@@ -3,44 +3,38 @@ name: CI
 on:
     - pull_request
 
+permissions:
+    contents: read
+
 jobs:
     code-quality:
         name: Code quality checks
         runs-on: ubuntu-latest
 
         steps:
-            - uses: actions/checkout@v2
+            - uses: actions/checkout@85e6279cec87321a52edac9c87bce653a07cf6c2
               with:
                   fetch-depth: 1
 
             - name: Set up Python 3.11
-              uses: actions/setup-python@v2
+              uses: actions/setup-python@8d9ed9ac5c53483de85588cdf95a591a75ab9f55
               with:
                   python-version: 3.11.11
 
-            - uses: actions/cache@v3
+            - name: Install uv
+              uses: astral-sh/setup-uv@0c5e2b8115b80b4c7c5ddf6ffdd634974642d182 # v5.4.1
               with:
-                  path: ~/.cache/pip
-                  key: ${{ runner.os }}-pip-${{ hashFiles('setup.py') }}
-                  restore-keys: |
-                      ${{ runner.os }}-pip-
+                  enable-cache: true
+                  pyproject-file: 'pyproject.toml'
 
             - name: Install dev dependencies
+              shell: bash
               run: |
-                  python -m pip install -e .[dev]
-              if: steps.cache.outputs.cache-hit != 'true'
+                  UV_PROJECT_ENVIRONMENT=$pythonLocation uv sync --extra dev
 
-            - name: Check formatting with black
+            - name: Check formatting with ruff
               run: |
-                  black --check .
-
-            - name: Lint with flake8
-              run: |
-                  flake8 posthog --ignore E501,W503
-
-            - name: Check import order with isort
-              run: |
-                  isort --check-only .
+                  ruff format --check .
 
             - name: Check types with mypy
               run: |
@@ -54,19 +48,57 @@ jobs:
                 python-version: ['3.9', '3.10', '3.11', '3.12', '3.13']
 
         steps:
-            - uses: actions/checkout@v2
+            - uses: actions/checkout@85e6279cec87321a52edac9c87bce653a07cf6c2
               with:
                   fetch-depth: 1
 
             - name: Set up Python ${{ matrix.python-version }}
-              uses: actions/setup-python@v2
+              uses: actions/setup-python@8d9ed9ac5c53483de85588cdf95a591a75ab9f55
               with:
                   python-version: ${{ matrix.python-version }}
 
-            - name: Install requirements.txt dependencies with pip
+            - name: Install uv
+              uses: astral-sh/setup-uv@0c5e2b8115b80b4c7c5ddf6ffdd634974642d182 # v5.4.1
+              with:
+                  enable-cache: true
+                  pyproject-file: 'pyproject.toml'
+
+            - name: Install test dependencies
+              shell: bash
               run: |
-                  python -m pip install -e .[test]
+                  UV_PROJECT_ENVIRONMENT=$pythonLocation uv sync --extra test
 
             - name: Run posthog tests
               run: |
                   pytest --verbose --timeout=30
+
+    django5-integration:
+        name: Django 5 integration tests
+        runs-on: ubuntu-latest
+
+        steps:
+            - uses: actions/checkout@85e6279cec87321a52edac9c87bce653a07cf6c2
+              with:
+                  fetch-depth: 1
+
+            - name: Set up Python 3.12
+              uses: actions/setup-python@8d9ed9ac5c53483de85588cdf95a591a75ab9f55
+              with:
+                  python-version: 3.12
+
+            - name: Install uv
+              uses: astral-sh/setup-uv@0c5e2b8115b80b4c7c5ddf6ffdd634974642d182 # v5.4.1
+              with:
+                  enable-cache: true
+                  pyproject-file: 'integration_tests/django5/pyproject.toml'
+
+            - name: Install Django 5 test project dependencies
+              shell: bash
+              working-directory: integration_tests/django5
+              run: |
+                  UV_PROJECT_ENVIRONMENT=$pythonLocation uv sync
+
+            - name: Run Django 5 middleware integration tests
+              working-directory: integration_tests/django5
+              run: |
+                  uv run pytest test_middleware.py test_exception_capture.py --verbose

.github/workflows/generate-references.yml

@@ -0,0 +1,48 @@
+name: "Generate References"
+
+on:
+  workflow_dispatch:
+
+jobs:
+  docs-generation:
+    name: Generate references
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout the repository
+        uses: actions/checkout@85e6279cec87321a52edac9c87bce653a07cf6c2
+        with:
+          fetch-depth: 0
+          token: ${{ secrets.POSTHOG_BOT_PAT }}
+
+      - name: Set up Python
+        uses: actions/setup-python@8d9ed9ac5c53483de85588cdf95a591a75ab9f55
+        with:
+          python-version: 3.11.11
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@0c5e2b8115b80b4c7c5ddf6ffdd634974642d182 # v5.4.1
+        with:
+            enable-cache: true
+            pyproject-file: 'pyproject.toml'
+
+      - name: Generate references
+        run: |
+          uv run bin/docs generate-references
+
+      - name: Check for changes in references
+        id: changes
+        run: |
+          if [ -n "$(git status --porcelain references/)" ]; then
+            echo "changed=true" >> $GITHUB_OUTPUT
+            echo "New references generated in references directory:"
+            git status --porcelain references/
+          else
+            echo "changed=false" >> $GITHUB_OUTPUT
+            echo "No new references generated in references directory"
+          fi
+
+      - uses: stefanzweifel/git-auto-commit-action@778341af668090896ca464160c2def5d1d1a3eb0
+        if: steps.changes.outputs.changed == 'true'
+        with:
+          commit_message: "Update generated references"
+          file_pattern: references/

.github/workflows/release.yaml

@@ -1,43 +0,0 @@
-name: "Release"
-
-on:
-  push:
-    branches:
-      - master
-    paths:
-      - "posthog/version.py"
-  workflow_dispatch:
-
-jobs:
-  release:
-    name: Publish release
-    runs-on: ubuntu-latest
-    env:
-      TWINE_USERNAME: __token__
-      TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
-    steps:
-      - name: Checkout the repository
-        uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-          token: ${{ secrets.POSTHOG_BOT_GITHUB_TOKEN }}
-
-      - name: Set up Python
-        uses: actions/setup-python@v2
-
-      - name: Detect version
-        run: echo "REPO_VERSION=$(python3 posthog/version.py)" >> $GITHUB_ENV
-
-      - name: Prepare for building release
-        run: pip install -U pip setuptools wheel twine
-
-      - name: Push release to PyPI
-        run: make release && make release_analytics
-
-      - name: Create GitHub release
-        uses: actions/create-release@v1
-        env:
-          GITHUB_TOKEN: ${{ secrets.POSTHOG_BOT_GITHUB_TOKEN }}
-        with:
-          tag_name: v${{ env.REPO_VERSION }}
-          release_name: ${{ env.REPO_VERSION }}

.github/workflows/release.yml

@@ -0,0 +1,57 @@
+name: "Release"
+
+on:
+  push:
+    branches:
+      - master
+    paths:
+      - "posthog/version.py"
+  workflow_dispatch:
+
+jobs:
+  release:
+    name: Publish release
+    runs-on: ubuntu-latest
+    env:
+      TWINE_USERNAME: __token__
+      TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+    steps:
+      - name: Checkout the repository
+        uses: actions/checkout@85e6279cec87321a52edac9c87bce653a07cf6c2
+        with:
+          fetch-depth: 0
+          token: ${{ secrets.POSTHOG_BOT_PAT }}
+
+      - name: Set up Python
+        uses: actions/setup-python@8d9ed9ac5c53483de85588cdf95a591a75ab9f55
+        with:
+          python-version: 3.11.11
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@0c5e2b8115b80b4c7c5ddf6ffdd634974642d182 # v5.4.1
+        with:
+            enable-cache: true
+            pyproject-file: 'pyproject.toml'
+            
+      - name: Detect version
+        run: echo "REPO_VERSION=$(python3 posthog/version.py)" >> $GITHUB_ENV
+
+      - name: Prepare for building release
+        run: uv sync --extra dev
+
+      - name: Push releases to PyPI
+        run: uv run make release && uv run make release_analytics
+
+      - name: Create GitHub release
+        uses: actions/create-release@0cb9c9b65d5d1901c1f53e5e66eaf4afd303e70e # v1
+        env:
+          GITHUB_TOKEN: ${{ secrets.POSTHOG_BOT_PAT }}
+        with:
+          tag_name: v${{ env.REPO_VERSION }}
+          release_name: ${{ env.REPO_VERSION }}
+      
+      - name: Dispatch generate-references for posthog-python
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          gh workflow run generate-references.yml --ref master

.gitignore

@@ -17,3 +17,6 @@ posthog-analytics
 .coverage
 pyrightconfig.json
 .env
+.DS_Store
+posthog-python-references.json
+.claude/settings.local.json

.posthog/4aacaff7-1e62-4f50-ada6-c4a8a3b27831/research.json

@@ -0,0 +1,40 @@
+{
+  "actionabilityScore": 0.15,
+  "context": "The task description is too vague to evaluate. The repository already has a comprehensive, mature test suite with 11,000+ LOC of tests across pytest/unittest, async testing, AI integrations, and Django middleware. Without clarity on what new tests should be created, I cannot assess actionability.",
+  "keyFiles": [],
+  "blockers": [
+    "Task description is just the word 'test' - completely unclear what should be tested",
+    "Repository already has extensive existing test coverage - unclear if task is to expand, fix, or reorganize tests",
+    "No specification of what modules, features, or behaviors need testing"
+  ],
+  "questions": [
+    {
+      "id": "q1",
+      "question": "What specific modules, features, or code areas need new tests?",
+      "options": [
+        "Expand existing test coverage for a specific module (e.g., add tests for uncovered edge cases in consumer.py, request.py, or size-limited dict)",
+        "Create tests for a new feature or integration that's missing coverage",
+        "Reorganize or improve existing tests (e.g., add more async tests, increase parameterization, improve mocking strategy)"
+      ]
+    },
+    {
+      "id": "q2",
+      "question": "What is the goal or context for this test creation?",
+      "options": [
+        "Increase code coverage percentage to meet a specific threshold target",
+        "Improve test reliability or add missing test patterns (e.g., more parameterized tests, better async coverage)",
+        "Test a recently added feature or bug fix that lacks test coverage"
+      ]
+    },
+    {
+      "id": "q3",
+      "question": "Are there specific test requirements or constraints?",
+      "options": [
+        "Follow existing patterns (pytest/unittest, use mock.patch, AsyncMock, freezegun as needed)",
+        "Introduce new testing tools or frameworks (e.g., switch to pure pytest fixtures, add property-based testing with hypothesis)",
+        "Focus on specific test categories (unit tests, integration tests, performance tests, or AI-specific tests)"
+      ]
+    }
+  ],
+  "answered": false
+}

.pre-commit-config.yaml

@@ -1,9 +1,10 @@
 repos:
-  - repo: https://github.com/psf/black
-    rev: stable
-    hooks:
-    - id: black
-  - repo: https://github.com/pycqa/isort
-    rev: 5.7.0
-    hooks:
-      - id: isort
+- repo: https://github.com/astral-sh/ruff-pre-commit
+  # Ruff version.
+  rev: v0.11.12
+  hooks:
+    # Run the linter.
+    - id: ruff-check
+      args: [ --fix ]
+    # Run the formatter.
+    - id: ruff-format

BEFORE_SEND.md

@@ -0,0 +1,237 @@
+# Before Send Hook
+
+The `before_send` parameter allows you to modify or filter events before they are sent to PostHog. This is useful for:
+
+- **Privacy**: Removing or masking sensitive data (PII)
+- **Filtering**: Dropping unwanted events (test events, internal users, etc.)
+- **Enhancement**: Adding custom properties to all events
+- **Transformation**: Modifying event names or property formats
+
+## Basic Usage
+
+```python
+import posthog
+from typing import Optional, Dict, Any
+
+def my_before_send(event: Dict[str, Any]) -> Optional[Dict[str, Any]]:
+    """
+    Process event before sending to PostHog.
+    
+    Args:
+        event: The event dictionary containing 'event', 'distinct_id', 'properties', etc.
+    
+    Returns:
+        Modified event dictionary to send, or None to drop the event
+    """
+    # Your processing logic here
+    return event
+
+# Initialize client with before_send hook
+client = posthog.Client(
+    api_key="your-project-api-key",
+    before_send=my_before_send
+)
+```
+
+## Common Use Cases
+
+### 1. Filter Out Events
+
+```python
+from typing import Optional, Any
+
+def filter_events_by_property_or_event_name(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Drop events from internal users or test environments."""
+    properties = event.get("properties", {})
+    
+    # Choose some property from your events
+    event_source = properties.get("event_source", "")
+    if event_source.endswith("internal"):
+        return None  # Drop the event
+    
+    # Filter out test events
+    if event.get("event") == "test_event":
+        return None
+    
+    return event
+```
+
+### 2. Remove/Mask PII Data
+
+```python
+from typing import Optional, Any
+
+def scrub_pii(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Remove or mask personally identifiable information."""
+    properties = event.get("properties", {})
+    
+    # Mask email but keep domain for analytics
+    if "email" in properties:
+        email = properties["email"]
+        if "@" in email:
+            domain = email.split("@")[1]
+            properties["email"] = f"***@{domain}"
+        else:
+            properties["email"] = "***"
+    
+    # Remove sensitive fields entirely
+    sensitive_fields = ["my_business_info", "secret_things"]
+    for field in sensitive_fields:
+        properties.pop(field, None)
+    
+    return event
+```
+
+### 3. Add Custom Properties
+
+```python
+from typing import Optional, Any
+
+from datetime import datetime
+from typing import Optional, Any
+
+def add_context(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Add custom properties to all events."""
+    if "properties" not in event:
+        event["properties"] = {}
+    
+    event["properties"].update({
+        "app_version": "2.1.0",
+        "environment": "production", 
+        "processed_at": datetime.now().isoformat()
+    })
+    
+    return event
+```
+
+### 4. Transform Event Names
+
+```python
+from typing import Optional, Any
+
+def normalize_event_names(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Convert event names to a consistent format."""
+    original_event = event.get("event")
+    if original_event:
+        # Convert to snake_case
+        normalized = original_event.lower().replace(" ", "_").replace("-", "_")
+        event["event"] = f"app_{normalized}"
+    
+    return event
+```
+
+### 5. Log and drop in "dev" mode
+
+When running in local dev often, you want to log but drop all events
+
+
+```python
+from typing import Optional, Any
+
+def log_and_drop_all(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Convert event names to a consistent format."""
+    print(event)
+    
+    return None
+```
+
+### 6. Combined Processing
+
+```python
+from typing import Optional, Any
+
+def comprehensive_processor(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    """Apply multiple transformations in sequence."""
+    
+    # Step 1: Filter unwanted events
+    if should_drop_event(event):
+        return None
+    
+    # Step 2: Scrub PII
+    event = scrub_pii(event)
+    
+    # Step 3: Add context
+    event = add_context(event)
+    
+    # Step 4: Normalize names
+    event = normalize_event_names(event)
+    
+    return event
+
+def should_drop_event(event: dict[str, Any]) -> bool:
+    """Determine if event should be dropped."""
+    # Your filtering logic
+    return False
+```
+
+## Error Handling
+
+If your `before_send` function raises an exception, PostHog will:
+
+1. Log the error 
+2. Continue with the original, unmodified event
+3. Not crash your application
+
+```python
+from typing import Optional, Any
+
+def risky_before_send(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    # If this raises an exception, the original event will be sent
+    risky_operation()
+    return event
+```
+
+## Complete Example
+
+```python
+import posthog
+from typing import Optional, Any
+import re
+
+def production_before_send(event: dict[str, Any]) -> Optional[dict[str, Any]]:
+    try:
+        properties = event.get("properties", {})
+        
+        # 1. Filter out bot traffic
+        user_agent = properties.get("$user_agent", "")
+        if re.search(r'bot|crawler|spider', user_agent, re.I):
+            return None
+        
+        # 2. Filter out internal traffic  
+        ip = properties.get("$ip", "")
+        if ip.startswith("192.168.") or ip.startswith("10."):
+            return None
+        
+        # 3. Scrub email PII but keep domain
+        if "email" in properties:
+            email = properties["email"]
+            if "@" in email:
+                domain = email.split("@")[1]
+                properties["email"] = f"***@{domain}"
+        
+        # 4. Add custom context
+        properties.update({
+            "app_version": "1.0.0",
+            "build_number": "123"
+        })
+        
+        # 5. Normalize event name
+        if event.get("event"):
+            event["event"] = event["event"].lower().replace(" ", "_")
+        
+        return event
+        
+    except Exception as e:
+        # Log error but don't crash
+        print(f"Error in before_send: {e}")
+        return event  # Return original event on error
+
+# Usage
+client = posthog.Client(
+    api_key="your-api-key",
+    before_send=production_before_send
+)
+
+# All events will now be processed by your before_send function
+client.capture("user_123", "Page View", {"url": "/home"})
+```

CHANGELOG.md

@@ -1,3 +1,248 @@
+# 6.9.0 - 2025-11-06
+
+- feat(error-tracking): add local variables capture
+
+# 6.8.0 - 2025-11-03
+
+- feat(llma): send web search calls to be used for LLM cost calculations
+
+# 6.7.14 - 2025-11-03
+
+- fix(django): Handle request.user access in async middleware context to prevent SynchronousOnlyOperation errors in Django 5+ (fixes #355)
+- test(django): Add Django 5 integration test suite with real ASGI application testing async middleware behavior
+
+# 6.7.13 - 2025-11-02
+
+- fix(llma): cache cost calculation in the LangChain callback
+
+# 6.7.12 - 2025-11-02
+
+- fix(django): Restore process_exception method to capture view and downstream middleware exceptions (fixes #329)
+- fix(ai/langchain): Add LangChain 1.0+ compatibility for CallbackHandler imports (fixes #362)
+
+# 6.7.11 - 2025-10-28
+
+- feat(ai): Add `$ai_framework` property for framework integrations (e.g. LangChain)
+
+# 6.7.10 - 2025-10-24
+
+- fix(django): Make middleware truly hybrid - compatible with both sync (WSGI) and async (ASGI) Django stacks without breaking sync-only deployments
+
+# 6.7.9 - 2025-10-22
+
+- fix(flags): multi-condition flags with static cohorts returning wrong variants
+
+# 6.7.8 - 2025-10-16
+
+- fix(llma): missing async for OpenAI's streaming implementation
+
+# 6.7.7 - 2025-10-14
+
+- fix: remove deprecated attribute $exception_personURL from exception events
+
+# 6.7.6 - 2025-09-16
+
+- fix: don't sort condition sets with variant overrides to the top
+- fix: Prevent core Client methods from raising exceptions
+
+# 6.7.5 - 2025-09-16
+
+- feat: Django middleware now supports async request handling.
+
+# 6.7.4 - 2025-09-05
+
+- fix: Missing system prompts for some providers
+
+# 6.7.3 - 2025-09-04
+
+- fix: missing usage tokens in Gemini
+
+# 6.7.2 - 2025-09-03
+
+- fix: tool call results in streaming providers
+
+# 6.7.1 - 2025-09-01
+
+- fix: Add base64 inline image sanitization
+
+# 6.7.0 - 2025-08-26
+
+- feat: Add support for feature flag dependencies
+
+# 6.6.1 - 2025-08-21
+
+- fix: Prevent `NoneType` error when `group_properties` is `None`
+
+# 6.6.0 - 2025-08-15
+
+- feat: Add `flag_keys_to_evaluate` parameter to optimize feature flag evaluation performance by only evaluating specified flags
+- feat: Add `flag_keys_filter` option to `send_feature_flags` for selective flag evaluation in capture events
+
+# 6.5.0 - 2025-08-08
+
+- feat: Add `$context_tags` to an event to know which properties were included as tags
+
+# 6.4.1 - 2025-08-06
+
+- fix: Always pass project API key in `remote_config` requests for deterministic project routing
+
+# 6.4.0 - 2025-08-05
+
+- feat: support Vertex AI for Gemini
+
+# 6.3.4 - 2025-08-04
+
+- fix: set `$ai_tools` for all providers and `$ai_output_choices` for all non-streaming provider flows properly
+
+# 6.3.3 - 2025-08-01
+
+- fix: `get_feature_flag_result` now correctly returns FeatureFlagResult when payload is empty string instead of None
+
+# 6.3.2 - 2025-07-31
+
+- fix: Anthropic's tool calls are now handled properly
+
+# 6.3.0 - 2025-07-22
+
+- feat: Enhanced `send_feature_flags` parameter to accept `SendFeatureFlagsOptions` object for declarative control over local/remote evaluation and custom properties
+
+# 6.2.1 - 2025-07-21
+
+- feat: make `posthog_client` an optional argument in PostHog AI providers wrappers (`posthog.ai.*`), intuitively using the default client as the default
+
+# 6.1.1 - 2025-07-16
+
+- fix: correctly capture exceptions processed by Django from views or middleware
+
+# 6.1.0 - 2025-07-10
+
+- feat: decouple feature flag local evaluation from personal API keys; support decrypting remote config payloads without relying on the feature flags poller
+
+# 6.0.4 - 2025-07-09
+
+- fix: add POSTHOG_MW_CLIENT setting to django middleware, to support custom clients for exception capture.
+
+# 6.0.3 - 2025-07-07
+
+- feat: add a feature flag evaluation cache (local storage or redis) to support returning flag evaluations when the service is down
+
+# 6.0.2 - 2025-07-02
+
+- fix: send_feature_flags changed to default to false in `Client::capture_exception`
+
+# 6.0.1
+
+- fix: response `$process_person_profile` property when passed to capture
+
+# 6.0.0
+
+This release contains a number of major breaking changes:
+
+- feat: make distinct_id an optional parameter in posthog.capture and related functions
+- feat: make capture and related functions return `Optional[str]`, which is the UUID of the sent event, if it was sent
+- fix: remove `identify` (prefer `posthog.set()`), and `page` and `screen` (prefer `posthog.capture()`)
+- fix: delete exception-capture specific integrations module. Prefer the general-purpose django middleware as a replacement for the django `Integration`.
+
+To migrate to this version, you'll mostly just need to switch to using named keyword arguments, rather than positional ones. For example:
+
+```python
+# Old calling convention
+posthog.capture("user123", "button_clicked", {"button_id": "123"})
+# New calling convention
+posthog.capture(distinct_id="user123", event="button_clicked", properties={"button_id": "123"})
+
+# Better pattern
+with posthog.new_context():
+    posthog.identify_context("user123")
+
+    # The event name is the first argument, and can be passed positionally, or as a keyword argument in a later position
+    posthog.capture("button_pressed")
+```
+
+Generally, arguments are now appropriately typed, and docstrings have been updated. If something is unclear, please open an issue, or submit a PR!
+
+# 5.4.0 - 2025-06-20
+
+- feat: add support to session_id context on page method
+
+# 5.3.0 - 2025-06-19
+
+- fix: safely handle exception values
+
+# 5.2.0 - 2025-06-19
+
+- feat: construct artificial stack traces if no traceback is available on a captured exception
+
+## 5.1.0 - 2025-06-18
+
+- feat: session and distinct ID's can now be associated with contexts, and are used as such
+- feat: django http request middleware
+
+## 5.0.0 - 2025-06-16
+
+- fix: removed deprecated sentry integration
+
+## 4.10.0 - 2025-06-13
+
+- fix: no longer fail in autocapture.
+
+## 4.9.0 - 2025-06-13
+
+- feat(ai): track reasoning and cache tokens in the LangChain callback
+
+## 4.8.0 - 2025-06-10
+
+- fix: export scoped, rather than tracked, decorator
+- feat: allow use of contexts without error tracking
+
+## 4.7.0 - 2025-06-10
+
+- feat: add support for parse endpoint in responses API (no longer beta)
+
+## 4.6.2 - 2025-06-09
+
+- fix: replace `import posthog` with direct method imports
+
+## 4.6.1 - 2025-06-09
+
+- fix: replace `import posthog` in `posthoganalytics` package
+
+## 4.6.0 - 2025-06-09
+
+- feat: add additional user and request context to captured exceptions via the Django integration
+- feat: Add `setup()` function to initialise default client
+
+## 4.5.0 - 2025-06-09
+
+- feat: add before_send callback (#249)
+
+## 4.4.2- 2025-06-09
+
+- empty point release to fix release automation
+
+## 4.4.1 2025-06-09
+
+- empty point release to fix release automation
+
+## 4.4.0 - 2025-06-09
+
+- Use the new `/flags` endpoint for all feature flag evaluations (don't fall back to `/decide` at all)
+
+## 4.3.2 - 2025-06-06
+
+1. Add context management:
+
+- New context manager with `posthog.new_context()`
+- Tag functions: `posthog.tag()`, `posthog.get_tags()`, `posthog.clear_tags()`
+- Function decorator:
+  - `@posthog.scoped` - Creates context and captures exceptions thrown within the function
+- Automatic deduplication of exceptions to ensure each exception is only captured once
+
+2. fix: feature flag request use geoip_disable (#235)
+3. chore: pin actions versions (#210)
+4. fix: opinionated setup and clean fn fix (#240)
+5. fix: release action failed (#241)
+
 ## 4.2.0 - 2025-05-22
 
 Add support for google gemini

Makefile

@@ -1,6 +1,5 @@
 lint:
-	pylint --rcfile=.pylintrc --reports=y --exit-zero analytics | tee pylint.out
-	flake8 --max-complexity=10 --statistics analytics > flake8.out || true
+	uvx ruff format
 
 test:
 	coverage run -m pytest
@@ -17,22 +16,42 @@ release_analytics:
 	rm -rf posthoganalytics
 	mkdir posthoganalytics
 	cp -r posthog/* posthoganalytics/
-	find ./posthoganalytics -type f -exec sed -i '' -e 's/from posthog /from posthoganalytics /g' {} \;
-	find ./posthoganalytics -type f -exec sed -i '' -e 's/from posthog\./from posthoganalytics\./g' {} \;
+	find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthog /from posthoganalytics /g' {} \;
+	find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthog\./from posthoganalytics\./g' {} \;
+	find ./posthoganalytics -name "*.bak" -delete
 	rm -rf posthog
 	python setup_analytics.py sdist bdist_wheel
 	twine upload dist/*
 	mkdir posthog
-	find ./posthoganalytics -type f -exec sed -i '' -e 's/from posthoganalytics /from posthog /g' {} \;
-	find ./posthoganalytics -type f -exec sed -i '' -e 's/from posthoganalytics\./from posthog\./g' {} \;
+	find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthoganalytics /from posthog /g' {} \;
+	find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthoganalytics\./from posthog\./g' {} \;
+	find ./posthoganalytics -name "*.bak" -delete
 	cp -r posthoganalytics/* posthog/
 	rm -rf posthoganalytics
+	rm -f pyproject.toml
+	cp pyproject.toml.backup pyproject.toml
+	rm -f pyproject.toml.backup
 
 e2e_test:
 	.buildscripts/e2e.sh
 
-django_example:
-	python -m pip install -e ".[sentry]"
-	cd sentry_django_example && python manage.py runserver 8080
+prep_local:
+	rm -rf ../posthog-python-local
+	mkdir ../posthog-python-local
+	cp -r . ../posthog-python-local/
+	cd ../posthog-python-local && rm -rf dist build posthoganalytics .git
+	cd ../posthog-python-local && mkdir posthoganalytics
+	cd ../posthog-python-local && cp -r posthog/* posthoganalytics/
+	cd ../posthog-python-local && find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthog /from posthoganalytics /g' {} \;
+	cd ../posthog-python-local && find ./posthoganalytics -type f -name "*.py" -exec sed -i.bak -e 's/from posthog\./from posthoganalytics\./g' {} \;
+	cd ../posthog-python-local && find ./posthoganalytics -name "*.bak" -delete
+	cd ../posthog-python-local && rm -rf posthog
+	cd ../posthog-python-local && sed -i.bak 's/from version import VERSION/from posthoganalytics.version import VERSION/' setup_analytics.py
+	cd ../posthog-python-local && rm setup_analytics.py.bak
+	cd ../posthog-python-local && sed -i.bak 's/"posthog"/"posthoganalytics"/' setup.py
+	cd ../posthog-python-local && rm setup.py.bak
+	cd ../posthog-python-local && python -c "import setup_analytics" 2>/dev/null || true
+	@echo "Local copy created at ../posthog-python-local"
+	@echo "Install with: pip install -e ../posthog-python-local"
 
-.PHONY: test lint release e2e_test
+.PHONY: test lint release e2e_test prep_local

README.md

@@ -1,7 +1,14 @@
 # PostHog Python
 
-[![PyPI](https://img.shields.io/pypi/v/posthog)](https://pypi.org/project/posthog/)
-
+<p align="center">
+  <img alt="posthoglogo" src="https://user-images.githubusercontent.com/65415371/205059737-c8a4f836-4889-4654-902e-f302b187b6a0.png">
+</p>
+<p align="center">
+   <a href="https://pypi.org/project/posthog/"><img alt="pypi installs" src="https://img.shields.io/pypi/v/posthog"/></a>
+   <img alt="GitHub contributors" src="https://img.shields.io/github/contributors/posthog/posthog-python">
+  <img alt="GitHub commit activity" src="https://img.shields.io/github/commit-activity/m/posthog/posthog-python"/>
+  <img alt="GitHub closed issues" src="https://img.shields.io/github/issues-closed/posthog/posthog-python"/>
+</p>
 
 Please see the [Python integration docs](https://posthog.com/docs/integrations/python-integration) for details.
 
@@ -9,40 +16,51 @@ Please see the [Python integration docs](https://posthog.com/docs/integrations/p
 
 ### Testing Locally
 
-1. Run `python3 -m venv env` (creates virtual environment called "env")
-    * or `uv venv env`
+We recommend using [uv](https://docs.astral.sh/uv/). It's super fast.
+
+1. Run `uv venv env` (creates virtual environment called "env")
+    * or `python3 -m venv env`
 2. Run `source env/bin/activate` (activates the virtual environment)
-3. Run `python3 -m pip install -e ".[test]"` (installs the package in develop mode, along with test dependencies)
-    * or `uv pip install -e ".[test]"`
-4. Run `make test`
+3. Run `uv sync --extra dev --extra test` (installs the package in develop mode, along with test dependencies)
+    * or `pip install -e ".[dev,test]"`
+4. you have to run `pre-commit install` to have auto linting pre commit
+5. Run `make test`
   1. To run a specific test do `pytest -k test_no_api_key`
 
+## PostHog recommends `uv` so...
+
+```bash
+uv python install 3.9.19
+uv python pin 3.9.19
+uv venv
+source env/bin/activate
+uv sync --extra dev --extra test
+pre-commit install
+make test
+```
+
 ### Running Locally
 
 Assuming you have a [local version of PostHog](https://posthog.com/docs/developing-locally) running, you can run `python3 example.py` to see the library in action.
 
-### Running the Django Sentry Integration Locally
-
-There's a sample Django project included, called `sentry_django_example`, which explains how to use PostHog with Sentry.
-
-There's 2 places of importance (Changes required are all marked with TODO in the sample project directory)
-
-1. Settings.py
-    1. Input your Sentry DSN
-    2. Input your Sentry Org and ProjectID details into `PosthogIntegration()`
-    3. Add `POSTHOG_DJANGO` to settings.py. This allows the `PosthogDistinctIdMiddleware` to get the distinct_ids
-
-2. urls.py
-    1. This includes the `sentry-debug/` endpoint, which generates an exception
-
-To run things: `make django_example`. This installs the posthog-python library with the sentry-sdk add-on, and then runs the django app.
-Also start the PostHog app locally.
-Then navigate to `http://127.0.0.1:8080/sentry-debug/` and you should get an event in both Sentry and PostHog, with links to each other.
-
 ### Releasing Versions
 
-Updated are released using GitHub Actions: after bumping `version.py` in `master` and adding to `CHANGELOG.md`, go to [our release workflow's page](https://github.com/PostHog/posthog-python/actions/workflows/release.yaml) and dispatch it manually, using workflow from `master`.
+Updates are released automatically using GitHub Actions when `version.py` is updated on `master`. After bumping `version.py` in `master` and adding to `CHANGELOG.md`, the [release workflow](https://github.com/PostHog/posthog-python/blob/master/.github/workflows/release.yaml) will automatically trigger and deploy the new version.
 
-## Questions?
+If you need to check the latest runs or manually trigger a release, you can go to [our release workflow's page](https://github.com/PostHog/posthog-python/actions/workflows/release.yaml) and dispatch it manually, using workflow from `master`.
 
-### [Join our Slack community.](https://join.slack.com/t/posthogusers/shared_invite/enQtOTY0MzU5NjAwMDY3LTc2MWQ0OTZlNjhkODk3ZDI3NDVjMDE1YjgxY2I4ZjI4MzJhZmVmNjJkN2NmMGJmMzc2N2U3Yjc3ZjI5NGFlZDQ)
+
+### Testing changes locally with the PostHog app
+
+You can run `make prep_local`, and it'll create a new folder alongside the SDK repo one called `posthog-python-local`, which you can then import into the posthog project by changing pyproject.toml to look like this:
+```toml
+dependencies = [
+    ...
+    "posthoganalytics" #NOTE: no version number
+    ...
+]
+...
+[tools.uv.sources]
+posthoganalytics = { path = "../posthog-python-local" }
+```
+This'll let you build and test SDK changes fully locally, incorporating them into your local posthog app stack. It mainly takes care of the `posthog -> posthoganalytics` module renaming. You'll need to re-run `make prep_local` each time you make a change, and re-run `uv sync --active` in the posthog app project.

bin/docs

@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+#/ Usage: bin/docs
+#/ Description: Generate documentation for the PostHog Python SDK
+source bin/helpers/_utils.sh
+set_source_and_root_dir
+ensure_virtual_env
+
+exec python3 "$(dirname "$0")/docs_scripts/generate_json_schemas.py" "$@" 

bin/docs_scripts/doc_constant.py

@@ -0,0 +1,43 @@
+"""
+Constants for PostHog Python SDK documentation generation.
+"""
+
+from typing import Dict, Union
+from posthog.version import VERSION
+
+# Documentation generation metadata
+DOCUMENTATION_METADATA = {
+    "hogRef": "0.3",
+    "slugPrefix": "posthog-python",
+    "specUrl": "https://github.com/PostHog/posthog-python",
+}
+
+# Docstring parsing patterns for new format
+DOCSTRING_PATTERNS = {
+    "examples_section": r"Examples:\s*\n(.*?)(?=\n\s*\n\s*Category:|\Z)",
+    "args_section": r"Args:\s*\n(.*?)(?=\n\s*\n\s*Examples:|\n\s*\n\s*Details:|\n\s*\n\s*Category:|\Z)",
+    "details_section": r"Details:\s*\n(.*?)(?=\n\s*\n\s*Examples:|\n\s*\n\s*Category:|\Z)",
+    "category_section": r"Category:\s*\n\s*(.+?)\s*(?:\n|$)",
+    "code_block": r"```(?:python)?\n(.*?)```",
+    "param_description": r"^\s*{param_name}:\s*(.+?)(?=\n\s*\w+:|\Z)",
+    "args_marker": r"\n\s*Args:\s*\n",
+    "examples_marker": r"\n\s*Examples:\s*\n",
+    "details_marker": r"\n\s*Details:\s*\n",
+    "category_marker": r"\n\s*Category:\s*\n",
+}
+
+# Output file configuration
+OUTPUT_CONFIG: Dict[str, Union[str, int]] = {
+    "output_dir": "./references",
+    "filename": f"posthog-python-references-{VERSION}.json",
+    "filename_latest": "posthog-python-references-latest.json",
+    "indent": 2,
+}
+
+# Documentation structure defaults
+DOC_DEFAULTS = {
+    "showDocs": True,
+    "releaseTag": "public",
+    "return_type_void": "None",
+    "max_optional_params": 3,
+}

bin/docs_scripts/generate_json_schemas.py

@@ -0,0 +1,498 @@
+#!/usr/bin/env python3
+"""
+Generate comprehensive SDK documentation JSON from PostHog Python SDK.
+This script inspects the code and docstrings to create documentation in the specified format.
+"""
+
+import json
+import inspect
+import re
+from dataclasses import is_dataclass, fields
+from typing import get_origin, get_args, Union
+from textwrap import dedent
+from doc_constant import (
+    DOCUMENTATION_METADATA,
+    DOCSTRING_PATTERNS,
+    OUTPUT_CONFIG,
+    DOC_DEFAULTS,
+)
+import os
+
+
+def extract_examples_from_docstring(docstring: str) -> list:
+    """Extract code examples from docstring."""
+    if not docstring:
+        return []
+
+    examples = []
+
+    # Look for Examples section in the new format
+    examples_section_match = re.search(
+        DOCSTRING_PATTERNS["examples_section"], docstring, re.DOTALL
+    )
+    if examples_section_match:
+        examples_content = examples_section_match.group(1).strip()
+        # Extract code blocks from the Examples section
+        code_blocks = re.findall(
+            DOCSTRING_PATTERNS["code_block"], examples_content, re.DOTALL
+        )
+        for i, code_block in enumerate(code_blocks):
+            # Remove common leading whitespace while preserving relative indentation
+            code = dedent(code_block).strip()
+
+            # Extract name from first comment line if present
+            lines = code.split("\n")
+            name = f"Example {i + 1}"  # Default fallback
+
+            if lines and lines[0].strip().startswith("#"):
+                # Extract name from first comment, keep the comment in the code
+                comment_text = lines[0].strip()[1:].strip()
+                if comment_text:
+                    name = comment_text
+
+            examples.append({"id": f"example_{i + 1}", "name": name, "code": code})
+
+    return examples
+
+
+def extract_details_from_docstring(docstring: str) -> str:
+    """Extract details section from docstring."""
+    if not docstring:
+        return ""
+
+    # Look for Details section
+    details_match = re.search(
+        DOCSTRING_PATTERNS["details_section"], docstring, re.DOTALL
+    )
+    if details_match:
+        details_content = details_match.group(1).strip()
+        # Clean up formatting
+        return details_content.replace("\n", " ")
+
+    return ""
+
+
+def parse_docstring_tags(docstring: str) -> dict:
+    """Parse tags from docstring Category section."""
+    if not docstring:
+        return {}
+
+    tags = {}
+
+    # Extract Category section
+    category_match = re.search(DOCSTRING_PATTERNS["category_section"], docstring)
+    if category_match:
+        category_value = category_match.group(1).strip()
+        tags["category"] = category_value
+
+    return tags
+
+
+def extract_description_from_docstring(docstring: str) -> str:
+    """Extract main description from docstring."""
+    if not docstring:
+        return ""
+
+    # Clean up the docstring
+    cleaned = dedent(docstring).strip()
+
+    # Find the end of the description by looking for first section marker
+    # Check for Args:, Examples:, Details:, or Category: sections
+    section_patterns = [
+        DOCSTRING_PATTERNS["args_marker"],
+        DOCSTRING_PATTERNS["examples_marker"],
+        DOCSTRING_PATTERNS["details_marker"],
+        DOCSTRING_PATTERNS["category_marker"],
+    ]
+
+    end_pos = len(cleaned)
+    for pattern in section_patterns:
+        match = re.search(pattern, cleaned)
+        if match:
+            end_pos = min(end_pos, match.start())
+
+    # Extract description up to the first section marker
+    description = cleaned[:end_pos].strip()
+
+    # Remove one level of \n since it will be rendered as markdown
+    # and \n will be padded in later steps
+    description = description.replace("\n", " ")
+
+    return description
+
+
+def get_type_name(type_annotation) -> str:
+    """Convert type annotation to string name."""
+    if type_annotation is None or type_annotation is type(None):
+        return "any"
+
+    # Handle typing constructs
+    origin = get_origin(type_annotation)
+    if origin is not None:
+        # Handle Union types (including Optional)
+        if origin is Union:
+            args = get_args(type_annotation)
+            if len(args) == 2 and type(None) in args:
+                # This is Optional[Type] - get the non-None type
+                non_none_type = next(arg for arg in args if arg is not type(None))
+                return f"Optional[{get_type_name(non_none_type)}]"
+            else:
+                # Regular Union - list all types
+                type_names = [get_type_name(arg) for arg in args]
+                return f"Union[{', '.join(type_names)}]"
+
+        # Handle other generic types (List, Dict, etc.)
+        origin_name = getattr(origin, "__name__", str(origin))
+        args = get_args(type_annotation)
+        if args:
+            arg_names = [get_type_name(arg) for arg in args]
+            return f"{origin_name}[{', '.join(arg_names)}]"
+        else:
+            return origin_name
+
+    # Handle regular types
+    elif hasattr(type_annotation, "__name__"):
+        return type_annotation.__name__
+    else:
+        return str(type_annotation)
+
+
+def analyze_parameter(param: inspect.Parameter, docstring: str = "") -> dict:
+    """Analyze a function parameter and return its documentation."""
+    # Determine if parameter is optional (has default value)
+    is_optional = param.default == inspect.Parameter.empty
+
+    # Get the type annotation
+    type_annotation = param.annotation
+    param_type = "any"
+
+    if type_annotation != inspect.Parameter.empty:
+        # Handle Union/Optional types first
+        origin = get_origin(type_annotation)
+        if origin is Union:
+            args = get_args(type_annotation)
+            if len(args) == 2 and type(None) in args:
+                # This is Optional[Type]
+                non_none_type = next(arg for arg in args if arg is not type(None))
+                param_type = get_type_name(non_none_type)
+                is_optional = True
+            else:
+                # Other Union types, use first type
+                param_type = get_type_name(args[0]) if args else "any"
+        else:
+            param_type = get_type_name(type_annotation)
+    elif param.default != inspect.Parameter.empty:
+        # No type annotation, but has default value - infer type from default
+        param_type = get_type_name(type(param.default))
+
+    # Extract parameter description from Args section
+    param_description = ""
+    if docstring:
+        # Look for Args section and extract description for this parameter
+        args_section_match = re.search(
+            DOCSTRING_PATTERNS["args_section"], docstring, re.DOTALL
+        )
+        if args_section_match:
+            args_content = args_section_match.group(1)
+            # Look for the parameter description
+            param_pattern = DOCSTRING_PATTERNS["param_description"].format(
+                param_name=re.escape(param.name)
+            )
+            param_match = re.search(
+                param_pattern, args_content, re.MULTILINE | re.DOTALL
+            )
+            if param_match:
+                param_description = param_match.group(1).strip().replace("\n", " ")
+
+    param_info = {
+        "name": param.name,
+        "description": param_description,
+        "isOptional": is_optional,
+        "type": param_type,
+    }
+
+    return param_info
+
+
+def analyze_function(func, name: str) -> dict:
+    """Analyze a function and return its documentation."""
+    try:
+        sig = inspect.signature(func)
+        docstring = inspect.getdoc(func) or ""
+
+        # Skip functions with empty docstrings
+        if not docstring.strip():
+            return {}
+
+        # Extract parameters (excluding 'self')
+        params = []
+        for param_name, param in sig.parameters.items():
+            if param_name != "self":
+                params.append(analyze_parameter(param, docstring))
+
+        # Special handling for constructor
+        display_name = name
+        if name == "__init__":
+            display_name = func.__qualname__.split(".")[0]
+
+        # Parse tags from docstring
+        tags = parse_docstring_tags(docstring)
+
+        category = tags.get("category", None)
+
+        # Extract description
+        description = extract_description_from_docstring(docstring)
+
+        # Skip if no meaningful description
+        if not description.strip():
+            return {}
+
+        # Extract details section (only if it exists)
+        details = extract_details_from_docstring(docstring)
+
+        # Get examples from docstring, do not generate fallback examples
+        examples = extract_examples_from_docstring(docstring)
+        # If no examples, do not include the examples key or set to empty list
+
+        result = {
+            "id": name,
+            "title": display_name,
+            "description": description,
+            "details": details,
+            "category": category,
+            "params": params,
+            "showDocs": DOC_DEFAULTS["showDocs"],
+            "releaseTag": DOC_DEFAULTS["releaseTag"],
+            "returnType": {
+                "id": "return_type",
+                "name": get_type_name(sig.return_annotation)
+                if sig.return_annotation != inspect.Signature.empty
+                else DOC_DEFAULTS["return_type_void"],
+            },
+        }
+        if examples:
+            result["examples"] = examples
+        return result
+    except Exception as e:
+        print(f"Error analyzing function {name}: {e}")
+        return {}
+
+
+def analyze_class(cls) -> dict:
+    """Analyze a class and return its documentation."""
+    class_doc = inspect.getdoc(cls) or f"Class: {cls.__name__}"
+
+    # Get all public methods and constructor
+    functions = []
+    for method_name in dir(cls):
+        if method_name.startswith("_") and method_name != "__init__":
+            continue
+
+        method = getattr(cls, method_name)
+        if callable(method):
+            func_info = analyze_function(method, method_name)
+            if func_info:  # Only add if not None (empty docstring check)
+                functions.append(func_info)
+
+    return {
+        "id": cls.__name__,
+        "title": cls.__name__,
+        "description": extract_description_from_docstring(class_doc),
+        "functions": functions,
+    }
+
+
+def analyze_type(cls) -> dict:
+    """Analyze a type/dataclass and return its documentation."""
+    type_info = {
+        "id": cls.__name__,
+        "name": cls.__name__,
+        "path": f"{cls.__module__}.{cls.__name__}",
+        "properties": [],
+        "example": "",
+    }
+
+    if is_dataclass(cls):
+        # Handle dataclass
+        for field in fields(cls):
+            prop = {
+                "name": field.name,
+                "type": get_type_name(field.type),
+                "description": f"Field: {field.name}",
+            }
+            type_info["properties"].append(prop)
+    elif hasattr(cls, "__annotations__"):
+        # Handle TypedDict or annotated class
+        for field_name, field_type in cls.__annotations__.items():
+            prop = {
+                "name": field_name,
+                "type": get_type_name(field_type),
+                "description": f"Field: {field_name}",
+            }
+            type_info["properties"].append(prop)
+
+    return type_info
+
+
+def generate_sdk_documentation():
+    """Generate complete SDK documentation in the requested format."""
+
+    # Import PostHog components
+    import posthog
+    from posthog.client import Client
+    import posthog.types as types_module
+    import posthog.args as args_module
+    from posthog.version import VERSION
+
+    # Main SDK info
+    sdk_info = {
+        "version": VERSION,
+        "id": "posthog-python",
+        "title": "PostHog Python SDK",
+        "description": "Integrate PostHog into any python application.",
+        "slugPrefix": DOCUMENTATION_METADATA["slugPrefix"],
+        "specUrl": DOCUMENTATION_METADATA["specUrl"],
+    }
+
+    # Collect types
+    types_list = []
+
+    # Types from posthog.types
+    for name in dir(types_module):
+        obj = getattr(types_module, name)
+        if inspect.isclass(obj) and not name.startswith("_"):
+            try:
+                type_info = analyze_type(obj)
+                types_list.append(type_info)
+            except Exception as e:
+                print(f"Error analyzing type {name}: {e}")
+
+    # Types from posthog.args
+    for name in dir(args_module):
+        obj = getattr(args_module, name)
+        if inspect.isclass(obj) and not name.startswith("_"):
+            try:
+                type_info = analyze_type(obj)
+                types_list.append(type_info)
+            except Exception as e:
+                print(f"Error analyzing type {name}: {e}")
+
+    # Clean types of empty types
+
+    # Remove types that have no properties and no examples
+    # Remove types that have no properties and no examples
+    types_list = [
+        t for t in types_list if len(t["properties"]) > 0 or t["example"] != ""
+    ]
+
+    # Collect classes
+    classes_list = []
+
+    # Main PostHog class (renamed from Client)
+    client_class = analyze_class(Client)
+    client_class["id"] = "PostHog"
+    client_class["title"] = "PostHog"
+    classes_list.append(client_class)
+
+    # Global module functions (functions callable as posthog.function_name)
+    global_functions = []
+    for func_name in dir(posthog):
+        # Skip private functions and non-callables
+        if func_name.startswith("_") or not callable(getattr(posthog, func_name)):
+            continue
+
+        func = getattr(posthog, func_name)
+        # Only include functions actually defined in the posthog module (not imported)
+        # and exclude class references
+        if (
+            func_name not in ["Client", "Posthog"]
+            and hasattr(func, "__module__")
+            and func.__module__ == "posthog"
+        ):
+            try:
+                func_info = analyze_function(func, func_name)
+                if func_info:  # Only add if not None (has proper docstring)
+                    global_functions.append(func_info)
+            except Exception:
+                continue
+
+    # Add global functions as a "class"
+    if global_functions:
+        classes_list.append(
+            {
+                "id": "PostHogModule",
+                "title": "PostHog Module Functions",
+                "description": "Global functions available in the PostHog module",
+                "functions": global_functions,
+            }
+        )
+
+    # Collect categories from functions
+    categories = ["Initialization", "Identification", "Capture"]
+    seen_categories = set(categories)
+    for class_info in classes_list:
+        if "functions" in class_info:
+            for func in class_info["functions"]:
+                if (
+                    "category" in func
+                    and func["category"] not in seen_categories
+                    and func["category"]
+                ):
+                    categories.append(func["category"])
+                    seen_categories.add(func["category"])
+
+    # Create the final structure
+    result = {
+        "id": "posthog-python",
+        "hogRef": DOCUMENTATION_METADATA["hogRef"],
+        "info": sdk_info,
+        "types": types_list,
+        "classes": classes_list,
+        "categories": categories,
+    }
+
+    return result
+
+
+if __name__ == "__main__":
+    print("Generating PostHog Python SDK documentation...")
+
+    try:
+        documentation = generate_sdk_documentation()
+
+        # Ensure output directory exists
+        output_dir = str(OUTPUT_CONFIG["output_dir"])
+        os.makedirs(output_dir, exist_ok=True)
+
+        output_file = os.path.join(
+            str(OUTPUT_CONFIG["output_dir"]), str(OUTPUT_CONFIG["filename"])
+        )
+        output_file_latest = os.path.join(
+            str(OUTPUT_CONFIG["output_dir"]), str(OUTPUT_CONFIG["filename_latest"])
+        )
+
+        # Write to current version
+        with open(output_file, "w") as f:
+            json.dump(documentation, f, indent=int(OUTPUT_CONFIG["indent"]))
+        # Write to latest
+        with open(output_file_latest, "w") as f:
+            json.dump(documentation, f, indent=int(OUTPUT_CONFIG["indent"]))
+
+        print(f"✓ Generated {output_file}")
+
+        # Print summary
+        types_count = len(documentation["types"])
+        classes_count = len(documentation["classes"])
+
+        total_functions = sum(len(cls["functions"]) for cls in documentation["classes"])
+
+        print("📊 Documentation Summary:")
+        print(f"   • {types_count} types documented")
+        print(f"   • {classes_count} classes documented")
+        print(f"   • {total_functions} functions documented")
+
+    except Exception as e:
+        print(f"❌ Error generating documentation: {e}")
+        import traceback
+
+        traceback.print_exc()

bin/fmt

@@ -6,9 +6,7 @@ set_source_and_root_dir
 ensure_virtual_env
 
 if [[ "$1" == "--check" ]]; then
-  black --check .
-  isort --check-only .
+  ruff format --check .
 else
-  black .
-  isort .
+  ruff format .
 fi

example.py

@@ -1,107 +1,502 @@
 # PostHog Python library example
-import argparse
+#
+# This script demonstrates various PostHog Python SDK capabilities including:
+# - Basic event capture and user identification
+# - Feature flag local evaluation
+# - Feature flag payloads
+# - Context management and tagging
+#
+# Setup:
+# 1. Copy .env.example to .env and fill in your PostHog credentials
+# 2. Run this script and choose from the interactive menu
+
+import os
 
 import posthog
 
-# Add argument parsing
-parser = argparse.ArgumentParser(description="PostHog Python library example")
-parser.add_argument(
-    "--flag", default="person-on-events-enabled", help="Feature flag key to check (default: person-on-events-enabled)"
-)
-args = parser.parse_args()
 
-posthog.debug = True
+def load_env_file():
+    """Load environment variables from .env file if it exists."""
+    env_path = os.path.join(os.path.dirname(__file__), ".env")
+    if os.path.exists(env_path):
+        with open(env_path, "r") as f:
+            for line in f:
+                line = line.strip()
+                if line and not line.startswith("#") and "=" in line:
+                    key, value = line.split("=", 1)
+                    os.environ.setdefault(key.strip(), value.strip())
 
-# You can find this key on the /setup page in PostHog
-posthog.project_api_key = "phc_gtWmTq3Pgl06u4sZY3TRcoQfp42yfuXHKoe8ZVSR6Kh"
-posthog.personal_api_key = "phx_fiRCOQkTA3o2ePSdLrFDAILLHjMu2Mv52vUi8MNruIm"
 
-# Where you host PostHog, with no trailing /.
-# You can remove this line if you're using posthog.com
-posthog.host = "http://localhost:8000"
-posthog.poll_interval = 10
+# Load .env file if it exists
+load_env_file()
 
-print(
-    posthog.feature_enabled(
-        args.flag,  # Use the flag from command line arguments
-        "12345",
-        groups={"organization": str("0182ee91-8ef7-0000-4cb9-fedc5f00926a")},
-        group_properties={
-            "organization": {
-                "id": "0182ee91-8ef7-0000-4cb9-fedc5f00926a",
-                "created_at": "2022-06-30 11:44:52.984121+00:00",
-            }
-        },
+# Get configuration
+project_key = os.getenv("POSTHOG_PROJECT_API_KEY", "")
+personal_api_key = os.getenv("POSTHOG_PERSONAL_API_KEY", "")
+host = os.getenv("POSTHOG_HOST", "http://localhost:8000")
+
+# Check if credentials are provided
+if not project_key or not personal_api_key:
+    print("❌ Missing PostHog credentials!")
+    print(
+        "   Please set POSTHOG_PROJECT_API_KEY and POSTHOG_PERSONAL_API_KEY environment variables"
+    )
+    print("   or copy .env.example to .env and fill in your values")
+    exit(1)
+
+# Test authentication before proceeding
+print("🔑 Testing PostHog authentication...")
+
+try:
+    # Configure PostHog with credentials
+    posthog.debug = False  # Keep quiet during auth test
+    posthog.api_key = project_key
+    posthog.project_api_key = project_key
+    posthog.personal_api_key = personal_api_key
+    posthog.host = host
+    posthog.poll_interval = 10
+
+    # Test by attempting to get feature flags (this validates both keys)
+    # This will fail if credentials are invalid
+    test_flags = posthog.get_all_flags("test_user", only_evaluate_locally=True)
+
+    # If we get here without exception, credentials work
+    print("✅ Authentication successful!")
+    print(f"   Project API Key: {project_key[:9]}...")
+    print("   Personal API Key: [REDACTED]")
+    print(f"   Host: {host}\n\n")
+
+except Exception as e:
+    print("❌ Authentication failed!")
+    print(f"   Error: {e}")
+    print("\n   Please check your credentials:")
+    print("   - POSTHOG_PROJECT_API_KEY: Project API key from PostHog settings")
+    print(
+        "   - POSTHOG_PERSONAL_API_KEY: Personal API key (required for local evaluation)"
+    )
+    print("   - POSTHOG_HOST: Your PostHog instance URL")
+    exit(1)
+
+# Display menu and get user choice
+print("🚀 PostHog Python SDK Demo - Choose an example to run:\n")
+print("1. Identify and capture examples")
+print("2. Feature flag local evaluation examples")
+print("3. Feature flag payload examples")
+print("4. Flag dependencies examples")
+print("5. Context management and tagging examples")
+print("6. Run all examples")
+print("7. Exit")
+choice = input("\nEnter your choice (1-7): ").strip()
+
+if choice == "1":
+    print("\n" + "=" * 60)
+    print("IDENTIFY AND CAPTURE EXAMPLES")
+    print("=" * 60)
+
+    posthog.debug = True
+
+    # Capture an event
+    print("📊 Capturing events...")
+    posthog.capture(
+        "event",
+        distinct_id="distinct_id",
+        properties={"property1": "value", "property2": "value"},
+        send_feature_flags=True,
+    )
+
+    # Alias a previous distinct id with a new one
+    print("🔗 Creating alias...")
+    posthog.alias("distinct_id", "new_distinct_id")
+
+    posthog.capture(
+        "event2",
+        distinct_id="new_distinct_id",
+        properties={"property1": "value", "property2": "value"},
+    )
+    posthog.capture(
+        "event-with-groups",
+        distinct_id="new_distinct_id",
+        properties={"property1": "value", "property2": "value"},
+        groups={"company": "id:5"},
+    )
+
+    # Add properties to the person
+    print("👤 Identifying user...")
+    posthog.set(
+        distinct_id="new_distinct_id", properties={"email": "something@something.com"}
+    )
+
+    # Add properties to a group
+    print("🏢 Identifying group...")
+    posthog.group_identify("company", "id:5", {"employees": 11})
+
+    # Properties set only once to the person
+    print("🔒 Setting properties once...")
+    posthog.set_once(
+        distinct_id="new_distinct_id", properties={"self_serve_signup": True}
+    )
+
+    # This will not change the property (because it was already set)
+    posthog.set_once(
+        distinct_id="new_distinct_id", properties={"self_serve_signup": False}
+    )
+
+    print("🔄 Updating properties...")
+    posthog.set(distinct_id="new_distinct_id", properties={"current_browser": "Chrome"})
+    posthog.set(
+        distinct_id="new_distinct_id", properties={"current_browser": "Firefox"}
+    )
+
+elif choice == "2":
+    print("\n" + "=" * 60)
+    print("FEATURE FLAG LOCAL EVALUATION EXAMPLES")
+    print("=" * 60)
+
+    posthog.debug = True
+
+    print("🏁 Testing basic feature flags...")
+    print(
+        f"beta-feature for 'distinct_id': {posthog.feature_enabled('beta-feature', 'distinct_id')}"
+    )
+    print(
+        f"beta-feature for 'new_distinct_id': {posthog.feature_enabled('beta-feature', 'new_distinct_id')}"
+    )
+    print(
+        f"beta-feature with groups: {posthog.feature_enabled('beta-feature-groups', 'distinct_id', groups={'company': 'id:5'})}"
+    )
+
+    print("\n🌍 Testing location-based flags...")
+    # Assume test-flag has `City Name = Sydney` as a person property set
+    print(
+        f"Sydney user: {posthog.feature_enabled('test-flag', 'random_id_12345', person_properties={'$geoip_city_name': 'Sydney'})}"
+    )
+
+    print(
+        f"Sydney user (local only): {posthog.feature_enabled('test-flag', 'distinct_id_random_22', person_properties={'$geoip_city_name': 'Sydney'}, only_evaluate_locally=True)}"
+    )
+
+    print("\n📋 Getting all flags...")
+    print(f"All flags: {posthog.get_all_flags('distinct_id_random_22')}")
+    print(
+        f"All flags (local): {posthog.get_all_flags('distinct_id_random_22', only_evaluate_locally=True)}"
+    )
+    print(
+        f"All flags with properties: {posthog.get_all_flags('distinct_id_random_22', person_properties={'$geoip_city_name': 'Sydney'}, only_evaluate_locally=True)}"
+    )
+
+elif choice == "3":
+    print("\n" + "=" * 60)
+    print("FEATURE FLAG PAYLOAD EXAMPLES")
+    print("=" * 60)
+
+    posthog.debug = True
+
+    print("📦 Testing feature flag payloads...")
+    print(
+        f"beta-feature payload: {posthog.get_feature_flag_payload('beta-feature', 'distinct_id')}"
+    )
+    print(
+        f"All flags and payloads: {posthog.get_all_flags_and_payloads('distinct_id')}"
+    )
+    print(
+        f"Remote config payload: {posthog.get_remote_config_payload('encrypted_payload_flag_key')}"
+    )
+
+    # Get feature flag result with all details (enabled, variant, payload, key, reason)
+    print("\n🔍 Getting detailed flag result...")
+    result = posthog.get_feature_flag_result("beta-feature", "distinct_id")
+    if result:
+        print(f"Flag key: {result.key}")
+        print(f"Flag enabled: {result.enabled}")
+        print(f"Variant: {result.variant}")
+        print(f"Payload: {result.payload}")
+        print(f"Reason: {result.reason}")
+        # get_value() returns the variant if it exists, otherwise the enabled value
+        print(f"Value (variant or enabled): {result.get_value()}")
+
+elif choice == "4":
+    print("\n" + "=" * 60)
+    print("FLAG DEPENDENCIES EXAMPLES")
+    print("=" * 60)
+    print("🔗 Testing flag dependencies with local evaluation...")
+    print(
+        "   Flag structure: 'test-flag-dependency' depends on 'beta-feature' being enabled"
+    )
+    print("")
+    print("📋 Required setup (if 'test-flag-dependency' doesn't exist):")
+    print("   1. Create feature flag 'beta-feature':")
+    print("      - Condition: email contains '@example.com'")
+    print("      - Rollout: 100%")
+    print("   2. Create feature flag 'test-flag-dependency':")
+    print("      - Condition: flag 'beta-feature' is enabled")
+    print("      - Rollout: 100%")
+    print("")
+
+    posthog.debug = True
+
+    # Test @example.com user (should satisfy dependency if flags exist)
+    result1 = posthog.feature_enabled(
+        "test-flag-dependency",
+        "example_user",
+        person_properties={"email": "user@example.com"},
         only_evaluate_locally=True,
     )
-)
+    print(f"✅ @example.com user (test-flag-dependency): {result1}")
 
-
-# Capture an event
-posthog.capture("distinct_id", "event", {"property1": "value", "property2": "value"}, send_feature_flags=True)
-
-print(posthog.feature_enabled("beta-feature", "distinct_id"))
-print(posthog.feature_enabled("beta-feature-groups", "distinct_id", groups={"company": "id:5"}))
-
-print(posthog.feature_enabled("beta-feature", "distinct_id"))
-
-# get payload
-print(posthog.get_feature_flag_payload("beta-feature", "distinct_id"))
-print(posthog.get_all_flags_and_payloads("distinct_id"))
-exit()
-# # Alias a previous distinct id with a new one
-
-posthog.alias("distinct_id", "new_distinct_id")
-
-posthog.capture("new_distinct_id", "event2", {"property1": "value", "property2": "value"})
-posthog.capture(
-    "new_distinct_id", "event-with-groups", {"property1": "value", "property2": "value"}, groups={"company": "id:5"}
-)
-
-# # Add properties to the person
-posthog.identify("new_distinct_id", {"email": "something@something.com"})
-
-# Add properties to a group
-posthog.group_identify("company", "id:5", {"employees": 11})
-
-# properties set only once to the person
-posthog.set_once("new_distinct_id", {"self_serve_signup": True})
-
-
-posthog.set_once(
-    "new_distinct_id", {"self_serve_signup": False}
-)  # this will not change the property (because it was already set)
-
-posthog.set("new_distinct_id", {"current_browser": "Chrome"})
-posthog.set("new_distinct_id", {"current_browser": "Firefox"})
-
-
-# #############################################################################
-# Make sure you have a personal API key for the examples below
-
-# Local Evaluation
-
-# If flag has City=Sydney, this call doesn't go to `/decide`
-print(posthog.feature_enabled("test-flag", "distinct_id_random_22", person_properties={"$geoip_city_name": "Sydney"}))
-
-print(
-    posthog.feature_enabled(
-        "test-flag",
-        "distinct_id_random_22",
-        person_properties={"$geoip_city_name": "Sydney"},
+    # Test non-example.com user (dependency should not be satisfied)
+    result2 = posthog.feature_enabled(
+        "test-flag-dependency",
+        "regular_user",
+        person_properties={"email": "user@other.com"},
         only_evaluate_locally=True,
     )
-)
+    print(f"❌ Regular user (test-flag-dependency): {result2}")
 
-
-print(posthog.get_all_flags("distinct_id_random_22"))
-print(posthog.get_all_flags("distinct_id_random_22", only_evaluate_locally=True))
-print(
-    posthog.get_all_flags(
-        "distinct_id_random_22", person_properties={"$geoip_city_name": "Sydney"}, only_evaluate_locally=True
+    # Test beta-feature directly for comparison
+    beta1 = posthog.feature_enabled(
+        "beta-feature",
+        "example_user",
+        person_properties={"email": "user@example.com"},
+        only_evaluate_locally=True,
     )
-)
-print(posthog.get_remote_config_payload("encrypted_payload_flag_key"))
+    beta2 = posthog.feature_enabled(
+        "beta-feature",
+        "regular_user",
+        person_properties={"email": "user@other.com"},
+        only_evaluate_locally=True,
+    )
+    print(f"📊 Beta feature comparison - @example.com: {beta1}, regular: {beta2}")
 
+    print("\n🎯 Results Summary:")
+    print(
+        f"   - Flag dependencies evaluated locally: {'✅ YES' if result1 != result2 else '❌ NO'}"
+    )
+    print("   - Zero API calls needed: ✅ YES (all evaluated locally)")
+    print("   - Python SDK supports flag dependencies: ✅ YES")
+
+    print("\n" + "-" * 60)
+    print("PRODUCTION-STYLE MULTIVARIATE DEPENDENCY CHAIN")
+    print("-" * 60)
+    print("🔗 Testing complex multivariate flag dependencies...")
+    print(
+        "   Structure: multivariate-root-flag -> multivariate-intermediate-flag -> multivariate-leaf-flag"
+    )
+    print("")
+    print("📋 Required setup (if flags don't exist):")
+    print(
+        "   1. Create 'multivariate-leaf-flag' with fruit variants (pineapple, mango, papaya, kiwi)"
+    )
+    print("      - pineapple: email = 'pineapple@example.com'")
+    print("      - mango: email = 'mango@example.com'")
+    print(
+        "   2. Create 'multivariate-intermediate-flag' with color variants (blue, red)"
+    )
+    print("      - blue: depends on multivariate-leaf-flag = 'pineapple'")
+    print("      - red: depends on multivariate-leaf-flag = 'mango'")
+    print(
+        "   3. Create 'multivariate-root-flag' with show variants (breaking-bad, the-wire)"
+    )
+    print("      - breaking-bad: depends on multivariate-intermediate-flag = 'blue'")
+    print("      - the-wire: depends on multivariate-intermediate-flag = 'red'")
+    print("")
+
+    # Test pineapple -> blue -> breaking-bad chain
+    dependent_result3 = posthog.get_feature_flag(
+        "multivariate-root-flag",
+        "regular_user",
+        person_properties={"email": "pineapple@example.com"},
+        only_evaluate_locally=True,
+    )
+    if str(dependent_result3) != "breaking-bad":
+        print(
+            f"     ❌ Something went wrong evaluating 'multivariate-root-flag' with pineapple@example.com. Expected 'breaking-bad', got '{dependent_result3}'"
+        )
+    else:
+        print("✅ 'multivariate-root-flag' with email pineapple@example.com succeeded")
+
+    # Test mango -> red -> the-wire chain
+    dependent_result4 = posthog.get_feature_flag(
+        "multivariate-root-flag",
+        "regular_user",
+        person_properties={"email": "mango@example.com"},
+        only_evaluate_locally=True,
+    )
+    if str(dependent_result4) != "the-wire":
+        print(
+            f"     ❌ Something went wrong evaluating multivariate-root-flag with mango@example.com. Expected 'the-wire', got '{dependent_result4}'"
+        )
+    else:
+        print("✅ 'multivariate-root-flag' with email mango@example.com succeeded")
+
+    # Show the complete chain evaluation
+    print("\n🔍 Complete dependency chain evaluation:")
+    for email, expected_chain in [
+        ("pineapple@example.com", ["pineapple", "blue", "breaking-bad"]),
+        ("mango@example.com", ["mango", "red", "the-wire"]),
+    ]:
+        leaf = posthog.get_feature_flag(
+            "multivariate-leaf-flag",
+            "regular_user",
+            person_properties={"email": email},
+            only_evaluate_locally=True,
+        )
+        intermediate = posthog.get_feature_flag(
+            "multivariate-intermediate-flag",
+            "regular_user",
+            person_properties={"email": email},
+            only_evaluate_locally=True,
+        )
+        root = posthog.get_feature_flag(
+            "multivariate-root-flag",
+            "regular_user",
+            person_properties={"email": email},
+            only_evaluate_locally=True,
+        )
+
+        actual_chain = [str(leaf), str(intermediate), str(root)]
+        chain_success = actual_chain == expected_chain
+
+        print(f"   📧 {email}:")
+        print(f"      Expected: {' -> '.join(map(str, expected_chain))}")
+        print(f"      Actual:   {' -> '.join(map(str, actual_chain))}")
+        print(f"      Status:   {'✅ SUCCESS' if chain_success else '❌ FAILED'}")
+
+    print("\n🎯 Multivariate Chain Summary:")
+    print("   - Complex dependency chains: ✅ SUPPORTED")
+    print("   - Multivariate flag dependencies: ✅ SUPPORTED")
+    print("   - Local evaluation of chains: ✅ WORKING")
+
+elif choice == "5":
+    print("\n" + "=" * 60)
+    print("CONTEXT MANAGEMENT AND TAGGING EXAMPLES")
+    print("=" * 60)
+
+    posthog.debug = True
+
+    print("🏷️ Testing context management...")
+    print(
+        "You can add tags to a context, and these are automatically added to any events captured within that context."
+    )
+
+    # You can enter a new context using a with statement. Any exceptions thrown in the context will be captured,
+    # and tagged with the context tags. Other events captured will also be tagged with the context tags. By default,
+    # the new context inherits tags from the parent context.
+    try:
+        with posthog.new_context():
+            posthog.tag("transaction_id", "abc123")
+            posthog.tag("some_arbitrary_value", {"tags": "can be dicts"})
+
+            # This event will be captured with the tags set above
+            posthog.capture("order_processed")
+            print("✅ Event captured with inherited context tags")
+            # This exception will be captured with the tags set above
+            # raise Exception("Order processing failed")
+    except Exception as e:
+        print(f"Exception captured: {e}")
+
+    # Use fresh=True to start with a clean context (no inherited tags)
+    try:
+        with posthog.new_context(fresh=True):
+            posthog.tag("session_id", "xyz789")
+            # Only session_id tag will be present, no inherited tags
+            posthog.capture("session_event")
+            print("✅ Event captured with fresh context tags")
+            # raise Exception("Session handling failed")
+    except Exception as e:
+        print(f"Exception captured: {e}")
+
+    # You can also use the `@posthog.scoped()` decorator to enter a new context.
+    # By default, it inherits tags from the parent context
+    @posthog.scoped()
+    def process_order(order_id):
+        posthog.tag("order_id", order_id)
+        posthog.capture("order_step_completed")
+        print(f"✅ Order {order_id} processed with scoped context")
+        # Exception will be captured and tagged automatically
+        # raise Exception("Order processing failed")
+
+    # Use fresh=True to start with a clean context (no inherited tags)
+    @posthog.scoped(fresh=True)
+    def process_payment(payment_id):
+        posthog.tag("payment_id", payment_id)
+        posthog.capture("payment_processed")
+        print(f"✅ Payment {payment_id} processed with fresh scoped context")
+        # Only payment_id tag will be present, no inherited tags
+        # raise Exception("Payment processing failed")
+
+    process_order("12345")
+    process_payment("67890")
+
+elif choice == "6":
+    print("\n🔄 Running all examples...")
+
+    # Run example 1
+    print(f"\n{'🔸' * 20} IDENTIFY AND CAPTURE {'🔸' * 20}")
+    posthog.debug = True
+    print("📊 Capturing events...")
+    posthog.capture(
+        "event",
+        distinct_id="distinct_id",
+        properties={"property1": "value", "property2": "value"},
+        send_feature_flags=True,
+    )
+    print("🔗 Creating alias...")
+    posthog.alias("distinct_id", "new_distinct_id")
+    print("👤 Identifying user...")
+    posthog.set(
+        distinct_id="new_distinct_id", properties={"email": "something@something.com"}
+    )
+
+    # Run example 2
+    print(f"\n{'🔸' * 20} FEATURE FLAGS {'🔸' * 20}")
+    print("🏁 Testing basic feature flags...")
+    print(f"beta-feature: {posthog.feature_enabled('beta-feature', 'distinct_id')}")
+    print(
+        f"Sydney user: {posthog.feature_enabled('test-flag', 'random_id_12345', person_properties={'$geoip_city_name': 'Sydney'})}"
+    )
+
+    # Run example 3
+    print(f"\n{'🔸' * 20} PAYLOADS {'🔸' * 20}")
+    print("📦 Testing payloads...")
+    print(f"Payload: {posthog.get_feature_flag_payload('beta-feature', 'distinct_id')}")
+
+    # Run example 4
+    print(f"\n{'🔸' * 20} FLAG DEPENDENCIES {'🔸' * 20}")
+    print("🔗 Testing flag dependencies...")
+    result1 = posthog.feature_enabled(
+        "test-flag-dependency",
+        "demo_user",
+        person_properties={"email": "user@example.com"},
+        only_evaluate_locally=True,
+    )
+    result2 = posthog.feature_enabled(
+        "test-flag-dependency",
+        "demo_user2",
+        person_properties={"email": "user@other.com"},
+        only_evaluate_locally=True,
+    )
+    print(f"✅ @example.com user: {result1}, regular user: {result2}")
+
+    # Run example 5
+    print(f"\n{'🔸' * 20} CONTEXT MANAGEMENT {'🔸' * 20}")
+    print("🏷️ Testing context management...")
+    with posthog.new_context():
+        posthog.tag("demo_run", "all_examples")
+        posthog.capture("demo_completed")
+        print("✅ Demo completed with context tags")
+
+elif choice == "7":
+    print("👋 Goodbye!")
+    posthog.shutdown()
+    exit()
+
+else:
+    print("❌ Invalid choice. Please run again and select 1-7.")
+    posthog.shutdown()
+    exit()
+
+print("\n" + "=" * 60)
+print("✅ Example completed!")
+print("=" * 60)
 
 posthog.shutdown()

integration_tests/django5/.gitignore

@@ -0,0 +1,4 @@
+db.sqlite3
+*.pyc
+__pycache__/
+.pytest_cache/

integration_tests/django5/manage.py

@@ -1,12 +1,13 @@
 #!/usr/bin/env python
 """Django's command-line utility for administrative tasks."""
+
 import os
 import sys
 
 
 def main():
     """Run administrative tasks."""
-    os.environ.setdefault("DJANGO_SETTINGS_MODULE", "sentry_django_example.settings")
+    os.environ.setdefault("DJANGO_SETTINGS_MODULE", "testdjango.settings")
     try:
         from django.core.management import execute_from_command_line
     except ImportError as exc:

integration_tests/django5/pyproject.toml

@@ -0,0 +1,19 @@
+[project]
+name = "test-django5"
+version = "0.1.0"
+requires-python = ">=3.12"
+dependencies = [
+    "django~=5.2.7",
+    "uvicorn[standard]~=0.38.0",
+    "posthog",
+    "pytest~=8.4.2",
+    "pytest-asyncio~=1.2.0",
+    "pytest-django~=4.11.1",
+    "httpx~=0.28.1",
+]
+
+[tool.uv]
+required-version = ">=0.5"
+
+[tool.uv.sources]
+posthog = { path = "../..", editable = true }

integration_tests/django5/test_exception_capture.py

@@ -0,0 +1,111 @@
+"""
+Test that verifies exception capture functionality.
+
+These tests verify that exceptions are actually captured to PostHog, not just that
+500 responses are returned.
+
+Without process_exception(), view exceptions are NOT captured to PostHog (v6.7.11 and earlier).
+With process_exception(), Django calls this method to capture exceptions before
+converting them to 500 responses.
+"""
+
+import os
+import django
+
+# Setup Django before importing anything else
+os.environ.setdefault("DJANGO_SETTINGS_MODULE", "testdjango.settings")
+django.setup()
+
+import pytest  # noqa: E402
+from httpx import AsyncClient, ASGITransport  # noqa: E402
+from django.core.asgi import get_asgi_application  # noqa: E402
+
+
+@pytest.fixture(scope="session")
+def asgi_app():
+    """Shared ASGI application for all tests."""
+    return get_asgi_application()
+
+
+@pytest.mark.asyncio
+async def test_async_exception_is_captured(asgi_app):
+    """
+    Test that async view exceptions are captured to PostHog.
+
+    The middleware's process_exception() method ensures exceptions are captured.
+    Without it (v6.7.11 and earlier), exceptions are NOT captured even though 500 is returned.
+    """
+    from unittest.mock import patch
+
+    # Track captured exceptions
+    captured = []
+
+    def mock_capture(exception, **kwargs):
+        """Mock capture_exception to record calls."""
+        captured.append(
+            {
+                "exception": exception,
+                "type": type(exception).__name__,
+                "message": str(exception),
+            }
+        )
+
+    # Patch at the posthog module level where middleware imports from
+    with patch("posthog.capture_exception", side_effect=mock_capture):
+        async with AsyncClient(
+            transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+        ) as ac:
+            response = await ac.get("/test/async-exception")
+
+        # Django returns 500
+        assert response.status_code == 500
+
+        # CRITICAL: Verify PostHog captured the exception
+        assert len(captured) > 0, "Exception was NOT captured to PostHog!"
+
+        # Verify it's the right exception
+        exception_data = captured[0]
+        assert exception_data["type"] == "ValueError"
+        assert "Test exception from Django 5 async view" in exception_data["message"]
+
+
+@pytest.mark.asyncio
+async def test_sync_exception_is_captured(asgi_app):
+    """
+    Test that sync view exceptions are captured to PostHog.
+
+    The middleware's process_exception() method ensures exceptions are captured.
+    Without it (v6.7.11 and earlier), exceptions are NOT captured even though 500 is returned.
+    """
+    from unittest.mock import patch
+
+    # Track captured exceptions
+    captured = []
+
+    def mock_capture(exception, **kwargs):
+        """Mock capture_exception to record calls."""
+        captured.append(
+            {
+                "exception": exception,
+                "type": type(exception).__name__,
+                "message": str(exception),
+            }
+        )
+
+    # Patch at the posthog module level where middleware imports from
+    with patch("posthog.capture_exception", side_effect=mock_capture):
+        async with AsyncClient(
+            transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+        ) as ac:
+            response = await ac.get("/test/sync-exception")
+
+        # Django returns 500
+        assert response.status_code == 500
+
+        # CRITICAL: Verify PostHog captured the exception
+        assert len(captured) > 0, "Exception was NOT captured to PostHog!"
+
+        # Verify it's the right exception
+        exception_data = captured[0]
+        assert exception_data["type"] == "ValueError"
+        assert "Test exception from Django 5 sync view" in exception_data["message"]

integration_tests/django5/test_middleware.py

@@ -0,0 +1,170 @@
+"""
+Tests for PostHog Django middleware in async context.
+
+These tests verify that the middleware correctly handles:
+1. Async user access (request.auser() in Django 5)
+2. Exception capture in both sync and async views
+3. No SynchronousOnlyOperation errors in async context
+
+Tests run directly against the ASGI application without needing a server.
+"""
+
+import os
+import django
+
+# Setup Django before importing anything else
+os.environ.setdefault("DJANGO_SETTINGS_MODULE", "testdjango.settings")
+django.setup()
+
+import pytest  # noqa: E402
+from httpx import AsyncClient, ASGITransport  # noqa: E402
+from django.core.asgi import get_asgi_application  # noqa: E402
+
+
+@pytest.fixture(scope="session")
+def asgi_app():
+    """Shared ASGI application for all tests."""
+    return get_asgi_application()
+
+
+@pytest.mark.asyncio
+async def test_async_user_access(asgi_app):
+    """
+    Test that middleware can access request.user in async context.
+
+    In Django 5, this requires using await request.auser() instead of request.user
+    to avoid SynchronousOnlyOperation error.
+
+    Without authentication, request.user is AnonymousUser which doesn't
+    trigger the lazy loading bug. This test verifies the middleware works
+    in the common case.
+    """
+    async with AsyncClient(
+        transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+    ) as ac:
+        response = await ac.get("/test/async-user")
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["status"] == "success"
+    assert "django_version" in data
+
+
+@pytest.mark.django_db(transaction=True)
+@pytest.mark.asyncio
+async def test_async_authenticated_user_access(asgi_app):
+    """
+    Test that middleware can access an authenticated user in async context.
+
+    This is the critical test that triggers the SynchronousOnlyOperation bug
+    in v6.7.11. When AuthenticationMiddleware sets request.user to a
+    SimpleLazyObject wrapping a database query, accessing user.pk or user.email
+    in async context causes the error.
+
+    In v6.7.11, extract_request_user() does getattr(user, "is_authenticated", False)
+    which triggers the lazy object evaluation synchronously.
+
+    The fix uses await request.auser() instead to avoid this.
+    """
+    from django.contrib.auth import get_user_model
+    from django.test import Client
+    from asgiref.sync import sync_to_async
+    from django.test import override_settings
+
+    # Create a test user (must use sync_to_async since we're in async test)
+    User = get_user_model()
+
+    @sync_to_async
+    def create_or_get_user():
+        user, created = User.objects.get_or_create(
+            username="testuser",
+            defaults={
+                "email": "test@example.com",
+            },
+        )
+        if created:
+            user.set_password("testpass123")
+            user.save()
+        return user
+
+    user = await create_or_get_user()
+
+    # Create a session with authenticated user (sync operation)
+    @sync_to_async
+    def create_session():
+        client = Client()
+        client.force_login(user)
+        return client.cookies.get("sessionid")
+
+    session_cookie = await create_session()
+
+    if not session_cookie:
+        pytest.skip("Could not create authenticated session")
+
+    # Make request with session cookie - this should trigger the bug in v6.7.11
+    # Disable exception capture to see the SynchronousOnlyOperation clearly
+    with override_settings(POSTHOG_MW_CAPTURE_EXCEPTIONS=False):
+        async with AsyncClient(
+            transport=ASGITransport(app=asgi_app),
+            base_url="http://testserver",
+            cookies={"sessionid": session_cookie.value},
+        ) as ac:
+            response = await ac.get("/test/async-user")
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["status"] == "success"
+        assert data["user_authenticated"]
+
+
+@pytest.mark.asyncio
+async def test_sync_user_access(asgi_app):
+    """
+    Test that middleware works with sync views.
+
+    This should always work regardless of middleware version.
+    """
+    async with AsyncClient(
+        transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+    ) as ac:
+        response = await ac.get("/test/sync-user")
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["status"] == "success"
+
+
+@pytest.mark.asyncio
+async def test_async_exception_capture(asgi_app):
+    """
+    Test that middleware handles exceptions from async views.
+
+    The middleware's process_exception() method captures view exceptions to PostHog
+    before Django converts them to 500 responses. This test verifies the exception
+    causes a 500 response. See test_exception_capture.py for tests that verify
+    actual exception capture to PostHog.
+    """
+    async with AsyncClient(
+        transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+    ) as ac:
+        response = await ac.get("/test/async-exception")
+
+    # Django returns 500 for unhandled exceptions
+    assert response.status_code == 500
+
+
+@pytest.mark.asyncio
+async def test_sync_exception_capture(asgi_app):
+    """
+    Test that middleware handles exceptions from sync views.
+
+    The middleware's process_exception() method captures view exceptions to PostHog.
+    This test verifies the exception causes a 500 response.
+    """
+    async with AsyncClient(
+        transport=ASGITransport(app=asgi_app), base_url="http://testserver"
+    ) as ac:
+        response = await ac.get("/test/sync-exception")
+
+    # Django returns 500 for unhandled exceptions
+    assert response.status_code == 500

integration_tests/django5/testdjango/asgi.py

@@ -1,16 +1,16 @@
 """
-ASGI config for sentry_django_example project.
+ASGI config for testdjango project.
 
 It exposes the ASGI callable as a module-level variable named ``application``.
 
 For more information on this file, see
-https://docs.djangoproject.com/en/3.2/howto/deployment/asgi/
+https://docs.djangoproject.com/en/5.2/howto/deployment/asgi/
 """
 
 import os
 
 from django.core.asgi import get_asgi_application
 
-os.environ.setdefault("DJANGO_SETTINGS_MODULE", "sentry_django_example.settings")
+os.environ.setdefault("DJANGO_SETTINGS_MODULE", "testdjango.settings")
 
 application = get_asgi_application()

integration_tests/django5/testdjango/settings.py

@@ -0,0 +1,129 @@
+"""
+Django settings for testdjango project.
+
+Generated by 'django-admin startproject' using Django 5.2.7.
+
+For more information on this file, see
+https://docs.djangoproject.com/en/5.2/topics/settings/
+
+For the full list of settings and their values, see
+https://docs.djangoproject.com/en/5.2/ref/settings/
+"""
+
+from pathlib import Path
+
+# Build paths inside the project like this: BASE_DIR / 'subdir'.
+BASE_DIR = Path(__file__).resolve().parent.parent
+
+
+# Quick-start development settings - unsuitable for production
+# See https://docs.djangoproject.com/en/5.2/howto/deployment/checklist/
+
+# SECURITY WARNING: keep the secret key used in production secret!
+SECRET_KEY = "django-insecure-q5(&wfw@_lb)noyowbfl$2ls8c82hl__0f9s5(mohlh2)aas#3"
+
+# SECURITY WARNING: don't run with debug turned on in production!
+DEBUG = True
+
+ALLOWED_HOSTS = ["*"]
+
+
+# Application definition
+
+INSTALLED_APPS = [
+    "django.contrib.admin",
+    "django.contrib.auth",
+    "django.contrib.contenttypes",
+    "django.contrib.sessions",
+    "django.contrib.messages",
+    "django.contrib.staticfiles",
+]
+
+MIDDLEWARE = [
+    "django.middleware.security.SecurityMiddleware",
+    "django.contrib.sessions.middleware.SessionMiddleware",
+    "django.middleware.common.CommonMiddleware",
+    "django.middleware.csrf.CsrfViewMiddleware",
+    "django.contrib.auth.middleware.AuthenticationMiddleware",
+    "django.contrib.messages.middleware.MessageMiddleware",
+    "django.middleware.clickjacking.XFrameOptionsMiddleware",
+    "posthog.integrations.django.PosthogContextMiddleware",  # Test PostHog middleware
+]
+
+ROOT_URLCONF = "testdjango.urls"
+
+TEMPLATES = [
+    {
+        "BACKEND": "django.template.backends.django.DjangoTemplates",
+        "DIRS": [],
+        "APP_DIRS": True,
+        "OPTIONS": {
+            "context_processors": [
+                "django.template.context_processors.request",
+                "django.contrib.auth.context_processors.auth",
+                "django.contrib.messages.context_processors.messages",
+            ],
+        },
+    },
+]
+
+WSGI_APPLICATION = "testdjango.wsgi.application"
+
+
+# Database
+# https://docs.djangoproject.com/en/5.2/ref/settings/#databases
+
+DATABASES = {
+    "default": {
+        "ENGINE": "django.db.backends.sqlite3",
+        "NAME": BASE_DIR / "db.sqlite3",
+    }
+}
+
+
+# Password validation
+# https://docs.djangoproject.com/en/5.2/ref/settings/#auth-password-validators
+
+AUTH_PASSWORD_VALIDATORS = [
+    {
+        "NAME": "django.contrib.auth.password_validation.UserAttributeSimilarityValidator",
+    },
+    {
+        "NAME": "django.contrib.auth.password_validation.MinimumLengthValidator",
+    },
+    {
+        "NAME": "django.contrib.auth.password_validation.CommonPasswordValidator",
+    },
+    {
+        "NAME": "django.contrib.auth.password_validation.NumericPasswordValidator",
+    },
+]
+
+
+# Internationalization
+# https://docs.djangoproject.com/en/5.2/topics/i18n/
+
+LANGUAGE_CODE = "en-us"
+
+TIME_ZONE = "UTC"
+
+USE_I18N = True
+
+USE_TZ = True
+
+
+# Static files (CSS, JavaScript, Images)
+# https://docs.djangoproject.com/en/5.2/howto/static-files/
+
+STATIC_URL = "static/"
+
+# Default primary key field type
+# https://docs.djangoproject.com/en/5.2/ref/settings/#default-auto-field
+
+DEFAULT_AUTO_FIELD = "django.db.models.BigAutoField"
+
+
+# PostHog settings for testing
+POSTHOG_API_KEY = "test-key"
+POSTHOG_HOST = "https://app.posthog.com"
+POSTHOG_MW_CAPTURE_EXCEPTIONS = True

integration_tests/django5/testdjango/urls.py

@@ -1,7 +1,8 @@
-"""sentry_django_example URL Configuration
+"""
+URL configuration for testdjango project.
 
 The `urlpatterns` list routes URLs to views. For more information please see:
-    https://docs.djangoproject.com/en/3.2/topics/http/urls/
+    https://docs.djangoproject.com/en/5.2/topics/http/urls/
 Examples:
 Function views
     1. Add an import:  from my_app import views
@@ -16,13 +17,12 @@ Including another URLconf
 
 from django.contrib import admin
 from django.urls import path
-
-
-def trigger_error(request):
-    division_by_zero = 1 / 0
-
+from testdjango import views
 
 urlpatterns = [
     path("admin/", admin.site.urls),
-    path("sentry-debug/", trigger_error),
+    path("test/async-user", views.test_async_user),
+    path("test/sync-user", views.test_sync_user),
+    path("test/async-exception", views.test_async_exception),
+    path("test/sync-exception", views.test_sync_exception),
 ]

integration_tests/django5/testdjango/views.py

@@ -0,0 +1,50 @@
+"""
+Test views for validating PostHog middleware with Django 5 ASGI.
+"""
+
+from django.http import JsonResponse
+
+
+async def test_async_user(request):
+    """
+    Async view that tests middleware with request.user access.
+
+    The middleware will access request.user (SimpleLazyObject) via auser()
+    in async context. Without the fix, this causes SynchronousOnlyOperation.
+    """
+    # The middleware has already accessed request.user via auser()
+    # If we got here, the fix works!
+    user = await request.auser()
+
+    return JsonResponse(
+        {
+            "status": "success",
+            "message": "Django 5 async middleware test passed!",
+            "django_version": "5.x",
+            "user_authenticated": user.is_authenticated if user else False,
+            "note": "Middleware used await request.auser() successfully",
+        }
+    )
+
+
+def test_sync_user(request):
+    """Sync view for comparison."""
+    return JsonResponse(
+        {
+            "status": "success",
+            "message": "Sync view works",
+            "user_authenticated": request.user.is_authenticated
+            if hasattr(request, "user")
+            else False,
+        }
+    )
+
+
+async def test_async_exception(request):
+    """Async view that raises an exception for testing exception capture."""
+    raise ValueError("Test exception from Django 5 async view")
+
+
+def test_sync_exception(request):
+    """Sync view that raises an exception for testing exception capture."""
+    raise ValueError("Test exception from Django 5 sync view")

integration_tests/django5/testdjango/wsgi.py

@@ -1,16 +1,16 @@
 """
-WSGI config for sentry_django_example project.
+WSGI config for testdjango project.
 
 It exposes the WSGI callable as a module-level variable named ``application``.
 
 For more information on this file, see
-https://docs.djangoproject.com/en/3.2/howto/deployment/wsgi/
+https://docs.djangoproject.com/en/5.2/howto/deployment/wsgi/
 """
 
 import os
 
 from django.core.wsgi import get_wsgi_application
 
-os.environ.setdefault("DJANGO_SETTINGS_MODULE", "sentry_django_example.settings")
+os.environ.setdefault("DJANGO_SETTINGS_MODULE", "testdjango.settings")
 
 application = get_wsgi_application()

integration_tests/django5/uv.lock

@@ -0,0 +1,674 @@
+version = 1
+revision = 3
+requires-python = ">=3.12"
+
+[[package]]
+name = "anyio"
+version = "4.11.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "idna" },
+    { name = "sniffio" },
+    { name = "typing-extensions", marker = "python_full_version < '3.13'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/c6/78/7d432127c41b50bccba979505f272c16cbcadcc33645d5fa3a738110ae75/anyio-4.11.0.tar.gz", hash = "sha256:82a8d0b81e318cc5ce71a5f1f8b5c4e63619620b63141ef8c995fa0db95a57c4", size = 219094, upload-time = "2025-09-23T09:19:12.58Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/15/b3/9b1a8074496371342ec1e796a96f99c82c945a339cd81a8e73de28b4cf9e/anyio-4.11.0-py3-none-any.whl", hash = "sha256:0287e96f4d26d4149305414d4e3bc32f0dcd0862365a4bddea19d7a1ec38c4fc", size = 109097, upload-time = "2025-09-23T09:19:10.601Z" },
+]
+
+[[package]]
+name = "asgiref"
+version = "3.10.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/46/08/4dfec9b90758a59acc6be32ac82e98d1fbfc321cb5cfa410436dbacf821c/asgiref-3.10.0.tar.gz", hash = "sha256:d89f2d8cd8b56dada7d52fa7dc8075baa08fb836560710d38c292a7a3f78c04e", size = 37483, upload-time = "2025-10-05T09:15:06.557Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/17/9c/fc2331f538fbf7eedba64b2052e99ccf9ba9d6888e2f41441ee28847004b/asgiref-3.10.0-py3-none-any.whl", hash = "sha256:aef8a81283a34d0ab31630c9b7dfe70c812c95eba78171367ca8745e88124734", size = 24050, upload-time = "2025-10-05T09:15:05.11Z" },
+]
+
+[[package]]
+name = "backoff"
+version = "2.2.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/47/d7/5bbeb12c44d7c4f2fb5b56abce497eb5ed9f34d85701de869acedd602619/backoff-2.2.1.tar.gz", hash = "sha256:03f829f5bb1923180821643f8753b0502c3b682293992485b0eef2807afa5cba", size = 17001, upload-time = "2022-10-05T19:19:32.061Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/df/73/b6e24bd22e6720ca8ee9a85a0c4a2971af8497d8f3193fa05390cbd46e09/backoff-2.2.1-py3-none-any.whl", hash = "sha256:63579f9a0628e06278f7e47b7d7d5b6ce20dc65c5e96a6f3ca99a6adca0396e8", size = 15148, upload-time = "2022-10-05T19:19:30.546Z" },
+]
+
+[[package]]
+name = "certifi"
+version = "2025.10.5"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/4c/5b/b6ce21586237c77ce67d01dc5507039d444b630dd76611bbca2d8e5dcd91/certifi-2025.10.5.tar.gz", hash = "sha256:47c09d31ccf2acf0be3f701ea53595ee7e0b8fa08801c6624be771df09ae7b43", size = 164519, upload-time = "2025-10-05T04:12:15.808Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/e4/37/af0d2ef3967ac0d6113837b44a4f0bfe1328c2b9763bd5b1744520e5cfed/certifi-2025.10.5-py3-none-any.whl", hash = "sha256:0f212c2744a9bb6de0c56639a6f68afe01ecd92d91f14ae897c4fe7bbeeef0de", size = 163286, upload-time = "2025-10-05T04:12:14.03Z" },
+]
+
+[[package]]
+name = "charset-normalizer"
+version = "3.4.4"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/13/69/33ddede1939fdd074bce5434295f38fae7136463422fe4fd3e0e89b98062/charset_normalizer-3.4.4.tar.gz", hash = "sha256:94537985111c35f28720e43603b8e7b43a6ecfb2ce1d3058bbe955b73404e21a", size = 129418, upload-time = "2025-10-14T04:42:32.879Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f3/85/1637cd4af66fa687396e757dec650f28025f2a2f5a5531a3208dc0ec43f2/charset_normalizer-3.4.4-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:0a98e6759f854bd25a58a73fa88833fba3b7c491169f86ce1180c948ab3fd394", size = 208425, upload-time = "2025-10-14T04:40:53.353Z" },
+    { url = "https://files.pythonhosted.org/packages/9d/6a/04130023fef2a0d9c62d0bae2649b69f7b7d8d24ea5536feef50551029df/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b5b290ccc2a263e8d185130284f8501e3e36c5e02750fc6b6bdeb2e9e96f1e25", size = 148162, upload-time = "2025-10-14T04:40:54.558Z" },
+    { url = "https://files.pythonhosted.org/packages/78/29/62328d79aa60da22c9e0b9a66539feae06ca0f5a4171ac4f7dc285b83688/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:74bb723680f9f7a6234dcf67aea57e708ec1fbdf5699fb91dfd6f511b0a320ef", size = 144558, upload-time = "2025-10-14T04:40:55.677Z" },
+    { url = "https://files.pythonhosted.org/packages/86/bb/b32194a4bf15b88403537c2e120b817c61cd4ecffa9b6876e941c3ee38fe/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:f1e34719c6ed0b92f418c7c780480b26b5d9c50349e9a9af7d76bf757530350d", size = 161497, upload-time = "2025-10-14T04:40:57.217Z" },
+    { url = "https://files.pythonhosted.org/packages/19/89/a54c82b253d5b9b111dc74aca196ba5ccfcca8242d0fb64146d4d3183ff1/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:2437418e20515acec67d86e12bf70056a33abdacb5cb1655042f6538d6b085a8", size = 159240, upload-time = "2025-10-14T04:40:58.358Z" },
+    { url = "https://files.pythonhosted.org/packages/c0/10/d20b513afe03acc89ec33948320a5544d31f21b05368436d580dec4e234d/charset_normalizer-3.4.4-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:11d694519d7f29d6cd09f6ac70028dba10f92f6cdd059096db198c283794ac86", size = 153471, upload-time = "2025-10-14T04:40:59.468Z" },
+    { url = "https://files.pythonhosted.org/packages/61/fa/fbf177b55bdd727010f9c0a3c49eefa1d10f960e5f09d1d887bf93c2e698/charset_normalizer-3.4.4-cp312-cp312-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:ac1c4a689edcc530fc9d9aa11f5774b9e2f33f9a0c6a57864e90908f5208d30a", size = 150864, upload-time = "2025-10-14T04:41:00.623Z" },
+    { url = "https://files.pythonhosted.org/packages/05/12/9fbc6a4d39c0198adeebbde20b619790e9236557ca59fc40e0e3cebe6f40/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:21d142cc6c0ec30d2efee5068ca36c128a30b0f2c53c1c07bd78cb6bc1d3be5f", size = 150647, upload-time = "2025-10-14T04:41:01.754Z" },
+    { url = "https://files.pythonhosted.org/packages/ad/1f/6a9a593d52e3e8c5d2b167daf8c6b968808efb57ef4c210acb907c365bc4/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:5dbe56a36425d26d6cfb40ce79c314a2e4dd6211d51d6d2191c00bed34f354cc", size = 145110, upload-time = "2025-10-14T04:41:03.231Z" },
+    { url = "https://files.pythonhosted.org/packages/30/42/9a52c609e72471b0fc54386dc63c3781a387bb4fe61c20231a4ebcd58bdd/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5bfbb1b9acf3334612667b61bd3002196fe2a1eb4dd74d247e0f2a4d50ec9bbf", size = 162839, upload-time = "2025-10-14T04:41:04.715Z" },
+    { url = "https://files.pythonhosted.org/packages/c4/5b/c0682bbf9f11597073052628ddd38344a3d673fda35a36773f7d19344b23/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_riscv64.whl", hash = "sha256:d055ec1e26e441f6187acf818b73564e6e6282709e9bcb5b63f5b23068356a15", size = 150667, upload-time = "2025-10-14T04:41:05.827Z" },
+    { url = "https://files.pythonhosted.org/packages/e4/24/a41afeab6f990cf2daf6cb8c67419b63b48cf518e4f56022230840c9bfb2/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:af2d8c67d8e573d6de5bc30cdb27e9b95e49115cd9baad5ddbd1a6207aaa82a9", size = 160535, upload-time = "2025-10-14T04:41:06.938Z" },
+    { url = "https://files.pythonhosted.org/packages/2a/e5/6a4ce77ed243c4a50a1fecca6aaaab419628c818a49434be428fe24c9957/charset_normalizer-3.4.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:780236ac706e66881f3b7f2f32dfe90507a09e67d1d454c762cf642e6e1586e0", size = 154816, upload-time = "2025-10-14T04:41:08.101Z" },
+    { url = "https://files.pythonhosted.org/packages/a8/ef/89297262b8092b312d29cdb2517cb1237e51db8ecef2e9af5edbe7b683b1/charset_normalizer-3.4.4-cp312-cp312-win32.whl", hash = "sha256:5833d2c39d8896e4e19b689ffc198f08ea58116bee26dea51e362ecc7cd3ed26", size = 99694, upload-time = "2025-10-14T04:41:09.23Z" },
+    { url = "https://files.pythonhosted.org/packages/3d/2d/1e5ed9dd3b3803994c155cd9aacb60c82c331bad84daf75bcb9c91b3295e/charset_normalizer-3.4.4-cp312-cp312-win_amd64.whl", hash = "sha256:a79cfe37875f822425b89a82333404539ae63dbdddf97f84dcbc3d339aae9525", size = 107131, upload-time = "2025-10-14T04:41:10.467Z" },
+    { url = "https://files.pythonhosted.org/packages/d0/d9/0ed4c7098a861482a7b6a95603edce4c0d9db2311af23da1fb2b75ec26fc/charset_normalizer-3.4.4-cp312-cp312-win_arm64.whl", hash = "sha256:376bec83a63b8021bb5c8ea75e21c4ccb86e7e45ca4eb81146091b56599b80c3", size = 100390, upload-time = "2025-10-14T04:41:11.915Z" },
+    { url = "https://files.pythonhosted.org/packages/97/45/4b3a1239bbacd321068ea6e7ac28875b03ab8bc0aa0966452db17cd36714/charset_normalizer-3.4.4-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:e1f185f86a6f3403aa2420e815904c67b2f9ebc443f045edd0de921108345794", size = 208091, upload-time = "2025-10-14T04:41:13.346Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/62/73a6d7450829655a35bb88a88fca7d736f9882a27eacdca2c6d505b57e2e/charset_normalizer-3.4.4-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6b39f987ae8ccdf0d2642338faf2abb1862340facc796048b604ef14919e55ed", size = 147936, upload-time = "2025-10-14T04:41:14.461Z" },
+    { url = "https://files.pythonhosted.org/packages/89/c5/adb8c8b3d6625bef6d88b251bbb0d95f8205831b987631ab0c8bb5d937c2/charset_normalizer-3.4.4-cp313-cp313-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:3162d5d8ce1bb98dd51af660f2121c55d0fa541b46dff7bb9b9f86ea1d87de72", size = 144180, upload-time = "2025-10-14T04:41:15.588Z" },
+    { url = "https://files.pythonhosted.org/packages/91/ed/9706e4070682d1cc219050b6048bfd293ccf67b3d4f5a4f39207453d4b99/charset_normalizer-3.4.4-cp313-cp313-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:81d5eb2a312700f4ecaa977a8235b634ce853200e828fbadf3a9c50bab278328", size = 161346, upload-time = "2025-10-14T04:41:16.738Z" },
+    { url = "https://files.pythonhosted.org/packages/d5/0d/031f0d95e4972901a2f6f09ef055751805ff541511dc1252ba3ca1f80cf5/charset_normalizer-3.4.4-cp313-cp313-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:5bd2293095d766545ec1a8f612559f6b40abc0eb18bb2f5d1171872d34036ede", size = 158874, upload-time = "2025-10-14T04:41:17.923Z" },
+    { url = "https://files.pythonhosted.org/packages/f5/83/6ab5883f57c9c801ce5e5677242328aa45592be8a00644310a008d04f922/charset_normalizer-3.4.4-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a8a8b89589086a25749f471e6a900d3f662d1d3b6e2e59dcecf787b1cc3a1894", size = 153076, upload-time = "2025-10-14T04:41:19.106Z" },
+    { url = "https://files.pythonhosted.org/packages/75/1e/5ff781ddf5260e387d6419959ee89ef13878229732732ee73cdae01800f2/charset_normalizer-3.4.4-cp313-cp313-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:bc7637e2f80d8530ee4a78e878bce464f70087ce73cf7c1caf142416923b98f1", size = 150601, upload-time = "2025-10-14T04:41:20.245Z" },
+    { url = "https://files.pythonhosted.org/packages/d7/57/71be810965493d3510a6ca79b90c19e48696fb1ff964da319334b12677f0/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:f8bf04158c6b607d747e93949aa60618b61312fe647a6369f88ce2ff16043490", size = 150376, upload-time = "2025-10-14T04:41:21.398Z" },
+    { url = "https://files.pythonhosted.org/packages/e5/d5/c3d057a78c181d007014feb7e9f2e65905a6c4ef182c0ddf0de2924edd65/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:554af85e960429cf30784dd47447d5125aaa3b99a6f0683589dbd27e2f45da44", size = 144825, upload-time = "2025-10-14T04:41:22.583Z" },
+    { url = "https://files.pythonhosted.org/packages/e6/8c/d0406294828d4976f275ffbe66f00266c4b3136b7506941d87c00cab5272/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:74018750915ee7ad843a774364e13a3db91682f26142baddf775342c3f5b1133", size = 162583, upload-time = "2025-10-14T04:41:23.754Z" },
+    { url = "https://files.pythonhosted.org/packages/d7/24/e2aa1f18c8f15c4c0e932d9287b8609dd30ad56dbe41d926bd846e22fb8d/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_riscv64.whl", hash = "sha256:c0463276121fdee9c49b98908b3a89c39be45d86d1dbaa22957e38f6321d4ce3", size = 150366, upload-time = "2025-10-14T04:41:25.27Z" },
+    { url = "https://files.pythonhosted.org/packages/e4/5b/1e6160c7739aad1e2df054300cc618b06bf784a7a164b0f238360721ab86/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:362d61fd13843997c1c446760ef36f240cf81d3ebf74ac62652aebaf7838561e", size = 160300, upload-time = "2025-10-14T04:41:26.725Z" },
+    { url = "https://files.pythonhosted.org/packages/7a/10/f882167cd207fbdd743e55534d5d9620e095089d176d55cb22d5322f2afd/charset_normalizer-3.4.4-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9a26f18905b8dd5d685d6d07b0cdf98a79f3c7a918906af7cc143ea2e164c8bc", size = 154465, upload-time = "2025-10-14T04:41:28.322Z" },
+    { url = "https://files.pythonhosted.org/packages/89/66/c7a9e1b7429be72123441bfdbaf2bc13faab3f90b933f664db506dea5915/charset_normalizer-3.4.4-cp313-cp313-win32.whl", hash = "sha256:9b35f4c90079ff2e2edc5b26c0c77925e5d2d255c42c74fdb70fb49b172726ac", size = 99404, upload-time = "2025-10-14T04:41:29.95Z" },
+    { url = "https://files.pythonhosted.org/packages/c4/26/b9924fa27db384bdcd97ab83b4f0a8058d96ad9626ead570674d5e737d90/charset_normalizer-3.4.4-cp313-cp313-win_amd64.whl", hash = "sha256:b435cba5f4f750aa6c0a0d92c541fb79f69a387c91e61f1795227e4ed9cece14", size = 107092, upload-time = "2025-10-14T04:41:31.188Z" },
+    { url = "https://files.pythonhosted.org/packages/af/8f/3ed4bfa0c0c72a7ca17f0380cd9e4dd842b09f664e780c13cff1dcf2ef1b/charset_normalizer-3.4.4-cp313-cp313-win_arm64.whl", hash = "sha256:542d2cee80be6f80247095cc36c418f7bddd14f4a6de45af91dfad36d817bba2", size = 100408, upload-time = "2025-10-14T04:41:32.624Z" },
+    { url = "https://files.pythonhosted.org/packages/2a/35/7051599bd493e62411d6ede36fd5af83a38f37c4767b92884df7301db25d/charset_normalizer-3.4.4-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:da3326d9e65ef63a817ecbcc0df6e94463713b754fe293eaa03da99befb9a5bd", size = 207746, upload-time = "2025-10-14T04:41:33.773Z" },
+    { url = "https://files.pythonhosted.org/packages/10/9a/97c8d48ef10d6cd4fcead2415523221624bf58bcf68a802721a6bc807c8f/charset_normalizer-3.4.4-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8af65f14dc14a79b924524b1e7fffe304517b2bff5a58bf64f30b98bbc5079eb", size = 147889, upload-time = "2025-10-14T04:41:34.897Z" },
+    { url = "https://files.pythonhosted.org/packages/10/bf/979224a919a1b606c82bd2c5fa49b5c6d5727aa47b4312bb27b1734f53cd/charset_normalizer-3.4.4-cp314-cp314-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:74664978bb272435107de04e36db5a9735e78232b85b77d45cfb38f758efd33e", size = 143641, upload-time = "2025-10-14T04:41:36.116Z" },
+    { url = "https://files.pythonhosted.org/packages/ba/33/0ad65587441fc730dc7bd90e9716b30b4702dc7b617e6ba4997dc8651495/charset_normalizer-3.4.4-cp314-cp314-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:752944c7ffbfdd10c074dc58ec2d5a8a4cd9493b314d367c14d24c17684ddd14", size = 160779, upload-time = "2025-10-14T04:41:37.229Z" },
+    { url = "https://files.pythonhosted.org/packages/67/ed/331d6b249259ee71ddea93f6f2f0a56cfebd46938bde6fcc6f7b9a3d0e09/charset_normalizer-3.4.4-cp314-cp314-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:d1f13550535ad8cff21b8d757a3257963e951d96e20ec82ab44bc64aeb62a191", size = 159035, upload-time = "2025-10-14T04:41:38.368Z" },
+    { url = "https://files.pythonhosted.org/packages/67/ff/f6b948ca32e4f2a4576aa129d8bed61f2e0543bf9f5f2b7fc3758ed005c9/charset_normalizer-3.4.4-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ecaae4149d99b1c9e7b88bb03e3221956f68fd6d50be2ef061b2381b61d20838", size = 152542, upload-time = "2025-10-14T04:41:39.862Z" },
+    { url = "https://files.pythonhosted.org/packages/16/85/276033dcbcc369eb176594de22728541a925b2632f9716428c851b149e83/charset_normalizer-3.4.4-cp314-cp314-manylinux_2_31_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:cb6254dc36b47a990e59e1068afacdcd02958bdcce30bb50cc1700a8b9d624a6", size = 149524, upload-time = "2025-10-14T04:41:41.319Z" },
+    { url = "https://files.pythonhosted.org/packages/9e/f2/6a2a1f722b6aba37050e626530a46a68f74e63683947a8acff92569f979a/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:c8ae8a0f02f57a6e61203a31428fa1d677cbe50c93622b4149d5c0f319c1d19e", size = 150395, upload-time = "2025-10-14T04:41:42.539Z" },
+    { url = "https://files.pythonhosted.org/packages/60/bb/2186cb2f2bbaea6338cad15ce23a67f9b0672929744381e28b0592676824/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_armv7l.whl", hash = "sha256:47cc91b2f4dd2833fddaedd2893006b0106129d4b94fdb6af1f4ce5a9965577c", size = 143680, upload-time = "2025-10-14T04:41:43.661Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/a5/bf6f13b772fbb2a90360eb620d52ed8f796f3c5caee8398c3b2eb7b1c60d/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_ppc64le.whl", hash = "sha256:82004af6c302b5d3ab2cfc4cc5f29db16123b1a8417f2e25f9066f91d4411090", size = 162045, upload-time = "2025-10-14T04:41:44.821Z" },
+    { url = "https://files.pythonhosted.org/packages/df/c5/d1be898bf0dc3ef9030c3825e5d3b83f2c528d207d246cbabe245966808d/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_riscv64.whl", hash = "sha256:2b7d8f6c26245217bd2ad053761201e9f9680f8ce52f0fcd8d0755aeae5b2152", size = 149687, upload-time = "2025-10-14T04:41:46.442Z" },
+    { url = "https://files.pythonhosted.org/packages/a5/42/90c1f7b9341eef50c8a1cb3f098ac43b0508413f33affd762855f67a410e/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_s390x.whl", hash = "sha256:799a7a5e4fb2d5898c60b640fd4981d6a25f1c11790935a44ce38c54e985f828", size = 160014, upload-time = "2025-10-14T04:41:47.631Z" },
+    { url = "https://files.pythonhosted.org/packages/76/be/4d3ee471e8145d12795ab655ece37baed0929462a86e72372fd25859047c/charset_normalizer-3.4.4-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:99ae2cffebb06e6c22bdc25801d7b30f503cc87dbd283479e7b606f70aff57ec", size = 154044, upload-time = "2025-10-14T04:41:48.81Z" },
+    { url = "https://files.pythonhosted.org/packages/b0/6f/8f7af07237c34a1defe7defc565a9bc1807762f672c0fde711a4b22bf9c0/charset_normalizer-3.4.4-cp314-cp314-win32.whl", hash = "sha256:f9d332f8c2a2fcbffe1378594431458ddbef721c1769d78e2cbc06280d8155f9", size = 99940, upload-time = "2025-10-14T04:41:49.946Z" },
+    { url = "https://files.pythonhosted.org/packages/4b/51/8ade005e5ca5b0d80fb4aff72a3775b325bdc3d27408c8113811a7cbe640/charset_normalizer-3.4.4-cp314-cp314-win_amd64.whl", hash = "sha256:8a6562c3700cce886c5be75ade4a5db4214fda19fede41d9792d100288d8f94c", size = 107104, upload-time = "2025-10-14T04:41:51.051Z" },
+    { url = "https://files.pythonhosted.org/packages/da/5f/6b8f83a55bb8278772c5ae54a577f3099025f9ade59d0136ac24a0df4bde/charset_normalizer-3.4.4-cp314-cp314-win_arm64.whl", hash = "sha256:de00632ca48df9daf77a2c65a484531649261ec9f25489917f09e455cb09ddb2", size = 100743, upload-time = "2025-10-14T04:41:52.122Z" },
+    { url = "https://files.pythonhosted.org/packages/0a/4c/925909008ed5a988ccbb72dcc897407e5d6d3bd72410d69e051fc0c14647/charset_normalizer-3.4.4-py3-none-any.whl", hash = "sha256:7a32c560861a02ff789ad905a2fe94e3f840803362c84fecf1851cb4cf3dc37f", size = 53402, upload-time = "2025-10-14T04:42:31.76Z" },
+]
+
+[[package]]
+name = "click"
+version = "8.3.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/46/61/de6cd827efad202d7057d93e0fed9294b96952e188f7384832791c7b2254/click-8.3.0.tar.gz", hash = "sha256:e7b8232224eba16f4ebe410c25ced9f7875cb5f3263ffc93cc3e8da705e229c4", size = 276943, upload-time = "2025-09-18T17:32:23.696Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/db/d3/9dcc0f5797f070ec8edf30fbadfb200e71d9db6b84d211e3b2085a7589a0/click-8.3.0-py3-none-any.whl", hash = "sha256:9b9f285302c6e3064f4330c05f05b81945b2a39544279343e6e7c5f27a9baddc", size = 107295, upload-time = "2025-09-18T17:32:22.42Z" },
+]
+
+[[package]]
+name = "colorama"
+version = "0.4.6"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.6.tar.gz", hash = "sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44", size = 27697, upload-time = "2022-10-25T02:36:22.414Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/d1/d6/3965ed04c63042e047cb6a3e6ed1a63a35087b6a609aa3a15ed8ac56c221/colorama-0.4.6-py2.py3-none-any.whl", hash = "sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6", size = 25335, upload-time = "2022-10-25T02:36:20.889Z" },
+]
+
+[[package]]
+name = "distro"
+version = "1.9.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/fc/f8/98eea607f65de6527f8a2e8885fc8015d3e6f5775df186e443e0964a11c3/distro-1.9.0.tar.gz", hash = "sha256:2fa77c6fd8940f116ee1d6b94a2f90b13b5ea8d019b98bc8bafdcabcdd9bdbed", size = 60722, upload-time = "2023-12-24T09:54:32.31Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/12/b3/231ffd4ab1fc9d679809f356cebee130ac7daa00d6d6f3206dd4fd137e9e/distro-1.9.0-py3-none-any.whl", hash = "sha256:7bffd925d65168f85027d8da9af6bddab658135b840670a223589bc0c8ef02b2", size = 20277, upload-time = "2023-12-24T09:54:30.421Z" },
+]
+
+[[package]]
+name = "django"
+version = "5.2.7"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "asgiref" },
+    { name = "sqlparse" },
+    { name = "tzdata", marker = "sys_platform == 'win32'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b1/96/bd84e2bb997994de8bcda47ae4560991084e86536541d7214393880f01a8/django-5.2.7.tar.gz", hash = "sha256:e0f6f12e2551b1716a95a63a1366ca91bbcd7be059862c1b18f989b1da356cdd", size = 10865812, upload-time = "2025-10-01T14:22:12.081Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/8f/ef/81f3372b5dd35d8d354321155d1a38894b2b766f576d0abffac4d8ae78d9/django-5.2.7-py3-none-any.whl", hash = "sha256:59a13a6515f787dec9d97a0438cd2efac78c8aca1c80025244b0fe507fe0754b", size = 8307145, upload-time = "2025-10-01T14:22:49.476Z" },
+]
+
+[[package]]
+name = "h11"
+version = "0.16.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/01/ee/02a2c011bdab74c6fb3c75474d40b3052059d95df7e73351460c8588d963/h11-0.16.0.tar.gz", hash = "sha256:4e35b956cf45792e4caa5885e69fba00bdbc6ffafbfa020300e549b208ee5ff1", size = 101250, upload-time = "2025-04-24T03:35:25.427Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl", hash = "sha256:63cf8bbe7522de3bf65932fda1d9c2772064ffb3dae62d55932da54b31cb6c86", size = 37515, upload-time = "2025-04-24T03:35:24.344Z" },
+]
+
+[[package]]
+name = "httpcore"
+version = "1.0.9"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "certifi" },
+    { name = "h11" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/06/94/82699a10bca87a5556c9c59b5963f2d039dbd239f25bc2a63907a05a14cb/httpcore-1.0.9.tar.gz", hash = "sha256:6e34463af53fd2ab5d807f399a9b45ea31c3dfa2276f15a2c3f00afff6e176e8", size = 85484, upload-time = "2025-04-24T22:06:22.219Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl", hash = "sha256:2d400746a40668fc9dec9810239072b40b4484b640a8c38fd654a024c7a1bf55", size = 78784, upload-time = "2025-04-24T22:06:20.566Z" },
+]
+
+[[package]]
+name = "httptools"
+version = "0.7.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/b5/46/120a669232c7bdedb9d52d4aeae7e6c7dfe151e99dc70802e2fc7a5e1993/httptools-0.7.1.tar.gz", hash = "sha256:abd72556974f8e7c74a259655924a717a2365b236c882c3f6f8a45fe94703ac9", size = 258961, upload-time = "2025-10-10T03:55:08.559Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/53/7f/403e5d787dc4942316e515e949b0c8a013d84078a915910e9f391ba9b3ed/httptools-0.7.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:38e0c83a2ea9746ebbd643bdfb521b9aa4a91703e2cd705c20443405d2fd16a5", size = 206280, upload-time = "2025-10-10T03:54:39.274Z" },
+    { url = "https://files.pythonhosted.org/packages/2a/0d/7f3fd28e2ce311ccc998c388dd1c53b18120fda3b70ebb022b135dc9839b/httptools-0.7.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:f25bbaf1235e27704f1a7b86cd3304eabc04f569c828101d94a0e605ef7205a5", size = 110004, upload-time = "2025-10-10T03:54:40.403Z" },
+    { url = "https://files.pythonhosted.org/packages/84/a6/b3965e1e146ef5762870bbe76117876ceba51a201e18cc31f5703e454596/httptools-0.7.1-cp312-cp312-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:2c15f37ef679ab9ecc06bfc4e6e8628c32a8e4b305459de7cf6785acd57e4d03", size = 517655, upload-time = "2025-10-10T03:54:41.347Z" },
+    { url = "https://files.pythonhosted.org/packages/11/7d/71fee6f1844e6fa378f2eddde6c3e41ce3a1fb4b2d81118dd544e3441ec0/httptools-0.7.1-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:7fe6e96090df46b36ccfaf746f03034e5ab723162bc51b0a4cf58305324036f2", size = 511440, upload-time = "2025-10-10T03:54:42.452Z" },
+    { url = "https://files.pythonhosted.org/packages/22/a5/079d216712a4f3ffa24af4a0381b108aa9c45b7a5cc6eb141f81726b1823/httptools-0.7.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:f72fdbae2dbc6e68b8239defb48e6a5937b12218e6ffc2c7846cc37befa84362", size = 495186, upload-time = "2025-10-10T03:54:43.937Z" },
+    { url = "https://files.pythonhosted.org/packages/e9/9e/025ad7b65278745dee3bd0ebf9314934c4592560878308a6121f7f812084/httptools-0.7.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e99c7b90a29fd82fea9ef57943d501a16f3404d7b9ee81799d41639bdaae412c", size = 499192, upload-time = "2025-10-10T03:54:45.003Z" },
+    { url = "https://files.pythonhosted.org/packages/6d/de/40a8f202b987d43afc4d54689600ff03ce65680ede2f31df348d7f368b8f/httptools-0.7.1-cp312-cp312-win_amd64.whl", hash = "sha256:3e14f530fefa7499334a79b0cf7e7cd2992870eb893526fb097d51b4f2d0f321", size = 86694, upload-time = "2025-10-10T03:54:45.923Z" },
+    { url = "https://files.pythonhosted.org/packages/09/8f/c77b1fcbfd262d422f12da02feb0d218fa228d52485b77b953832105bb90/httptools-0.7.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:6babce6cfa2a99545c60bfef8bee0cc0545413cb0018f617c8059a30ad985de3", size = 202889, upload-time = "2025-10-10T03:54:47.089Z" },
+    { url = "https://files.pythonhosted.org/packages/0a/1a/22887f53602feaa066354867bc49a68fc295c2293433177ee90870a7d517/httptools-0.7.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:601b7628de7504077dd3dcb3791c6b8694bbd967148a6d1f01806509254fb1ca", size = 108180, upload-time = "2025-10-10T03:54:48.052Z" },
+    { url = "https://files.pythonhosted.org/packages/32/6a/6aaa91937f0010d288d3d124ca2946d48d60c3a5ee7ca62afe870e3ea011/httptools-0.7.1-cp313-cp313-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:04c6c0e6c5fb0739c5b8a9eb046d298650a0ff38cf42537fc372b28dc7e4472c", size = 478596, upload-time = "2025-10-10T03:54:48.919Z" },
+    { url = "https://files.pythonhosted.org/packages/6d/70/023d7ce117993107be88d2cbca566a7c1323ccbaf0af7eabf2064fe356f6/httptools-0.7.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:69d4f9705c405ae3ee83d6a12283dc9feba8cc6aaec671b412917e644ab4fa66", size = 473268, upload-time = "2025-10-10T03:54:49.993Z" },
+    { url = "https://files.pythonhosted.org/packages/32/4d/9dd616c38da088e3f436e9a616e1d0cc66544b8cdac405cc4e81c8679fc7/httptools-0.7.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:44c8f4347d4b31269c8a9205d8a5ee2df5322b09bbbd30f8f862185bb6b05346", size = 455517, upload-time = "2025-10-10T03:54:51.066Z" },
+    { url = "https://files.pythonhosted.org/packages/1d/3a/a6c595c310b7df958e739aae88724e24f9246a514d909547778d776799be/httptools-0.7.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:465275d76db4d554918aba40bf1cbebe324670f3dfc979eaffaa5d108e2ed650", size = 458337, upload-time = "2025-10-10T03:54:52.196Z" },
+    { url = "https://files.pythonhosted.org/packages/fd/82/88e8d6d2c51edc1cc391b6e044c6c435b6aebe97b1abc33db1b0b24cd582/httptools-0.7.1-cp313-cp313-win_amd64.whl", hash = "sha256:322d00c2068d125bd570f7bf78b2d367dad02b919d8581d7476d8b75b294e3e6", size = 85743, upload-time = "2025-10-10T03:54:53.448Z" },
+    { url = "https://files.pythonhosted.org/packages/34/50/9d095fcbb6de2d523e027a2f304d4551855c2f46e0b82befd718b8b20056/httptools-0.7.1-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:c08fe65728b8d70b6923ce31e3956f859d5e1e8548e6f22ec520a962c6757270", size = 203619, upload-time = "2025-10-10T03:54:54.321Z" },
+    { url = "https://files.pythonhosted.org/packages/07/f0/89720dc5139ae54b03f861b5e2c55a37dba9a5da7d51e1e824a1f343627f/httptools-0.7.1-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:7aea2e3c3953521c3c51106ee11487a910d45586e351202474d45472db7d72d3", size = 108714, upload-time = "2025-10-10T03:54:55.163Z" },
+    { url = "https://files.pythonhosted.org/packages/b3/cb/eea88506f191fb552c11787c23f9a405f4c7b0c5799bf73f2249cd4f5228/httptools-0.7.1-cp314-cp314-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:0e68b8582f4ea9166be62926077a3334064d422cf08ab87d8b74664f8e9058e1", size = 472909, upload-time = "2025-10-10T03:54:56.056Z" },
+    { url = "https://files.pythonhosted.org/packages/e0/4a/a548bdfae6369c0d078bab5769f7b66f17f1bfaa6fa28f81d6be6959066b/httptools-0.7.1-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:df091cf961a3be783d6aebae963cc9b71e00d57fa6f149025075217bc6a55a7b", size = 470831, upload-time = "2025-10-10T03:54:57.219Z" },
+    { url = "https://files.pythonhosted.org/packages/4d/31/14df99e1c43bd132eec921c2e7e11cda7852f65619bc0fc5bdc2d0cb126c/httptools-0.7.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:f084813239e1eb403ddacd06a30de3d3e09a9b76e7894dcda2b22f8a726e9c60", size = 452631, upload-time = "2025-10-10T03:54:58.219Z" },
+    { url = "https://files.pythonhosted.org/packages/22/d2/b7e131f7be8d854d48cb6d048113c30f9a46dca0c9a8b08fcb3fcd588cdc/httptools-0.7.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:7347714368fb2b335e9063bc2b96f2f87a9ceffcd9758ac295f8bbcd3ffbc0ca", size = 452910, upload-time = "2025-10-10T03:54:59.366Z" },
+    { url = "https://files.pythonhosted.org/packages/53/cf/878f3b91e4e6e011eff6d1fa9ca39f7eb17d19c9d7971b04873734112f30/httptools-0.7.1-cp314-cp314-win_amd64.whl", hash = "sha256:cfabda2a5bb85aa2a904ce06d974a3f30fb36cc63d7feaddec05d2050acede96", size = 88205, upload-time = "2025-10-10T03:55:00.389Z" },
+]
+
+[[package]]
+name = "httpx"
+version = "0.28.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "anyio" },
+    { name = "certifi" },
+    { name = "httpcore" },
+    { name = "idna" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b1/df/48c586a5fe32a0f01324ee087459e112ebb7224f646c0b5023f5e79e9956/httpx-0.28.1.tar.gz", hash = "sha256:75e98c5f16b0f35b567856f597f06ff2270a374470a5c2392242528e3e3e42fc", size = 141406, upload-time = "2024-12-06T15:37:23.222Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl", hash = "sha256:d909fcccc110f8c7faf814ca82a9a4d816bc5a6dbfea25d6591d6985b8ba59ad", size = 73517, upload-time = "2024-12-06T15:37:21.509Z" },
+]
+
+[[package]]
+name = "idna"
+version = "3.11"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/6f/6d/0703ccc57f3a7233505399edb88de3cbd678da106337b9fcde432b65ed60/idna-3.11.tar.gz", hash = "sha256:795dafcc9c04ed0c1fb032c2aa73654d8e8c5023a7df64a53f39190ada629902", size = 194582, upload-time = "2025-10-12T14:55:20.501Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/0e/61/66938bbb5fc52dbdf84594873d5b51fb1f7c7794e9c0f5bd885f30bc507b/idna-3.11-py3-none-any.whl", hash = "sha256:771a87f49d9defaf64091e6e6fe9c18d4833f140bd19464795bc32d966ca37ea", size = 71008, upload-time = "2025-10-12T14:55:18.883Z" },
+]
+
+[[package]]
+name = "iniconfig"
+version = "2.3.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/72/34/14ca021ce8e5dfedc35312d08ba8bf51fdd999c576889fc2c24cb97f4f10/iniconfig-2.3.0.tar.gz", hash = "sha256:c76315c77db068650d49c5b56314774a7804df16fee4402c1f19d6d15d8c4730", size = 20503, upload-time = "2025-10-18T21:55:43.219Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/cb/b1/3846dd7f199d53cb17f49cba7e651e9ce294d8497c8c150530ed11865bb8/iniconfig-2.3.0-py3-none-any.whl", hash = "sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12", size = 7484, upload-time = "2025-10-18T21:55:41.639Z" },
+]
+
+[[package]]
+name = "packaging"
+version = "25.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/a1/d4/1fc4078c65507b51b96ca8f8c3ba19e6a61c8253c72794544580a7b6c24d/packaging-25.0.tar.gz", hash = "sha256:d443872c98d677bf60f6a1f2f8c1cb748e8fe762d2bf9d3148b5599295b0fc4f", size = 165727, upload-time = "2025-04-19T11:48:59.673Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/20/12/38679034af332785aac8774540895e234f4d07f7545804097de4b666afd8/packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484", size = 66469, upload-time = "2025-04-19T11:48:57.875Z" },
+]
+
+[[package]]
+name = "pluggy"
+version = "1.6.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" },
+]
+
+[[package]]
+name = "posthog"
+source = { editable = "../" }
+dependencies = [
+    { name = "backoff" },
+    { name = "distro" },
+    { name = "python-dateutil" },
+    { name = "requests" },
+    { name = "six" },
+    { name = "typing-extensions" },
+]
+
+[package.metadata]
+requires-dist = [
+    { name = "anthropic", marker = "extra == 'test'" },
+    { name = "backoff", specifier = ">=1.10.0" },
+    { name = "coverage", marker = "extra == 'test'" },
+    { name = "distro", specifier = ">=1.5.0" },
+    { name = "django", marker = "extra == 'test'" },
+    { name = "django-stubs", marker = "extra == 'dev'" },
+    { name = "freezegun", marker = "extra == 'test'", specifier = "==1.5.1" },
+    { name = "google-genai", marker = "extra == 'test'" },
+    { name = "langchain", marker = "extra == 'langchain'", specifier = ">=0.2.0" },
+    { name = "langchain-anthropic", marker = "extra == 'test'", specifier = ">=0.3.15" },
+    { name = "langchain-community", marker = "extra == 'test'", specifier = ">=0.3.25" },
+    { name = "langchain-core", marker = "extra == 'test'", specifier = ">=0.3.65" },
+    { name = "langchain-openai", marker = "extra == 'test'", specifier = ">=0.3.22" },
+    { name = "langgraph", marker = "extra == 'test'", specifier = ">=0.4.8" },
+    { name = "lxml", marker = "extra == 'dev'" },
+    { name = "mock", marker = "extra == 'test'", specifier = ">=2.0.0" },
+    { name = "mypy", marker = "extra == 'dev'" },
+    { name = "mypy-baseline", marker = "extra == 'dev'" },
+    { name = "openai", marker = "extra == 'test'" },
+    { name = "packaging", marker = "extra == 'dev'" },
+    { name = "parameterized", marker = "extra == 'test'", specifier = ">=0.8.1" },
+    { name = "pre-commit", marker = "extra == 'dev'" },
+    { name = "pydantic", marker = "extra == 'dev'" },
+    { name = "pydantic", marker = "extra == 'test'" },
+    { name = "pytest", marker = "extra == 'test'" },
+    { name = "pytest-asyncio", marker = "extra == 'test'" },
+    { name = "pytest-timeout", marker = "extra == 'test'" },
+    { name = "python-dateutil", specifier = ">=2.2" },
+    { name = "requests", specifier = ">=2.7,<3.0" },
+    { name = "ruff", marker = "extra == 'dev'" },
+    { name = "setuptools", marker = "extra == 'dev'" },
+    { name = "six", specifier = ">=1.5" },
+    { name = "tomli", marker = "extra == 'dev'" },
+    { name = "tomli-w", marker = "extra == 'dev'" },
+    { name = "twine", marker = "extra == 'dev'" },
+    { name = "types-mock", marker = "extra == 'dev'" },
+    { name = "types-python-dateutil", marker = "extra == 'dev'" },
+    { name = "types-requests", marker = "extra == 'dev'" },
+    { name = "types-setuptools", marker = "extra == 'dev'" },
+    { name = "types-six", marker = "extra == 'dev'" },
+    { name = "typing-extensions", specifier = ">=4.2.0" },
+    { name = "wheel", marker = "extra == 'dev'" },
+]
+provides-extras = ["langchain", "dev", "test"]
+
+[[package]]
+name = "pygments"
+version = "2.19.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/b0/77/a5b8c569bf593b0140bde72ea885a803b82086995367bf2037de0159d924/pygments-2.19.2.tar.gz", hash = "sha256:636cb2477cec7f8952536970bc533bc43743542f70392ae026374600add5b887", size = 4968631, upload-time = "2025-06-21T13:39:12.283Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c7/21/705964c7812476f378728bdf590ca4b771ec72385c533964653c68e86bdc/pygments-2.19.2-py3-none-any.whl", hash = "sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b", size = 1225217, upload-time = "2025-06-21T13:39:07.939Z" },
+]
+
+[[package]]
+name = "pytest"
+version = "8.4.2"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "iniconfig" },
+    { name = "packaging" },
+    { name = "pluggy" },
+    { name = "pygments" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/a3/5c/00a0e072241553e1a7496d638deababa67c5058571567b92a7eaa258397c/pytest-8.4.2.tar.gz", hash = "sha256:86c0d0b93306b961d58d62a4db4879f27fe25513d4b969df351abdddb3c30e01", size = 1519618, upload-time = "2025-09-04T14:34:22.711Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a8/a4/20da314d277121d6534b3a980b29035dcd51e6744bd79075a6ce8fa4eb8d/pytest-8.4.2-py3-none-any.whl", hash = "sha256:872f880de3fc3a5bdc88a11b39c9710c3497a547cfa9320bc3c5e62fbf272e79", size = 365750, upload-time = "2025-09-04T14:34:20.226Z" },
+]
+
+[[package]]
+name = "pytest-asyncio"
+version = "1.2.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "pytest" },
+    { name = "typing-extensions", marker = "python_full_version < '3.13'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/42/86/9e3c5f48f7b7b638b216e4b9e645f54d199d7abbbab7a64a13b4e12ba10f/pytest_asyncio-1.2.0.tar.gz", hash = "sha256:c609a64a2a8768462d0c99811ddb8bd2583c33fd33cf7f21af1c142e824ffb57", size = 50119, upload-time = "2025-09-12T07:33:53.816Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/04/93/2fa34714b7a4ae72f2f8dad66ba17dd9a2c793220719e736dda28b7aec27/pytest_asyncio-1.2.0-py3-none-any.whl", hash = "sha256:8e17ae5e46d8e7efe51ab6494dd2010f4ca8dae51652aa3c8d55acf50bfb2e99", size = 15095, upload-time = "2025-09-12T07:33:52.639Z" },
+]
+
+[[package]]
+name = "pytest-django"
+version = "4.11.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "pytest" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b1/fb/55d580352db26eb3d59ad50c64321ddfe228d3d8ac107db05387a2fadf3a/pytest_django-4.11.1.tar.gz", hash = "sha256:a949141a1ee103cb0e7a20f1451d355f83f5e4a5d07bdd4dcfdd1fd0ff227991", size = 86202, upload-time = "2025-04-03T18:56:09.338Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/be/ac/bd0608d229ec808e51a21044f3f2f27b9a37e7a0ebaca7247882e67876af/pytest_django-4.11.1-py3-none-any.whl", hash = "sha256:1b63773f648aa3d8541000c26929c1ea63934be1cfa674c76436966d73fe6a10", size = 25281, upload-time = "2025-04-03T18:56:07.678Z" },
+]
+
+[[package]]
+name = "python-dateutil"
+version = "2.9.0.post0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "six" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/66/c0/0c8b6ad9f17a802ee498c46e004a0eb49bc148f2fd230864601a86dcf6db/python-dateutil-2.9.0.post0.tar.gz", hash = "sha256:37dd54208da7e1cd875388217d5e00ebd4179249f90fb72437e91a35459a0ad3", size = 342432, upload-time = "2024-03-01T18:36:20.211Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ec/57/56b9bcc3c9c6a792fcbaf139543cee77261f3651ca9da0c93f5c1221264b/python_dateutil-2.9.0.post0-py2.py3-none-any.whl", hash = "sha256:a8b2bc7bffae282281c8140a97d3aa9c14da0b136dfe83f850eea9a5f7470427", size = 229892, upload-time = "2024-03-01T18:36:18.57Z" },
+]
+
+[[package]]
+name = "python-dotenv"
+version = "1.2.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f0/26/19cadc79a718c5edbec86fd4919a6b6d3f681039a2f6d66d14be94e75fb9/python_dotenv-1.2.1.tar.gz", hash = "sha256:42667e897e16ab0d66954af0e60a9caa94f0fd4ecf3aaf6d2d260eec1aa36ad6", size = 44221, upload-time = "2025-10-26T15:12:10.434Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/14/1b/a298b06749107c305e1fe0f814c6c74aea7b2f1e10989cb30f544a1b3253/python_dotenv-1.2.1-py3-none-any.whl", hash = "sha256:b81ee9561e9ca4004139c6cbba3a238c32b03e4894671e181b671e8cb8425d61", size = 21230, upload-time = "2025-10-26T15:12:09.109Z" },
+]
+
+[[package]]
+name = "pyyaml"
+version = "6.0.3"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/05/8e/961c0007c59b8dd7729d542c61a4d537767a59645b82a0b521206e1e25c2/pyyaml-6.0.3.tar.gz", hash = "sha256:d76623373421df22fb4cf8817020cbb7ef15c725b9d5e45f17e189bfc384190f", size = 130960, upload-time = "2025-09-25T21:33:16.546Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/d1/33/422b98d2195232ca1826284a76852ad5a86fe23e31b009c9886b2d0fb8b2/pyyaml-6.0.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7f047e29dcae44602496db43be01ad42fc6f1cc0d8cd6c83d342306c32270196", size = 182063, upload-time = "2025-09-25T21:32:11.445Z" },
+    { url = "https://files.pythonhosted.org/packages/89/a0/6cf41a19a1f2f3feab0e9c0b74134aa2ce6849093d5517a0c550fe37a648/pyyaml-6.0.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fc09d0aa354569bc501d4e787133afc08552722d3ab34836a80547331bb5d4a0", size = 173973, upload-time = "2025-09-25T21:32:12.492Z" },
+    { url = "https://files.pythonhosted.org/packages/ed/23/7a778b6bd0b9a8039df8b1b1d80e2e2ad78aa04171592c8a5c43a56a6af4/pyyaml-6.0.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9149cad251584d5fb4981be1ecde53a1ca46c891a79788c0df828d2f166bda28", size = 775116, upload-time = "2025-09-25T21:32:13.652Z" },
+    { url = "https://files.pythonhosted.org/packages/65/30/d7353c338e12baef4ecc1b09e877c1970bd3382789c159b4f89d6a70dc09/pyyaml-6.0.3-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:5fdec68f91a0c6739b380c83b951e2c72ac0197ace422360e6d5a959d8d97b2c", size = 844011, upload-time = "2025-09-25T21:32:15.21Z" },
+    { url = "https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ba1cc08a7ccde2d2ec775841541641e4548226580ab850948cbfda66a1befcdc", size = 807870, upload-time = "2025-09-25T21:32:16.431Z" },
+    { url = "https://files.pythonhosted.org/packages/05/c0/b3be26a015601b822b97d9149ff8cb5ead58c66f981e04fedf4e762f4bd4/pyyaml-6.0.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:8dc52c23056b9ddd46818a57b78404882310fb473d63f17b07d5c40421e47f8e", size = 761089, upload-time = "2025-09-25T21:32:17.56Z" },
+    { url = "https://files.pythonhosted.org/packages/be/8e/98435a21d1d4b46590d5459a22d88128103f8da4c2d4cb8f14f2a96504e1/pyyaml-6.0.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:41715c910c881bc081f1e8872880d3c650acf13dfa8214bad49ed4cede7c34ea", size = 790181, upload-time = "2025-09-25T21:32:18.834Z" },
+    { url = "https://files.pythonhosted.org/packages/74/93/7baea19427dcfbe1e5a372d81473250b379f04b1bd3c4c5ff825e2327202/pyyaml-6.0.3-cp312-cp312-win32.whl", hash = "sha256:96b533f0e99f6579b3d4d4995707cf36df9100d67e0c8303a0c55b27b5f99bc5", size = 137658, upload-time = "2025-09-25T21:32:20.209Z" },
+    { url = "https://files.pythonhosted.org/packages/86/bf/899e81e4cce32febab4fb42bb97dcdf66bc135272882d1987881a4b519e9/pyyaml-6.0.3-cp312-cp312-win_amd64.whl", hash = "sha256:5fcd34e47f6e0b794d17de1b4ff496c00986e1c83f7ab2fb8fcfe9616ff7477b", size = 154003, upload-time = "2025-09-25T21:32:21.167Z" },
+    { url = "https://files.pythonhosted.org/packages/1a/08/67bd04656199bbb51dbed1439b7f27601dfb576fb864099c7ef0c3e55531/pyyaml-6.0.3-cp312-cp312-win_arm64.whl", hash = "sha256:64386e5e707d03a7e172c0701abfb7e10f0fb753ee1d773128192742712a98fd", size = 140344, upload-time = "2025-09-25T21:32:22.617Z" },
+    { url = "https://files.pythonhosted.org/packages/d1/11/0fd08f8192109f7169db964b5707a2f1e8b745d4e239b784a5a1dd80d1db/pyyaml-6.0.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:8da9669d359f02c0b91ccc01cac4a67f16afec0dac22c2ad09f46bee0697eba8", size = 181669, upload-time = "2025-09-25T21:32:23.673Z" },
+    { url = "https://files.pythonhosted.org/packages/b1/16/95309993f1d3748cd644e02e38b75d50cbc0d9561d21f390a76242ce073f/pyyaml-6.0.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:2283a07e2c21a2aa78d9c4442724ec1eb15f5e42a723b99cb3d822d48f5f7ad1", size = 173252, upload-time = "2025-09-25T21:32:25.149Z" },
+    { url = "https://files.pythonhosted.org/packages/50/31/b20f376d3f810b9b2371e72ef5adb33879b25edb7a6d072cb7ca0c486398/pyyaml-6.0.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ee2922902c45ae8ccada2c5b501ab86c36525b883eff4255313a253a3160861c", size = 767081, upload-time = "2025-09-25T21:32:26.575Z" },
+    { url = "https://files.pythonhosted.org/packages/49/1e/a55ca81e949270d5d4432fbbd19dfea5321eda7c41a849d443dc92fd1ff7/pyyaml-6.0.3-cp313-cp313-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:a33284e20b78bd4a18c8c2282d549d10bc8408a2a7ff57653c0cf0b9be0afce5", size = 841159, upload-time = "2025-09-25T21:32:27.727Z" },
+    { url = "https://files.pythonhosted.org/packages/74/27/e5b8f34d02d9995b80abcef563ea1f8b56d20134d8f4e5e81733b1feceb2/pyyaml-6.0.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0f29edc409a6392443abf94b9cf89ce99889a1dd5376d94316ae5145dfedd5d6", size = 801626, upload-time = "2025-09-25T21:32:28.878Z" },
+    { url = "https://files.pythonhosted.org/packages/f9/11/ba845c23988798f40e52ba45f34849aa8a1f2d4af4b798588010792ebad6/pyyaml-6.0.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:f7057c9a337546edc7973c0d3ba84ddcdf0daa14533c2065749c9075001090e6", size = 753613, upload-time = "2025-09-25T21:32:30.178Z" },
+    { url = "https://files.pythonhosted.org/packages/3d/e0/7966e1a7bfc0a45bf0a7fb6b98ea03fc9b8d84fa7f2229e9659680b69ee3/pyyaml-6.0.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:eda16858a3cab07b80edaf74336ece1f986ba330fdb8ee0d6c0d68fe82bc96be", size = 794115, upload-time = "2025-09-25T21:32:31.353Z" },
+    { url = "https://files.pythonhosted.org/packages/de/94/980b50a6531b3019e45ddeada0626d45fa85cbe22300844a7983285bed3b/pyyaml-6.0.3-cp313-cp313-win32.whl", hash = "sha256:d0eae10f8159e8fdad514efdc92d74fd8d682c933a6dd088030f3834bc8e6b26", size = 137427, upload-time = "2025-09-25T21:32:32.58Z" },
+    { url = "https://files.pythonhosted.org/packages/97/c9/39d5b874e8b28845e4ec2202b5da735d0199dbe5b8fb85f91398814a9a46/pyyaml-6.0.3-cp313-cp313-win_amd64.whl", hash = "sha256:79005a0d97d5ddabfeeea4cf676af11e647e41d81c9a7722a193022accdb6b7c", size = 154090, upload-time = "2025-09-25T21:32:33.659Z" },
+    { url = "https://files.pythonhosted.org/packages/73/e8/2bdf3ca2090f68bb3d75b44da7bbc71843b19c9f2b9cb9b0f4ab7a5a4329/pyyaml-6.0.3-cp313-cp313-win_arm64.whl", hash = "sha256:5498cd1645aa724a7c71c8f378eb29ebe23da2fc0d7a08071d89469bf1d2defb", size = 140246, upload-time = "2025-09-25T21:32:34.663Z" },
+    { url = "https://files.pythonhosted.org/packages/9d/8c/f4bd7f6465179953d3ac9bc44ac1a8a3e6122cf8ada906b4f96c60172d43/pyyaml-6.0.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:8d1fab6bb153a416f9aeb4b8763bc0f22a5586065f86f7664fc23339fc1c1fac", size = 181814, upload-time = "2025-09-25T21:32:35.712Z" },
+    { url = "https://files.pythonhosted.org/packages/bd/9c/4d95bb87eb2063d20db7b60faa3840c1b18025517ae857371c4dd55a6b3a/pyyaml-6.0.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:34d5fcd24b8445fadc33f9cf348c1047101756fd760b4dacb5c3e99755703310", size = 173809, upload-time = "2025-09-25T21:32:36.789Z" },
+    { url = "https://files.pythonhosted.org/packages/92/b5/47e807c2623074914e29dabd16cbbdd4bf5e9b2db9f8090fa64411fc5382/pyyaml-6.0.3-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:501a031947e3a9025ed4405a168e6ef5ae3126c59f90ce0cd6f2bfc477be31b7", size = 766454, upload-time = "2025-09-25T21:32:37.966Z" },
+    { url = "https://files.pythonhosted.org/packages/02/9e/e5e9b168be58564121efb3de6859c452fccde0ab093d8438905899a3a483/pyyaml-6.0.3-cp314-cp314-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:b3bc83488de33889877a0f2543ade9f70c67d66d9ebb4ac959502e12de895788", size = 836355, upload-time = "2025-09-25T21:32:39.178Z" },
+    { url = "https://files.pythonhosted.org/packages/88/f9/16491d7ed2a919954993e48aa941b200f38040928474c9e85ea9e64222c3/pyyaml-6.0.3-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c458b6d084f9b935061bc36216e8a69a7e293a2f1e68bf956dcd9e6cbcd143f5", size = 794175, upload-time = "2025-09-25T21:32:40.865Z" },
+    { url = "https://files.pythonhosted.org/packages/dd/3f/5989debef34dc6397317802b527dbbafb2b4760878a53d4166579111411e/pyyaml-6.0.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:7c6610def4f163542a622a73fb39f534f8c101d690126992300bf3207eab9764", size = 755228, upload-time = "2025-09-25T21:32:42.084Z" },
+    { url = "https://files.pythonhosted.org/packages/d7/ce/af88a49043cd2e265be63d083fc75b27b6ed062f5f9fd6cdc223ad62f03e/pyyaml-6.0.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:5190d403f121660ce8d1d2c1bb2ef1bd05b5f68533fc5c2ea899bd15f4399b35", size = 789194, upload-time = "2025-09-25T21:32:43.362Z" },
+    { url = "https://files.pythonhosted.org/packages/23/20/bb6982b26a40bb43951265ba29d4c246ef0ff59c9fdcdf0ed04e0687de4d/pyyaml-6.0.3-cp314-cp314-win_amd64.whl", hash = "sha256:4a2e8cebe2ff6ab7d1050ecd59c25d4c8bd7e6f400f5f82b96557ac0abafd0ac", size = 156429, upload-time = "2025-09-25T21:32:57.844Z" },
+    { url = "https://files.pythonhosted.org/packages/f4/f4/a4541072bb9422c8a883ab55255f918fa378ecf083f5b85e87fc2b4eda1b/pyyaml-6.0.3-cp314-cp314-win_arm64.whl", hash = "sha256:93dda82c9c22deb0a405ea4dc5f2d0cda384168e466364dec6255b293923b2f3", size = 143912, upload-time = "2025-09-25T21:32:59.247Z" },
+    { url = "https://files.pythonhosted.org/packages/7c/f9/07dd09ae774e4616edf6cda684ee78f97777bdd15847253637a6f052a62f/pyyaml-6.0.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:02893d100e99e03eda1c8fd5c441d8c60103fd175728e23e431db1b589cf5ab3", size = 189108, upload-time = "2025-09-25T21:32:44.377Z" },
+    { url = "https://files.pythonhosted.org/packages/4e/78/8d08c9fb7ce09ad8c38ad533c1191cf27f7ae1effe5bb9400a46d9437fcf/pyyaml-6.0.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:c1ff362665ae507275af2853520967820d9124984e0f7466736aea23d8611fba", size = 183641, upload-time = "2025-09-25T21:32:45.407Z" },
+    { url = "https://files.pythonhosted.org/packages/7b/5b/3babb19104a46945cf816d047db2788bcaf8c94527a805610b0289a01c6b/pyyaml-6.0.3-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6adc77889b628398debc7b65c073bcb99c4a0237b248cacaf3fe8a557563ef6c", size = 831901, upload-time = "2025-09-25T21:32:48.83Z" },
+    { url = "https://files.pythonhosted.org/packages/8b/cc/dff0684d8dc44da4d22a13f35f073d558c268780ce3c6ba1b87055bb0b87/pyyaml-6.0.3-cp314-cp314t-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:a80cb027f6b349846a3bf6d73b5e95e782175e52f22108cfa17876aaeff93702", size = 861132, upload-time = "2025-09-25T21:32:50.149Z" },
+    { url = "https://files.pythonhosted.org/packages/b1/5e/f77dc6b9036943e285ba76b49e118d9ea929885becb0a29ba8a7c75e29fe/pyyaml-6.0.3-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:00c4bdeba853cc34e7dd471f16b4114f4162dc03e6b7afcc2128711f0eca823c", size = 839261, upload-time = "2025-09-25T21:32:51.808Z" },
+    { url = "https://files.pythonhosted.org/packages/ce/88/a9db1376aa2a228197c58b37302f284b5617f56a5d959fd1763fb1675ce6/pyyaml-6.0.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:66e1674c3ef6f541c35191caae2d429b967b99e02040f5ba928632d9a7f0f065", size = 805272, upload-time = "2025-09-25T21:32:52.941Z" },
+    { url = "https://files.pythonhosted.org/packages/da/92/1446574745d74df0c92e6aa4a7b0b3130706a4142b2d1a5869f2eaa423c6/pyyaml-6.0.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:16249ee61e95f858e83976573de0f5b2893b3677ba71c9dd36b9cf8be9ac6d65", size = 829923, upload-time = "2025-09-25T21:32:54.537Z" },
+    { url = "https://files.pythonhosted.org/packages/f0/7a/1c7270340330e575b92f397352af856a8c06f230aa3e76f86b39d01b416a/pyyaml-6.0.3-cp314-cp314t-win_amd64.whl", hash = "sha256:4ad1906908f2f5ae4e5a8ddfce73c320c2a1429ec52eafd27138b7f1cbe341c9", size = 174062, upload-time = "2025-09-25T21:32:55.767Z" },
+    { url = "https://files.pythonhosted.org/packages/f1/12/de94a39c2ef588c7e6455cfbe7343d3b2dc9d6b6b2f40c4c6565744c873d/pyyaml-6.0.3-cp314-cp314t-win_arm64.whl", hash = "sha256:ebc55a14a21cb14062aa4162f906cd962b28e2e9ea38f9b4391244cd8de4ae0b", size = 149341, upload-time = "2025-09-25T21:32:56.828Z" },
+]
+
+[[package]]
+name = "requests"
+version = "2.32.5"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "certifi" },
+    { name = "charset-normalizer" },
+    { name = "idna" },
+    { name = "urllib3" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/c9/74/b3ff8e6c8446842c3f5c837e9c3dfcfe2018ea6ecef224c710c85ef728f4/requests-2.32.5.tar.gz", hash = "sha256:dbba0bac56e100853db0ea71b82b4dfd5fe2bf6d3754a8893c3af500cec7d7cf", size = 134517, upload-time = "2025-08-18T20:46:02.573Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/1e/db/4254e3eabe8020b458f1a747140d32277ec7a271daf1d235b70dc0b4e6e3/requests-2.32.5-py3-none-any.whl", hash = "sha256:2462f94637a34fd532264295e186976db0f5d453d1cdd31473c85a6a161affb6", size = 64738, upload-time = "2025-08-18T20:46:00.542Z" },
+]
+
+[[package]]
+name = "six"
+version = "1.17.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/94/e7/b2c673351809dca68a0e064b6af791aa332cf192da575fd474ed7d6f16a2/six-1.17.0.tar.gz", hash = "sha256:ff70335d468e7eb6ec65b95b99d3a2836546063f63acc5171de367e834932a81", size = 34031, upload-time = "2024-12-04T17:35:28.174Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b7/ce/149a00dd41f10bc29e5921b496af8b574d8413afcd5e30dfa0ed46c2cc5e/six-1.17.0-py2.py3-none-any.whl", hash = "sha256:4721f391ed90541fddacab5acf947aa0d3dc7d27b2e1e8eda2be8970586c3274", size = 11050, upload-time = "2024-12-04T17:35:26.475Z" },
+]
+
+[[package]]
+name = "sniffio"
+version = "1.3.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/a2/87/a6771e1546d97e7e041b6ae58d80074f81b7d5121207425c964ddf5cfdbd/sniffio-1.3.1.tar.gz", hash = "sha256:f4324edc670a0f49750a81b895f35c3adb843cca46f0530f79fc1babb23789dc", size = 20372, upload-time = "2024-02-25T23:20:04.057Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/e9/44/75a9c9421471a6c4805dbf2356f7c181a29c1879239abab1ea2cc8f38b40/sniffio-1.3.1-py3-none-any.whl", hash = "sha256:2f6da418d1f1e0fddd844478f41680e794e6051915791a034ff65e5f100525a2", size = 10235, upload-time = "2024-02-25T23:20:01.196Z" },
+]
+
+[[package]]
+name = "sqlparse"
+version = "0.5.3"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/e5/40/edede8dd6977b0d3da179a342c198ed100dd2aba4be081861ee5911e4da4/sqlparse-0.5.3.tar.gz", hash = "sha256:09f67787f56a0b16ecdbde1bfc7f5d9c3371ca683cfeaa8e6ff60b4807ec9272", size = 84999, upload-time = "2024-12-10T12:05:30.728Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a9/5c/bfd6bd0bf979426d405cc6e71eceb8701b148b16c21d2dc3c261efc61c7b/sqlparse-0.5.3-py3-none-any.whl", hash = "sha256:cf2196ed3418f3ba5de6af7e82c694a9fbdbfecccdfc72e281548517081f16ca", size = 44415, upload-time = "2024-12-10T12:05:27.824Z" },
+]
+
+[[package]]
+name = "test-django5"
+version = "0.1.0"
+source = { virtual = "." }
+dependencies = [
+    { name = "django" },
+    { name = "httpx" },
+    { name = "posthog" },
+    { name = "pytest" },
+    { name = "pytest-asyncio" },
+    { name = "pytest-django" },
+    { name = "uvicorn", extra = ["standard"] },
+]
+
+[package.metadata]
+requires-dist = [
+    { name = "django", specifier = "~=5.2.7" },
+    { name = "httpx", specifier = "~=0.28.1" },
+    { name = "posthog", editable = "../" },
+    { name = "pytest", specifier = "~=8.4.2" },
+    { name = "pytest-asyncio", specifier = "~=1.2.0" },
+    { name = "pytest-django", specifier = "~=4.11.1" },
+    { name = "uvicorn", extras = ["standard"], specifier = "~=0.38.0" },
+]
+
+[[package]]
+name = "typing-extensions"
+version = "4.15.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/72/94/1a15dd82efb362ac84269196e94cf00f187f7ed21c242792a923cdb1c61f/typing_extensions-4.15.0.tar.gz", hash = "sha256:0cea48d173cc12fa28ecabc3b837ea3cf6f38c6d1136f85cbaaf598984861466", size = 109391, upload-time = "2025-08-25T13:49:26.313Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl", hash = "sha256:f0fa19c6845758ab08074a0cfa8b7aecb71c999ca73d62883bc25cc018c4e548", size = 44614, upload-time = "2025-08-25T13:49:24.86Z" },
+]
+
+[[package]]
+name = "tzdata"
+version = "2025.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/95/32/1a225d6164441be760d75c2c42e2780dc0873fe382da3e98a2e1e48361e5/tzdata-2025.2.tar.gz", hash = "sha256:b60a638fcc0daffadf82fe0f57e53d06bdec2f36c4df66280ae79bce6bd6f2b9", size = 196380, upload-time = "2025-03-23T13:54:43.652Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/5c/23/c7abc0ca0a1526a0774eca151daeb8de62ec457e77262b66b359c3c7679e/tzdata-2025.2-py2.py3-none-any.whl", hash = "sha256:1a403fada01ff9221ca8044d701868fa132215d84beb92242d9acd2147f667a8", size = 347839, upload-time = "2025-03-23T13:54:41.845Z" },
+]
+
+[[package]]
+name = "urllib3"
+version = "2.5.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/15/22/9ee70a2574a4f4599c47dd506532914ce044817c7752a79b6a51286319bc/urllib3-2.5.0.tar.gz", hash = "sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760", size = 393185, upload-time = "2025-06-18T14:07:41.644Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a7/c2/fe1e52489ae3122415c51f387e221dd0773709bad6c6cdaa599e8a2c5185/urllib3-2.5.0-py3-none-any.whl", hash = "sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc", size = 129795, upload-time = "2025-06-18T14:07:40.39Z" },
+]
+
+[[package]]
+name = "uvicorn"
+version = "0.38.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "click" },
+    { name = "h11" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/cb/ce/f06b84e2697fef4688ca63bdb2fdf113ca0a3be33f94488f2cadb690b0cf/uvicorn-0.38.0.tar.gz", hash = "sha256:fd97093bdd120a2609fc0d3afe931d4d4ad688b6e75f0f929fde1bc36fe0e91d", size = 80605, upload-time = "2025-10-18T13:46:44.63Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ee/d9/d88e73ca598f4f6ff671fb5fde8a32925c2e08a637303a1d12883c7305fa/uvicorn-0.38.0-py3-none-any.whl", hash = "sha256:48c0afd214ceb59340075b4a052ea1ee91c16fbc2a9b1469cca0e54566977b02", size = 68109, upload-time = "2025-10-18T13:46:42.958Z" },
+]
+
+[package.optional-dependencies]
+standard = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "httptools" },
+    { name = "python-dotenv" },
+    { name = "pyyaml" },
+    { name = "uvloop", marker = "platform_python_implementation != 'PyPy' and sys_platform != 'cygwin' and sys_platform != 'win32'" },
+    { name = "watchfiles" },
+    { name = "websockets" },
+]
+
+[[package]]
+name = "uvloop"
+version = "0.22.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/06/f0/18d39dbd1971d6d62c4629cc7fa67f74821b0dc1f5a77af43719de7936a7/uvloop-0.22.1.tar.gz", hash = "sha256:6c84bae345b9147082b17371e3dd5d42775bddce91f885499017f4607fdaf39f", size = 2443250, upload-time = "2025-10-16T22:17:19.342Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/3d/ff/7f72e8170be527b4977b033239a83a68d5c881cc4775fca255c677f7ac5d/uvloop-0.22.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:fe94b4564e865d968414598eea1a6de60adba0c040ba4ed05ac1300de402cd42", size = 1359936, upload-time = "2025-10-16T22:16:29.436Z" },
+    { url = "https://files.pythonhosted.org/packages/c3/c6/e5d433f88fd54d81ef4be58b2b7b0cea13c442454a1db703a1eea0db1a59/uvloop-0.22.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:51eb9bd88391483410daad430813d982010f9c9c89512321f5b60e2cddbdddd6", size = 752769, upload-time = "2025-10-16T22:16:30.493Z" },
+    { url = "https://files.pythonhosted.org/packages/24/68/a6ac446820273e71aa762fa21cdcc09861edd3536ff47c5cd3b7afb10eeb/uvloop-0.22.1-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:700e674a166ca5778255e0e1dc4e9d79ab2acc57b9171b79e65feba7184b3370", size = 4317413, upload-time = "2025-10-16T22:16:31.644Z" },
+    { url = "https://files.pythonhosted.org/packages/5f/6f/e62b4dfc7ad6518e7eff2516f680d02a0f6eb62c0c212e152ca708a0085e/uvloop-0.22.1-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7b5b1ac819a3f946d3b2ee07f09149578ae76066d70b44df3fa990add49a82e4", size = 4426307, upload-time = "2025-10-16T22:16:32.917Z" },
+    { url = "https://files.pythonhosted.org/packages/90/60/97362554ac21e20e81bcef1150cb2a7e4ffdaf8ea1e5b2e8bf7a053caa18/uvloop-0.22.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:e047cc068570bac9866237739607d1313b9253c3051ad84738cbb095be0537b2", size = 4131970, upload-time = "2025-10-16T22:16:34.015Z" },
+    { url = "https://files.pythonhosted.org/packages/99/39/6b3f7d234ba3964c428a6e40006340f53ba37993f46ed6e111c6e9141d18/uvloop-0.22.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:512fec6815e2dd45161054592441ef76c830eddaad55c8aa30952e6fe1ed07c0", size = 4296343, upload-time = "2025-10-16T22:16:35.149Z" },
+    { url = "https://files.pythonhosted.org/packages/89/8c/182a2a593195bfd39842ea68ebc084e20c850806117213f5a299dfc513d9/uvloop-0.22.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:561577354eb94200d75aca23fbde86ee11be36b00e52a4eaf8f50fb0c86b7705", size = 1358611, upload-time = "2025-10-16T22:16:36.833Z" },
+    { url = "https://files.pythonhosted.org/packages/d2/14/e301ee96a6dc95224b6f1162cd3312f6d1217be3907b79173b06785f2fe7/uvloop-0.22.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:1cdf5192ab3e674ca26da2eada35b288d2fa49fdd0f357a19f0e7c4e7d5077c8", size = 751811, upload-time = "2025-10-16T22:16:38.275Z" },
+    { url = "https://files.pythonhosted.org/packages/b7/02/654426ce265ac19e2980bfd9ea6590ca96a56f10c76e63801a2df01c0486/uvloop-0.22.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6e2ea3d6190a2968f4a14a23019d3b16870dd2190cd69c8180f7c632d21de68d", size = 4288562, upload-time = "2025-10-16T22:16:39.375Z" },
+    { url = "https://files.pythonhosted.org/packages/15/c0/0be24758891ef825f2065cd5db8741aaddabe3e248ee6acc5e8a80f04005/uvloop-0.22.1-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0530a5fbad9c9e4ee3f2b33b148c6a64d47bbad8000ea63704fa8260f4cf728e", size = 4366890, upload-time = "2025-10-16T22:16:40.547Z" },
+    { url = "https://files.pythonhosted.org/packages/d2/53/8369e5219a5855869bcee5f4d317f6da0e2c669aecf0ef7d371e3d084449/uvloop-0.22.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:bc5ef13bbc10b5335792360623cc378d52d7e62c2de64660616478c32cd0598e", size = 4119472, upload-time = "2025-10-16T22:16:41.694Z" },
+    { url = "https://files.pythonhosted.org/packages/f8/ba/d69adbe699b768f6b29a5eec7b47dd610bd17a69de51b251126a801369ea/uvloop-0.22.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:1f38ec5e3f18c8a10ded09742f7fb8de0108796eb673f30ce7762ce1b8550cad", size = 4239051, upload-time = "2025-10-16T22:16:43.224Z" },
+    { url = "https://files.pythonhosted.org/packages/90/cd/b62bdeaa429758aee8de8b00ac0dd26593a9de93d302bff3d21439e9791d/uvloop-0.22.1-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:3879b88423ec7e97cd4eba2a443aa26ed4e59b45e6b76aabf13fe2f27023a142", size = 1362067, upload-time = "2025-10-16T22:16:44.503Z" },
+    { url = "https://files.pythonhosted.org/packages/0d/f8/a132124dfda0777e489ca86732e85e69afcd1ff7686647000050ba670689/uvloop-0.22.1-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:4baa86acedf1d62115c1dc6ad1e17134476688f08c6efd8a2ab076e815665c74", size = 752423, upload-time = "2025-10-16T22:16:45.968Z" },
+    { url = "https://files.pythonhosted.org/packages/a3/94/94af78c156f88da4b3a733773ad5ba0b164393e357cc4bd0ab2e2677a7d6/uvloop-0.22.1-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:297c27d8003520596236bdb2335e6b3f649480bd09e00d1e3a99144b691d2a35", size = 4272437, upload-time = "2025-10-16T22:16:47.451Z" },
+    { url = "https://files.pythonhosted.org/packages/b5/35/60249e9fd07b32c665192cec7af29e06c7cd96fa1d08b84f012a56a0b38e/uvloop-0.22.1-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c1955d5a1dd43198244d47664a5858082a3239766a839b2102a269aaff7a4e25", size = 4292101, upload-time = "2025-10-16T22:16:49.318Z" },
+    { url = "https://files.pythonhosted.org/packages/02/62/67d382dfcb25d0a98ce73c11ed1a6fba5037a1a1d533dcbb7cab033a2636/uvloop-0.22.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:b31dc2fccbd42adc73bc4e7cdbae4fc5086cf378979e53ca5d0301838c5682c6", size = 4114158, upload-time = "2025-10-16T22:16:50.517Z" },
+    { url = "https://files.pythonhosted.org/packages/f0/7a/f1171b4a882a5d13c8b7576f348acfe6074d72eaf52cccef752f748d4a9f/uvloop-0.22.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:93f617675b2d03af4e72a5333ef89450dfaa5321303ede6e67ba9c9d26878079", size = 4177360, upload-time = "2025-10-16T22:16:52.646Z" },
+    { url = "https://files.pythonhosted.org/packages/79/7b/b01414f31546caf0919da80ad57cbfe24c56b151d12af68cee1b04922ca8/uvloop-0.22.1-cp314-cp314t-macosx_10_13_universal2.whl", hash = "sha256:37554f70528f60cad66945b885eb01f1bb514f132d92b6eeed1c90fd54ed6289", size = 1454790, upload-time = "2025-10-16T22:16:54.355Z" },
+    { url = "https://files.pythonhosted.org/packages/d4/31/0bb232318dd838cad3fa8fb0c68c8b40e1145b32025581975e18b11fab40/uvloop-0.22.1-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:b76324e2dc033a0b2f435f33eb88ff9913c156ef78e153fb210e03c13da746b3", size = 796783, upload-time = "2025-10-16T22:16:55.906Z" },
+    { url = "https://files.pythonhosted.org/packages/42/38/c9b09f3271a7a723a5de69f8e237ab8e7803183131bc57c890db0b6bb872/uvloop-0.22.1-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:badb4d8e58ee08dad957002027830d5c3b06aea446a6a3744483c2b3b745345c", size = 4647548, upload-time = "2025-10-16T22:16:57.008Z" },
+    { url = "https://files.pythonhosted.org/packages/c1/37/945b4ca0ac27e3dc4952642d4c900edd030b3da6c9634875af6e13ae80e5/uvloop-0.22.1-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b91328c72635f6f9e0282e4a57da7470c7350ab1c9f48546c0f2866205349d21", size = 4467065, upload-time = "2025-10-16T22:16:58.206Z" },
+    { url = "https://files.pythonhosted.org/packages/97/cc/48d232f33d60e2e2e0b42f4e73455b146b76ebe216487e862700457fbf3c/uvloop-0.22.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:daf620c2995d193449393d6c62131b3fbd40a63bf7b307a1527856ace637fe88", size = 4328384, upload-time = "2025-10-16T22:16:59.36Z" },
+    { url = "https://files.pythonhosted.org/packages/e4/16/c1fd27e9549f3c4baf1dc9c20c456cd2f822dbf8de9f463824b0c0357e06/uvloop-0.22.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:6cde23eeda1a25c75b2e07d39970f3374105d5eafbaab2a4482be82f272d5a5e", size = 4296730, upload-time = "2025-10-16T22:17:00.744Z" },
+]
+
+[[package]]
+name = "watchfiles"
+version = "1.1.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "anyio" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/c2/c9/8869df9b2a2d6c59d79220a4db37679e74f807c559ffe5265e08b227a210/watchfiles-1.1.1.tar.gz", hash = "sha256:a173cb5c16c4f40ab19cecf48a534c409f7ea983ab8fed0741304a1c0a31b3f2", size = 94440, upload-time = "2025-10-14T15:06:21.08Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/74/d5/f039e7e3c639d9b1d09b07ea412a6806d38123f0508e5f9b48a87b0a76cc/watchfiles-1.1.1-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:8c89f9f2f740a6b7dcc753140dd5e1ab9215966f7a3530d0c0705c83b401bd7d", size = 404745, upload-time = "2025-10-14T15:04:46.731Z" },
+    { url = "https://files.pythonhosted.org/packages/a5/96/a881a13aa1349827490dab2d363c8039527060cfcc2c92cc6d13d1b1049e/watchfiles-1.1.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:bd404be08018c37350f0d6e34676bd1e2889990117a2b90070b3007f172d0610", size = 391769, upload-time = "2025-10-14T15:04:48.003Z" },
+    { url = "https://files.pythonhosted.org/packages/4b/5b/d3b460364aeb8da471c1989238ea0e56bec24b6042a68046adf3d9ddb01c/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8526e8f916bb5b9a0a777c8317c23ce65de259422bba5b31325a6fa6029d33af", size = 449374, upload-time = "2025-10-14T15:04:49.179Z" },
+    { url = "https://files.pythonhosted.org/packages/b9/44/5769cb62d4ed055cb17417c0a109a92f007114a4e07f30812a73a4efdb11/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:2edc3553362b1c38d9f06242416a5d8e9fe235c204a4072e988ce2e5bb1f69f6", size = 459485, upload-time = "2025-10-14T15:04:50.155Z" },
+    { url = "https://files.pythonhosted.org/packages/19/0c/286b6301ded2eccd4ffd0041a1b726afda999926cf720aab63adb68a1e36/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:30f7da3fb3f2844259cba4720c3fc7138eb0f7b659c38f3bfa65084c7fc7abce", size = 488813, upload-time = "2025-10-14T15:04:51.059Z" },
+    { url = "https://files.pythonhosted.org/packages/c7/2b/8530ed41112dd4a22f4dcfdb5ccf6a1baad1ff6eed8dc5a5f09e7e8c41c7/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f8979280bdafff686ba5e4d8f97840f929a87ed9cdf133cbbd42f7766774d2aa", size = 594816, upload-time = "2025-10-14T15:04:52.031Z" },
+    { url = "https://files.pythonhosted.org/packages/ce/d2/f5f9fb49489f184f18470d4f99f4e862a4b3e9ac2865688eb2099e3d837a/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:dcc5c24523771db3a294c77d94771abcfcb82a0e0ee8efd910c37c59ec1b31bb", size = 475186, upload-time = "2025-10-14T15:04:53.064Z" },
+    { url = "https://files.pythonhosted.org/packages/cf/68/5707da262a119fb06fbe214d82dd1fe4a6f4af32d2d14de368d0349eb52a/watchfiles-1.1.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1db5d7ae38ff20153d542460752ff397fcf5c96090c1230803713cf3147a6803", size = 456812, upload-time = "2025-10-14T15:04:55.174Z" },
+    { url = "https://files.pythonhosted.org/packages/66/ab/3cbb8756323e8f9b6f9acb9ef4ec26d42b2109bce830cc1f3468df20511d/watchfiles-1.1.1-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:28475ddbde92df1874b6c5c8aaeb24ad5be47a11f87cde5a28ef3835932e3e94", size = 630196, upload-time = "2025-10-14T15:04:56.22Z" },
+    { url = "https://files.pythonhosted.org/packages/78/46/7152ec29b8335f80167928944a94955015a345440f524d2dfe63fc2f437b/watchfiles-1.1.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:36193ed342f5b9842edd3532729a2ad55c4160ffcfa3700e0d54be496b70dd43", size = 622657, upload-time = "2025-10-14T15:04:57.521Z" },
+    { url = "https://files.pythonhosted.org/packages/0a/bf/95895e78dd75efe9a7f31733607f384b42eb5feb54bd2eb6ed57cc2e94f4/watchfiles-1.1.1-cp312-cp312-win32.whl", hash = "sha256:859e43a1951717cc8de7f4c77674a6d389b106361585951d9e69572823f311d9", size = 272042, upload-time = "2025-10-14T15:04:59.046Z" },
+    { url = "https://files.pythonhosted.org/packages/87/0a/90eb755f568de2688cb220171c4191df932232c20946966c27a59c400850/watchfiles-1.1.1-cp312-cp312-win_amd64.whl", hash = "sha256:91d4c9a823a8c987cce8fa2690923b069966dabb196dd8d137ea2cede885fde9", size = 288410, upload-time = "2025-10-14T15:05:00.081Z" },
+    { url = "https://files.pythonhosted.org/packages/36/76/f322701530586922fbd6723c4f91ace21364924822a8772c549483abed13/watchfiles-1.1.1-cp312-cp312-win_arm64.whl", hash = "sha256:a625815d4a2bdca61953dbba5a39d60164451ef34c88d751f6c368c3ea73d404", size = 278209, upload-time = "2025-10-14T15:05:01.168Z" },
+    { url = "https://files.pythonhosted.org/packages/bb/f4/f750b29225fe77139f7ae5de89d4949f5a99f934c65a1f1c0b248f26f747/watchfiles-1.1.1-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:130e4876309e8686a5e37dba7d5e9bc77e6ed908266996ca26572437a5271e18", size = 404321, upload-time = "2025-10-14T15:05:02.063Z" },
+    { url = "https://files.pythonhosted.org/packages/2b/f9/f07a295cde762644aa4c4bb0f88921d2d141af45e735b965fb2e87858328/watchfiles-1.1.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:5f3bde70f157f84ece3765b42b4a52c6ac1a50334903c6eaf765362f6ccca88a", size = 391783, upload-time = "2025-10-14T15:05:03.052Z" },
+    { url = "https://files.pythonhosted.org/packages/bc/11/fc2502457e0bea39a5c958d86d2cb69e407a4d00b85735ca724bfa6e0d1a/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:14e0b1fe858430fc0251737ef3824c54027bedb8c37c38114488b8e131cf8219", size = 449279, upload-time = "2025-10-14T15:05:04.004Z" },
+    { url = "https://files.pythonhosted.org/packages/e3/1f/d66bc15ea0b728df3ed96a539c777acfcad0eb78555ad9efcaa1274688f0/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f27db948078f3823a6bb3b465180db8ebecf26dd5dae6f6180bd87383b6b4428", size = 459405, upload-time = "2025-10-14T15:05:04.942Z" },
+    { url = "https://files.pythonhosted.org/packages/be/90/9f4a65c0aec3ccf032703e6db02d89a157462fbb2cf20dd415128251cac0/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:059098c3a429f62fc98e8ec62b982230ef2c8df68c79e826e37b895bc359a9c0", size = 488976, upload-time = "2025-10-14T15:05:05.905Z" },
+    { url = "https://files.pythonhosted.org/packages/37/57/ee347af605d867f712be7029bb94c8c071732a4b44792e3176fa3c612d39/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bfb5862016acc9b869bb57284e6cb35fdf8e22fe59f7548858e2f971d045f150", size = 595506, upload-time = "2025-10-14T15:05:06.906Z" },
+    { url = "https://files.pythonhosted.org/packages/a8/78/cc5ab0b86c122047f75e8fc471c67a04dee395daf847d3e59381996c8707/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:319b27255aacd9923b8a276bb14d21a5f7ff82564c744235fc5eae58d95422ae", size = 474936, upload-time = "2025-10-14T15:05:07.906Z" },
+    { url = "https://files.pythonhosted.org/packages/62/da/def65b170a3815af7bd40a3e7010bf6ab53089ef1b75d05dd5385b87cf08/watchfiles-1.1.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c755367e51db90e75b19454b680903631d41f9e3607fbd941d296a020c2d752d", size = 456147, upload-time = "2025-10-14T15:05:09.138Z" },
+    { url = "https://files.pythonhosted.org/packages/57/99/da6573ba71166e82d288d4df0839128004c67d2778d3b566c138695f5c0b/watchfiles-1.1.1-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:c22c776292a23bfc7237a98f791b9ad3144b02116ff10d820829ce62dff46d0b", size = 630007, upload-time = "2025-10-14T15:05:10.117Z" },
+    { url = "https://files.pythonhosted.org/packages/a8/51/7439c4dd39511368849eb1e53279cd3454b4a4dbace80bab88feeb83c6b5/watchfiles-1.1.1-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:3a476189be23c3686bc2f4321dd501cb329c0a0469e77b7b534ee10129ae6374", size = 622280, upload-time = "2025-10-14T15:05:11.146Z" },
+    { url = "https://files.pythonhosted.org/packages/95/9c/8ed97d4bba5db6fdcdb2b298d3898f2dd5c20f6b73aee04eabe56c59677e/watchfiles-1.1.1-cp313-cp313-win32.whl", hash = "sha256:bf0a91bfb5574a2f7fc223cf95eeea79abfefa404bf1ea5e339c0c1560ae99a0", size = 272056, upload-time = "2025-10-14T15:05:12.156Z" },
+    { url = "https://files.pythonhosted.org/packages/1f/f3/c14e28429f744a260d8ceae18bf58c1d5fa56b50d006a7a9f80e1882cb0d/watchfiles-1.1.1-cp313-cp313-win_amd64.whl", hash = "sha256:52e06553899e11e8074503c8e716d574adeeb7e68913115c4b3653c53f9bae42", size = 288162, upload-time = "2025-10-14T15:05:13.208Z" },
+    { url = "https://files.pythonhosted.org/packages/dc/61/fe0e56c40d5cd29523e398d31153218718c5786b5e636d9ae8ae79453d27/watchfiles-1.1.1-cp313-cp313-win_arm64.whl", hash = "sha256:ac3cc5759570cd02662b15fbcd9d917f7ecd47efe0d6b40474eafd246f91ea18", size = 277909, upload-time = "2025-10-14T15:05:14.49Z" },
+    { url = "https://files.pythonhosted.org/packages/79/42/e0a7d749626f1e28c7108a99fb9bf524b501bbbeb9b261ceecde644d5a07/watchfiles-1.1.1-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:563b116874a9a7ce6f96f87cd0b94f7faf92d08d0021e837796f0a14318ef8da", size = 403389, upload-time = "2025-10-14T15:05:15.777Z" },
+    { url = "https://files.pythonhosted.org/packages/15/49/08732f90ce0fbbc13913f9f215c689cfc9ced345fb1bcd8829a50007cc8d/watchfiles-1.1.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:3ad9fe1dae4ab4212d8c91e80b832425e24f421703b5a42ef2e4a1e215aff051", size = 389964, upload-time = "2025-10-14T15:05:16.85Z" },
+    { url = "https://files.pythonhosted.org/packages/27/0d/7c315d4bd5f2538910491a0393c56bf70d333d51bc5b34bee8e68e8cea19/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ce70f96a46b894b36eba678f153f052967a0d06d5b5a19b336ab0dbbd029f73e", size = 448114, upload-time = "2025-10-14T15:05:17.876Z" },
+    { url = "https://files.pythonhosted.org/packages/c3/24/9e096de47a4d11bc4df41e9d1e61776393eac4cb6eb11b3e23315b78b2cc/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:cb467c999c2eff23a6417e58d75e5828716f42ed8289fe6b77a7e5a91036ca70", size = 460264, upload-time = "2025-10-14T15:05:18.962Z" },
+    { url = "https://files.pythonhosted.org/packages/cc/0f/e8dea6375f1d3ba5fcb0b3583e2b493e77379834c74fd5a22d66d85d6540/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:836398932192dae4146c8f6f737d74baeac8b70ce14831a239bdb1ca882fc261", size = 487877, upload-time = "2025-10-14T15:05:20.094Z" },
+    { url = "https://files.pythonhosted.org/packages/ac/5b/df24cfc6424a12deb41503b64d42fbea6b8cb357ec62ca84a5a3476f654a/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:743185e7372b7bc7c389e1badcc606931a827112fbbd37f14c537320fca08620", size = 595176, upload-time = "2025-10-14T15:05:21.134Z" },
+    { url = "https://files.pythonhosted.org/packages/8f/b5/853b6757f7347de4e9b37e8cc3289283fb983cba1ab4d2d7144694871d9c/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:afaeff7696e0ad9f02cbb8f56365ff4686ab205fcf9c4c5b6fdfaaa16549dd04", size = 473577, upload-time = "2025-10-14T15:05:22.306Z" },
+    { url = "https://files.pythonhosted.org/packages/e1/f7/0a4467be0a56e80447c8529c9fce5b38eab4f513cb3d9bf82e7392a5696b/watchfiles-1.1.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3f7eb7da0eb23aa2ba036d4f616d46906013a68caf61b7fdbe42fc8b25132e77", size = 455425, upload-time = "2025-10-14T15:05:23.348Z" },
+    { url = "https://files.pythonhosted.org/packages/8e/e0/82583485ea00137ddf69bc84a2db88bd92ab4a6e3c405e5fb878ead8d0e7/watchfiles-1.1.1-cp313-cp313t-musllinux_1_1_aarch64.whl", hash = "sha256:831a62658609f0e5c64178211c942ace999517f5770fe9436be4c2faeba0c0ef", size = 628826, upload-time = "2025-10-14T15:05:24.398Z" },
+    { url = "https://files.pythonhosted.org/packages/28/9a/a785356fccf9fae84c0cc90570f11702ae9571036fb25932f1242c82191c/watchfiles-1.1.1-cp313-cp313t-musllinux_1_1_x86_64.whl", hash = "sha256:f9a2ae5c91cecc9edd47e041a930490c31c3afb1f5e6d71de3dc671bfaca02bf", size = 622208, upload-time = "2025-10-14T15:05:25.45Z" },
+    { url = "https://files.pythonhosted.org/packages/c3/f4/0872229324ef69b2c3edec35e84bd57a1289e7d3fe74588048ed8947a323/watchfiles-1.1.1-cp314-cp314-macosx_10_12_x86_64.whl", hash = "sha256:d1715143123baeeaeadec0528bb7441103979a1d5f6fd0e1f915383fea7ea6d5", size = 404315, upload-time = "2025-10-14T15:05:26.501Z" },
+    { url = "https://files.pythonhosted.org/packages/7b/22/16d5331eaed1cb107b873f6ae1b69e9ced582fcf0c59a50cd84f403b1c32/watchfiles-1.1.1-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:39574d6370c4579d7f5d0ad940ce5b20db0e4117444e39b6d8f99db5676c52fd", size = 390869, upload-time = "2025-10-14T15:05:27.649Z" },
+    { url = "https://files.pythonhosted.org/packages/b2/7e/5643bfff5acb6539b18483128fdc0ef2cccc94a5b8fbda130c823e8ed636/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7365b92c2e69ee952902e8f70f3ba6360d0d596d9299d55d7d386df84b6941fb", size = 449919, upload-time = "2025-10-14T15:05:28.701Z" },
+    { url = "https://files.pythonhosted.org/packages/51/2e/c410993ba5025a9f9357c376f48976ef0e1b1aefb73b97a5ae01a5972755/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:bfff9740c69c0e4ed32416f013f3c45e2ae42ccedd1167ef2d805c000b6c71a5", size = 460845, upload-time = "2025-10-14T15:05:30.064Z" },
+    { url = "https://files.pythonhosted.org/packages/8e/a4/2df3b404469122e8680f0fcd06079317e48db58a2da2950fb45020947734/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b27cf2eb1dda37b2089e3907d8ea92922b673c0c427886d4edc6b94d8dfe5db3", size = 489027, upload-time = "2025-10-14T15:05:31.064Z" },
+    { url = "https://files.pythonhosted.org/packages/ea/84/4587ba5b1f267167ee715b7f66e6382cca6938e0a4b870adad93e44747e6/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:526e86aced14a65a5b0ec50827c745597c782ff46b571dbfe46192ab9e0b3c33", size = 595615, upload-time = "2025-10-14T15:05:32.074Z" },
+    { url = "https://files.pythonhosted.org/packages/6a/0f/c6988c91d06e93cd0bb3d4a808bcf32375ca1904609835c3031799e3ecae/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:04e78dd0b6352db95507fd8cb46f39d185cf8c74e4cf1e4fbad1d3df96faf510", size = 474836, upload-time = "2025-10-14T15:05:33.209Z" },
+    { url = "https://files.pythonhosted.org/packages/b4/36/ded8aebea91919485b7bbabbd14f5f359326cb5ec218cd67074d1e426d74/watchfiles-1.1.1-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5c85794a4cfa094714fb9c08d4a218375b2b95b8ed1666e8677c349906246c05", size = 455099, upload-time = "2025-10-14T15:05:34.189Z" },
+    { url = "https://files.pythonhosted.org/packages/98/e0/8c9bdba88af756a2fce230dd365fab2baf927ba42cd47521ee7498fd5211/watchfiles-1.1.1-cp314-cp314-musllinux_1_1_aarch64.whl", hash = "sha256:74d5012b7630714b66be7b7b7a78855ef7ad58e8650c73afc4c076a1f480a8d6", size = 630626, upload-time = "2025-10-14T15:05:35.216Z" },
+    { url = "https://files.pythonhosted.org/packages/2a/84/a95db05354bf2d19e438520d92a8ca475e578c647f78f53197f5a2f17aaf/watchfiles-1.1.1-cp314-cp314-musllinux_1_1_x86_64.whl", hash = "sha256:8fbe85cb3201c7d380d3d0b90e63d520f15d6afe217165d7f98c9c649654db81", size = 622519, upload-time = "2025-10-14T15:05:36.259Z" },
+    { url = "https://files.pythonhosted.org/packages/1d/ce/d8acdc8de545de995c339be67711e474c77d643555a9bb74a9334252bd55/watchfiles-1.1.1-cp314-cp314-win32.whl", hash = "sha256:3fa0b59c92278b5a7800d3ee7733da9d096d4aabcfabb9a928918bd276ef9b9b", size = 272078, upload-time = "2025-10-14T15:05:37.63Z" },
+    { url = "https://files.pythonhosted.org/packages/c4/c9/a74487f72d0451524be827e8edec251da0cc1fcf111646a511ae752e1a3d/watchfiles-1.1.1-cp314-cp314-win_amd64.whl", hash = "sha256:c2047d0b6cea13b3316bdbafbfa0c4228ae593d995030fda39089d36e64fc03a", size = 287664, upload-time = "2025-10-14T15:05:38.95Z" },
+    { url = "https://files.pythonhosted.org/packages/df/b8/8ac000702cdd496cdce998c6f4ee0ca1f15977bba51bdf07d872ebdfc34c/watchfiles-1.1.1-cp314-cp314-win_arm64.whl", hash = "sha256:842178b126593addc05acf6fce960d28bc5fae7afbaa2c6c1b3a7b9460e5be02", size = 277154, upload-time = "2025-10-14T15:05:39.954Z" },
+    { url = "https://files.pythonhosted.org/packages/47/a8/e3af2184707c29f0f14b1963c0aace6529f9d1b8582d5b99f31bbf42f59e/watchfiles-1.1.1-cp314-cp314t-macosx_10_12_x86_64.whl", hash = "sha256:88863fbbc1a7312972f1c511f202eb30866370ebb8493aef2812b9ff28156a21", size = 403820, upload-time = "2025-10-14T15:05:40.932Z" },
+    { url = "https://files.pythonhosted.org/packages/c0/ec/e47e307c2f4bd75f9f9e8afbe3876679b18e1bcec449beca132a1c5ffb2d/watchfiles-1.1.1-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:55c7475190662e202c08c6c0f4d9e345a29367438cf8e8037f3155e10a88d5a5", size = 390510, upload-time = "2025-10-14T15:05:41.945Z" },
+    { url = "https://files.pythonhosted.org/packages/d5/a0/ad235642118090f66e7b2f18fd5c42082418404a79205cdfca50b6309c13/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3f53fa183d53a1d7a8852277c92b967ae99c2d4dcee2bfacff8868e6e30b15f7", size = 448408, upload-time = "2025-10-14T15:05:43.385Z" },
+    { url = "https://files.pythonhosted.org/packages/df/85/97fa10fd5ff3332ae17e7e40e20784e419e28521549780869f1413742e9d/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:6aae418a8b323732fa89721d86f39ec8f092fc2af67f4217a2b07fd3e93c6101", size = 458968, upload-time = "2025-10-14T15:05:44.404Z" },
+    { url = "https://files.pythonhosted.org/packages/47/c2/9059c2e8966ea5ce678166617a7f75ecba6164375f3b288e50a40dc6d489/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f096076119da54a6080e8920cbdaac3dbee667eb91dcc5e5b78840b87415bd44", size = 488096, upload-time = "2025-10-14T15:05:45.398Z" },
+    { url = "https://files.pythonhosted.org/packages/94/44/d90a9ec8ac309bc26db808a13e7bfc0e4e78b6fc051078a554e132e80160/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:00485f441d183717038ed2e887a7c868154f216877653121068107b227a2f64c", size = 596040, upload-time = "2025-10-14T15:05:46.502Z" },
+    { url = "https://files.pythonhosted.org/packages/95/68/4e3479b20ca305cfc561db3ed207a8a1c745ee32bf24f2026a129d0ddb6e/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:a55f3e9e493158d7bfdb60a1165035f1cf7d320914e7b7ea83fe22c6023b58fc", size = 473847, upload-time = "2025-10-14T15:05:47.484Z" },
+    { url = "https://files.pythonhosted.org/packages/4f/55/2af26693fd15165c4ff7857e38330e1b61ab8c37d15dc79118cdba115b7a/watchfiles-1.1.1-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8c91ed27800188c2ae96d16e3149f199d62f86c7af5f5f4d2c61a3ed8cd3666c", size = 455072, upload-time = "2025-10-14T15:05:48.928Z" },
+    { url = "https://files.pythonhosted.org/packages/66/1d/d0d200b10c9311ec25d2273f8aad8c3ef7cc7ea11808022501811208a750/watchfiles-1.1.1-cp314-cp314t-musllinux_1_1_aarch64.whl", hash = "sha256:311ff15a0bae3714ffb603e6ba6dbfba4065ab60865d15a6ec544133bdb21099", size = 629104, upload-time = "2025-10-14T15:05:49.908Z" },
+    { url = "https://files.pythonhosted.org/packages/e3/bd/fa9bb053192491b3867ba07d2343d9f2252e00811567d30ae8d0f78136fe/watchfiles-1.1.1-cp314-cp314t-musllinux_1_1_x86_64.whl", hash = "sha256:a916a2932da8f8ab582f242c065f5c81bed3462849ca79ee357dd9551b0e9b01", size = 622112, upload-time = "2025-10-14T15:05:50.941Z" },
+]
+
+[[package]]
+name = "websockets"
+version = "15.0.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/21/e6/26d09fab466b7ca9c7737474c52be4f76a40301b08362eb2dbc19dcc16c1/websockets-15.0.1.tar.gz", hash = "sha256:82544de02076bafba038ce055ee6412d68da13ab47f0c60cab827346de828dee", size = 177016, upload-time = "2025-03-05T20:03:41.606Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/51/6b/4545a0d843594f5d0771e86463606a3988b5a09ca5123136f8a76580dd63/websockets-15.0.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:3e90baa811a5d73f3ca0bcbf32064d663ed81318ab225ee4f427ad4e26e5aff3", size = 175437, upload-time = "2025-03-05T20:02:16.706Z" },
+    { url = "https://files.pythonhosted.org/packages/f4/71/809a0f5f6a06522af902e0f2ea2757f71ead94610010cf570ab5c98e99ed/websockets-15.0.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:592f1a9fe869c778694f0aa806ba0374e97648ab57936f092fd9d87f8bc03665", size = 173096, upload-time = "2025-03-05T20:02:18.832Z" },
+    { url = "https://files.pythonhosted.org/packages/3d/69/1a681dd6f02180916f116894181eab8b2e25b31e484c5d0eae637ec01f7c/websockets-15.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:0701bc3cfcb9164d04a14b149fd74be7347a530ad3bbf15ab2c678a2cd3dd9a2", size = 173332, upload-time = "2025-03-05T20:02:20.187Z" },
+    { url = "https://files.pythonhosted.org/packages/a6/02/0073b3952f5bce97eafbb35757f8d0d54812b6174ed8dd952aa08429bcc3/websockets-15.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e8b56bdcdb4505c8078cb6c7157d9811a85790f2f2b3632c7d1462ab5783d215", size = 183152, upload-time = "2025-03-05T20:02:22.286Z" },
+    { url = "https://files.pythonhosted.org/packages/74/45/c205c8480eafd114b428284840da0b1be9ffd0e4f87338dc95dc6ff961a1/websockets-15.0.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0af68c55afbd5f07986df82831c7bff04846928ea8d1fd7f30052638788bc9b5", size = 182096, upload-time = "2025-03-05T20:02:24.368Z" },
+    { url = "https://files.pythonhosted.org/packages/14/8f/aa61f528fba38578ec553c145857a181384c72b98156f858ca5c8e82d9d3/websockets-15.0.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:64dee438fed052b52e4f98f76c5790513235efaa1ef7f3f2192c392cd7c91b65", size = 182523, upload-time = "2025-03-05T20:02:25.669Z" },
+    { url = "https://files.pythonhosted.org/packages/ec/6d/0267396610add5bc0d0d3e77f546d4cd287200804fe02323797de77dbce9/websockets-15.0.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:d5f6b181bb38171a8ad1d6aa58a67a6aa9d4b38d0f8c5f496b9e42561dfc62fe", size = 182790, upload-time = "2025-03-05T20:02:26.99Z" },
+    { url = "https://files.pythonhosted.org/packages/02/05/c68c5adbf679cf610ae2f74a9b871ae84564462955d991178f95a1ddb7dd/websockets-15.0.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:5d54b09eba2bada6011aea5375542a157637b91029687eb4fdb2dab11059c1b4", size = 182165, upload-time = "2025-03-05T20:02:30.291Z" },
+    { url = "https://files.pythonhosted.org/packages/29/93/bb672df7b2f5faac89761cb5fa34f5cec45a4026c383a4b5761c6cea5c16/websockets-15.0.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:3be571a8b5afed347da347bfcf27ba12b069d9d7f42cb8c7028b5e98bbb12597", size = 182160, upload-time = "2025-03-05T20:02:31.634Z" },
+    { url = "https://files.pythonhosted.org/packages/ff/83/de1f7709376dc3ca9b7eeb4b9a07b4526b14876b6d372a4dc62312bebee0/websockets-15.0.1-cp312-cp312-win32.whl", hash = "sha256:c338ffa0520bdb12fbc527265235639fb76e7bc7faafbb93f6ba80d9c06578a9", size = 176395, upload-time = "2025-03-05T20:02:33.017Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/71/abf2ebc3bbfa40f391ce1428c7168fb20582d0ff57019b69ea20fa698043/websockets-15.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:fcd5cf9e305d7b8338754470cf69cf81f420459dbae8a3b40cee57417f4614a7", size = 176841, upload-time = "2025-03-05T20:02:34.498Z" },
+    { url = "https://files.pythonhosted.org/packages/cb/9f/51f0cf64471a9d2b4d0fc6c534f323b664e7095640c34562f5182e5a7195/websockets-15.0.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:ee443ef070bb3b6ed74514f5efaa37a252af57c90eb33b956d35c8e9c10a1931", size = 175440, upload-time = "2025-03-05T20:02:36.695Z" },
+    { url = "https://files.pythonhosted.org/packages/8a/05/aa116ec9943c718905997412c5989f7ed671bc0188ee2ba89520e8765d7b/websockets-15.0.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:5a939de6b7b4e18ca683218320fc67ea886038265fd1ed30173f5ce3f8e85675", size = 173098, upload-time = "2025-03-05T20:02:37.985Z" },
+    { url = "https://files.pythonhosted.org/packages/ff/0b/33cef55ff24f2d92924923c99926dcce78e7bd922d649467f0eda8368923/websockets-15.0.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:746ee8dba912cd6fc889a8147168991d50ed70447bf18bcda7039f7d2e3d9151", size = 173329, upload-time = "2025-03-05T20:02:39.298Z" },
+    { url = "https://files.pythonhosted.org/packages/31/1d/063b25dcc01faa8fada1469bdf769de3768b7044eac9d41f734fd7b6ad6d/websockets-15.0.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:595b6c3969023ecf9041b2936ac3827e4623bfa3ccf007575f04c5a6aa318c22", size = 183111, upload-time = "2025-03-05T20:02:40.595Z" },
+    { url = "https://files.pythonhosted.org/packages/93/53/9a87ee494a51bf63e4ec9241c1ccc4f7c2f45fff85d5bde2ff74fcb68b9e/websockets-15.0.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3c714d2fc58b5ca3e285461a4cc0c9a66bd0e24c5da9911e30158286c9b5be7f", size = 182054, upload-time = "2025-03-05T20:02:41.926Z" },
+    { url = "https://files.pythonhosted.org/packages/ff/b2/83a6ddf56cdcbad4e3d841fcc55d6ba7d19aeb89c50f24dd7e859ec0805f/websockets-15.0.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0f3c1e2ab208db911594ae5b4f79addeb3501604a165019dd221c0bdcabe4db8", size = 182496, upload-time = "2025-03-05T20:02:43.304Z" },
+    { url = "https://files.pythonhosted.org/packages/98/41/e7038944ed0abf34c45aa4635ba28136f06052e08fc2168520bb8b25149f/websockets-15.0.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:229cf1d3ca6c1804400b0a9790dc66528e08a6a1feec0d5040e8b9eb14422375", size = 182829, upload-time = "2025-03-05T20:02:48.812Z" },
+    { url = "https://files.pythonhosted.org/packages/e0/17/de15b6158680c7623c6ef0db361da965ab25d813ae54fcfeae2e5b9ef910/websockets-15.0.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:756c56e867a90fb00177d530dca4b097dd753cde348448a1012ed6c5131f8b7d", size = 182217, upload-time = "2025-03-05T20:02:50.14Z" },
+    { url = "https://files.pythonhosted.org/packages/33/2b/1f168cb6041853eef0362fb9554c3824367c5560cbdaad89ac40f8c2edfc/websockets-15.0.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:558d023b3df0bffe50a04e710bc87742de35060580a293c2a984299ed83bc4e4", size = 182195, upload-time = "2025-03-05T20:02:51.561Z" },
+    { url = "https://files.pythonhosted.org/packages/86/eb/20b6cdf273913d0ad05a6a14aed4b9a85591c18a987a3d47f20fa13dcc47/websockets-15.0.1-cp313-cp313-win32.whl", hash = "sha256:ba9e56e8ceeeedb2e080147ba85ffcd5cd0711b89576b83784d8605a7df455fa", size = 176393, upload-time = "2025-03-05T20:02:53.814Z" },
+    { url = "https://files.pythonhosted.org/packages/1b/6c/c65773d6cab416a64d191d6ee8a8b1c68a09970ea6909d16965d26bfed1e/websockets-15.0.1-cp313-cp313-win_amd64.whl", hash = "sha256:e09473f095a819042ecb2ab9465aee615bd9c2028e4ef7d933600a8401c79561", size = 176837, upload-time = "2025-03-05T20:02:55.237Z" },
+    { url = "https://files.pythonhosted.org/packages/fa/a8/5b41e0da817d64113292ab1f8247140aac61cbf6cfd085d6a0fa77f4984f/websockets-15.0.1-py3-none-any.whl", hash = "sha256:f7a866fbc1e97b5c617ee4116daaa09b722101d4a3c170c787450ba409f9736f", size = 169743, upload-time = "2025-03-05T20:03:39.41Z" },
+]

mypy-baseline.txt

@@ -1,7 +1,6 @@
 posthog/utils.py:0: error: Library stubs not installed for "six"  [import-untyped]
 posthog/utils.py:0: error: Library stubs not installed for "dateutil.tz"  [import-untyped]
 posthog/utils.py:0: error: Statement is unreachable  [unreachable]
-posthog/utils.py:0: error: Argument 1 to "join" of "str" has incompatible type "AttributeError"; expected "Iterable[str]"  [arg-type]
 posthog/request.py:0: error: Library stubs not installed for "requests"  [import-untyped]
 posthog/request.py:0: note: Hint: "python3 -m pip install types-requests"
 posthog/request.py:0: error: Library stubs not installed for "dateutil.tz"  [import-untyped]
@@ -23,19 +22,19 @@ posthog/client.py:0: error: Library stubs not installed for "six"  [import-untyp
 posthog/client.py:0: note: Hint: "python3 -m pip install types-six"
 posthog/client.py:0: error: Name "queue" already defined (by an import)  [no-redef]
 posthog/client.py:0: error: Need type annotation for "queue"  [var-annotated]
-posthog/client.py:0: error: Item "None" of "Any | None" has no attribute "get"  [union-attr]
-simulator.py:0: error: Unexpected keyword argument "anonymous_id" for "capture"  [call-arg]
-posthog/__init__.py:0: note: "capture" defined here
-simulator.py:0: error: Unexpected keyword argument "anonymous_id" for "identify"  [call-arg]
-posthog/__init__.py:0: note: "identify" defined here
-simulator.py:0: error: Unexpected keyword argument "traits" for "identify"  [call-arg]
-posthog/__init__.py:0: note: "identify" defined here
-example.py:0: error: Statement is unreachable  [unreachable]
-posthog/sentry/posthog_integration.py:0: error: Statement is unreachable  [unreachable]
-posthog/ai/utils.py:0: error: Need type annotation for "output" (hint: "output: list[<type>] = ...")  [var-annotated]
-posthog/ai/utils.py:0: error: Function "builtins.any" is not valid as a type  [valid-type]
-posthog/ai/utils.py:0: note: Perhaps you meant "typing.Any" instead of "any"?
-posthog/ai/utils.py:0: error: Function "builtins.any" is not valid as a type  [valid-type]
-posthog/ai/utils.py:0: note: Perhaps you meant "typing.Any" instead of "any"?
-sentry_django_example/sentry_django_example/settings.py:0: error: Need type annotation for "ALLOWED_HOSTS" (hint: "ALLOWED_HOSTS: list[<type>] = ...")  [var-annotated]
-sentry_django_example/sentry_django_example/settings.py:0: error: Incompatible types in assignment (expression has type "str", variable has type "None")  [assignment]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "Any | list[Any]", variable has type "None")  [assignment]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "dict[Any, Any]", variable has type "None")  [assignment]
+posthog/client.py:0: error: "None" has no attribute "__iter__" (not iterable)  [attr-defined]
+posthog/client.py:0: error: Statement is unreachable  [unreachable]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "Any | dict[Any, Any]", variable has type "None")  [assignment]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "Any | dict[Any, Any]", variable has type "None")  [assignment]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "dict[Never, Never]", variable has type "None")  [assignment]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "dict[Never, Never]", variable has type "None")  [assignment]
+posthog/client.py:0: error: Right operand of "and" is never evaluated  [unreachable]
+posthog/client.py:0: error: Incompatible types in assignment (expression has type "Poller", variable has type "None")  [assignment]
+posthog/client.py:0: error: "None" has no attribute "start"  [attr-defined]
+posthog/client.py:0: error: "None" has no attribute "get"  [attr-defined]
+posthog/client.py:0: error: Statement is unreachable  [unreachable]
+posthog/client.py:0: error: Statement is unreachable  [unreachable]
+posthog/client.py:0: error: Name "urlparse" already defined (possibly by an import)  [no-redef]
+posthog/client.py:0: error: Name "parse_qs" already defined (possibly by an import)  [no-redef]

mypy.ini

@@ -9,7 +9,7 @@ check_untyped_defs = True
 warn_unreachable = True
 strict_equality = True
 ignore_missing_imports = True
-exclude = env/.*|venv/.*
+exclude = env/.*|venv/.*|build/.*
 
 [mypy-django.*]
 ignore_missing_imports = True

posthog/ai/anthropic/__init__.py

@@ -1,6 +1,17 @@
 from .anthropic import Anthropic
 from .anthropic_async import AsyncAnthropic
-from .anthropic_providers import AnthropicBedrock, AnthropicVertex, AsyncAnthropicBedrock, AsyncAnthropicVertex
+from .anthropic_providers import (
+    AnthropicBedrock,
+    AnthropicVertex,
+    AsyncAnthropicBedrock,
+    AsyncAnthropicVertex,
+)
+from .anthropic_converter import (
+    format_anthropic_response,
+    format_anthropic_input,
+    extract_anthropic_tools,
+    format_anthropic_streaming_content,
+)
 
 __all__ = [
     "Anthropic",
@@ -9,4 +20,8 @@ __all__ = [
     "AsyncAnthropicBedrock",
     "AnthropicVertex",
     "AsyncAnthropicVertex",
+    "format_anthropic_response",
+    "format_anthropic_input",
+    "extract_anthropic_tools",
+    "format_anthropic_streaming_content",
 ]

posthog/ai/anthropic/anthropic.py

@@ -2,14 +2,29 @@ try:
     import anthropic
     from anthropic.resources import Messages
 except ImportError:
-    raise ModuleNotFoundError("Please install the Anthropic SDK to use this feature: 'pip install anthropic'")
+    raise ModuleNotFoundError(
+        "Please install the Anthropic SDK to use this feature: 'pip install anthropic'"
+    )
 
 import time
 import uuid
-from typing import Any, Dict, Optional
+from typing import Any, Dict, List, Optional
 
-from posthog.ai.utils import call_llm_and_track_usage, get_model_params, merge_system_prompt, with_privacy_mode
+from posthog.ai.types import StreamingContentBlock, TokenUsage, ToolInProgress
+from posthog.ai.utils import (
+    call_llm_and_track_usage,
+    merge_usage_stats,
+)
+from posthog.ai.anthropic.anthropic_converter import (
+    extract_anthropic_usage_from_event,
+    handle_anthropic_content_block_start,
+    handle_anthropic_text_delta,
+    handle_anthropic_tool_delta,
+    finalize_anthropic_tool_input,
+)
+from posthog.ai.sanitization import sanitize_anthropic
 from posthog.client import Client as PostHogClient
+from posthog import setup
 
 
 class Anthropic(anthropic.Anthropic):
@@ -19,14 +34,14 @@ class Anthropic(anthropic.Anthropic):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             posthog_client: PostHog client for tracking usage
             **kwargs: Additional arguments passed to the Anthropic client
         """
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = WrappedMessages(self)
 
 
@@ -53,6 +68,7 @@ class WrappedMessages(Messages):
             posthog_groups: Optional group analytics properties
             **kwargs: Arguments passed to Anthropic's messages.create
         """
+
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
@@ -110,35 +126,66 @@ class WrappedMessages(Messages):
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {"input_tokens": 0, "output_tokens": 0}
-        accumulated_content = []
+        usage_stats: TokenUsage = TokenUsage(input_tokens=0, output_tokens=0)
+        accumulated_content = ""
+        content_blocks: List[StreamingContentBlock] = []
+        tools_in_progress: Dict[str, ToolInProgress] = {}
+        current_text_block: Optional[StreamingContentBlock] = None
         response = super().create(**kwargs)
 
         def generator():
             nonlocal usage_stats
-            nonlocal accumulated_content  # noqa: F824
+            nonlocal accumulated_content
+            nonlocal content_blocks
+            nonlocal tools_in_progress
+            nonlocal current_text_block
+
             try:
                 for event in response:
-                    if hasattr(event, "usage") and event.usage:
-                        usage_stats = {
-                            k: getattr(event.usage, k, 0)
-                            for k in [
-                                "input_tokens",
-                                "output_tokens",
-                                "cache_read_input_tokens",
-                                "cache_creation_input_tokens",
-                            ]
-                        }
+                    # Extract usage stats from event
+                    event_usage = extract_anthropic_usage_from_event(event)
+                    merge_usage_stats(usage_stats, event_usage)
 
-                    if hasattr(event, "content") and event.content:
-                        accumulated_content.append(event.content)
+                    # Handle content block start events
+                    if hasattr(event, "type") and event.type == "content_block_start":
+                        block, tool = handle_anthropic_content_block_start(event)
+
+                        if block:
+                            content_blocks.append(block)
+
+                            if block.get("type") == "text":
+                                current_text_block = block
+                            else:
+                                current_text_block = None
+
+                        if tool:
+                            tool_id = tool["block"].get("id")
+                            if tool_id:
+                                tools_in_progress[tool_id] = tool
+
+                    # Handle text delta events
+                    delta_text = handle_anthropic_text_delta(event, current_text_block)
+
+                    if delta_text:
+                        accumulated_content += delta_text
+
+                    # Handle tool input delta events
+                    handle_anthropic_tool_delta(
+                        event, content_blocks, tools_in_progress
+                    )
+
+                    # Handle content block stop events
+                    if hasattr(event, "type") and event.type == "content_block_stop":
+                        current_text_block = None
+                        finalize_anthropic_tool_input(
+                            event, content_blocks, tools_in_progress
+                        )
 
                     yield event
 
             finally:
                 end_time = time.time()
                 latency = end_time - start_time
-                output = "".join(accumulated_content)
 
                 self._capture_streaming_event(
                     posthog_distinct_id,
@@ -149,7 +196,8 @@ class WrappedMessages(Messages):
                     kwargs,
                     usage_stats,
                     latency,
-                    output,
+                    content_blocks,
+                    accumulated_content,
                 )
 
         return generator()
@@ -162,45 +210,39 @@ class WrappedMessages(Messages):
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
-        output: str,
+        content_blocks: List[StreamingContentBlock],
+        accumulated_content: str,
     ):
-        if posthog_trace_id is None:
-            posthog_trace_id = str(uuid.uuid4())
+        from posthog.ai.types import StreamingEventData
+        from posthog.ai.anthropic.anthropic_converter import (
+            format_anthropic_streaming_input,
+            format_anthropic_streaming_output_complete,
+        )
+        from posthog.ai.utils import capture_streaming_event
 
-        event_properties = {
-            "$ai_provider": "anthropic",
-            "$ai_model": kwargs.get("model"),
-            "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                merge_system_prompt(kwargs, "anthropic"),
+        # Prepare standardized event data
+        formatted_input = format_anthropic_streaming_input(kwargs)
+        sanitized_input = sanitize_anthropic(formatted_input)
+
+        event_data = StreamingEventData(
+            provider="anthropic",
+            model=kwargs.get("model", "unknown"),
+            base_url=str(self._client.base_url),
+            kwargs=kwargs,
+            formatted_input=sanitized_input,
+            formatted_output=format_anthropic_streaming_output_complete(
+                content_blocks, accumulated_content
             ),
-            "$ai_output_choices": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                [{"content": output, "role": "assistant"}],
-            ),
-            "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("output_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
-            "$ai_cache_creation_input_tokens": usage_stats.get("cache_creation_input_tokens", 0),
-            "$ai_latency": latency,
-            "$ai_trace_id": posthog_trace_id,
-            "$ai_base_url": str(self._client.base_url),
-            **(posthog_properties or {}),
-        }
+            usage_stats=usage_stats,
+            latency=latency,
+            distinct_id=posthog_distinct_id,
+            trace_id=posthog_trace_id,
+            properties=posthog_properties,
+            privacy_mode=posthog_privacy_mode,
+            groups=posthog_groups,
+        )
 
-        if posthog_distinct_id is None:
-            event_properties["$process_person_profile"] = False
-
-        if hasattr(self._client._ph_client, "capture"):
-            self._client._ph_client.capture(
-                distinct_id=posthog_distinct_id or posthog_trace_id,
-                event="$ai_generation",
-                properties=event_properties,
-                groups=posthog_groups,
-            )
+        # Use the common capture function
+        capture_streaming_event(self._client._ph_client, event_data)

posthog/ai/anthropic/anthropic_async.py

@@ -2,13 +2,28 @@ try:
     import anthropic
     from anthropic.resources import AsyncMessages
 except ImportError:
-    raise ModuleNotFoundError("Please install the Anthropic SDK to use this feature: 'pip install anthropic'")
+    raise ModuleNotFoundError(
+        "Please install the Anthropic SDK to use this feature: 'pip install anthropic'"
+    )
 
 import time
 import uuid
-from typing import Any, Dict, Optional
+from typing import Any, Dict, List, Optional
 
-from posthog.ai.utils import call_llm_and_track_usage_async, get_model_params, merge_system_prompt, with_privacy_mode
+from posthog import setup
+from posthog.ai.types import StreamingContentBlock, TokenUsage, ToolInProgress
+from posthog.ai.utils import (
+    call_llm_and_track_usage_async,
+    merge_usage_stats,
+)
+from posthog.ai.anthropic.anthropic_converter import (
+    extract_anthropic_usage_from_event,
+    handle_anthropic_content_block_start,
+    handle_anthropic_text_delta,
+    handle_anthropic_tool_delta,
+    finalize_anthropic_tool_input,
+)
+from posthog.ai.sanitization import sanitize_anthropic
 from posthog.client import Client as PostHogClient
 
 
@@ -19,14 +34,14 @@ class AsyncAnthropic(anthropic.AsyncAnthropic):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             posthog_client: PostHog client for tracking usage
             **kwargs: Additional arguments passed to the Anthropic client
         """
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = AsyncWrappedMessages(self)
 
 
@@ -53,6 +68,7 @@ class AsyncWrappedMessages(AsyncMessages):
             posthog_groups: Optional group analytics properties
             **kwargs: Arguments passed to Anthropic's messages.create
         """
+
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
@@ -110,35 +126,66 @@ class AsyncWrappedMessages(AsyncMessages):
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {"input_tokens": 0, "output_tokens": 0}
-        accumulated_content = []
+        usage_stats: TokenUsage = TokenUsage(input_tokens=0, output_tokens=0)
+        accumulated_content = ""
+        content_blocks: List[StreamingContentBlock] = []
+        tools_in_progress: Dict[str, ToolInProgress] = {}
+        current_text_block: Optional[StreamingContentBlock] = None
         response = await super().create(**kwargs)
 
         async def generator():
             nonlocal usage_stats
-            nonlocal accumulated_content  # noqa: F824
+            nonlocal accumulated_content
+            nonlocal content_blocks
+            nonlocal tools_in_progress
+            nonlocal current_text_block
+
             try:
                 async for event in response:
-                    if hasattr(event, "usage") and event.usage:
-                        usage_stats = {
-                            k: getattr(event.usage, k, 0)
-                            for k in [
-                                "input_tokens",
-                                "output_tokens",
-                                "cache_read_input_tokens",
-                                "cache_creation_input_tokens",
-                            ]
-                        }
+                    # Extract usage stats from event
+                    event_usage = extract_anthropic_usage_from_event(event)
+                    merge_usage_stats(usage_stats, event_usage)
 
-                    if hasattr(event, "content") and event.content:
-                        accumulated_content.append(event.content)
+                    # Handle content block start events
+                    if hasattr(event, "type") and event.type == "content_block_start":
+                        block, tool = handle_anthropic_content_block_start(event)
+
+                        if block:
+                            content_blocks.append(block)
+
+                            if block.get("type") == "text":
+                                current_text_block = block
+                            else:
+                                current_text_block = None
+
+                        if tool:
+                            tool_id = tool["block"].get("id")
+                            if tool_id:
+                                tools_in_progress[tool_id] = tool
+
+                    # Handle text delta events
+                    delta_text = handle_anthropic_text_delta(event, current_text_block)
+
+                    if delta_text:
+                        accumulated_content += delta_text
+
+                    # Handle tool input delta events
+                    handle_anthropic_tool_delta(
+                        event, content_blocks, tools_in_progress
+                    )
+
+                    # Handle content block stop events
+                    if hasattr(event, "type") and event.type == "content_block_stop":
+                        current_text_block = None
+                        finalize_anthropic_tool_input(
+                            event, content_blocks, tools_in_progress
+                        )
 
                     yield event
 
             finally:
                 end_time = time.time()
                 latency = end_time - start_time
-                output = "".join(accumulated_content)
 
                 await self._capture_streaming_event(
                     posthog_distinct_id,
@@ -149,7 +196,8 @@ class AsyncWrappedMessages(AsyncMessages):
                     kwargs,
                     usage_stats,
                     latency,
-                    output,
+                    content_blocks,
+                    accumulated_content,
                 )
 
         return generator()
@@ -162,45 +210,39 @@ class AsyncWrappedMessages(AsyncMessages):
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
-        output: str,
+        content_blocks: List[StreamingContentBlock],
+        accumulated_content: str,
     ):
-        if posthog_trace_id is None:
-            posthog_trace_id = str(uuid.uuid4())
+        from posthog.ai.types import StreamingEventData
+        from posthog.ai.anthropic.anthropic_converter import (
+            format_anthropic_streaming_input,
+            format_anthropic_streaming_output_complete,
+        )
+        from posthog.ai.utils import capture_streaming_event
 
-        event_properties = {
-            "$ai_provider": "anthropic",
-            "$ai_model": kwargs.get("model"),
-            "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                merge_system_prompt(kwargs, "anthropic"),
+        # Prepare standardized event data
+        formatted_input = format_anthropic_streaming_input(kwargs)
+        sanitized_input = sanitize_anthropic(formatted_input)
+
+        event_data = StreamingEventData(
+            provider="anthropic",
+            model=kwargs.get("model", "unknown"),
+            base_url=str(self._client.base_url),
+            kwargs=kwargs,
+            formatted_input=sanitized_input,
+            formatted_output=format_anthropic_streaming_output_complete(
+                content_blocks, accumulated_content
             ),
-            "$ai_output_choices": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                [{"content": output, "role": "assistant"}],
-            ),
-            "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("output_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
-            "$ai_cache_creation_input_tokens": usage_stats.get("cache_creation_input_tokens", 0),
-            "$ai_latency": latency,
-            "$ai_trace_id": posthog_trace_id,
-            "$ai_base_url": str(self._client.base_url),
-            **(posthog_properties or {}),
-        }
+            usage_stats=usage_stats,
+            latency=latency,
+            distinct_id=posthog_distinct_id,
+            trace_id=posthog_trace_id,
+            properties=posthog_properties,
+            privacy_mode=posthog_privacy_mode,
+            groups=posthog_groups,
+        )
 
-        if posthog_distinct_id is None:
-            event_properties["$process_person_profile"] = False
-
-        if hasattr(self._client._ph_client, "capture"):
-            self._client._ph_client.capture(
-                distinct_id=posthog_distinct_id or posthog_trace_id,
-                event="$ai_generation",
-                properties=event_properties,
-                groups=posthog_groups,
-            )
+        # Use the common capture function
+        capture_streaming_event(self._client._ph_client, event_data)

posthog/ai/anthropic/anthropic_converter.py

@@ -0,0 +1,443 @@
+"""
+Anthropic-specific conversion utilities.
+
+This module handles the conversion of Anthropic API responses and inputs
+into standardized formats for PostHog tracking.
+"""
+
+import json
+from typing import Any, Dict, List, Optional, Tuple
+
+from posthog.ai.types import (
+    FormattedContentItem,
+    FormattedFunctionCall,
+    FormattedMessage,
+    FormattedTextContent,
+    StreamingContentBlock,
+    TokenUsage,
+    ToolInProgress,
+)
+
+
+def format_anthropic_response(response: Any) -> List[FormattedMessage]:
+    """
+    Format an Anthropic response into standardized message format.
+
+    Args:
+        response: The response object from Anthropic API
+
+    Returns:
+        List of formatted messages with role and content
+    """
+
+    output: List[FormattedMessage] = []
+
+    if response is None:
+        return output
+
+    content: List[FormattedContentItem] = []
+
+    # Process content blocks from the response
+    if hasattr(response, "content"):
+        for choice in response.content:
+            if (
+                hasattr(choice, "type")
+                and choice.type == "text"
+                and hasattr(choice, "text")
+                and choice.text
+            ):
+                text_content: FormattedTextContent = {
+                    "type": "text",
+                    "text": choice.text,
+                }
+                content.append(text_content)
+
+            elif (
+                hasattr(choice, "type")
+                and choice.type == "tool_use"
+                and hasattr(choice, "name")
+                and hasattr(choice, "id")
+            ):
+                function_call: FormattedFunctionCall = {
+                    "type": "function",
+                    "id": choice.id,
+                    "function": {
+                        "name": choice.name,
+                        "arguments": getattr(choice, "input", {}),
+                    },
+                }
+                content.append(function_call)
+
+    if content:
+        message: FormattedMessage = {
+            "role": "assistant",
+            "content": content,
+        }
+        output.append(message)
+
+    return output
+
+
+def format_anthropic_input(
+    messages: List[Dict[str, Any]], system: Optional[str] = None
+) -> List[FormattedMessage]:
+    """
+    Format Anthropic input messages with optional system prompt.
+
+    Args:
+        messages: List of message dictionaries
+        system: Optional system prompt to prepend
+
+    Returns:
+        List of formatted messages
+    """
+
+    formatted_messages: List[FormattedMessage] = []
+
+    # Add system message if provided
+    if system is not None:
+        formatted_messages.append({"role": "system", "content": system})
+
+    # Add user messages
+    if messages:
+        for msg in messages:
+            # Messages are already in the correct format, just ensure type safety
+            formatted_msg: FormattedMessage = {
+                "role": msg.get("role", "user"),
+                "content": msg.get("content", ""),
+            }
+            formatted_messages.append(formatted_msg)
+
+    return formatted_messages
+
+
+def extract_anthropic_tools(kwargs: Dict[str, Any]) -> Optional[Any]:
+    """
+    Extract tool definitions from Anthropic API kwargs.
+
+    Args:
+        kwargs: Keyword arguments passed to Anthropic API
+
+    Returns:
+        Tool definitions if present, None otherwise
+    """
+
+    return kwargs.get("tools", None)
+
+
+def format_anthropic_streaming_content(
+    content_blocks: List[StreamingContentBlock],
+) -> List[FormattedContentItem]:
+    """
+    Format content blocks from Anthropic streaming response.
+
+    Used by streaming handlers to format accumulated content blocks.
+
+    Args:
+        content_blocks: List of content block dictionaries from streaming
+
+    Returns:
+        List of formatted content items
+    """
+
+    formatted: List[FormattedContentItem] = []
+
+    for block in content_blocks:
+        if block.get("type") == "text":
+            formatted.append(
+                {
+                    "type": "text",
+                    "text": block.get("text") or "",
+                }
+            )
+
+        elif block.get("type") == "function":
+            formatted.append(
+                {
+                    "type": "function",
+                    "id": block.get("id"),
+                    "function": block.get("function") or {},
+                }
+            )
+
+    return formatted
+
+
+def extract_anthropic_web_search_count(response: Any) -> int:
+    """
+    Extract web search count from Anthropic response.
+
+    Anthropic provides exact web search counts via usage.server_tool_use.web_search_requests.
+
+    Args:
+        response: The response from Anthropic API
+
+    Returns:
+        Number of web search requests (0 if none)
+    """
+    if not hasattr(response, "usage"):
+        return 0
+
+    if not hasattr(response.usage, "server_tool_use"):
+        return 0
+
+    server_tool_use = response.usage.server_tool_use
+
+    if hasattr(server_tool_use, "web_search_requests"):
+        return max(0, int(getattr(server_tool_use, "web_search_requests", 0)))
+
+    return 0
+
+
+def extract_anthropic_usage_from_response(response: Any) -> TokenUsage:
+    """
+    Extract usage from a full Anthropic response (non-streaming).
+
+    Args:
+        response: The complete response from Anthropic API
+
+    Returns:
+        TokenUsage with standardized usage
+    """
+    if not hasattr(response, "usage"):
+        return TokenUsage(input_tokens=0, output_tokens=0)
+
+    result = TokenUsage(
+        input_tokens=getattr(response.usage, "input_tokens", 0),
+        output_tokens=getattr(response.usage, "output_tokens", 0),
+    )
+
+    if hasattr(response.usage, "cache_read_input_tokens"):
+        cache_read = response.usage.cache_read_input_tokens
+        if cache_read and cache_read > 0:
+            result["cache_read_input_tokens"] = cache_read
+
+    if hasattr(response.usage, "cache_creation_input_tokens"):
+        cache_creation = response.usage.cache_creation_input_tokens
+        if cache_creation and cache_creation > 0:
+            result["cache_creation_input_tokens"] = cache_creation
+
+    web_search_count = extract_anthropic_web_search_count(response)
+    if web_search_count > 0:
+        result["web_search_count"] = web_search_count
+
+    return result
+
+
+def extract_anthropic_usage_from_event(event: Any) -> TokenUsage:
+    """
+    Extract usage statistics from an Anthropic streaming event.
+
+    Args:
+        event: Streaming event from Anthropic API
+
+    Returns:
+        Dictionary of usage statistics
+    """
+
+    usage: TokenUsage = TokenUsage()
+
+    # Handle usage stats from message_start event
+    if hasattr(event, "type") and event.type == "message_start":
+        if hasattr(event, "message") and hasattr(event.message, "usage"):
+            usage["input_tokens"] = getattr(event.message.usage, "input_tokens", 0)
+            usage["cache_creation_input_tokens"] = getattr(
+                event.message.usage, "cache_creation_input_tokens", 0
+            )
+            usage["cache_read_input_tokens"] = getattr(
+                event.message.usage, "cache_read_input_tokens", 0
+            )
+
+    # Handle usage stats from message_delta event
+    if hasattr(event, "usage") and event.usage:
+        usage["output_tokens"] = getattr(event.usage, "output_tokens", 0)
+
+        # Extract web search count from usage
+        if hasattr(event.usage, "server_tool_use"):
+            server_tool_use = event.usage.server_tool_use
+            if hasattr(server_tool_use, "web_search_requests"):
+                web_search_count = int(
+                    getattr(server_tool_use, "web_search_requests", 0)
+                )
+                if web_search_count > 0:
+                    usage["web_search_count"] = web_search_count
+
+    return usage
+
+
+def handle_anthropic_content_block_start(
+    event: Any,
+) -> Tuple[Optional[StreamingContentBlock], Optional[ToolInProgress]]:
+    """
+    Handle content block start event from Anthropic streaming.
+
+    Args:
+        event: Content block start event
+
+    Returns:
+        Tuple of (content_block, tool_in_progress)
+    """
+
+    if not (hasattr(event, "type") and event.type == "content_block_start"):
+        return None, None
+
+    if not hasattr(event, "content_block"):
+        return None, None
+
+    block = event.content_block
+
+    if not hasattr(block, "type"):
+        return None, None
+
+    if block.type == "text":
+        content_block: StreamingContentBlock = {"type": "text", "text": ""}
+        return content_block, None
+
+    elif block.type == "tool_use":
+        tool_block: StreamingContentBlock = {
+            "type": "function",
+            "id": getattr(block, "id", ""),
+            "function": {"name": getattr(block, "name", ""), "arguments": {}},
+        }
+        tool_in_progress: ToolInProgress = {"block": tool_block, "input_string": ""}
+        return tool_block, tool_in_progress
+
+    return None, None
+
+
+def handle_anthropic_text_delta(
+    event: Any, current_block: Optional[StreamingContentBlock]
+) -> Optional[str]:
+    """
+    Handle text delta event from Anthropic streaming.
+
+    Args:
+        event: Delta event
+        current_block: Current text block being accumulated
+
+    Returns:
+        Text delta if present
+    """
+
+    if hasattr(event, "delta") and hasattr(event.delta, "text"):
+        delta_text = event.delta.text or ""
+
+        if current_block is not None and current_block.get("type") == "text":
+            text_val = current_block.get("text")
+            if text_val is not None:
+                current_block["text"] = text_val + delta_text
+            else:
+                current_block["text"] = delta_text
+
+        return delta_text
+
+    return None
+
+
+def handle_anthropic_tool_delta(
+    event: Any,
+    content_blocks: List[StreamingContentBlock],
+    tools_in_progress: Dict[str, ToolInProgress],
+) -> None:
+    """
+    Handle tool input delta event from Anthropic streaming.
+
+    Args:
+        event: Tool delta event
+        content_blocks: List of content blocks
+        tools_in_progress: Dictionary tracking tools being accumulated
+    """
+
+    if not (hasattr(event, "type") and event.type == "content_block_delta"):
+        return
+
+    if not (
+        hasattr(event, "delta")
+        and hasattr(event.delta, "type")
+        and event.delta.type == "input_json_delta"
+    ):
+        return
+
+    if hasattr(event, "index") and event.index < len(content_blocks):
+        block = content_blocks[event.index]
+
+        if block.get("type") == "function" and block.get("id") in tools_in_progress:
+            tool = tools_in_progress[block["id"]]
+            partial_json = getattr(event.delta, "partial_json", "")
+            tool["input_string"] += partial_json
+
+
+def finalize_anthropic_tool_input(
+    event: Any,
+    content_blocks: List[StreamingContentBlock],
+    tools_in_progress: Dict[str, ToolInProgress],
+) -> None:
+    """
+    Finalize tool input when content block stops.
+
+    Args:
+        event: Content block stop event
+        content_blocks: List of content blocks
+        tools_in_progress: Dictionary tracking tools being accumulated
+    """
+
+    if not (hasattr(event, "type") and event.type == "content_block_stop"):
+        return
+
+    if hasattr(event, "index") and event.index < len(content_blocks):
+        block = content_blocks[event.index]
+
+        if block.get("type") == "function" and block.get("id") in tools_in_progress:
+            tool = tools_in_progress[block["id"]]
+
+            try:
+                block["function"]["arguments"] = json.loads(tool["input_string"])
+            except (json.JSONDecodeError, Exception):
+                # Keep empty dict if parsing fails
+                pass
+
+            del tools_in_progress[block["id"]]
+
+
+def format_anthropic_streaming_input(kwargs: Dict[str, Any]) -> Any:
+    """
+    Format Anthropic streaming input using system prompt merging.
+
+    Args:
+        kwargs: Keyword arguments passed to Anthropic API
+
+    Returns:
+        Formatted input ready for PostHog tracking
+    """
+    from posthog.ai.utils import merge_system_prompt
+
+    return merge_system_prompt(kwargs, "anthropic")
+
+
+def format_anthropic_streaming_output_complete(
+    content_blocks: List[StreamingContentBlock], accumulated_content: str
+) -> List[FormattedMessage]:
+    """
+    Format complete Anthropic streaming output.
+
+    Combines existing logic for formatting content blocks with fallback to accumulated content.
+
+    Args:
+        content_blocks: List of content blocks accumulated during streaming
+        accumulated_content: Raw accumulated text content as fallback
+
+    Returns:
+        Formatted messages ready for PostHog tracking
+    """
+    formatted_content = format_anthropic_streaming_content(content_blocks)
+
+    if formatted_content:
+        return [{"role": "assistant", "content": formatted_content}]
+    else:
+        # Fallback to accumulated content if no blocks
+        return [
+            {
+                "role": "assistant",
+                "content": [{"type": "text", "text": accumulated_content}],
+            }
+        ]

posthog/ai/anthropic/anthropic_providers.py

@@ -1,11 +1,16 @@
 try:
     import anthropic
 except ImportError:
-    raise ModuleNotFoundError("Please install the Anthropic SDK to use this feature: 'pip install anthropic'")
+    raise ModuleNotFoundError(
+        "Please install the Anthropic SDK to use this feature: 'pip install anthropic'"
+    )
+
+from typing import Optional
 
 from posthog.ai.anthropic.anthropic import WrappedMessages
 from posthog.ai.anthropic.anthropic_async import AsyncWrappedMessages
 from posthog.client import Client as PostHogClient
+from posthog import setup
 
 
 class AnthropicBedrock(anthropic.AnthropicBedrock):
@@ -15,9 +20,9 @@ class AnthropicBedrock(anthropic.AnthropicBedrock):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = WrappedMessages(self)
 
 
@@ -28,9 +33,9 @@ class AsyncAnthropicBedrock(anthropic.AsyncAnthropicBedrock):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = AsyncWrappedMessages(self)
 
 
@@ -41,9 +46,9 @@ class AnthropicVertex(anthropic.AnthropicVertex):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = WrappedMessages(self)
 
 
@@ -54,7 +59,7 @@ class AsyncAnthropicVertex(anthropic.AsyncAnthropicVertex):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
         self.messages = AsyncWrappedMessages(self)

posthog/ai/gemini/__init__.py

@@ -1,4 +1,9 @@
 from .gemini import Client
+from .gemini_converter import (
+    format_gemini_input,
+    format_gemini_response,
+    extract_gemini_tools,
+)
 
 
 # Create a genai-like module for perfect drop-in replacement
@@ -8,4 +13,10 @@ class _GenAI:
 
 genai = _GenAI()
 
-__all__ = ["Client", "genai"]
+__all__ = [
+    "Client",
+    "genai",
+    "format_gemini_input",
+    "format_gemini_response",
+    "extract_gemini_tools",
+]

posthog/ai/gemini/gemini.py

@@ -3,12 +3,28 @@ import time
 import uuid
 from typing import Any, Dict, Optional
 
+from posthog.ai.types import TokenUsage, StreamingEventData
+from posthog.ai.utils import merge_system_prompt
+
 try:
     from google import genai
 except ImportError:
-    raise ModuleNotFoundError("Please install the Google Gemini SDK to use this feature: 'pip install google-genai'")
+    raise ModuleNotFoundError(
+        "Please install the Google Gemini SDK to use this feature: 'pip install google-genai'"
+    )
 
-from posthog.ai.utils import call_llm_and_track_usage, get_model_params, with_privacy_mode
+from posthog import setup
+from posthog.ai.utils import (
+    call_llm_and_track_usage,
+    capture_streaming_event,
+    merge_usage_stats,
+)
+from posthog.ai.gemini.gemini_converter import (
+    extract_gemini_usage_from_chunk,
+    extract_gemini_content_from_chunk,
+    format_gemini_streaming_output,
+)
+from posthog.ai.sanitization import sanitize_gemini
 from posthog.client import Client as PostHogClient
 
 
@@ -30,9 +46,17 @@ class Client:
         )
     """
 
+    _ph_client: PostHogClient
+
     def __init__(
         self,
         api_key: Optional[str] = None,
+        vertexai: Optional[bool] = None,
+        credentials: Optional[Any] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        debug_config: Optional[Any] = None,
+        http_options: Optional[Any] = None,
         posthog_client: Optional[PostHogClient] = None,
         posthog_distinct_id: Optional[str] = None,
         posthog_properties: Optional[Dict[str, Any]] = None,
@@ -42,7 +66,13 @@ class Client:
     ):
         """
         Args:
-            api_key: Google AI API key. If not provided, will use GOOGLE_API_KEY or API_KEY environment variable
+            api_key: Google AI API key. If not provided, will use GOOGLE_API_KEY or API_KEY environment variable (not required for Vertex AI)
+            vertexai: Whether to use Vertex AI authentication
+            credentials: Vertex AI credentials object
+            project: GCP project ID for Vertex AI
+            location: GCP location for Vertex AI
+            debug_config: Debug configuration for the client
+            http_options: HTTP options for the client
             posthog_client: PostHog client for tracking usage
             posthog_distinct_id: Default distinct ID for all calls (can be overridden per call)
             posthog_properties: Default properties for all calls (can be overridden per call)
@@ -50,12 +80,21 @@ class Client:
             posthog_groups: Default groups for all calls (can be overridden per call)
             **kwargs: Additional arguments (for future compatibility)
         """
-        if posthog_client is None:
+
+        self._ph_client = posthog_client or setup()
+
+        if self._ph_client is None:
             raise ValueError("posthog_client is required for PostHog tracking")
 
         self.models = Models(
             api_key=api_key,
-            posthog_client=posthog_client,
+            vertexai=vertexai,
+            credentials=credentials,
+            project=project,
+            location=location,
+            debug_config=debug_config,
+            http_options=http_options,
+            posthog_client=self._ph_client,
             posthog_distinct_id=posthog_distinct_id,
             posthog_properties=posthog_properties,
             posthog_privacy_mode=posthog_privacy_mode,
@@ -74,6 +113,12 @@ class Models:
     def __init__(
         self,
         api_key: Optional[str] = None,
+        vertexai: Optional[bool] = None,
+        credentials: Optional[Any] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        debug_config: Optional[Any] = None,
+        http_options: Optional[Any] = None,
         posthog_client: Optional[PostHogClient] = None,
         posthog_distinct_id: Optional[str] = None,
         posthog_properties: Optional[Dict[str, Any]] = None,
@@ -83,7 +128,13 @@ class Models:
     ):
         """
         Args:
-            api_key: Google AI API key. If not provided, will use GOOGLE_API_KEY or API_KEY environment variable
+            api_key: Google AI API key. If not provided, will use GOOGLE_API_KEY or API_KEY environment variable (not required for Vertex AI)
+            vertexai: Whether to use Vertex AI authentication
+            credentials: Vertex AI credentials object
+            project: GCP project ID for Vertex AI
+            location: GCP location for Vertex AI
+            debug_config: Debug configuration for the client
+            http_options: HTTP options for the client
             posthog_client: PostHog client for tracking usage
             posthog_distinct_id: Default distinct ID for all calls
             posthog_properties: Default properties for all calls
@@ -91,10 +142,11 @@ class Models:
             posthog_groups: Default groups for all calls
             **kwargs: Additional arguments (for future compatibility)
         """
-        if posthog_client is None:
-            raise ValueError("posthog_client is required for PostHog tracking")
 
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
+
+        if self._ph_client is None:
+            raise ValueError("posthog_client is required for PostHog tracking")
 
         # Store default PostHog settings
         self._default_distinct_id = posthog_distinct_id
@@ -102,16 +154,46 @@ class Models:
         self._default_privacy_mode = posthog_privacy_mode
         self._default_groups = posthog_groups
 
-        # Handle API key - try parameter first, then environment variables
-        if api_key is None:
-            api_key = os.environ.get("GOOGLE_API_KEY") or os.environ.get("API_KEY")
+        # Build genai.Client arguments
+        client_args: Dict[str, Any] = {}
 
-        if api_key is None:
-            raise ValueError(
-                "API key must be provided either as parameter or via GOOGLE_API_KEY/API_KEY environment variable"
-            )
+        # Add Vertex AI parameters if provided
+        if vertexai is not None:
+            client_args["vertexai"] = vertexai
 
-        self._client = genai.Client(api_key=api_key)
+        if credentials is not None:
+            client_args["credentials"] = credentials
+
+        if project is not None:
+            client_args["project"] = project
+
+        if location is not None:
+            client_args["location"] = location
+
+        if debug_config is not None:
+            client_args["debug_config"] = debug_config
+
+        if http_options is not None:
+            client_args["http_options"] = http_options
+
+        # Handle API key authentication
+        if vertexai:
+            # For Vertex AI, api_key is optional
+            if api_key is not None:
+                client_args["api_key"] = api_key
+        else:
+            # For non-Vertex AI mode, api_key is required (backwards compatibility)
+            if api_key is None:
+                api_key = os.environ.get("GOOGLE_API_KEY") or os.environ.get("API_KEY")
+
+            if api_key is None:
+                raise ValueError(
+                    "API key must be provided either as parameter or via GOOGLE_API_KEY/API_KEY environment variable"
+                )
+
+            client_args["api_key"] = api_key
+
+        self._client = genai.Client(**client_args)
         self._base_url = "https://generativelanguage.googleapis.com"
 
     def _merge_posthog_params(
@@ -123,13 +205,23 @@ class Models:
         call_groups: Optional[Dict[str, Any]],
     ):
         """Merge call-level PostHog parameters with client defaults."""
+
         # Use call-level values if provided, otherwise fall back to defaults
-        distinct_id = call_distinct_id if call_distinct_id is not None else self._default_distinct_id
-        privacy_mode = call_privacy_mode if call_privacy_mode is not None else self._default_privacy_mode
+        distinct_id = (
+            call_distinct_id
+            if call_distinct_id is not None
+            else self._default_distinct_id
+        )
+        privacy_mode = (
+            call_privacy_mode
+            if call_privacy_mode is not None
+            else self._default_privacy_mode
+        )
         groups = call_groups if call_groups is not None else self._default_groups
 
         # Merge properties: default properties + call properties (call properties override)
         properties = dict(self._default_properties)
+
         if call_properties:
             properties.update(call_properties)
 
@@ -165,9 +257,16 @@ class Models:
             posthog_groups: Group analytics properties (overrides client default)
             **kwargs: Arguments passed to Gemini's generate_content
         """
+
         # Merge PostHog parameters
-        distinct_id, trace_id, properties, privacy_mode, groups = self._merge_posthog_params(
-            posthog_distinct_id, posthog_trace_id, posthog_properties, posthog_privacy_mode, posthog_groups
+        distinct_id, trace_id, properties, privacy_mode, groups = (
+            self._merge_posthog_params(
+                posthog_distinct_id,
+                posthog_trace_id,
+                posthog_properties,
+                posthog_privacy_mode,
+                posthog_groups,
+            )
         )
 
         kwargs_with_contents = {"model": model, "contents": contents, **kwargs}
@@ -197,7 +296,7 @@ class Models:
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {"input_tokens": 0, "output_tokens": 0}
+        usage_stats: TokenUsage = TokenUsage(input_tokens=0, output_tokens=0)
         accumulated_content = []
 
         kwargs_without_stream = {"model": model, "contents": contents, **kwargs}
@@ -208,21 +307,24 @@ class Models:
             nonlocal accumulated_content  # noqa: F824
             try:
                 for chunk in response:
-                    if hasattr(chunk, "usage_metadata") and chunk.usage_metadata:
-                        usage_stats = {
-                            "input_tokens": getattr(chunk.usage_metadata, "prompt_token_count", 0),
-                            "output_tokens": getattr(chunk.usage_metadata, "candidates_token_count", 0),
-                        }
+                    # Extract usage stats from chunk
+                    chunk_usage = extract_gemini_usage_from_chunk(chunk)
 
-                    if hasattr(chunk, "text") and chunk.text:
-                        accumulated_content.append(chunk.text)
+                    if chunk_usage:
+                        # Gemini reports cumulative totals, not incremental values
+                        merge_usage_stats(usage_stats, chunk_usage, mode="cumulative")
+
+                    # Extract content from chunk (now returns content blocks)
+                    content_block = extract_gemini_content_from_chunk(chunk)
+
+                    if content_block is not None:
+                        accumulated_content.append(content_block)
 
                     yield chunk
 
             finally:
                 end_time = time.time()
                 latency = end_time - start_time
-                output = "".join(accumulated_content)
 
                 self._capture_streaming_event(
                     model,
@@ -235,7 +337,7 @@ class Models:
                     kwargs,
                     usage_stats,
                     latency,
-                    output,
+                    accumulated_content,
                 )
 
         return generator()
@@ -250,63 +352,39 @@ class Models:
         privacy_mode: bool,
         groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
-        output: str,
+        output: Any,
     ):
-        if trace_id is None:
-            trace_id = str(uuid.uuid4())
+        # Prepare standardized event data
+        formatted_input = self._format_input(contents, **kwargs)
+        sanitized_input = sanitize_gemini(formatted_input)
 
-        event_properties = {
-            "$ai_provider": "gemini",
-            "$ai_model": model,
-            "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(
-                self._ph_client,
-                privacy_mode,
-                self._format_input(contents),
-            ),
-            "$ai_output_choices": with_privacy_mode(
-                self._ph_client,
-                privacy_mode,
-                [{"content": output, "role": "assistant"}],
-            ),
-            "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("output_tokens", 0),
-            "$ai_latency": latency,
-            "$ai_trace_id": trace_id,
-            "$ai_base_url": self._base_url,
-            **(properties or {}),
-        }
+        event_data = StreamingEventData(
+            provider="gemini",
+            model=model,
+            base_url=self._base_url,
+            kwargs=kwargs,
+            formatted_input=sanitized_input,
+            formatted_output=format_gemini_streaming_output(output),
+            usage_stats=usage_stats,
+            latency=latency,
+            distinct_id=distinct_id,
+            trace_id=trace_id,
+            properties=properties,
+            privacy_mode=privacy_mode,
+            groups=groups,
+        )
 
-        if distinct_id is None:
-            event_properties["$process_person_profile"] = False
+        # Use the common capture function
+        capture_streaming_event(self._ph_client, event_data)
 
-        if hasattr(self._ph_client, "capture"):
-            self._ph_client.capture(
-                distinct_id=distinct_id,
-                event="$ai_generation",
-                properties=event_properties,
-                groups=groups,
-            )
-
-    def _format_input(self, contents):
+    def _format_input(self, contents, **kwargs):
         """Format input contents for PostHog tracking"""
-        if isinstance(contents, str):
-            return [{"role": "user", "content": contents}]
-        elif isinstance(contents, list):
-            formatted = []
-            for item in contents:
-                if isinstance(item, str):
-                    formatted.append({"role": "user", "content": item})
-                elif hasattr(item, "text"):
-                    formatted.append({"role": "user", "content": item.text})
-                else:
-                    formatted.append({"role": "user", "content": str(item)})
-            return formatted
-        else:
-            return [{"role": "user", "content": str(contents)}]
+
+        # Create kwargs dict with contents for merge_system_prompt
+        input_kwargs = {"contents": contents, **kwargs}
+        return merge_system_prompt(input_kwargs, "gemini")
 
     def generate_content_stream(
         self,
@@ -320,8 +398,14 @@ class Models:
         **kwargs: Any,
     ):
         # Merge PostHog parameters
-        distinct_id, trace_id, properties, privacy_mode, groups = self._merge_posthog_params(
-            posthog_distinct_id, posthog_trace_id, posthog_properties, posthog_privacy_mode, posthog_groups
+        distinct_id, trace_id, properties, privacy_mode, groups = (
+            self._merge_posthog_params(
+                posthog_distinct_id,
+                posthog_trace_id,
+                posthog_properties,
+                posthog_privacy_mode,
+                posthog_groups,
+            )
         )
 
         return self._generate_content_streaming(

posthog/ai/gemini/gemini_converter.py

@@ -0,0 +1,586 @@
+"""
+Gemini-specific conversion utilities.
+
+This module handles the conversion of Gemini API responses and inputs
+into standardized formats for PostHog tracking.
+"""
+
+from typing import Any, Dict, List, Optional, TypedDict, Union
+
+from posthog.ai.types import (
+    FormattedContentItem,
+    FormattedMessage,
+    TokenUsage,
+)
+
+
+class GeminiPart(TypedDict, total=False):
+    """Represents a part in a Gemini message."""
+
+    text: str
+
+
+class GeminiMessage(TypedDict, total=False):
+    """Represents a Gemini message with various possible fields."""
+
+    role: str
+    parts: List[Union[GeminiPart, Dict[str, Any]]]
+    content: Union[str, List[Any]]
+    text: str
+
+
+def _extract_text_from_parts(parts: List[Any]) -> str:
+    """
+    Extract and concatenate text from a parts array.
+
+    Args:
+        parts: List of parts that may contain text content
+
+    Returns:
+        Concatenated text from all parts
+    """
+
+    content_parts = []
+
+    for part in parts:
+        if isinstance(part, dict) and "text" in part:
+            content_parts.append(part["text"])
+
+        elif isinstance(part, str):
+            content_parts.append(part)
+
+        elif hasattr(part, "text"):
+            # Get the text attribute value
+            text_value = getattr(part, "text", "")
+            content_parts.append(text_value if text_value else str(part))
+
+        else:
+            content_parts.append(str(part))
+
+    return "".join(content_parts)
+
+
+def _format_dict_message(item: Dict[str, Any]) -> FormattedMessage:
+    """
+    Format a dictionary message into standardized format.
+
+    Args:
+        item: Dictionary containing message data
+
+    Returns:
+        Formatted message with role and content
+    """
+
+    # Handle dict format with parts array (Gemini-specific format)
+    if "parts" in item and isinstance(item["parts"], list):
+        content = _extract_text_from_parts(item["parts"])
+        return {"role": item.get("role", "user"), "content": content}
+
+    # Handle dict with content field
+    if "content" in item:
+        content = item["content"]
+
+        if isinstance(content, list):
+            # If content is a list, extract text from it
+            content = _extract_text_from_parts(content)
+
+        elif not isinstance(content, str):
+            content = str(content)
+
+        return {"role": item.get("role", "user"), "content": content}
+
+    # Handle dict with text field
+    if "text" in item:
+        return {"role": item.get("role", "user"), "content": item["text"]}
+
+    # Fallback to string representation
+    return {"role": "user", "content": str(item)}
+
+
+def _format_object_message(item: Any) -> FormattedMessage:
+    """
+    Format an object (with attributes) into standardized format.
+
+    Args:
+        item: Object that may have text or parts attributes
+
+    Returns:
+        Formatted message with role and content
+    """
+
+    # Handle object with parts attribute
+    if hasattr(item, "parts") and hasattr(item.parts, "__iter__"):
+        content = _extract_text_from_parts(item.parts)
+        role = getattr(item, "role", "user") if hasattr(item, "role") else "user"
+
+        # Ensure role is a string
+        if not isinstance(role, str):
+            role = "user"
+
+        return {"role": role, "content": content}
+
+    # Handle object with text attribute
+    if hasattr(item, "text"):
+        role = getattr(item, "role", "user") if hasattr(item, "role") else "user"
+
+        # Ensure role is a string
+        if not isinstance(role, str):
+            role = "user"
+
+        return {"role": role, "content": item.text}
+
+    # Handle object with content attribute
+    if hasattr(item, "content"):
+        role = getattr(item, "role", "user") if hasattr(item, "role") else "user"
+
+        # Ensure role is a string
+        if not isinstance(role, str):
+            role = "user"
+
+        content = item.content
+
+        if isinstance(content, list):
+            content = _extract_text_from_parts(content)
+
+        elif not isinstance(content, str):
+            content = str(content)
+        return {"role": role, "content": content}
+
+    # Fallback to string representation
+    return {"role": "user", "content": str(item)}
+
+
+def format_gemini_response(response: Any) -> List[FormattedMessage]:
+    """
+    Format a Gemini response into standardized message format.
+
+    Args:
+        response: The response object from Gemini API
+
+    Returns:
+        List of formatted messages with role and content
+    """
+
+    output: List[FormattedMessage] = []
+
+    if response is None:
+        return output
+
+    if hasattr(response, "candidates") and response.candidates:
+        for candidate in response.candidates:
+            if hasattr(candidate, "content") and candidate.content:
+                content: List[FormattedContentItem] = []
+
+                if hasattr(candidate.content, "parts") and candidate.content.parts:
+                    for part in candidate.content.parts:
+                        if hasattr(part, "text") and part.text:
+                            content.append(
+                                {
+                                    "type": "text",
+                                    "text": part.text,
+                                }
+                            )
+
+                        elif hasattr(part, "function_call") and part.function_call:
+                            function_call = part.function_call
+                            content.append(
+                                {
+                                    "type": "function",
+                                    "function": {
+                                        "name": function_call.name,
+                                        "arguments": function_call.args,
+                                    },
+                                }
+                            )
+
+                if content:
+                    output.append(
+                        {
+                            "role": "assistant",
+                            "content": content,
+                        }
+                    )
+
+            elif hasattr(candidate, "text") and candidate.text:
+                output.append(
+                    {
+                        "role": "assistant",
+                        "content": [{"type": "text", "text": candidate.text}],
+                    }
+                )
+
+    elif hasattr(response, "text") and response.text:
+        output.append(
+            {
+                "role": "assistant",
+                "content": [{"type": "text", "text": response.text}],
+            }
+        )
+
+    return output
+
+
+def extract_gemini_system_instruction(config: Any) -> Optional[str]:
+    """
+    Extract system instruction from Gemini config parameter.
+
+    Args:
+        config: Config object or dict that may contain system instruction
+
+    Returns:
+        System instruction string if present, None otherwise
+    """
+    if config is None:
+        return None
+
+    # Handle different config formats
+    if hasattr(config, "system_instruction"):
+        return config.system_instruction
+    elif isinstance(config, dict) and "system_instruction" in config:
+        return config["system_instruction"]
+    elif isinstance(config, dict) and "systemInstruction" in config:
+        return config["systemInstruction"]
+
+    return None
+
+
+def extract_gemini_tools(kwargs: Dict[str, Any]) -> Optional[Any]:
+    """
+    Extract tool definitions from Gemini API kwargs.
+
+    Args:
+        kwargs: Keyword arguments passed to Gemini API
+
+    Returns:
+        Tool definitions if present, None otherwise
+    """
+
+    if "config" in kwargs and hasattr(kwargs["config"], "tools"):
+        return kwargs["config"].tools
+
+    return None
+
+
+def format_gemini_input_with_system(
+    contents: Any, config: Any = None
+) -> List[FormattedMessage]:
+    """
+    Format Gemini input contents into standardized message format, including system instruction handling.
+
+    Args:
+        contents: Input contents in various possible formats
+        config: Config object or dict that may contain system instruction
+
+    Returns:
+        List of formatted messages with role and content fields, with system message prepended if needed
+    """
+    formatted_messages = format_gemini_input(contents)
+
+    # Check if system instruction is provided in config parameter
+    system_instruction = extract_gemini_system_instruction(config)
+
+    if system_instruction is not None:
+        has_system = any(msg.get("role") == "system" for msg in formatted_messages)
+        if not has_system:
+            from posthog.ai.types import FormattedMessage
+
+            system_message: FormattedMessage = {
+                "role": "system",
+                "content": system_instruction,
+            }
+            formatted_messages = [system_message] + list(formatted_messages)
+
+    return formatted_messages
+
+
+def format_gemini_input(contents: Any) -> List[FormattedMessage]:
+    """
+    Format Gemini input contents into standardized message format for PostHog tracking.
+
+    This function handles various input formats:
+    - String inputs
+    - List of strings, dicts, or objects
+    - Single dict or object
+    - Gemini-specific format with parts array
+
+    Args:
+        contents: Input contents in various possible formats
+
+    Returns:
+        List of formatted messages with role and content fields
+    """
+
+    # Handle string input
+    if isinstance(contents, str):
+        return [{"role": "user", "content": contents}]
+
+    # Handle list input
+    if isinstance(contents, list):
+        formatted: List[FormattedMessage] = []
+
+        for item in contents:
+            if isinstance(item, str):
+                formatted.append({"role": "user", "content": item})
+
+            elif isinstance(item, dict):
+                formatted.append(_format_dict_message(item))
+
+            else:
+                formatted.append(_format_object_message(item))
+
+        return formatted
+
+    # Handle single dict input
+    if isinstance(contents, dict):
+        return [_format_dict_message(contents)]
+
+    # Handle single object input
+    return [_format_object_message(contents)]
+
+
+def extract_gemini_web_search_count(response: Any) -> int:
+    """
+    Extract web search count from Gemini response.
+
+    Gemini bills per request that uses grounding, not per query.
+    Returns 1 if grounding_metadata is present with actual search data, 0 otherwise.
+
+    Args:
+        response: The response from Gemini API
+
+    Returns:
+        1 if web search/grounding was used, 0 otherwise
+    """
+
+    # Check for grounding_metadata in candidates
+    if hasattr(response, "candidates"):
+        for candidate in response.candidates:
+            if (
+                hasattr(candidate, "grounding_metadata")
+                and candidate.grounding_metadata
+            ):
+                grounding_metadata = candidate.grounding_metadata
+
+                # Check if web_search_queries exists and is non-empty
+                if hasattr(grounding_metadata, "web_search_queries"):
+                    queries = grounding_metadata.web_search_queries
+
+                    if queries is not None and len(queries) > 0:
+                        return 1
+
+                # Check if grounding_chunks exists and is non-empty
+                if hasattr(grounding_metadata, "grounding_chunks"):
+                    chunks = grounding_metadata.grounding_chunks
+
+                    if chunks is not None and len(chunks) > 0:
+                        return 1
+
+            # Also check for google_search or grounding in function call names
+            if hasattr(candidate, "content") and candidate.content:
+                if hasattr(candidate.content, "parts") and candidate.content.parts:
+                    for part in candidate.content.parts:
+                        if hasattr(part, "function_call") and part.function_call:
+                            function_name = getattr(
+                                part.function_call, "name", ""
+                            ).lower()
+
+                            if (
+                                "google_search" in function_name
+                                or "grounding" in function_name
+                            ):
+                                return 1
+
+    return 0
+
+
+def _extract_usage_from_metadata(metadata: Any) -> TokenUsage:
+    """
+    Common logic to extract usage from Gemini metadata.
+    Used by both streaming and non-streaming paths.
+
+    Args:
+        metadata: usage_metadata from Gemini response or chunk
+
+    Returns:
+        TokenUsage with standardized usage
+    """
+    usage = TokenUsage(
+        input_tokens=getattr(metadata, "prompt_token_count", 0),
+        output_tokens=getattr(metadata, "candidates_token_count", 0),
+    )
+
+    # Add cache tokens if present (don't add if 0)
+    if hasattr(metadata, "cached_content_token_count"):
+        cache_tokens = metadata.cached_content_token_count
+        if cache_tokens and cache_tokens > 0:
+            usage["cache_read_input_tokens"] = cache_tokens
+
+    # Add reasoning tokens if present (don't add if 0)
+    if hasattr(metadata, "thoughts_token_count"):
+        reasoning_tokens = metadata.thoughts_token_count
+        if reasoning_tokens and reasoning_tokens > 0:
+            usage["reasoning_tokens"] = reasoning_tokens
+
+    return usage
+
+
+def extract_gemini_usage_from_response(response: Any) -> TokenUsage:
+    """
+    Extract usage statistics from a full Gemini response (non-streaming).
+
+    Args:
+        response: The complete response from Gemini API
+
+    Returns:
+        TokenUsage with standardized usage statistics
+    """
+    if not hasattr(response, "usage_metadata") or not response.usage_metadata:
+        return TokenUsage(input_tokens=0, output_tokens=0)
+
+    usage = _extract_usage_from_metadata(response.usage_metadata)
+
+    # Add web search count if present
+    web_search_count = extract_gemini_web_search_count(response)
+    if web_search_count > 0:
+        usage["web_search_count"] = web_search_count
+
+    return usage
+
+
+def extract_gemini_usage_from_chunk(chunk: Any) -> TokenUsage:
+    """
+    Extract usage statistics from a Gemini streaming chunk.
+
+    Args:
+        chunk: Streaming chunk from Gemini API
+
+    Returns:
+        TokenUsage with standardized usage statistics
+    """
+
+    usage: TokenUsage = TokenUsage()
+
+    # Extract web search count from the chunk before checking for usage_metadata
+    # Web search indicators can appear on any chunk, not just those with usage data
+    web_search_count = extract_gemini_web_search_count(chunk)
+    if web_search_count > 0:
+        usage["web_search_count"] = web_search_count
+
+    if not hasattr(chunk, "usage_metadata") or not chunk.usage_metadata:
+        return usage
+
+    usage_from_metadata = _extract_usage_from_metadata(chunk.usage_metadata)
+
+    # Merge the usage from metadata with any web search count we found
+    usage.update(usage_from_metadata)
+
+    return usage
+
+
+def extract_gemini_content_from_chunk(chunk: Any) -> Optional[Dict[str, Any]]:
+    """
+    Extract content (text or function call) from a Gemini streaming chunk.
+
+    Args:
+        chunk: Streaming chunk from Gemini API
+
+    Returns:
+        Content block dictionary if present, None otherwise
+    """
+
+    # Check for text content
+    if hasattr(chunk, "text") and chunk.text:
+        return {"type": "text", "text": chunk.text}
+
+    # Check for function calls in candidates
+    if hasattr(chunk, "candidates") and chunk.candidates:
+        for candidate in chunk.candidates:
+            if hasattr(candidate, "content") and candidate.content:
+                if hasattr(candidate.content, "parts") and candidate.content.parts:
+                    for part in candidate.content.parts:
+                        # Check for function_call part
+                        if hasattr(part, "function_call") and part.function_call:
+                            function_call = part.function_call
+                            return {
+                                "type": "function",
+                                "function": {
+                                    "name": function_call.name,
+                                    "arguments": function_call.args,
+                                },
+                            }
+                        # Also check for text in parts
+                        elif hasattr(part, "text") and part.text:
+                            return {"type": "text", "text": part.text}
+
+    return None
+
+
+def format_gemini_streaming_output(
+    accumulated_content: Union[str, List[Any]],
+) -> List[FormattedMessage]:
+    """
+    Format the final output from Gemini streaming.
+
+    Args:
+        accumulated_content: Accumulated content from streaming (string, list of strings, or list of content blocks)
+
+    Returns:
+        List of formatted messages
+    """
+
+    # Handle legacy string input (backward compatibility)
+    if isinstance(accumulated_content, str):
+        return [
+            {
+                "role": "assistant",
+                "content": [{"type": "text", "text": accumulated_content}],
+            }
+        ]
+
+    # Handle list input
+    if isinstance(accumulated_content, list):
+        content: List[FormattedContentItem] = []
+        text_parts = []
+
+        for item in accumulated_content:
+            if isinstance(item, str):
+                # Legacy support: accumulate strings
+                text_parts.append(item)
+            elif isinstance(item, dict):
+                # New format: content blocks
+                if item.get("type") == "text":
+                    text_parts.append(item.get("text", ""))
+                elif item.get("type") == "function":
+                    # If we have accumulated text, add it first
+                    if text_parts:
+                        content.append(
+                            {
+                                "type": "text",
+                                "text": "".join(text_parts),
+                            }
+                        )
+                        text_parts = []
+
+                    # Add the function call
+                    content.append(
+                        {
+                            "type": "function",
+                            "function": item.get("function", {}),
+                        }
+                    )
+
+        # Add any remaining text
+        if text_parts:
+            content.append(
+                {
+                    "type": "text",
+                    "text": "".join(text_parts),
+                }
+            )
+
+        # If we have content, return it
+        if content:
+            return [{"role": "assistant", "content": content}]
+
+    # Fallback for empty or unexpected input
+    return [{"role": "assistant", "content": [{"type": "text", "text": ""}]}]

posthog/ai/langchain/callbacks.py

@@ -1,8 +1,11 @@
 try:
     import langchain  # noqa: F401
 except ImportError:
-    raise ModuleNotFoundError("Please install LangChain to use this feature: 'pip install langchain'")
+    raise ModuleNotFoundError(
+        "Please install LangChain to use this feature: 'pip install langchain'"
+    )
 
+import json
 import logging
 import time
 from dataclasses import dataclass
@@ -12,21 +15,35 @@ from typing import (
     List,
     Optional,
     Sequence,
-    Tuple,
     Union,
     cast,
 )
 from uuid import UUID
 
-from langchain.callbacks.base import BaseCallbackHandler
-from langchain.schema.agent import AgentAction, AgentFinish
+try:
+    # LangChain 1.0+ and modern 0.x with langchain-core
+    from langchain_core.callbacks.base import BaseCallbackHandler
+    from langchain_core.agents import AgentAction, AgentFinish
+except (ImportError, ModuleNotFoundError):
+    # Fallback for older LangChain versions
+    from langchain.callbacks.base import BaseCallbackHandler
+    from langchain.schema.agent import AgentAction, AgentFinish
 from langchain_core.documents import Document
-from langchain_core.messages import AIMessage, BaseMessage, FunctionMessage, HumanMessage, SystemMessage, ToolMessage
+from langchain_core.messages import (
+    AIMessage,
+    BaseMessage,
+    FunctionMessage,
+    HumanMessage,
+    SystemMessage,
+    ToolMessage,
+    ToolCall,
+)
 from langchain_core.outputs import ChatGeneration, LLMResult
 from pydantic import BaseModel
 
-from posthog import default_client
+from posthog import setup
 from posthog.ai.utils import get_model_params, with_privacy_mode
+from posthog.ai.sanitization import sanitize_langchain
 from posthog.client import Client
 
 log = logging.getLogger("posthog")
@@ -73,10 +90,10 @@ class CallbackHandler(BaseCallbackHandler):
     The PostHog LLM observability callback handler for LangChain.
     """
 
-    _client: Client
+    _ph_client: Client
     """PostHog client instance."""
 
-    _distinct_id: Optional[Union[str, int, float, UUID]]
+    _distinct_id: Optional[Union[str, int, UUID]]
     """Distinct ID of the user to associate the trace with."""
 
     _trace_id: Optional[Union[str, int, float, UUID]]
@@ -104,7 +121,7 @@ class CallbackHandler(BaseCallbackHandler):
         self,
         client: Optional[Client] = None,
         *,
-        distinct_id: Optional[Union[str, int, float, UUID]] = None,
+        distinct_id: Optional[Union[str, int, UUID]] = None,
         trace_id: Optional[Union[str, int, float, UUID]] = None,
         properties: Optional[Dict[str, Any]] = None,
         privacy_mode: bool = False,
@@ -119,10 +136,7 @@ class CallbackHandler(BaseCallbackHandler):
             privacy_mode: Whether to redact the input and output of the trace.
             groups: Optional additional PostHog groups to use for the trace.
         """
-        posthog_client = client or default_client
-        if posthog_client is None:
-            raise ValueError("PostHog client is required")
-        self._client = posthog_client
+        self._ph_client = client or setup()
         self._distinct_id = distinct_id
         self._trace_id = trace_id
         self._properties = properties or {}
@@ -143,7 +157,9 @@ class CallbackHandler(BaseCallbackHandler):
     ):
         self._log_debug_event("on_chain_start", run_id, parent_run_id, inputs=inputs)
         self._set_parent_of_run(run_id, parent_run_id)
-        self._set_trace_or_span_metadata(serialized, inputs, run_id, parent_run_id, **kwargs)
+        self._set_trace_or_span_metadata(
+            serialized, inputs, run_id, parent_run_id, **kwargs
+        )
 
     def on_chain_end(
         self,
@@ -176,9 +192,13 @@ class CallbackHandler(BaseCallbackHandler):
         parent_run_id: Optional[UUID] = None,
         **kwargs,
     ):
-        self._log_debug_event("on_chat_model_start", run_id, parent_run_id, messages=messages)
+        self._log_debug_event(
+            "on_chat_model_start", run_id, parent_run_id, messages=messages
+        )
         self._set_parent_of_run(run_id, parent_run_id)
-        input = [_convert_message_to_dict(message) for row in messages for message in row]
+        input = [
+            _convert_message_to_dict(message) for row in messages for message in row
+        ]
         self._set_llm_metadata(serialized, run_id, input, **kwargs)
 
     def on_llm_start(
@@ -216,7 +236,9 @@ class CallbackHandler(BaseCallbackHandler):
         """
         The callback works for both streaming and non-streaming runs. For streaming runs, the chain must set `stream_usage=True` in the LLM.
         """
-        self._log_debug_event("on_llm_end", run_id, parent_run_id, response=response, kwargs=kwargs)
+        self._log_debug_event(
+            "on_llm_end", run_id, parent_run_id, response=response, kwargs=kwargs
+        )
         self._pop_run_and_capture_generation(run_id, parent_run_id, response)
 
     def on_llm_error(
@@ -240,9 +262,13 @@ class CallbackHandler(BaseCallbackHandler):
         metadata: Optional[Dict[str, Any]] = None,
         **kwargs: Any,
     ) -> Any:
-        self._log_debug_event("on_tool_start", run_id, parent_run_id, input_str=input_str)
+        self._log_debug_event(
+            "on_tool_start", run_id, parent_run_id, input_str=input_str
+        )
         self._set_parent_of_run(run_id, parent_run_id)
-        self._set_trace_or_span_metadata(serialized, input_str, run_id, parent_run_id, **kwargs)
+        self._set_trace_or_span_metadata(
+            serialized, input_str, run_id, parent_run_id, **kwargs
+        )
 
     def on_tool_end(
         self,
@@ -279,7 +305,9 @@ class CallbackHandler(BaseCallbackHandler):
     ) -> Any:
         self._log_debug_event("on_retriever_start", run_id, parent_run_id, query=query)
         self._set_parent_of_run(run_id, parent_run_id)
-        self._set_trace_or_span_metadata(serialized, query, run_id, parent_run_id, **kwargs)
+        self._set_trace_or_span_metadata(
+            serialized, query, run_id, parent_run_id, **kwargs
+        )
 
     def on_retriever_end(
         self,
@@ -289,7 +317,9 @@ class CallbackHandler(BaseCallbackHandler):
         parent_run_id: Optional[UUID] = None,
         **kwargs: Any,
     ):
-        self._log_debug_event("on_retriever_end", run_id, parent_run_id, documents=documents)
+        self._log_debug_event(
+            "on_retriever_end", run_id, parent_run_id, documents=documents
+        )
         self._pop_run_and_capture_trace_or_span(run_id, parent_run_id, documents)
 
     def on_retriever_error(
@@ -364,7 +394,9 @@ class CallbackHandler(BaseCallbackHandler):
     ):
         default_name = "trace" if parent_run_id is None else "span"
         run_name = _get_langchain_run_name(serialized, **kwargs) or default_name
-        self._runs[run_id] = SpanMetadata(name=run_name, input=input, start_time=time.time(), end_time=None)
+        self._runs[run_id] = SpanMetadata(
+            name=run_name, input=input, start_time=time.time(), end_time=None
+        )
 
     def _set_llm_metadata(
         self,
@@ -376,7 +408,9 @@ class CallbackHandler(BaseCallbackHandler):
         **kwargs,
     ):
         run_name = _get_langchain_run_name(serialized, **kwargs) or "generation"
-        generation = GenerationMetadata(name=run_name, input=messages, start_time=time.time(), end_time=None)
+        generation = GenerationMetadata(
+            name=run_name, input=messages, start_time=time.time(), end_time=None
+        )
         if isinstance(invocation_params, dict):
             generation.model_params = get_model_params(invocation_params)
             if tools := invocation_params.get("tools"):
@@ -410,7 +444,9 @@ class CallbackHandler(BaseCallbackHandler):
             return run_id
         return trace_id
 
-    def _get_parent_run_id(self, trace_id: Any, run_id: UUID, parent_run_id: Optional[UUID]):
+    def _get_parent_run_id(
+        self, trace_id: Any, run_id: UUID, parent_run_id: Optional[UUID]
+    ):
         """
         Replace the parent run ID with the trace ID for second level runs when a custom trace ID is set.
         """
@@ -418,14 +454,18 @@ class CallbackHandler(BaseCallbackHandler):
             return trace_id
         return parent_run_id
 
-    def _pop_run_and_capture_trace_or_span(self, run_id: UUID, parent_run_id: Optional[UUID], outputs: Any):
+    def _pop_run_and_capture_trace_or_span(
+        self, run_id: UUID, parent_run_id: Optional[UUID], outputs: Any
+    ):
         trace_id = self._get_trace_id(run_id)
         self._pop_parent_of_run(run_id)
         run = self._pop_run_metadata(run_id)
         if not run:
             return
         if isinstance(run, GenerationMetadata):
-            log.warning(f"Run {run_id} is a generation, but attempted to be captured as a trace or span.")
+            log.warning(
+                f"Run {run_id} is a generation, but attempted to be captured as a trace or span."
+            )
             return
         self._capture_trace_or_span(
             trace_id,
@@ -446,10 +486,13 @@ class CallbackHandler(BaseCallbackHandler):
         event_name = "$ai_trace" if parent_run_id is None else "$ai_span"
         event_properties = {
             "$ai_trace_id": trace_id,
-            "$ai_input_state": with_privacy_mode(self._client, self._privacy_mode, run.input),
+            "$ai_input_state": with_privacy_mode(
+                self._ph_client, self._privacy_mode, sanitize_langchain(run.input)
+            ),
             "$ai_latency": run.latency,
             "$ai_span_name": run.name,
             "$ai_span_id": run_id,
+            "$ai_framework": "langchain",
         }
         if parent_run_id is not None:
             event_properties["$ai_parent_id"] = parent_run_id
@@ -460,12 +503,14 @@ class CallbackHandler(BaseCallbackHandler):
             event_properties["$ai_error"] = _stringify_exception(outputs)
             event_properties["$ai_is_error"] = True
         elif outputs is not None:
-            event_properties["$ai_output_state"] = with_privacy_mode(self._client, self._privacy_mode, outputs)
+            event_properties["$ai_output_state"] = with_privacy_mode(
+                self._ph_client, self._privacy_mode, outputs
+            )
 
         if self._distinct_id is None:
             event_properties["$process_person_profile"] = False
 
-        self._client.capture(
+        self._ph_client.capture(
             distinct_id=self._distinct_id or run_id,
             event=event_name,
             properties=event_properties,
@@ -484,7 +529,9 @@ class CallbackHandler(BaseCallbackHandler):
         if not run:
             return
         if not isinstance(run, GenerationMetadata):
-            log.warning(f"Run {run_id} is not a generation, but attempted to be captured as a generation.")
+            log.warning(
+                f"Run {run_id} is not a generation, but attempted to be captured as a generation."
+            )
             return
         self._capture_generation(
             trace_id,
@@ -510,17 +557,17 @@ class CallbackHandler(BaseCallbackHandler):
             "$ai_provider": run.provider,
             "$ai_model": run.model,
             "$ai_model_parameters": run.model_params,
-            "$ai_input": with_privacy_mode(self._client, self._privacy_mode, run.input),
+            "$ai_input": with_privacy_mode(
+                self._ph_client, self._privacy_mode, sanitize_langchain(run.input)
+            ),
             "$ai_http_status": 200,
             "$ai_latency": run.latency,
             "$ai_base_url": run.base_url,
+            "$ai_framework": "langchain",
         }
+
         if run.tools:
-            event_properties["$ai_tools"] = with_privacy_mode(
-                self._client,
-                self._privacy_mode,
-                run.tools,
-            )
+            event_properties["$ai_tools"] = run.tools
 
         if isinstance(output, BaseException):
             event_properties["$ai_http_status"] = _get_http_status(output)
@@ -528,9 +575,14 @@ class CallbackHandler(BaseCallbackHandler):
             event_properties["$ai_is_error"] = True
         else:
             # Add usage
-            input_tokens, output_tokens = _parse_usage(output)
-            event_properties["$ai_input_tokens"] = input_tokens
-            event_properties["$ai_output_tokens"] = output_tokens
+            usage = _parse_usage(output)
+            event_properties["$ai_input_tokens"] = usage.input_tokens
+            event_properties["$ai_output_tokens"] = usage.output_tokens
+            event_properties["$ai_cache_creation_input_tokens"] = (
+                usage.cache_write_tokens
+            )
+            event_properties["$ai_cache_read_input_tokens"] = usage.cache_read_tokens
+            event_properties["$ai_reasoning_tokens"] = usage.reasoning_tokens
 
             # Generation results
             generation_result = output.generations[-1]
@@ -540,8 +592,13 @@ class CallbackHandler(BaseCallbackHandler):
                     for generation in generation_result
                 ]
             else:
-                completions = [_extract_raw_esponse(generation) for generation in generation_result]
-            event_properties["$ai_output_choices"] = with_privacy_mode(self._client, self._privacy_mode, completions)
+                completions = [
+                    _extract_raw_response(generation)
+                    for generation in generation_result
+                ]
+            event_properties["$ai_output_choices"] = with_privacy_mode(
+                self._ph_client, self._privacy_mode, completions
+            )
 
         if self._properties:
             event_properties.update(self._properties)
@@ -549,7 +606,7 @@ class CallbackHandler(BaseCallbackHandler):
         if self._distinct_id is None:
             event_properties["$process_person_profile"] = False
 
-        self._client.capture(
+        self._ph_client.capture(
             distinct_id=self._distinct_id or trace_id,
             event="$ai_generation",
             properties=event_properties,
@@ -568,7 +625,7 @@ class CallbackHandler(BaseCallbackHandler):
         )
 
 
-def _extract_raw_esponse(last_response):
+def _extract_raw_response(last_response):
     """Extract the response from the last response of the LLM call."""
     # We return the text of the response if not empty
     if last_response.text is not None and last_response.text.strip() != "":
@@ -581,12 +638,35 @@ def _extract_raw_esponse(last_response):
         return ""
 
 
-def _convert_message_to_dict(message: BaseMessage) -> Dict[str, Any]:
+def _convert_lc_tool_calls_to_oai(
+    tool_calls: list[ToolCall],
+) -> list[dict[str, Any]]:
+    try:
+        return [
+            {
+                "type": "function",
+                "id": tool_call["id"],
+                "function": {
+                    "name": tool_call["name"],
+                    "arguments": json.dumps(tool_call["args"]),
+                },
+            }
+            for tool_call in tool_calls
+        ]
+    except KeyError:
+        return tool_calls
+
+
+def _convert_message_to_dict(message: BaseMessage) -> dict[str, Any]:
     # assistant message
     if isinstance(message, HumanMessage):
         message_dict = {"role": "user", "content": message.content}
     elif isinstance(message, AIMessage):
         message_dict = {"role": "assistant", "content": message.content}
+        if message.tool_calls:
+            message_dict["tool_calls"] = _convert_lc_tool_calls_to_oai(
+                message.tool_calls
+            )
     elif isinstance(message, SystemMessage):
         message_dict = {"role": "system", "content": message.content}
     elif isinstance(message, ToolMessage):
@@ -599,12 +679,24 @@ def _convert_message_to_dict(message: BaseMessage) -> Dict[str, Any]:
     if message.additional_kwargs:
         message_dict.update(message.additional_kwargs)
 
+    if "content" in message_dict and not message_dict["content"]:
+        message_dict["content"] = ""
+
     return message_dict
 
 
+@dataclass
+class ModelUsage:
+    input_tokens: Optional[int]
+    output_tokens: Optional[int]
+    cache_write_tokens: Optional[int]
+    cache_read_tokens: Optional[int]
+    reasoning_tokens: Optional[int]
+
+
 def _parse_usage_model(
-    usage: Union[BaseModel, Dict],
-) -> Tuple[Union[int, None], Union[int, None]]:
+    usage: Union[BaseModel, dict],
+) -> ModelUsage:
     if isinstance(usage, BaseModel):
         usage = usage.__dict__
 
@@ -612,15 +704,23 @@ def _parse_usage_model(
         # https://pypi.org/project/langchain-anthropic/ (works also for Bedrock-Anthropic)
         ("input_tokens", "input"),
         ("output_tokens", "output"),
+        ("cache_creation_input_tokens", "cache_write"),
+        ("cache_read_input_tokens", "cache_read"),
         # https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/get-token-count
         ("prompt_token_count", "input"),
         ("candidates_token_count", "output"),
+        ("cached_content_token_count", "cache_read"),
+        ("thoughts_token_count", "reasoning"),
         # Bedrock: https://docs.aws.amazon.com/bedrock/latest/userguide/monitoring-cw.html#runtime-cloudwatch-metrics
         ("inputTokenCount", "input"),
         ("outputTokenCount", "output"),
+        ("cacheCreationInputTokenCount", "cache_write"),
+        ("cacheReadInputTokenCount", "cache_read"),
         # Bedrock Anthropic
         ("prompt_tokens", "input"),
         ("completion_tokens", "output"),
+        ("cache_creation_input_tokens", "cache_write"),
+        ("cache_read_input_tokens", "cache_read"),
         # langchain-ibm https://pypi.org/project/langchain-ibm/
         ("input_token_count", "input"),
         ("generated_token_count", "output"),
@@ -631,18 +731,59 @@ def _parse_usage_model(
         if model_key in usage:
             captured_count = usage[model_key]
             final_count = (
-                sum(captured_count) if isinstance(captured_count, list) else captured_count
+                sum(captured_count)
+                if isinstance(captured_count, list)
+                else captured_count
             )  # For Bedrock, the token count is a list when streamed
 
             parsed_usage[type_key] = final_count
 
-    return parsed_usage.get("input"), parsed_usage.get("output")
+    # Caching (OpenAI & langchain 0.3.9+)
+    if "input_token_details" in usage and isinstance(
+        usage["input_token_details"], dict
+    ):
+        parsed_usage["cache_write"] = usage["input_token_details"].get("cache_creation")
+        parsed_usage["cache_read"] = usage["input_token_details"].get("cache_read")
+
+    # Reasoning (OpenAI & langchain 0.3.9+)
+    if "output_token_details" in usage and isinstance(
+        usage["output_token_details"], dict
+    ):
+        parsed_usage["reasoning"] = usage["output_token_details"].get("reasoning")
+
+    field_mapping = {
+        "input": "input_tokens",
+        "output": "output_tokens",
+        "cache_write": "cache_write_tokens",
+        "cache_read": "cache_read_tokens",
+        "reasoning": "reasoning_tokens",
+    }
+    normalized_usage = ModelUsage(
+        **{
+            dataclass_key: parsed_usage.get(mapped_key) or 0
+            for mapped_key, dataclass_key in field_mapping.items()
+        },
+    )
+    # In LangChain, input_tokens is the sum of input and cache read tokens.
+    # Our cost calculation expects them to be separate, for Anthropic.
+    if normalized_usage.input_tokens and normalized_usage.cache_read_tokens:
+        normalized_usage.input_tokens = max(
+            normalized_usage.input_tokens - normalized_usage.cache_read_tokens, 0
+        )
+    return normalized_usage
 
 
-def _parse_usage(response: LLMResult):
+def _parse_usage(response: LLMResult) -> ModelUsage:
     # langchain-anthropic uses the usage field
     llm_usage_keys = ["token_usage", "usage"]
-    llm_usage: Tuple[Union[int, None], Union[int, None]] = (None, None)
+    llm_usage: ModelUsage = ModelUsage(
+        input_tokens=None,
+        output_tokens=None,
+        cache_write_tokens=None,
+        cache_read_tokens=None,
+        reasoning_tokens=None,
+    )
+
     if response.llm_output is not None:
         for key in llm_usage_keys:
             if response.llm_output.get(key):
@@ -656,8 +797,12 @@ def _parse_usage(response: LLMResult):
                 break
 
             for generation_chunk in generation:
-                if generation_chunk.generation_info and ("usage_metadata" in generation_chunk.generation_info):
-                    llm_usage = _parse_usage_model(generation_chunk.generation_info["usage_metadata"])
+                if generation_chunk.generation_info and (
+                    "usage_metadata" in generation_chunk.generation_info
+                ):
+                    llm_usage = _parse_usage_model(
+                        generation_chunk.generation_info["usage_metadata"]
+                    )
                     break
 
                 message_chunk = getattr(generation_chunk, "message", {})
@@ -669,13 +814,19 @@ def _parse_usage(response: LLMResult):
                     else None
                 )
                 bedrock_titan_usage = (
-                    response_metadata.get("amazon-bedrock-invocationMetrics", None)  # for Bedrock-Titan
+                    response_metadata.get(
+                        "amazon-bedrock-invocationMetrics", None
+                    )  # for Bedrock-Titan
                     if isinstance(response_metadata, dict)
                     else None
                 )
-                ollama_usage = getattr(message_chunk, "usage_metadata", None)  # for Ollama
+                ollama_usage = getattr(
+                    message_chunk, "usage_metadata", None
+                )  # for Ollama
 
-                chunk_usage = bedrock_anthropic_usage or bedrock_titan_usage or ollama_usage
+                chunk_usage = (
+                    bedrock_anthropic_usage or bedrock_titan_usage or ollama_usage
+                )
                 if chunk_usage:
                     llm_usage = _parse_usage_model(chunk_usage)
                     break
@@ -691,7 +842,9 @@ def _get_http_status(error: BaseException) -> int:
     return status_code
 
 
-def _get_langchain_run_name(serialized: Optional[Dict[str, Any]], **kwargs: Any) -> Optional[str]:
+def _get_langchain_run_name(
+    serialized: Optional[Dict[str, Any]], **kwargs: Any
+) -> Optional[str]:
     """Retrieve the name of a serialized LangChain runnable.
 
     The prioritization for the determination of the run name is as follows:

posthog/ai/openai/__init__.py

@@ -1,5 +1,20 @@
 from .openai import OpenAI
 from .openai_async import AsyncOpenAI
 from .openai_providers import AsyncAzureOpenAI, AzureOpenAI
+from .openai_converter import (
+    format_openai_response,
+    format_openai_input,
+    extract_openai_tools,
+    format_openai_streaming_content,
+)
 
-__all__ = ["OpenAI", "AsyncOpenAI", "AzureOpenAI", "AsyncAzureOpenAI"]
+__all__ = [
+    "OpenAI",
+    "AsyncOpenAI",
+    "AzureOpenAI",
+    "AsyncAzureOpenAI",
+    "format_openai_response",
+    "format_openai_input",
+    "extract_openai_tools",
+    "format_openai_streaming_content",
+]

posthog/ai/openai/openai.py

@@ -2,13 +2,30 @@ import time
 import uuid
 from typing import Any, Dict, List, Optional
 
+from posthog.ai.types import TokenUsage
+
 try:
     import openai
 except ImportError:
-    raise ModuleNotFoundError("Please install the OpenAI SDK to use this feature: 'pip install openai'")
+    raise ModuleNotFoundError(
+        "Please install the OpenAI SDK to use this feature: 'pip install openai'"
+    )
 
-from posthog.ai.utils import call_llm_and_track_usage, get_model_params, with_privacy_mode
+from posthog.ai.utils import (
+    call_llm_and_track_usage,
+    extract_available_tool_calls,
+    merge_usage_stats,
+    with_privacy_mode,
+)
+from posthog.ai.openai.openai_converter import (
+    extract_openai_usage_from_chunk,
+    extract_openai_content_from_chunk,
+    extract_openai_tool_calls_from_chunk,
+    accumulate_openai_tool_calls,
+)
+from posthog.ai.sanitization import sanitize_openai, sanitize_openai_response
 from posthog.client import Client as PostHogClient
+from posthog import setup
 
 
 class OpenAI(openai.OpenAI):
@@ -18,16 +35,16 @@ class OpenAI(openai.OpenAI):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             api_key: OpenAI API key.
-            posthog_client: If provided, events will be captured via this client instead
-                            of the global posthog.
+            posthog_client: If provided, events will be captured via this client instead of the global `posthog`.
             **openai_config: Any additional keyword args to set on openai (e.g. organization="xxx").
         """
+
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
 
         # Store original objects after parent initialization (only if they exist)
         self._original_chat = getattr(self, "chat", None)
@@ -105,7 +122,7 @@ class WrappedResponses:
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {}
+        usage_stats: TokenUsage = TokenUsage()
         final_content = []
         response = self._original.create(**kwargs)
 
@@ -115,31 +132,17 @@ class WrappedResponses:
 
             try:
                 for chunk in response:
-                    if hasattr(chunk, "type") and chunk.type == "response.completed":
-                        res = chunk.response
-                        if res.output and len(res.output) > 0:
-                            final_content.append(res.output[0])
+                    # Extract usage stats from chunk
+                    chunk_usage = extract_openai_usage_from_chunk(chunk, "responses")
 
-                    if hasattr(chunk, "usage") and chunk.usage:
-                        usage_stats = {
-                            k: getattr(chunk.usage, k, 0)
-                            for k in [
-                                "input_tokens",
-                                "output_tokens",
-                                "total_tokens",
-                            ]
-                        }
+                    if chunk_usage:
+                        merge_usage_stats(usage_stats, chunk_usage)
 
-                        # Add support for cached tokens
-                        if hasattr(chunk.usage, "output_tokens_details") and hasattr(
-                            chunk.usage.output_tokens_details, "reasoning_tokens"
-                        ):
-                            usage_stats["reasoning_tokens"] = chunk.usage.output_tokens_details.reasoning_tokens
+                    # Extract content from chunk
+                    content = extract_openai_content_from_chunk(chunk, "responses")
 
-                        if hasattr(chunk.usage, "input_tokens_details") and hasattr(
-                            chunk.usage.input_tokens_details, "cached_tokens"
-                        ):
-                            usage_stats["cache_read_input_tokens"] = chunk.usage.input_tokens_details.cached_tokens
+                    if content is not None:
+                        final_content.append(content)
 
                     yield chunk
 
@@ -157,6 +160,7 @@ class WrappedResponses:
                     usage_stats,
                     latency,
                     output,
+                    None,  # Responses API doesn't have tools
                 )
 
         return generator()
@@ -169,52 +173,76 @@ class WrappedResponses:
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
         output: Any,
-        tool_calls: Optional[List[Dict[str, Any]]] = None,
+        available_tool_calls: Optional[List[Dict[str, Any]]] = None,
     ):
-        if posthog_trace_id is None:
-            posthog_trace_id = str(uuid.uuid4())
+        from posthog.ai.types import StreamingEventData
+        from posthog.ai.openai.openai_converter import (
+            format_openai_streaming_input,
+            format_openai_streaming_output,
+        )
+        from posthog.ai.utils import capture_streaming_event
 
-        event_properties = {
-            "$ai_provider": "openai",
-            "$ai_model": kwargs.get("model"),
-            "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("input")),
-            "$ai_output_choices": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                output,
-            ),
-            "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("output_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
-            "$ai_reasoning_tokens": usage_stats.get("reasoning_tokens", 0),
-            "$ai_latency": latency,
-            "$ai_trace_id": posthog_trace_id,
-            "$ai_base_url": str(self._client.base_url),
-            **(posthog_properties or {}),
-        }
+        # Prepare standardized event data
+        formatted_input = format_openai_streaming_input(kwargs, "responses")
+        sanitized_input = sanitize_openai_response(formatted_input)
 
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                tool_calls,
-            )
+        event_data = StreamingEventData(
+            provider="openai",
+            model=kwargs.get("model", "unknown"),
+            base_url=str(self._client.base_url),
+            kwargs=kwargs,
+            formatted_input=sanitized_input,
+            formatted_output=format_openai_streaming_output(output, "responses"),
+            usage_stats=usage_stats,
+            latency=latency,
+            distinct_id=posthog_distinct_id,
+            trace_id=posthog_trace_id,
+            properties=posthog_properties,
+            privacy_mode=posthog_privacy_mode,
+            groups=posthog_groups,
+        )
 
-        if posthog_distinct_id is None:
-            event_properties["$process_person_profile"] = False
+        # Use the common capture function
+        capture_streaming_event(self._client._ph_client, event_data)
 
-        if hasattr(self._client._ph_client, "capture"):
-            self._client._ph_client.capture(
-                distinct_id=posthog_distinct_id or posthog_trace_id,
-                event="$ai_generation",
-                properties=event_properties,
-                groups=posthog_groups,
-            )
+    def parse(
+        self,
+        posthog_distinct_id: Optional[str] = None,
+        posthog_trace_id: Optional[str] = None,
+        posthog_properties: Optional[Dict[str, Any]] = None,
+        posthog_privacy_mode: bool = False,
+        posthog_groups: Optional[Dict[str, Any]] = None,
+        **kwargs: Any,
+    ):
+        """
+        Parse structured output using OpenAI's 'responses.parse' method, but also track usage in PostHog.
+
+        Args:
+            posthog_distinct_id: Optional ID to associate with the usage event.
+            posthog_trace_id: Optional trace UUID for linking events.
+            posthog_properties: Optional dictionary of extra properties to include in the event.
+            posthog_privacy_mode: Whether to anonymize the input and output.
+            posthog_groups: Optional dictionary of groups to associate with the event.
+            **kwargs: Any additional parameters for the OpenAI Responses Parse API.
+
+        Returns:
+            The response from OpenAI's responses.parse call.
+        """
+        return call_llm_and_track_usage(
+            posthog_distinct_id,
+            self._client._ph_client,
+            "openai",
+            posthog_trace_id,
+            posthog_properties,
+            posthog_privacy_mode,
+            posthog_groups,
+            self._client.base_url,
+            self._original.parse,
+            **kwargs,
+        )
 
 
 class WrappedChat:
@@ -289,9 +317,9 @@ class WrappedCompletions:
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {}
+        usage_stats: TokenUsage = TokenUsage()
         accumulated_content = []
-        accumulated_tools = {}
+        accumulated_tool_calls: Dict[int, Dict[str, Any]] = {}
         if "stream_options" not in kwargs:
             kwargs["stream_options"] = {}
         kwargs["stream_options"]["include_usage"] = True
@@ -300,56 +328,42 @@ class WrappedCompletions:
         def generator():
             nonlocal usage_stats
             nonlocal accumulated_content  # noqa: F824
-            nonlocal accumulated_tools  # noqa: F824
+            nonlocal accumulated_tool_calls
 
             try:
                 for chunk in response:
-                    if hasattr(chunk, "usage") and chunk.usage:
-                        usage_stats = {
-                            k: getattr(chunk.usage, k, 0)
-                            for k in [
-                                "prompt_tokens",
-                                "completion_tokens",
-                                "total_tokens",
-                            ]
-                        }
+                    # Extract usage stats from chunk
+                    chunk_usage = extract_openai_usage_from_chunk(chunk, "chat")
 
-                        # Add support for cached tokens
-                        if hasattr(chunk.usage, "prompt_tokens_details") and hasattr(
-                            chunk.usage.prompt_tokens_details, "cached_tokens"
-                        ):
-                            usage_stats["cache_read_input_tokens"] = chunk.usage.prompt_tokens_details.cached_tokens
+                    if chunk_usage:
+                        merge_usage_stats(usage_stats, chunk_usage)
 
-                        if hasattr(chunk.usage, "output_tokens_details") and hasattr(
-                            chunk.usage.output_tokens_details, "reasoning_tokens"
-                        ):
-                            usage_stats["reasoning_tokens"] = chunk.usage.output_tokens_details.reasoning_tokens
+                    # Extract content from chunk
+                    content = extract_openai_content_from_chunk(chunk, "chat")
 
-                    if hasattr(chunk, "choices") and chunk.choices and len(chunk.choices) > 0:
-                        if chunk.choices[0].delta and chunk.choices[0].delta.content:
-                            content = chunk.choices[0].delta.content
-                            if content:
-                                accumulated_content.append(content)
+                    if content is not None:
+                        accumulated_content.append(content)
 
-                        # Process tool calls
-                        tool_calls = getattr(chunk.choices[0].delta, "tool_calls", None)
-                        if tool_calls:
-                            for tool_call in tool_calls:
-                                index = tool_call.index
-                                if index not in accumulated_tools:
-                                    accumulated_tools[index] = tool_call
-                                else:
-                                    # Append arguments for existing tool calls
-                                    if hasattr(tool_call, "function") and hasattr(tool_call.function, "arguments"):
-                                        accumulated_tools[index].function.arguments += tool_call.function.arguments
+                    # Extract and accumulate tool calls from chunk
+                    chunk_tool_calls = extract_openai_tool_calls_from_chunk(chunk)
+                    if chunk_tool_calls:
+                        accumulate_openai_tool_calls(
+                            accumulated_tool_calls, chunk_tool_calls
+                        )
 
                     yield chunk
 
             finally:
                 end_time = time.time()
                 latency = end_time - start_time
-                output = "".join(accumulated_content)
-                tools = list(accumulated_tools.values()) if accumulated_tools else None
+
+                # Convert accumulated tool calls dict to list
+                tool_calls_list = (
+                    list(accumulated_tool_calls.values())
+                    if accumulated_tool_calls
+                    else None
+                )
+
                 self._capture_streaming_event(
                     posthog_distinct_id,
                     posthog_trace_id,
@@ -359,8 +373,9 @@ class WrappedCompletions:
                     kwargs,
                     usage_stats,
                     latency,
-                    output,
-                    tools,
+                    accumulated_content,
+                    tool_calls_list,
+                    extract_available_tool_calls("openai", kwargs),
                 )
 
         return generator()
@@ -373,52 +388,41 @@ class WrappedCompletions:
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
         output: Any,
         tool_calls: Optional[List[Dict[str, Any]]] = None,
+        available_tool_calls: Optional[List[Dict[str, Any]]] = None,
     ):
-        if posthog_trace_id is None:
-            posthog_trace_id = str(uuid.uuid4())
+        from posthog.ai.types import StreamingEventData
+        from posthog.ai.openai.openai_converter import (
+            format_openai_streaming_input,
+            format_openai_streaming_output,
+        )
+        from posthog.ai.utils import capture_streaming_event
 
-        event_properties = {
-            "$ai_provider": "openai",
-            "$ai_model": kwargs.get("model"),
-            "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("messages")),
-            "$ai_output_choices": with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                [{"content": output, "role": "assistant"}],
-            ),
-            "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("prompt_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("completion_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
-            "$ai_reasoning_tokens": usage_stats.get("reasoning_tokens", 0),
-            "$ai_latency": latency,
-            "$ai_trace_id": posthog_trace_id,
-            "$ai_base_url": str(self._client.base_url),
-            **(posthog_properties or {}),
-        }
+        # Prepare standardized event data
+        formatted_input = format_openai_streaming_input(kwargs, "chat")
+        sanitized_input = sanitize_openai(formatted_input)
 
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                tool_calls,
-            )
+        event_data = StreamingEventData(
+            provider="openai",
+            model=kwargs.get("model", "unknown"),
+            base_url=str(self._client.base_url),
+            kwargs=kwargs,
+            formatted_input=sanitized_input,
+            formatted_output=format_openai_streaming_output(output, "chat", tool_calls),
+            usage_stats=usage_stats,
+            latency=latency,
+            distinct_id=posthog_distinct_id,
+            trace_id=posthog_trace_id,
+            properties=posthog_properties,
+            privacy_mode=posthog_privacy_mode,
+            groups=posthog_groups,
+        )
 
-        if posthog_distinct_id is None:
-            event_properties["$process_person_profile"] = False
-
-        if hasattr(self._client._ph_client, "capture"):
-            self._client._ph_client.capture(
-                distinct_id=posthog_distinct_id or posthog_trace_id,
-                event="$ai_generation",
-                properties=event_properties,
-                groups=posthog_groups,
-            )
+        # Use the common capture function
+        capture_streaming_event(self._client._ph_client, event_data)
 
 
 class WrappedEmbeddings:
@@ -455,6 +459,7 @@ class WrappedEmbeddings:
         Returns:
             The response from OpenAI's embeddings.create call.
         """
+
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
@@ -476,7 +481,11 @@ class WrappedEmbeddings:
         event_properties = {
             "$ai_provider": "openai",
             "$ai_model": kwargs.get("model"),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("input")),
+            "$ai_input": with_privacy_mode(
+                self._client._ph_client,
+                posthog_privacy_mode,
+                sanitize_openai_response(kwargs.get("input")),
+            ),
             "$ai_http_status": 200,
             "$ai_input_tokens": usage_stats.get("prompt_tokens", 0),
             "$ai_latency": latency,

posthog/ai/openai/openai_async.py

@@ -2,12 +2,31 @@ import time
 import uuid
 from typing import Any, Dict, List, Optional
 
+from posthog.ai.types import TokenUsage
+
 try:
     import openai
 except ImportError:
-    raise ModuleNotFoundError("Please install the OpenAI SDK to use this feature: 'pip install openai'")
+    raise ModuleNotFoundError(
+        "Please install the OpenAI SDK to use this feature: 'pip install openai'"
+    )
 
-from posthog.ai.utils import call_llm_and_track_usage_async, get_model_params, with_privacy_mode
+from posthog import setup
+from posthog.ai.utils import (
+    call_llm_and_track_usage_async,
+    extract_available_tool_calls,
+    get_model_params,
+    merge_usage_stats,
+    with_privacy_mode,
+)
+from posthog.ai.openai.openai_converter import (
+    extract_openai_usage_from_chunk,
+    extract_openai_content_from_chunk,
+    extract_openai_tool_calls_from_chunk,
+    accumulate_openai_tool_calls,
+    format_openai_streaming_output,
+)
+from posthog.ai.sanitization import sanitize_openai, sanitize_openai_response
 from posthog.client import Client as PostHogClient
 
 
@@ -18,7 +37,7 @@ class AsyncOpenAI(openai.AsyncOpenAI):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             api_key: OpenAI API key.
@@ -26,8 +45,9 @@ class AsyncOpenAI(openai.AsyncOpenAI):
                             of the global posthog.
             **openai_config: Any additional keyword args to set on openai (e.g. organization="xxx").
         """
+
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
 
         # Store original objects after parent initialization (only if they exist)
         self._original_chat = getattr(self, "chat", None)
@@ -58,6 +78,7 @@ class WrappedResponses:
 
     def __getattr__(self, name):
         """Fallback to original responses object for any methods we don't explicitly handle."""
+
         return getattr(self._original, name)
 
     async def create(
@@ -105,7 +126,7 @@ class WrappedResponses:
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {}
+        usage_stats: TokenUsage = TokenUsage()
         final_content = []
         response = await self._original.create(**kwargs)
 
@@ -115,31 +136,17 @@ class WrappedResponses:
 
             try:
                 async for chunk in response:
-                    if hasattr(chunk, "type") and chunk.type == "response.completed":
-                        res = chunk.response
-                        if res.output and len(res.output) > 0:
-                            final_content.append(res.output[0])
+                    # Extract usage stats from chunk
+                    chunk_usage = extract_openai_usage_from_chunk(chunk, "responses")
 
-                    if hasattr(chunk, "usage") and chunk.usage:
-                        usage_stats = {
-                            k: getattr(chunk.usage, k, 0)
-                            for k in [
-                                "input_tokens",
-                                "output_tokens",
-                                "total_tokens",
-                            ]
-                        }
+                    if chunk_usage:
+                        merge_usage_stats(usage_stats, chunk_usage)
 
-                        # Add support for cached tokens
-                        if hasattr(chunk.usage, "output_tokens_details") and hasattr(
-                            chunk.usage.output_tokens_details, "reasoning_tokens"
-                        ):
-                            usage_stats["reasoning_tokens"] = chunk.usage.output_tokens_details.reasoning_tokens
+                    # Extract content from chunk
+                    content = extract_openai_content_from_chunk(chunk, "responses")
 
-                        if hasattr(chunk.usage, "input_tokens_details") and hasattr(
-                            chunk.usage.input_tokens_details, "cached_tokens"
-                        ):
-                            usage_stats["cache_read_input_tokens"] = chunk.usage.input_tokens_details.cached_tokens
+                    if content is not None:
+                        final_content.append(content)
 
                     yield chunk
 
@@ -147,6 +154,7 @@ class WrappedResponses:
                 end_time = time.time()
                 latency = end_time - start_time
                 output = final_content
+
                 await self._capture_streaming_event(
                     posthog_distinct_id,
                     posthog_trace_id,
@@ -157,6 +165,7 @@ class WrappedResponses:
                     usage_stats,
                     latency,
                     output,
+                    extract_available_tool_calls("openai", kwargs),
                 )
 
         return async_generator()
@@ -169,10 +178,10 @@ class WrappedResponses:
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
         output: Any,
-        tool_calls: Optional[List[Dict[str, Any]]] = None,
+        available_tool_calls: Optional[List[Dict[str, Any]]] = None,
     ):
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
@@ -181,16 +190,22 @@ class WrappedResponses:
             "$ai_provider": "openai",
             "$ai_model": kwargs.get("model"),
             "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("input")),
+            "$ai_input": with_privacy_mode(
+                self._client._ph_client,
+                posthog_privacy_mode,
+                sanitize_openai_response(kwargs.get("input")),
+            ),
             "$ai_output_choices": with_privacy_mode(
                 self._client._ph_client,
                 posthog_privacy_mode,
-                output,
+                format_openai_streaming_output(output, "responses"),
             ),
             "$ai_http_status": 200,
             "$ai_input_tokens": usage_stats.get("input_tokens", 0),
             "$ai_output_tokens": usage_stats.get("output_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
+            "$ai_cache_read_input_tokens": usage_stats.get(
+                "cache_read_input_tokens", 0
+            ),
             "$ai_reasoning_tokens": usage_stats.get("reasoning_tokens", 0),
             "$ai_latency": latency,
             "$ai_trace_id": posthog_trace_id,
@@ -198,12 +213,17 @@ class WrappedResponses:
             **(posthog_properties or {}),
         }
 
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                tool_calls,
-            )
+        # Add web search count if present
+        web_search_count = usage_stats.get("web_search_count")
+        if (
+            web_search_count is not None
+            and isinstance(web_search_count, int)
+            and web_search_count > 0
+        ):
+            event_properties["$ai_web_search_count"] = web_search_count
+
+        if available_tool_calls:
+            event_properties["$ai_tools"] = available_tool_calls
 
         if posthog_distinct_id is None:
             event_properties["$process_person_profile"] = False
@@ -216,6 +236,42 @@ class WrappedResponses:
                 groups=posthog_groups,
             )
 
+    async def parse(
+        self,
+        posthog_distinct_id: Optional[str] = None,
+        posthog_trace_id: Optional[str] = None,
+        posthog_properties: Optional[Dict[str, Any]] = None,
+        posthog_privacy_mode: bool = False,
+        posthog_groups: Optional[Dict[str, Any]] = None,
+        **kwargs: Any,
+    ):
+        """
+        Parse structured output using OpenAI's 'responses.parse' method, but also track usage in PostHog.
+
+        Args:
+            posthog_distinct_id: Optional ID to associate with the usage event.
+            posthog_trace_id: Optional trace UUID for linking events.
+            posthog_properties: Optional dictionary of extra properties to include in the event.
+            posthog_privacy_mode: Whether to anonymize the input and output.
+            posthog_groups: Optional dictionary of groups to associate with the event.
+            **kwargs: Any additional parameters for the OpenAI Responses Parse API.
+
+        Returns:
+            The response from OpenAI's responses.parse call.
+        """
+        return await call_llm_and_track_usage_async(
+            posthog_distinct_id,
+            self._client._ph_client,
+            "openai",
+            posthog_trace_id,
+            posthog_properties,
+            posthog_privacy_mode,
+            posthog_groups,
+            self._client.base_url,
+            self._original.parse,
+            **kwargs,
+        )
+
 
 class WrappedChat:
     """Async wrapper for OpenAI chat that tracks usage in PostHog."""
@@ -291,9 +347,9 @@ class WrappedCompletions:
         **kwargs: Any,
     ):
         start_time = time.time()
-        usage_stats: Dict[str, int] = {}
+        usage_stats: TokenUsage = TokenUsage()
         accumulated_content = []
-        accumulated_tools = {}
+        accumulated_tool_calls: Dict[int, Dict[str, Any]] = {}
 
         if "stream_options" not in kwargs:
             kwargs["stream_options"] = {}
@@ -303,56 +359,40 @@ class WrappedCompletions:
         async def async_generator():
             nonlocal usage_stats
             nonlocal accumulated_content  # noqa: F824
-            nonlocal accumulated_tools  # noqa: F824
+            nonlocal accumulated_tool_calls
 
             try:
                 async for chunk in response:
-                    if hasattr(chunk, "usage") and chunk.usage:
-                        usage_stats = {
-                            k: getattr(chunk.usage, k, 0)
-                            for k in [
-                                "prompt_tokens",
-                                "completion_tokens",
-                                "total_tokens",
-                            ]
-                        }
+                    # Extract usage stats from chunk
+                    chunk_usage = extract_openai_usage_from_chunk(chunk, "chat")
+                    if chunk_usage:
+                        merge_usage_stats(usage_stats, chunk_usage)
 
-                        # Add support for cached tokens
-                        if hasattr(chunk.usage, "prompt_tokens_details") and hasattr(
-                            chunk.usage.prompt_tokens_details, "cached_tokens"
-                        ):
-                            usage_stats["cache_read_input_tokens"] = chunk.usage.prompt_tokens_details.cached_tokens
+                    # Extract content from chunk
+                    content = extract_openai_content_from_chunk(chunk, "chat")
+                    if content is not None:
+                        accumulated_content.append(content)
 
-                        if hasattr(chunk.usage, "output_tokens_details") and hasattr(
-                            chunk.usage.output_tokens_details, "reasoning_tokens"
-                        ):
-                            usage_stats["reasoning_tokens"] = chunk.usage.output_tokens_details.reasoning_tokens
-
-                    if hasattr(chunk, "choices") and chunk.choices and len(chunk.choices) > 0:
-                        if chunk.choices[0].delta and chunk.choices[0].delta.content:
-                            content = chunk.choices[0].delta.content
-                            if content:
-                                accumulated_content.append(content)
-
-                        # Process tool calls
-                        tool_calls = getattr(chunk.choices[0].delta, "tool_calls", None)
-                        if tool_calls:
-                            for tool_call in tool_calls:
-                                index = tool_call.index
-                                if index not in accumulated_tools:
-                                    accumulated_tools[index] = tool_call
-                                else:
-                                    # Append arguments for existing tool calls
-                                    if hasattr(tool_call, "function") and hasattr(tool_call.function, "arguments"):
-                                        accumulated_tools[index].function.arguments += tool_call.function.arguments
+                    # Extract and accumulate tool calls from chunk
+                    chunk_tool_calls = extract_openai_tool_calls_from_chunk(chunk)
+                    if chunk_tool_calls:
+                        accumulate_openai_tool_calls(
+                            accumulated_tool_calls, chunk_tool_calls
+                        )
 
                     yield chunk
 
             finally:
                 end_time = time.time()
                 latency = end_time - start_time
-                output = "".join(accumulated_content)
-                tools = list(accumulated_tools.values()) if accumulated_tools else None
+
+                # Convert accumulated tool calls dict to list
+                tool_calls_list = (
+                    list(accumulated_tool_calls.values())
+                    if accumulated_tool_calls
+                    else None
+                )
+
                 await self._capture_streaming_event(
                     posthog_distinct_id,
                     posthog_trace_id,
@@ -362,8 +402,9 @@ class WrappedCompletions:
                     kwargs,
                     usage_stats,
                     latency,
-                    output,
-                    tools,
+                    accumulated_content,
+                    tool_calls_list,
+                    extract_available_tool_calls("openai", kwargs),
                 )
 
         return async_generator()
@@ -376,10 +417,11 @@ class WrappedCompletions:
         posthog_privacy_mode: bool,
         posthog_groups: Optional[Dict[str, Any]],
         kwargs: Dict[str, Any],
-        usage_stats: Dict[str, int],
+        usage_stats: TokenUsage,
         latency: float,
         output: Any,
         tool_calls: Optional[List[Dict[str, Any]]] = None,
+        available_tool_calls: Optional[List[Dict[str, Any]]] = None,
     ):
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
@@ -388,16 +430,22 @@ class WrappedCompletions:
             "$ai_provider": "openai",
             "$ai_model": kwargs.get("model"),
             "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("messages")),
+            "$ai_input": with_privacy_mode(
+                self._client._ph_client,
+                posthog_privacy_mode,
+                sanitize_openai(kwargs.get("messages")),
+            ),
             "$ai_output_choices": with_privacy_mode(
                 self._client._ph_client,
                 posthog_privacy_mode,
-                [{"content": output, "role": "assistant"}],
+                format_openai_streaming_output(output, "chat", tool_calls),
             ),
             "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("prompt_tokens", 0),
-            "$ai_output_tokens": usage_stats.get("completion_tokens", 0),
-            "$ai_cache_read_input_tokens": usage_stats.get("cache_read_input_tokens", 0),
+            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
+            "$ai_output_tokens": usage_stats.get("output_tokens", 0),
+            "$ai_cache_read_input_tokens": usage_stats.get(
+                "cache_read_input_tokens", 0
+            ),
             "$ai_reasoning_tokens": usage_stats.get("reasoning_tokens", 0),
             "$ai_latency": latency,
             "$ai_trace_id": posthog_trace_id,
@@ -405,12 +453,18 @@ class WrappedCompletions:
             **(posthog_properties or {}),
         }
 
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(
-                self._client._ph_client,
-                posthog_privacy_mode,
-                tool_calls,
-            )
+        # Add web search count if present
+        web_search_count = usage_stats.get("web_search_count")
+
+        if (
+            web_search_count is not None
+            and isinstance(web_search_count, int)
+            and web_search_count > 0
+        ):
+            event_properties["$ai_web_search_count"] = web_search_count
+
+        if available_tool_calls:
+            event_properties["$ai_tools"] = available_tool_calls
 
         if posthog_distinct_id is None:
             event_properties["$process_person_profile"] = False
@@ -433,6 +487,7 @@ class WrappedEmbeddings:
 
     def __getattr__(self, name):
         """Fallback to original embeddings object for any methods we don't explicitly handle."""
+
         return getattr(self._original, name)
 
     async def create(
@@ -458,6 +513,7 @@ class WrappedEmbeddings:
         Returns:
             The response from OpenAI's embeddings.create call.
         """
+
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
@@ -466,12 +522,13 @@ class WrappedEmbeddings:
         end_time = time.time()
 
         # Extract usage statistics if available
-        usage_stats = {}
+        usage_stats: TokenUsage = TokenUsage()
+
         if hasattr(response, "usage") and response.usage:
-            usage_stats = {
-                "prompt_tokens": getattr(response.usage, "prompt_tokens", 0),
-                "total_tokens": getattr(response.usage, "total_tokens", 0),
-            }
+            usage_stats = TokenUsage(
+                input_tokens=getattr(response.usage, "prompt_tokens", 0),
+                output_tokens=getattr(response.usage, "completion_tokens", 0),
+            )
 
         latency = end_time - start_time
 
@@ -479,9 +536,13 @@ class WrappedEmbeddings:
         event_properties = {
             "$ai_provider": "openai",
             "$ai_model": kwargs.get("model"),
-            "$ai_input": with_privacy_mode(self._client._ph_client, posthog_privacy_mode, kwargs.get("input")),
+            "$ai_input": with_privacy_mode(
+                self._client._ph_client,
+                posthog_privacy_mode,
+                sanitize_openai_response(kwargs.get("input")),
+            ),
             "$ai_http_status": 200,
-            "$ai_input_tokens": usage_stats.get("prompt_tokens", 0),
+            "$ai_input_tokens": usage_stats.get("input_tokens", 0),
             "$ai_latency": latency,
             "$ai_trace_id": posthog_trace_id,
             "$ai_base_url": str(self._client.base_url),
@@ -512,6 +573,7 @@ class WrappedBeta:
 
     def __getattr__(self, name):
         """Fallback to original beta object for any methods we don't explicitly handle."""
+
         return getattr(self._original, name)
 
     @property
@@ -528,6 +590,7 @@ class WrappedBetaChat:
 
     def __getattr__(self, name):
         """Fallback to original beta chat object for any methods we don't explicitly handle."""
+
         return getattr(self._original, name)
 
     @property
@@ -544,6 +607,7 @@ class WrappedBetaCompletions:
 
     def __getattr__(self, name):
         """Fallback to original beta completions object for any methods we don't explicitly handle."""
+
         return getattr(self._original, name)
 
     async def parse(

posthog/ai/openai/openai_converter.py

@@ -0,0 +1,735 @@
+"""
+OpenAI-specific conversion utilities.
+
+This module handles the conversion of OpenAI API responses and inputs
+into standardized formats for PostHog tracking. It supports both
+Chat Completions API and Responses API formats.
+"""
+
+from typing import Any, Dict, List, Optional
+
+from posthog.ai.types import (
+    FormattedContentItem,
+    FormattedFunctionCall,
+    FormattedImageContent,
+    FormattedMessage,
+    FormattedTextContent,
+    TokenUsage,
+)
+
+
+def format_openai_response(response: Any) -> List[FormattedMessage]:
+    """
+    Format an OpenAI response into standardized message format.
+
+    Handles both Chat Completions API and Responses API formats.
+
+    Args:
+        response: The response object from OpenAI API
+
+    Returns:
+        List of formatted messages with role and content
+    """
+
+    output: List[FormattedMessage] = []
+
+    if response is None:
+        return output
+
+    # Handle Chat Completions response format
+    if hasattr(response, "choices"):
+        content: List[FormattedContentItem] = []
+        role = "assistant"
+
+        for choice in response.choices:
+            if hasattr(choice, "message") and choice.message:
+                if choice.message.role:
+                    role = choice.message.role
+
+                if choice.message.content:
+                    content.append(
+                        {
+                            "type": "text",
+                            "text": choice.message.content,
+                        }
+                    )
+
+                if hasattr(choice.message, "tool_calls") and choice.message.tool_calls:
+                    for tool_call in choice.message.tool_calls:
+                        content.append(
+                            {
+                                "type": "function",
+                                "id": tool_call.id,
+                                "function": {
+                                    "name": tool_call.function.name,
+                                    "arguments": tool_call.function.arguments,
+                                },
+                            }
+                        )
+
+        if content:
+            output.append(
+                {
+                    "role": role,
+                    "content": content,
+                }
+            )
+
+    # Handle Responses API format
+    if hasattr(response, "output"):
+        content = []
+        role = "assistant"
+
+        for item in response.output:
+            if item.type == "message":
+                role = item.role
+
+                if hasattr(item, "content") and isinstance(item.content, list):
+                    for content_item in item.content:
+                        if (
+                            hasattr(content_item, "type")
+                            and content_item.type == "output_text"
+                            and hasattr(content_item, "text")
+                        ):
+                            content.append(
+                                {
+                                    "type": "text",
+                                    "text": content_item.text,
+                                }
+                            )
+
+                        elif hasattr(content_item, "text"):
+                            content.append({"type": "text", "text": content_item.text})
+
+                        elif (
+                            hasattr(content_item, "type")
+                            and content_item.type == "input_image"
+                            and hasattr(content_item, "image_url")
+                        ):
+                            image_content: FormattedImageContent = {
+                                "type": "image",
+                                "image": content_item.image_url,
+                            }
+                            content.append(image_content)
+
+                elif hasattr(item, "content"):
+                    text_content = {"type": "text", "text": str(item.content)}
+                    content.append(text_content)
+
+            elif hasattr(item, "type") and item.type == "function_call":
+                content.append(
+                    {
+                        "type": "function",
+                        "id": getattr(item, "call_id", getattr(item, "id", "")),
+                        "function": {
+                            "name": item.name,
+                            "arguments": getattr(item, "arguments", {}),
+                        },
+                    }
+                )
+
+        if content:
+            output.append(
+                {
+                    "role": role,
+                    "content": content,
+                }
+            )
+
+    return output
+
+
+def format_openai_input(
+    messages: Optional[List[Dict[str, Any]]] = None, input_data: Optional[Any] = None
+) -> List[FormattedMessage]:
+    """
+    Format OpenAI input messages.
+
+    Handles both messages parameter (Chat Completions) and input parameter (Responses API).
+
+    Args:
+        messages: List of message dictionaries for Chat Completions API
+        input_data: Input data for Responses API
+
+    Returns:
+        List of formatted messages
+    """
+
+    formatted_messages: List[FormattedMessage] = []
+
+    # Handle Chat Completions API format
+    if messages is not None:
+        for msg in messages:
+            formatted_messages.append(
+                {
+                    "role": msg.get("role", "user"),
+                    "content": msg.get("content", ""),
+                }
+            )
+
+    # Handle Responses API format
+    if input_data is not None:
+        if isinstance(input_data, list):
+            for item in input_data:
+                role = "user"
+                content = ""
+
+                if isinstance(item, dict):
+                    role = item.get("role", "user")
+                    content = item.get("content", "")
+
+                elif isinstance(item, str):
+                    content = item
+
+                else:
+                    content = str(item)
+
+                formatted_messages.append({"role": role, "content": content})
+
+        elif isinstance(input_data, str):
+            formatted_messages.append({"role": "user", "content": input_data})
+
+        else:
+            formatted_messages.append({"role": "user", "content": str(input_data)})
+
+    return formatted_messages
+
+
+def extract_openai_tools(kwargs: Dict[str, Any]) -> Optional[Any]:
+    """
+    Extract tool definitions from OpenAI API kwargs.
+
+    Args:
+        kwargs: Keyword arguments passed to OpenAI API
+
+    Returns:
+        Tool definitions if present, None otherwise
+    """
+
+    # Check for tools parameter (newer API)
+    if "tools" in kwargs:
+        return kwargs["tools"]
+
+    # Check for functions parameter (older API)
+    if "functions" in kwargs:
+        return kwargs["functions"]
+
+    return None
+
+
+def format_openai_streaming_content(
+    accumulated_content: str, tool_calls: Optional[List[Dict[str, Any]]] = None
+) -> List[FormattedContentItem]:
+    """
+    Format content from OpenAI streaming response.
+
+    Used by streaming handlers to format accumulated content.
+
+    Args:
+        accumulated_content: Accumulated text content from streaming
+        tool_calls: Optional list of tool calls accumulated during streaming
+
+    Returns:
+        List of formatted content items
+    """
+    formatted: List[FormattedContentItem] = []
+
+    # Add text content if present
+    if accumulated_content:
+        text_content: FormattedTextContent = {
+            "type": "text",
+            "text": accumulated_content,
+        }
+        formatted.append(text_content)
+
+    # Add tool calls if present
+    if tool_calls:
+        for tool_call in tool_calls:
+            function_call: FormattedFunctionCall = {
+                "type": "function",
+                "id": tool_call.get("id"),
+                "function": tool_call.get("function", {}),
+            }
+            formatted.append(function_call)
+
+    return formatted
+
+
+def extract_openai_web_search_count(response: Any) -> int:
+    """
+    Extract web search count from OpenAI response.
+
+    Uses a two-tier detection strategy:
+    1. Priority 1 (exact count): Check for output[].type == "web_search_call" (Responses API)
+    2. Priority 2 (binary detection): Check for various web search indicators:
+       - Root-level citations, search_results, or usage.search_context_size (Perplexity)
+       - Annotations with type "url_citation" in choices/output (including delta for streaming)
+
+    Args:
+        response: The response from OpenAI API
+
+    Returns:
+        Number of web search requests (exact count or binary 1/0)
+    """
+
+    # Priority 1: Check for exact count in Responses API output
+    if hasattr(response, "output"):
+        web_search_count = 0
+
+        for item in response.output:
+            if hasattr(item, "type") and item.type == "web_search_call":
+                web_search_count += 1
+
+        web_search_count = max(0, web_search_count)
+
+        if web_search_count > 0:
+            return web_search_count
+
+    # Priority 2: Binary detection (returns 1 or 0)
+
+    # Check root-level indicators (Perplexity)
+    if hasattr(response, "citations"):
+        citations = getattr(response, "citations")
+
+        if citations and len(citations) > 0:
+            return 1
+
+    if hasattr(response, "search_results"):
+        search_results = getattr(response, "search_results")
+
+        if search_results and len(search_results) > 0:
+            return 1
+
+    if hasattr(response, "usage") and hasattr(response.usage, "search_context_size"):
+        if response.usage.search_context_size:
+            return 1
+
+    # Check for url_citation annotations in choices (Chat Completions)
+    if hasattr(response, "choices"):
+        for choice in response.choices:
+            # Check message.annotations (non-streaming or final chunk)
+            if hasattr(choice, "message") and hasattr(choice.message, "annotations"):
+                annotations = choice.message.annotations
+
+                if annotations:
+                    for annotation in annotations:
+                        # Support both dict and object formats
+                        annotation_type = (
+                            annotation.get("type")
+                            if isinstance(annotation, dict)
+                            else getattr(annotation, "type", None)
+                        )
+
+                        if annotation_type == "url_citation":
+                            return 1
+
+            # Check delta.annotations (streaming chunks)
+            if hasattr(choice, "delta") and hasattr(choice.delta, "annotations"):
+                annotations = choice.delta.annotations
+
+                if annotations:
+                    for annotation in annotations:
+                        # Support both dict and object formats
+                        annotation_type = (
+                            annotation.get("type")
+                            if isinstance(annotation, dict)
+                            else getattr(annotation, "type", None)
+                        )
+
+                        if annotation_type == "url_citation":
+                            return 1
+
+    # Check for url_citation annotations in output (Responses API)
+    if hasattr(response, "output"):
+        for item in response.output:
+            if hasattr(item, "content") and isinstance(item.content, list):
+                for content_item in item.content:
+                    if hasattr(content_item, "annotations"):
+                        annotations = content_item.annotations
+
+                        if annotations:
+                            for annotation in annotations:
+                                # Support both dict and object formats
+                                annotation_type = (
+                                    annotation.get("type")
+                                    if isinstance(annotation, dict)
+                                    else getattr(annotation, "type", None)
+                                )
+
+                                if annotation_type == "url_citation":
+                                    return 1
+
+    return 0
+
+
+def extract_openai_usage_from_response(response: Any) -> TokenUsage:
+    """
+    Extract usage statistics from a full OpenAI response (non-streaming).
+    Handles both Chat Completions and Responses API.
+
+    Args:
+        response: The complete response from OpenAI API
+
+    Returns:
+        TokenUsage with standardized usage statistics
+    """
+    if not hasattr(response, "usage"):
+        return TokenUsage(input_tokens=0, output_tokens=0)
+
+    cached_tokens = 0
+    input_tokens = 0
+    output_tokens = 0
+    reasoning_tokens = 0
+
+    # Responses API format
+    if hasattr(response.usage, "input_tokens"):
+        input_tokens = response.usage.input_tokens
+    if hasattr(response.usage, "output_tokens"):
+        output_tokens = response.usage.output_tokens
+    if hasattr(response.usage, "input_tokens_details") and hasattr(
+        response.usage.input_tokens_details, "cached_tokens"
+    ):
+        cached_tokens = response.usage.input_tokens_details.cached_tokens
+    if hasattr(response.usage, "output_tokens_details") and hasattr(
+        response.usage.output_tokens_details, "reasoning_tokens"
+    ):
+        reasoning_tokens = response.usage.output_tokens_details.reasoning_tokens
+
+    # Chat Completions format
+    if hasattr(response.usage, "prompt_tokens"):
+        input_tokens = response.usage.prompt_tokens
+    if hasattr(response.usage, "completion_tokens"):
+        output_tokens = response.usage.completion_tokens
+    if hasattr(response.usage, "prompt_tokens_details") and hasattr(
+        response.usage.prompt_tokens_details, "cached_tokens"
+    ):
+        cached_tokens = response.usage.prompt_tokens_details.cached_tokens
+    if hasattr(response.usage, "completion_tokens_details") and hasattr(
+        response.usage.completion_tokens_details, "reasoning_tokens"
+    ):
+        reasoning_tokens = response.usage.completion_tokens_details.reasoning_tokens
+
+    result = TokenUsage(
+        input_tokens=input_tokens,
+        output_tokens=output_tokens,
+    )
+
+    if cached_tokens > 0:
+        result["cache_read_input_tokens"] = cached_tokens
+    if reasoning_tokens > 0:
+        result["reasoning_tokens"] = reasoning_tokens
+
+    web_search_count = extract_openai_web_search_count(response)
+    if web_search_count > 0:
+        result["web_search_count"] = web_search_count
+
+    return result
+
+
+def extract_openai_usage_from_chunk(
+    chunk: Any, provider_type: str = "chat"
+) -> TokenUsage:
+    """
+    Extract usage statistics from an OpenAI streaming chunk.
+
+    Handles both Chat Completions and Responses API formats.
+
+    Args:
+        chunk: Streaming chunk from OpenAI API
+        provider_type: Either "chat" or "responses" to handle different API formats
+
+    Returns:
+        Dictionary of usage statistics
+    """
+
+    usage: TokenUsage = TokenUsage()
+
+    if provider_type == "chat":
+        # Extract web search count from the chunk before checking for usage
+        # Web search indicators (citations, annotations) can appear on any chunk,
+        # not just those with usage data
+        web_search_count = extract_openai_web_search_count(chunk)
+        if web_search_count > 0:
+            usage["web_search_count"] = web_search_count
+
+        if not hasattr(chunk, "usage") or not chunk.usage:
+            return usage
+
+        # Chat Completions API uses prompt_tokens and completion_tokens
+        # Standardize to input_tokens and output_tokens
+        usage["input_tokens"] = getattr(chunk.usage, "prompt_tokens", 0)
+        usage["output_tokens"] = getattr(chunk.usage, "completion_tokens", 0)
+
+        # Handle cached tokens
+        if hasattr(chunk.usage, "prompt_tokens_details") and hasattr(
+            chunk.usage.prompt_tokens_details, "cached_tokens"
+        ):
+            usage["cache_read_input_tokens"] = (
+                chunk.usage.prompt_tokens_details.cached_tokens
+            )
+
+        # Handle reasoning tokens
+        if hasattr(chunk.usage, "completion_tokens_details") and hasattr(
+            chunk.usage.completion_tokens_details, "reasoning_tokens"
+        ):
+            usage["reasoning_tokens"] = (
+                chunk.usage.completion_tokens_details.reasoning_tokens
+            )
+
+    elif provider_type == "responses":
+        # For Responses API, usage is only in chunk.response.usage for completed events
+        if hasattr(chunk, "type") and chunk.type == "response.completed":
+            if (
+                hasattr(chunk, "response")
+                and hasattr(chunk.response, "usage")
+                and chunk.response.usage
+            ):
+                response_usage = chunk.response.usage
+                usage["input_tokens"] = getattr(response_usage, "input_tokens", 0)
+                usage["output_tokens"] = getattr(response_usage, "output_tokens", 0)
+
+                # Handle cached tokens
+                if hasattr(response_usage, "input_tokens_details") and hasattr(
+                    response_usage.input_tokens_details, "cached_tokens"
+                ):
+                    usage["cache_read_input_tokens"] = (
+                        response_usage.input_tokens_details.cached_tokens
+                    )
+
+                # Handle reasoning tokens
+                if hasattr(response_usage, "output_tokens_details") and hasattr(
+                    response_usage.output_tokens_details, "reasoning_tokens"
+                ):
+                    usage["reasoning_tokens"] = (
+                        response_usage.output_tokens_details.reasoning_tokens
+                    )
+
+                # Extract web search count from the complete response
+                if hasattr(chunk, "response"):
+                    web_search_count = extract_openai_web_search_count(chunk.response)
+                    if web_search_count > 0:
+                        usage["web_search_count"] = web_search_count
+
+    return usage
+
+
+def extract_openai_content_from_chunk(
+    chunk: Any, provider_type: str = "chat"
+) -> Optional[str]:
+    """
+    Extract content from an OpenAI streaming chunk.
+
+    Handles both Chat Completions and Responses API formats.
+
+    Args:
+        chunk: Streaming chunk from OpenAI API
+        provider_type: Either "chat" or "responses" to handle different API formats
+
+    Returns:
+        Text content if present, None otherwise
+    """
+
+    if provider_type == "chat":
+        # Chat Completions API format
+        if (
+            hasattr(chunk, "choices")
+            and chunk.choices
+            and len(chunk.choices) > 0
+            and chunk.choices[0].delta
+            and chunk.choices[0].delta.content
+        ):
+            return chunk.choices[0].delta.content
+
+    elif provider_type == "responses":
+        # Responses API format
+        if hasattr(chunk, "type") and chunk.type == "response.completed":
+            if hasattr(chunk, "response") and chunk.response:
+                res = chunk.response
+                if res.output and len(res.output) > 0:
+                    # Return the full output for responses
+                    return res.output[0]
+
+    return None
+
+
+def extract_openai_tool_calls_from_chunk(chunk: Any) -> Optional[List[Dict[str, Any]]]:
+    """
+    Extract tool calls from an OpenAI streaming chunk.
+
+    Args:
+        chunk: Streaming chunk from OpenAI API
+
+    Returns:
+        List of tool call deltas if present, None otherwise
+    """
+    if (
+        hasattr(chunk, "choices")
+        and chunk.choices
+        and len(chunk.choices) > 0
+        and chunk.choices[0].delta
+        and hasattr(chunk.choices[0].delta, "tool_calls")
+        and chunk.choices[0].delta.tool_calls
+    ):
+        tool_calls = []
+        for tool_call in chunk.choices[0].delta.tool_calls:
+            tc_dict = {
+                "index": getattr(tool_call, "index", None),
+            }
+
+            if hasattr(tool_call, "id") and tool_call.id:
+                tc_dict["id"] = tool_call.id
+
+            if hasattr(tool_call, "type") and tool_call.type:
+                tc_dict["type"] = tool_call.type
+
+            if hasattr(tool_call, "function") and tool_call.function:
+                function_dict = {}
+                if hasattr(tool_call.function, "name") and tool_call.function.name:
+                    function_dict["name"] = tool_call.function.name
+                if (
+                    hasattr(tool_call.function, "arguments")
+                    and tool_call.function.arguments
+                ):
+                    function_dict["arguments"] = tool_call.function.arguments
+                tc_dict["function"] = function_dict
+
+            tool_calls.append(tc_dict)
+        return tool_calls
+
+    return None
+
+
+def accumulate_openai_tool_calls(
+    accumulated_tool_calls: Dict[int, Dict[str, Any]],
+    chunk_tool_calls: List[Dict[str, Any]],
+) -> None:
+    """
+    Accumulate tool calls from streaming chunks.
+
+    OpenAI sends tool calls incrementally:
+    - First chunk has id, type, function.name and partial function.arguments
+    - Subsequent chunks have more function.arguments
+
+    Args:
+        accumulated_tool_calls: Dictionary mapping index to accumulated tool call data
+        chunk_tool_calls: List of tool call deltas from current chunk
+    """
+    for tool_call_delta in chunk_tool_calls:
+        index = tool_call_delta.get("index")
+        if index is None:
+            continue
+
+        # Initialize tool call if first time seeing this index
+        if index not in accumulated_tool_calls:
+            accumulated_tool_calls[index] = {
+                "id": "",
+                "type": "function",
+                "function": {
+                    "name": "",
+                    "arguments": "",
+                },
+            }
+
+        # Update with new data from delta
+        tc = accumulated_tool_calls[index]
+
+        if "id" in tool_call_delta and tool_call_delta["id"]:
+            tc["id"] = tool_call_delta["id"]
+
+        if "type" in tool_call_delta and tool_call_delta["type"]:
+            tc["type"] = tool_call_delta["type"]
+
+        if "function" in tool_call_delta:
+            func_delta = tool_call_delta["function"]
+            if "name" in func_delta and func_delta["name"]:
+                tc["function"]["name"] = func_delta["name"]
+            if "arguments" in func_delta and func_delta["arguments"]:
+                # Arguments are sent incrementally, concatenate them
+                tc["function"]["arguments"] += func_delta["arguments"]
+
+
+def format_openai_streaming_output(
+    accumulated_content: Any,
+    provider_type: str = "chat",
+    tool_calls: Optional[List[Dict[str, Any]]] = None,
+) -> List[FormattedMessage]:
+    """
+    Format the final output from OpenAI streaming.
+
+    Args:
+        accumulated_content: Accumulated content from streaming (string for chat, list for responses)
+        provider_type: Either "chat" or "responses" to handle different API formats
+        tool_calls: Optional list of accumulated tool calls
+
+    Returns:
+        List of formatted messages
+    """
+
+    if provider_type == "chat":
+        content_items: List[FormattedContentItem] = []
+
+        # Add text content if present
+        if isinstance(accumulated_content, str) and accumulated_content:
+            content_items.append({"type": "text", "text": accumulated_content})
+        elif isinstance(accumulated_content, list):
+            # If it's a list of strings, join them
+            text = "".join(str(item) for item in accumulated_content if item)
+            if text:
+                content_items.append({"type": "text", "text": text})
+
+        # Add tool calls if present
+        if tool_calls:
+            for tool_call in tool_calls:
+                if "function" in tool_call:
+                    function_call: FormattedFunctionCall = {
+                        "type": "function",
+                        "id": tool_call.get("id", ""),
+                        "function": tool_call["function"],
+                    }
+                    content_items.append(function_call)
+
+        # Return formatted message with content
+        if content_items:
+            return [{"role": "assistant", "content": content_items}]
+        else:
+            # Empty response
+            return [{"role": "assistant", "content": []}]
+
+    elif provider_type == "responses":
+        # Responses API: accumulated_content is a list of output items
+        if isinstance(accumulated_content, list) and accumulated_content:
+            # The output is already formatted, just return it
+            return accumulated_content
+        elif isinstance(accumulated_content, str):
+            return [
+                {
+                    "role": "assistant",
+                    "content": [{"type": "text", "text": accumulated_content}],
+                }
+            ]
+
+    # Fallback for any other format
+    return [
+        {
+            "role": "assistant",
+            "content": [{"type": "text", "text": str(accumulated_content)}],
+        }
+    ]
+
+
+def format_openai_streaming_input(
+    kwargs: Dict[str, Any], api_type: str = "chat"
+) -> Any:
+    """
+    Format OpenAI streaming input based on API type.
+
+    Args:
+        kwargs: Keyword arguments passed to OpenAI API
+        api_type: Either "chat" or "responses"
+
+    Returns:
+        Formatted input ready for PostHog tracking
+    """
+    from posthog.ai.utils import merge_system_prompt
+
+    return merge_system_prompt(kwargs, "openai")

posthog/ai/openai/openai_providers.py

@@ -1,14 +1,24 @@
 try:
     import openai
 except ImportError:
-    raise ModuleNotFoundError("Please install the Open AI SDK to use this feature: 'pip install openai'")
+    raise ModuleNotFoundError(
+        "Please install the Open AI SDK to use this feature: 'pip install openai'"
+    )
 
-from posthog.ai.openai.openai import WrappedBeta, WrappedChat, WrappedEmbeddings, WrappedResponses
+from posthog.ai.openai.openai import (
+    WrappedBeta,
+    WrappedChat,
+    WrappedEmbeddings,
+    WrappedResponses,
+)
 from posthog.ai.openai.openai_async import WrappedBeta as AsyncWrappedBeta
 from posthog.ai.openai.openai_async import WrappedChat as AsyncWrappedChat
 from posthog.ai.openai.openai_async import WrappedEmbeddings as AsyncWrappedEmbeddings
 from posthog.ai.openai.openai_async import WrappedResponses as AsyncWrappedResponses
+from typing import Optional
+
 from posthog.client import Client as PostHogClient
+from posthog import setup
 
 
 class AzureOpenAI(openai.AzureOpenAI):
@@ -18,7 +28,7 @@ class AzureOpenAI(openai.AzureOpenAI):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             api_key: Azure OpenAI API key.
@@ -27,7 +37,7 @@ class AzureOpenAI(openai.AzureOpenAI):
             **openai_config: Any additional keyword args to set on Azure OpenAI (e.g. azure_endpoint="xxx").
         """
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
 
         # Store original objects after parent initialization (only if they exist)
         self._original_chat = getattr(self, "chat", None)
@@ -56,7 +66,7 @@ class AsyncAzureOpenAI(openai.AsyncAzureOpenAI):
 
     _ph_client: PostHogClient
 
-    def __init__(self, posthog_client: PostHogClient, **kwargs):
+    def __init__(self, posthog_client: Optional[PostHogClient] = None, **kwargs):
         """
         Args:
             api_key: Azure OpenAI API key.
@@ -65,7 +75,7 @@ class AsyncAzureOpenAI(openai.AsyncAzureOpenAI):
             **openai_config: Any additional keyword args to set on Azure OpenAI (e.g. azure_endpoint="xxx").
         """
         super().__init__(**kwargs)
-        self._ph_client = posthog_client
+        self._ph_client = posthog_client or setup()
 
         # Store original objects after parent initialization (only if they exist)
         self._original_chat = getattr(self, "chat", None)

posthog/ai/sanitization.py

@@ -0,0 +1,226 @@
+import re
+from typing import Any
+from urllib.parse import urlparse
+
+REDACTED_IMAGE_PLACEHOLDER = "[base64 image redacted]"
+
+
+def is_base64_data_url(text: str) -> bool:
+    return re.match(r"^data:([^;]+);base64,", text) is not None
+
+
+def is_valid_url(text: str) -> bool:
+    try:
+        result = urlparse(text)
+        return bool(result.scheme and result.netloc)
+    except Exception:
+        pass
+
+    return text.startswith(("/", "./", "../"))
+
+
+def is_raw_base64(text: str) -> bool:
+    if is_valid_url(text):
+        return False
+
+    return len(text) > 20 and re.match(r"^[A-Za-z0-9+/]+=*$", text) is not None
+
+
+def redact_base64_data_url(value: Any) -> Any:
+    if not isinstance(value, str):
+        return value
+
+    if is_base64_data_url(value):
+        return REDACTED_IMAGE_PLACEHOLDER
+
+    if is_raw_base64(value):
+        return REDACTED_IMAGE_PLACEHOLDER
+
+    return value
+
+
+def process_messages(messages: Any, transform_content_func) -> Any:
+    if not messages:
+        return messages
+
+    def process_content(content: Any) -> Any:
+        if isinstance(content, str):
+            return content
+
+        if not content:
+            return content
+
+        if isinstance(content, list):
+            return [transform_content_func(item) for item in content]
+
+        return transform_content_func(content)
+
+    def process_message(msg: Any) -> Any:
+        if not isinstance(msg, dict) or "content" not in msg:
+            return msg
+        return {**msg, "content": process_content(msg["content"])}
+
+    if isinstance(messages, list):
+        return [process_message(msg) for msg in messages]
+
+    return process_message(messages)
+
+
+def sanitize_openai_image(item: Any) -> Any:
+    if not isinstance(item, dict):
+        return item
+
+    if (
+        item.get("type") == "image_url"
+        and isinstance(item.get("image_url"), dict)
+        and "url" in item["image_url"]
+    ):
+        return {
+            **item,
+            "image_url": {
+                **item["image_url"],
+                "url": redact_base64_data_url(item["image_url"]["url"]),
+            },
+        }
+
+    return item
+
+
+def sanitize_openai_response_image(item: Any) -> Any:
+    if not isinstance(item, dict):
+        return item
+
+    if item.get("type") == "input_image" and "image_url" in item:
+        return {
+            **item,
+            "image_url": redact_base64_data_url(item["image_url"]),
+        }
+
+    return item
+
+
+def sanitize_anthropic_image(item: Any) -> Any:
+    if not isinstance(item, dict):
+        return item
+
+    if (
+        item.get("type") == "image"
+        and isinstance(item.get("source"), dict)
+        and item["source"].get("type") == "base64"
+        and "data" in item["source"]
+    ):
+        # For Anthropic, if the source type is "base64", we should always redact the data
+        # The provider is explicitly telling us this is base64 data
+        return {
+            **item,
+            "source": {
+                **item["source"],
+                "data": REDACTED_IMAGE_PLACEHOLDER,
+            },
+        }
+
+    return item
+
+
+def sanitize_gemini_part(part: Any) -> Any:
+    if not isinstance(part, dict):
+        return part
+
+    if (
+        "inline_data" in part
+        and isinstance(part["inline_data"], dict)
+        and "data" in part["inline_data"]
+    ):
+        # For Gemini, the inline_data structure indicates base64 data
+        # We should redact any string data in this context
+        return {
+            **part,
+            "inline_data": {
+                **part["inline_data"],
+                "data": REDACTED_IMAGE_PLACEHOLDER,
+            },
+        }
+
+    return part
+
+
+def process_gemini_item(item: Any) -> Any:
+    if not isinstance(item, dict):
+        return item
+
+    if "parts" in item and item["parts"]:
+        parts = item["parts"]
+        if isinstance(parts, list):
+            parts = [sanitize_gemini_part(part) for part in parts]
+        else:
+            parts = sanitize_gemini_part(parts)
+
+        return {**item, "parts": parts}
+
+    return item
+
+
+def sanitize_langchain_image(item: Any) -> Any:
+    if not isinstance(item, dict):
+        return item
+
+    if (
+        item.get("type") == "image_url"
+        and isinstance(item.get("image_url"), dict)
+        and "url" in item["image_url"]
+    ):
+        return {
+            **item,
+            "image_url": {
+                **item["image_url"],
+                "url": redact_base64_data_url(item["image_url"]["url"]),
+            },
+        }
+
+    if item.get("type") == "image" and "data" in item:
+        return {**item, "data": redact_base64_data_url(item["data"])}
+
+    if (
+        item.get("type") == "image"
+        and isinstance(item.get("source"), dict)
+        and "data" in item["source"]
+    ):
+        # Anthropic style - raw base64 in structured format, always redact
+        return {
+            **item,
+            "source": {
+                **item["source"],
+                "data": REDACTED_IMAGE_PLACEHOLDER,
+            },
+        }
+
+    if item.get("type") == "media" and "data" in item:
+        return {**item, "data": redact_base64_data_url(item["data"])}
+
+    return item
+
+
+def sanitize_openai(data: Any) -> Any:
+    return process_messages(data, sanitize_openai_image)
+
+
+def sanitize_openai_response(data: Any) -> Any:
+    return process_messages(data, sanitize_openai_response_image)
+
+
+def sanitize_anthropic(data: Any) -> Any:
+    return process_messages(data, sanitize_anthropic_image)
+
+
+def sanitize_gemini(data: Any) -> Any:
+    if not data:
+        return data
+
+    if isinstance(data, list):
+        return [process_gemini_item(item) for item in data]
+
+    return process_gemini_item(data)
+
+
+def sanitize_langchain(data: Any) -> Any:
+    return process_messages(data, sanitize_langchain_image)

posthog/ai/types.py

@@ -0,0 +1,125 @@
+"""
+Common type definitions for PostHog AI SDK.
+
+These types are used for formatting messages and responses across different AI providers
+(Anthropic, OpenAI, Gemini, etc.) to ensure consistency in tracking and data structure.
+"""
+
+from typing import Any, Dict, List, Optional, TypedDict, Union
+
+
+class FormattedTextContent(TypedDict):
+    """Formatted text content item."""
+
+    type: str  # Literal["text"]
+    text: str
+
+
+class FormattedFunctionCall(TypedDict, total=False):
+    """Formatted function/tool call content item."""
+
+    type: str  # Literal["function"]
+    id: Optional[str]
+    function: Dict[str, Any]  # Contains 'name' and 'arguments'
+
+
+class FormattedImageContent(TypedDict):
+    """Formatted image content item."""
+
+    type: str  # Literal["image"]
+    image: str
+
+
+# Union type for all formatted content items
+FormattedContentItem = Union[
+    FormattedTextContent,
+    FormattedFunctionCall,
+    FormattedImageContent,
+    Dict[str, Any],  # Fallback for unknown content types
+]
+
+
+class FormattedMessage(TypedDict):
+    """
+    Standardized message format for PostHog tracking.
+
+    Used across all providers to ensure consistent message structure
+    when sending events to PostHog.
+    """
+
+    role: str
+    content: Union[str, List[FormattedContentItem], Any]
+
+
+class TokenUsage(TypedDict, total=False):
+    """
+    Token usage information for AI model responses.
+
+    Different providers may populate different fields.
+    """
+
+    input_tokens: int
+    output_tokens: int
+    cache_read_input_tokens: Optional[int]
+    cache_creation_input_tokens: Optional[int]
+    reasoning_tokens: Optional[int]
+    web_search_count: Optional[int]
+
+
+class ProviderResponse(TypedDict, total=False):
+    """
+    Standardized provider response format.
+
+    Used for consistent response formatting across all providers.
+    """
+
+    messages: List[FormattedMessage]
+    usage: TokenUsage
+    error: Optional[str]
+
+
+class StreamingContentBlock(TypedDict, total=False):
+    """
+    Content block used during streaming to accumulate content.
+
+    Used for tracking text and function calls as they stream in.
+    """
+
+    type: str  # "text" or "function"
+    text: Optional[str]
+    id: Optional[str]
+    function: Optional[Dict[str, Any]]
+
+
+class ToolInProgress(TypedDict):
+    """
+    Tracks a tool/function call being accumulated during streaming.
+
+    Used by Anthropic to accumulate JSON input for tools.
+    """
+
+    block: StreamingContentBlock
+    input_string: str
+
+
+class StreamingEventData(TypedDict):
+    """
+    Standardized data for streaming events across all providers.
+
+    This type ensures consistent data structure when capturing streaming events,
+    with all provider-specific formatting already completed.
+    """
+
+    provider: str  # "openai", "anthropic", "gemini"
+    model: str
+    base_url: str
+    kwargs: Dict[str, Any]  # Original kwargs for tool extraction and special handling
+    formatted_input: Any  # Provider-formatted input ready for tracking
+    formatted_output: Any  # Provider-formatted output ready for tracking
+    usage_stats: TokenUsage
+    latency: float
+    distinct_id: Optional[str]
+    trace_id: Optional[str]
+    properties: Optional[Dict[str, Any]]
+    privacy_mode: bool
+    groups: Optional[Dict[str, Any]]

posthog/ai/utils.py

@@ -1,10 +1,83 @@
 import time
 import uuid
-from typing import Any, Callable, Dict, List, Optional
-
-from httpx import URL
+from typing import Any, Callable, Dict, List, Optional, cast
 
 from posthog.client import Client as PostHogClient
+from posthog.ai.types import FormattedMessage, StreamingEventData, TokenUsage
+from posthog.ai.sanitization import (
+    sanitize_openai,
+    sanitize_anthropic,
+    sanitize_gemini,
+    sanitize_langchain,
+)
+
+
+def merge_usage_stats(
+    target: TokenUsage, source: TokenUsage, mode: str = "incremental"
+) -> None:
+    """
+    Merge streaming usage statistics into target dict, handling None values.
+
+    Supports two modes:
+    - "incremental": Add source values to target (for APIs that report new tokens)
+    - "cumulative": Replace target with source values (for APIs that report totals)
+
+    Args:
+        target: Dictionary to update with usage stats
+        source: TokenUsage that may contain None values
+        mode: Either "incremental" or "cumulative"
+    """
+    if mode == "incremental":
+        # Add new values to existing totals
+        source_input = source.get("input_tokens")
+        if source_input is not None:
+            current = target.get("input_tokens") or 0
+            target["input_tokens"] = current + source_input
+
+        source_output = source.get("output_tokens")
+        if source_output is not None:
+            current = target.get("output_tokens") or 0
+            target["output_tokens"] = current + source_output
+
+        source_cache_read = source.get("cache_read_input_tokens")
+        if source_cache_read is not None:
+            current = target.get("cache_read_input_tokens") or 0
+            target["cache_read_input_tokens"] = current + source_cache_read
+
+        source_cache_creation = source.get("cache_creation_input_tokens")
+        if source_cache_creation is not None:
+            current = target.get("cache_creation_input_tokens") or 0
+            target["cache_creation_input_tokens"] = current + source_cache_creation
+
+        source_reasoning = source.get("reasoning_tokens")
+        if source_reasoning is not None:
+            current = target.get("reasoning_tokens") or 0
+            target["reasoning_tokens"] = current + source_reasoning
+
+        source_web_search = source.get("web_search_count")
+        if source_web_search is not None:
+            current = target.get("web_search_count") or 0
+            target["web_search_count"] = max(current, source_web_search)
+
+    elif mode == "cumulative":
+        # Replace with latest values (already cumulative)
+        if source.get("input_tokens") is not None:
+            target["input_tokens"] = source["input_tokens"]
+        if source.get("output_tokens") is not None:
+            target["output_tokens"] = source["output_tokens"]
+        if source.get("cache_read_input_tokens") is not None:
+            target["cache_read_input_tokens"] = source["cache_read_input_tokens"]
+        if source.get("cache_creation_input_tokens") is not None:
+            target["cache_creation_input_tokens"] = source[
+                "cache_creation_input_tokens"
+            ]
+        if source.get("reasoning_tokens") is not None:
+            target["reasoning_tokens"] = source["reasoning_tokens"]
+        if source.get("web_search_count") is not None:
+            target["web_search_count"] = source["web_search_count"]
+
+    else:
+        raise ValueError(f"Invalid mode: {mode}. Must be 'incremental' or 'cumulative'")
 
 
 def get_model_params(kwargs: Dict[str, Any]) -> Dict[str, Any]:
@@ -29,274 +102,135 @@ def get_model_params(kwargs: Dict[str, Any]) -> Dict[str, Any]:
     return model_params
 
 
-def get_usage(response, provider: str) -> Dict[str, Any]:
+def get_usage(response, provider: str) -> TokenUsage:
+    """
+    Extract usage statistics from response based on provider.
+    Delegates to provider-specific converter functions.
+    """
     if provider == "anthropic":
-        return {
-            "input_tokens": response.usage.input_tokens,
-            "output_tokens": response.usage.output_tokens,
-            "cache_read_input_tokens": response.usage.cache_read_input_tokens,
-            "cache_creation_input_tokens": response.usage.cache_creation_input_tokens,
-        }
+        from posthog.ai.anthropic.anthropic_converter import (
+            extract_anthropic_usage_from_response,
+        )
+
+        return extract_anthropic_usage_from_response(response)
     elif provider == "openai":
-        cached_tokens = 0
-        input_tokens = 0
-        output_tokens = 0
-        reasoning_tokens = 0
+        from posthog.ai.openai.openai_converter import (
+            extract_openai_usage_from_response,
+        )
 
-        # responses api
-        if hasattr(response.usage, "input_tokens"):
-            input_tokens = response.usage.input_tokens
-        if hasattr(response.usage, "output_tokens"):
-            output_tokens = response.usage.output_tokens
-        if hasattr(response.usage, "input_tokens_details") and hasattr(
-            response.usage.input_tokens_details, "cached_tokens"
-        ):
-            cached_tokens = response.usage.input_tokens_details.cached_tokens
-        if hasattr(response.usage, "output_tokens_details") and hasattr(
-            response.usage.output_tokens_details, "reasoning_tokens"
-        ):
-            reasoning_tokens = response.usage.output_tokens_details.reasoning_tokens
-
-        # chat completions
-        if hasattr(response.usage, "prompt_tokens"):
-            input_tokens = response.usage.prompt_tokens
-        if hasattr(response.usage, "completion_tokens"):
-            output_tokens = response.usage.completion_tokens
-        if hasattr(response.usage, "prompt_tokens_details") and hasattr(
-            response.usage.prompt_tokens_details, "cached_tokens"
-        ):
-            cached_tokens = response.usage.prompt_tokens_details.cached_tokens
-
-        return {
-            "input_tokens": input_tokens,
-            "output_tokens": output_tokens,
-            "cache_read_input_tokens": cached_tokens,
-            "reasoning_tokens": reasoning_tokens,
-        }
+        return extract_openai_usage_from_response(response)
     elif provider == "gemini":
-        input_tokens = 0
-        output_tokens = 0
+        from posthog.ai.gemini.gemini_converter import (
+            extract_gemini_usage_from_response,
+        )
 
-        if hasattr(response, "usage_metadata") and response.usage_metadata:
-            input_tokens = getattr(response.usage_metadata, "prompt_token_count", 0)
-            output_tokens = getattr(response.usage_metadata, "candidates_token_count", 0)
+        return extract_gemini_usage_from_response(response)
 
-        return {
-            "input_tokens": input_tokens,
-            "output_tokens": output_tokens,
-            "cache_read_input_tokens": 0,
-            "cache_creation_input_tokens": 0,
-            "reasoning_tokens": 0,
-        }
-    return {
-        "input_tokens": 0,
-        "output_tokens": 0,
-        "cache_read_input_tokens": 0,
-        "cache_creation_input_tokens": 0,
-        "reasoning_tokens": 0,
-    }
+    return TokenUsage(input_tokens=0, output_tokens=0)
 
 
 def format_response(response, provider: str):
     """
     Format a regular (non-streaming) response.
     """
-    output = []
-    if response is None:
-        return output
     if provider == "anthropic":
-        return format_response_anthropic(response)
+        from posthog.ai.anthropic.anthropic_converter import format_anthropic_response
+
+        return format_anthropic_response(response)
     elif provider == "openai":
-        return format_response_openai(response)
+        from posthog.ai.openai.openai_converter import format_openai_response
+
+        return format_openai_response(response)
     elif provider == "gemini":
-        return format_response_gemini(response)
-    return output
+        from posthog.ai.gemini.gemini_converter import format_gemini_response
+
+        return format_gemini_response(response)
+    return []
 
 
-def format_response_anthropic(response):
-    output = []
-    for choice in response.content:
-        if choice.text:
-            output.append(
-                {
-                    "role": "assistant",
-                    "content": choice.text,
-                }
-            )
-    return output
-
-
-def format_response_openai(response):
-    output = []
-    if hasattr(response, "choices"):
-        for choice in response.choices:
-            # Handle Chat Completions response format
-            if hasattr(choice, "message") and choice.message and choice.message.content:
-                output.append(
-                    {
-                        "content": choice.message.content,
-                        "role": choice.message.role,
-                    }
-                )
-    # Handle Responses API format
-    if hasattr(response, "output"):
-        for item in response.output:
-            if item.type == "message":
-                # Extract text content from the content list
-                if hasattr(item, "content") and isinstance(item.content, list):
-                    for content_item in item.content:
-                        if (
-                            hasattr(content_item, "type")
-                            and content_item.type == "output_text"
-                            and hasattr(content_item, "text")
-                        ):
-                            output.append(
-                                {
-                                    "content": content_item.text,
-                                    "role": item.role,
-                                }
-                            )
-                        elif hasattr(content_item, "text"):
-                            output.append(
-                                {
-                                    "content": content_item.text,
-                                    "role": item.role,
-                                }
-                            )
-                        elif (
-                            hasattr(content_item, "type")
-                            and content_item.type == "input_image"
-                            and hasattr(content_item, "image_url")
-                        ):
-                            output.append(
-                                {
-                                    "content": {
-                                        "type": "image",
-                                        "image": content_item.image_url,
-                                    },
-                                    "role": item.role,
-                                }
-                            )
-                else:
-                    output.append(
-                        {
-                            "content": item.content,
-                            "role": item.role,
-                        }
-                    )
-    return output
-
-
-def format_response_gemini(response):
-    output = []
-    if hasattr(response, "candidates") and response.candidates:
-        for candidate in response.candidates:
-            if hasattr(candidate, "content") and candidate.content:
-                content_text = ""
-                if hasattr(candidate.content, "parts") and candidate.content.parts:
-                    for part in candidate.content.parts:
-                        if hasattr(part, "text") and part.text:
-                            content_text += part.text
-                if content_text:
-                    output.append(
-                        {
-                            "role": "assistant",
-                            "content": content_text,
-                        }
-                    )
-            elif hasattr(candidate, "text") and candidate.text:
-                output.append(
-                    {
-                        "role": "assistant",
-                        "content": candidate.text,
-                    }
-                )
-    elif hasattr(response, "text") and response.text:
-        output.append(
-            {
-                "role": "assistant",
-                "content": response.text,
-            }
-        )
-    return output
-
-
-def format_tool_calls(response, provider: str):
+def extract_available_tool_calls(provider: str, kwargs: Dict[str, Any]):
+    """
+    Extract available tool calls for the given provider.
+    """
     if provider == "anthropic":
-        if hasattr(response, "tools") and response.tools and len(response.tools) > 0:
-            return response.tools
-    elif provider == "openai":
-        # Handle both Chat Completions and Responses API
-        if hasattr(response, "choices") and response.choices:
-            # Check for tool_calls in message (Chat Completions format)
-            if (
-                hasattr(response.choices[0], "message")
-                and hasattr(response.choices[0].message, "tool_calls")
-                and response.choices[0].message.tool_calls
-            ):
-                return response.choices[0].message.tool_calls
+        from posthog.ai.anthropic.anthropic_converter import extract_anthropic_tools
 
-            # Check for tool_calls directly in response (Responses API format)
-            if hasattr(response.choices[0], "tool_calls") and response.choices[0].tool_calls:
-                return response.choices[0].tool_calls
+        return extract_anthropic_tools(kwargs)
+    elif provider == "gemini":
+        from posthog.ai.gemini.gemini_converter import extract_gemini_tools
+
+        return extract_gemini_tools(kwargs)
+    elif provider == "openai":
+        from posthog.ai.openai.openai_converter import extract_openai_tools
+
+        return extract_openai_tools(kwargs)
     return None
 
 
-def merge_system_prompt(kwargs: Dict[str, Any], provider: str):
-    messages: List[Dict[str, Any]] = []
+def merge_system_prompt(
+    kwargs: Dict[str, Any], provider: str
+) -> List[FormattedMessage]:
+    """
+    Merge system prompts and format messages for the given provider.
+    """
     if provider == "anthropic":
+        from posthog.ai.anthropic.anthropic_converter import format_anthropic_input
+
         messages = kwargs.get("messages") or []
-        if kwargs.get("system") is None:
-            return messages
-        return [{"role": "system", "content": kwargs.get("system")}] + messages
+        system = kwargs.get("system")
+        return format_anthropic_input(messages, system)
     elif provider == "gemini":
+        from posthog.ai.gemini.gemini_converter import format_gemini_input_with_system
+
         contents = kwargs.get("contents", [])
-        if isinstance(contents, str):
-            return [{"role": "user", "content": contents}]
-        elif isinstance(contents, list):
-            formatted = []
-            for item in contents:
-                if isinstance(item, str):
-                    formatted.append({"role": "user", "content": item})
-                elif hasattr(item, "text"):
-                    formatted.append({"role": "user", "content": item.text})
-                else:
-                    formatted.append({"role": "user", "content": str(item)})
-            return formatted
-        else:
-            return [{"role": "user", "content": str(contents)}]
+        config = kwargs.get("config")
+        return format_gemini_input_with_system(contents, config)
+    elif provider == "openai":
+        from posthog.ai.openai.openai_converter import format_openai_input
 
-    # For OpenAI, handle both Chat Completions and Responses API
-    if kwargs.get("messages") is not None:
-        messages = list(kwargs.get("messages", []))
+        # For OpenAI, handle both Chat Completions and Responses API
+        messages_param = kwargs.get("messages")
+        input_param = kwargs.get("input")
 
-    if kwargs.get("input") is not None:
-        input_data = kwargs.get("input")
-        if isinstance(input_data, list):
-            messages.extend(input_data)
-        else:
-            messages.append({"role": "user", "content": input_data})
+        # Get base formatted messages
+        messages = format_openai_input(messages_param, input_param)
 
-    # Check if system prompt is provided as a separate parameter
-    if kwargs.get("system") is not None:
-        has_system = any(msg.get("role") == "system" for msg in messages)
-        if not has_system:
-            messages = [{"role": "system", "content": kwargs.get("system")}] + messages
+        # Check if system prompt is provided as a separate parameter
+        if kwargs.get("system") is not None:
+            has_system = any(msg.get("role") == "system" for msg in messages)
+            if not has_system:
+                system_msg = cast(
+                    FormattedMessage,
+                    {"role": "system", "content": kwargs.get("system")},
+                )
+                messages = [system_msg] + messages
 
-    # For Responses API, add instructions to the system prompt if provided
-    if kwargs.get("instructions") is not None:
-        # Find the system message if it exists
-        system_idx = next((i for i, msg in enumerate(messages) if msg.get("role") == "system"), None)
+        # For Responses API, add instructions to the system prompt if provided
+        if kwargs.get("instructions") is not None:
+            # Find the system message if it exists
+            system_idx = next(
+                (i for i, msg in enumerate(messages) if msg.get("role") == "system"),
+                None,
+            )
 
-        if system_idx is not None:
-            # Append instructions to existing system message
-            system_content = messages[system_idx].get("content", "")
-            messages[system_idx]["content"] = f"{system_content}\n\n{kwargs.get('instructions')}"
-        else:
-            # Create a new system message with instructions
-            messages = [{"role": "system", "content": kwargs.get("instructions")}] + messages
+            if system_idx is not None:
+                # Append instructions to existing system message
+                system_content = messages[system_idx].get("content", "")
+                messages[system_idx]["content"] = (
+                    f"{system_content}\n\n{kwargs.get('instructions')}"
+                )
+            else:
+                # Create a new system message with instructions
+                instruction_msg = cast(
+                    FormattedMessage,
+                    {"role": "system", "content": kwargs.get("instructions")},
+                )
+                messages = [instruction_msg] + messages
 
-    return messages
+        return messages
+
+    # Default case - return empty list
+    return []
 
 
 def call_llm_and_track_usage(
@@ -307,7 +241,7 @@ def call_llm_and_track_usage(
     posthog_properties: Optional[Dict[str, Any]],
     posthog_privacy_mode: bool,
     posthog_groups: Optional[Dict[str, Any]],
-    base_url: URL,
+    base_url: str,
     call_method: Callable[..., Any],
     **kwargs: Any,
 ) -> Any:
@@ -319,14 +253,16 @@ def call_llm_and_track_usage(
     response = None
     error = None
     http_status = 200
-    usage: Dict[str, Any] = {}
-    error_params: Dict[str, any] = {}
+    usage: TokenUsage = TokenUsage()
+    error_params: Dict[str, Any] = {}
 
     try:
         response = call_method(**kwargs)
     except Exception as exc:
         error = exc
-        http_status = getattr(exc, "status_code", 0)  # default to 0 becuase its likely an SDK error
+        http_status = getattr(
+            exc, "status_code", 0
+        )  # default to 0 becuase its likely an SDK error
         error_params = {
             "$ai_is_error": True,
             "$ai_error": exc.__str__(),
@@ -338,16 +274,22 @@ def call_llm_and_track_usage(
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
-        if response and (hasattr(response, "usage") or (provider == "gemini" and hasattr(response, "usage_metadata"))):
+        if response and (
+            hasattr(response, "usage")
+            or (provider == "gemini" and hasattr(response, "usage_metadata"))
+        ):
             usage = get_usage(response, provider)
 
         messages = merge_system_prompt(kwargs, provider)
+        sanitized_messages = sanitize_messages(messages, provider)
 
         event_properties = {
             "$ai_provider": provider,
             "$ai_model": kwargs.get("model"),
             "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(ph_client, posthog_privacy_mode, messages),
+            "$ai_input": with_privacy_mode(
+                ph_client, posthog_privacy_mode, sanitized_messages
+            ),
             "$ai_output_choices": with_privacy_mode(
                 ph_client, posthog_privacy_mode, format_response(response, provider)
             ),
@@ -361,18 +303,26 @@ def call_llm_and_track_usage(
             **(error_params or {}),
         }
 
-        tool_calls = format_tool_calls(response, provider)
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(ph_client, posthog_privacy_mode, tool_calls)
+        available_tool_calls = extract_available_tool_calls(provider, kwargs)
 
-        if usage.get("cache_read_input_tokens") is not None and usage.get("cache_read_input_tokens", 0) > 0:
-            event_properties["$ai_cache_read_input_tokens"] = usage.get("cache_read_input_tokens", 0)
+        if available_tool_calls:
+            event_properties["$ai_tools"] = available_tool_calls
 
-        if usage.get("cache_creation_input_tokens") is not None and usage.get("cache_creation_input_tokens", 0) > 0:
-            event_properties["$ai_cache_creation_input_tokens"] = usage.get("cache_creation_input_tokens", 0)
+        cache_read = usage.get("cache_read_input_tokens")
+        if cache_read is not None and cache_read > 0:
+            event_properties["$ai_cache_read_input_tokens"] = cache_read
 
-        if usage.get("reasoning_tokens") is not None and usage.get("reasoning_tokens", 0) > 0:
-            event_properties["$ai_reasoning_tokens"] = usage.get("reasoning_tokens", 0)
+        cache_creation = usage.get("cache_creation_input_tokens")
+        if cache_creation is not None and cache_creation > 0:
+            event_properties["$ai_cache_creation_input_tokens"] = cache_creation
+
+        reasoning = usage.get("reasoning_tokens")
+        if reasoning is not None and reasoning > 0:
+            event_properties["$ai_reasoning_tokens"] = reasoning
+
+        web_search_count = usage.get("web_search_count")
+        if web_search_count is not None and web_search_count > 0:
+            event_properties["$ai_web_search_count"] = web_search_count
 
         if posthog_distinct_id is None:
             event_properties["$process_person_profile"] = False
@@ -406,7 +356,7 @@ async def call_llm_and_track_usage_async(
     posthog_properties: Optional[Dict[str, Any]],
     posthog_privacy_mode: bool,
     posthog_groups: Optional[Dict[str, Any]],
-    base_url: URL,
+    base_url: str,
     call_async_method: Callable[..., Any],
     **kwargs: Any,
 ) -> Any:
@@ -414,14 +364,16 @@ async def call_llm_and_track_usage_async(
     response = None
     error = None
     http_status = 200
-    usage: Dict[str, Any] = {}
-    error_params: Dict[str, any] = {}
+    usage: TokenUsage = TokenUsage()
+    error_params: Dict[str, Any] = {}
 
     try:
         response = await call_async_method(**kwargs)
     except Exception as exc:
         error = exc
-        http_status = getattr(exc, "status_code", 0)  # default to 0 because its likely an SDK error
+        http_status = getattr(
+            exc, "status_code", 0
+        )  # default to 0 because its likely an SDK error
         error_params = {
             "$ai_is_error": True,
             "$ai_error": exc.__str__(),
@@ -433,16 +385,22 @@ async def call_llm_and_track_usage_async(
         if posthog_trace_id is None:
             posthog_trace_id = str(uuid.uuid4())
 
-        if response and (hasattr(response, "usage") or (provider == "gemini" and hasattr(response, "usage_metadata"))):
+        if response and (
+            hasattr(response, "usage")
+            or (provider == "gemini" and hasattr(response, "usage_metadata"))
+        ):
             usage = get_usage(response, provider)
 
         messages = merge_system_prompt(kwargs, provider)
+        sanitized_messages = sanitize_messages(messages, provider)
 
         event_properties = {
             "$ai_provider": provider,
             "$ai_model": kwargs.get("model"),
             "$ai_model_parameters": get_model_params(kwargs),
-            "$ai_input": with_privacy_mode(ph_client, posthog_privacy_mode, messages),
+            "$ai_input": with_privacy_mode(
+                ph_client, posthog_privacy_mode, sanitized_messages
+            ),
             "$ai_output_choices": with_privacy_mode(
                 ph_client, posthog_privacy_mode, format_response(response, provider)
             ),
@@ -456,15 +414,26 @@ async def call_llm_and_track_usage_async(
             **(error_params or {}),
         }
 
-        tool_calls = format_tool_calls(response, provider)
-        if tool_calls:
-            event_properties["$ai_tools"] = with_privacy_mode(ph_client, posthog_privacy_mode, tool_calls)
+        available_tool_calls = extract_available_tool_calls(provider, kwargs)
 
-        if usage.get("cache_read_input_tokens") is not None and usage.get("cache_read_input_tokens", 0) > 0:
-            event_properties["$ai_cache_read_input_tokens"] = usage.get("cache_read_input_tokens", 0)
+        if available_tool_calls:
+            event_properties["$ai_tools"] = available_tool_calls
 
-        if usage.get("cache_creation_input_tokens") is not None and usage.get("cache_creation_input_tokens", 0) > 0:
-            event_properties["$ai_cache_creation_input_tokens"] = usage.get("cache_creation_input_tokens", 0)
+        cache_read = usage.get("cache_read_input_tokens")
+        if cache_read is not None and cache_read > 0:
+            event_properties["$ai_cache_read_input_tokens"] = cache_read
+
+        cache_creation = usage.get("cache_creation_input_tokens")
+        if cache_creation is not None and cache_creation > 0:
+            event_properties["$ai_cache_creation_input_tokens"] = cache_creation
+
+        reasoning = usage.get("reasoning_tokens")
+        if reasoning is not None and reasoning > 0:
+            event_properties["$ai_reasoning_tokens"] = reasoning
+
+        web_search_count = usage.get("web_search_count")
+        if web_search_count is not None and web_search_count > 0:
+            event_properties["$ai_web_search_count"] = web_search_count
 
         if posthog_distinct_id is None:
             event_properties["$process_person_profile"] = False
@@ -490,7 +459,131 @@ async def call_llm_and_track_usage_async(
     return response
 
 
+def sanitize_messages(data: Any, provider: str) -> Any:
+    """Sanitize messages using provider-specific sanitization functions."""
+    if provider == "anthropic":
+        return sanitize_anthropic(data)
+    elif provider == "openai":
+        return sanitize_openai(data)
+    elif provider == "gemini":
+        return sanitize_gemini(data)
+    elif provider == "langchain":
+        return sanitize_langchain(data)
+    return data
+
+
 def with_privacy_mode(ph_client: PostHogClient, privacy_mode: bool, value: Any):
     if ph_client.privacy_mode or privacy_mode:
         return None
     return value
+
+
+def capture_streaming_event(
+    ph_client: PostHogClient,
+    event_data: StreamingEventData,
+):
+    """
+    Unified streaming event capture for all LLM providers.
+
+    This function handles the common logic for capturing streaming events across all providers.
+    All provider-specific formatting should be done BEFORE calling this function.
+
+    The function handles:
+    - Building PostHog event properties
+    - Extracting and adding tools based on provider
+    - Applying privacy mode
+    - Adding special token fields (cache, reasoning)
+    - Provider-specific fields (e.g., OpenAI instructions)
+    - Sending the event to PostHog
+
+    Args:
+        ph_client: PostHog client instance
+        event_data: Standardized streaming event data containing all necessary information
+    """
+    trace_id = event_data.get("trace_id") or str(uuid.uuid4())
+
+    # Build base event properties
+    event_properties = {
+        "$ai_provider": event_data["provider"],
+        "$ai_model": event_data["model"],
+        "$ai_model_parameters": get_model_params(event_data["kwargs"]),
+        "$ai_input": with_privacy_mode(
+            ph_client,
+            event_data["privacy_mode"],
+            event_data["formatted_input"],
+        ),
+        "$ai_output_choices": with_privacy_mode(
+            ph_client,
+            event_data["privacy_mode"],
+            event_data["formatted_output"],
+        ),
+        "$ai_http_status": 200,
+        "$ai_input_tokens": event_data["usage_stats"].get("input_tokens", 0),
+        "$ai_output_tokens": event_data["usage_stats"].get("output_tokens", 0),
+        "$ai_latency": event_data["latency"],
+        "$ai_trace_id": trace_id,
+        "$ai_base_url": str(event_data["base_url"]),
+        **(event_data.get("properties") or {}),
+    }
+
+    # Extract and add tools based on provider
+    available_tools = extract_available_tool_calls(
+        event_data["provider"],
+        event_data["kwargs"],
+    )
+    if available_tools:
+        event_properties["$ai_tools"] = available_tools
+
+    # Add optional token fields
+    # For Anthropic, always include cache fields even if 0 (backward compatibility)
+    # For others, only include if present and non-zero
+    if event_data["provider"] == "anthropic":
+        # Anthropic always includes cache fields
+        cache_read = event_data["usage_stats"].get("cache_read_input_tokens", 0)
+        cache_creation = event_data["usage_stats"].get("cache_creation_input_tokens", 0)
+        event_properties["$ai_cache_read_input_tokens"] = cache_read
+        event_properties["$ai_cache_creation_input_tokens"] = cache_creation
+    else:
+        # Other providers only include if non-zero
+        optional_token_fields = [
+            "cache_read_input_tokens",
+            "cache_creation_input_tokens",
+            "reasoning_tokens",
+        ]
+
+        for field in optional_token_fields:
+            value = event_data["usage_stats"].get(field)
+            if value is not None and isinstance(value, int) and value > 0:
+                event_properties[f"$ai_{field}"] = value
+
+    # Add web search count if present (all providers)
+    web_search_count = event_data["usage_stats"].get("web_search_count")
+    if (
+        web_search_count is not None
+        and isinstance(web_search_count, int)
+        and web_search_count > 0
+    ):
+        event_properties["$ai_web_search_count"] = web_search_count
+
+    # Handle provider-specific fields
+    if (
+        event_data["provider"] == "openai"
+        and event_data["kwargs"].get("instructions") is not None
+    ):
+        event_properties["$ai_instructions"] = with_privacy_mode(
+            ph_client,
+            event_data["privacy_mode"],
+            event_data["kwargs"]["instructions"],
+        )
+
+    if event_data.get("distinct_id") is None:
+        event_properties["$process_person_profile"] = False
+
+    # Send event to PostHog
+    if hasattr(ph_client, "capture"):
+        ph_client.capture(
+            distinct_id=event_data.get("distinct_id") or trace_id,
+            event="$ai_generation",
+            properties=event_properties,
+            groups=event_data.get("groups"),
+        )

posthog/args.py

@@ -0,0 +1,71 @@
+from typing import TypedDict, Optional, Any, Dict, Union, Tuple, Type
+from types import TracebackType
+from typing_extensions import NotRequired  # For Python < 3.11 compatibility
+from datetime import datetime
+import numbers
+from uuid import UUID
+
+from posthog.types import SendFeatureFlagsOptions
+
+ID_TYPES = Union[numbers.Number, str, UUID, int]
+
+
+class OptionalCaptureArgs(TypedDict):
+    """Optional arguments for the capture method.
+
+    Args:
+        distinct_id: Unique identifier for the person associated with this event. If not set, the context
+            distinct_id is used, if available, otherwise a UUID is generated, and the event is marked
+            as personless. Setting context-level distinct_id's is recommended.
+        properties: Dictionary of properties to track with the event
+        timestamp: When the event occurred (defaults to current time)
+        uuid: Unique identifier for this specific event. If not provided, one is generated. The event
+            UUID is returned, so you can correlate it with actions in your app (like showing users an
+            error ID if you capture an exception).
+        groups: Group identifiers to associate with this event (format: {group_type: group_key})
+        send_feature_flags: Whether to include currently active feature flags in the event properties.
+            Can be a boolean (True/False) or a SendFeatureFlagsOptions object for advanced configuration.
+            Defaults to False.
+        disable_geoip: Whether to disable GeoIP lookup for this event. Defaults to False.
+    """
+
+    distinct_id: NotRequired[Optional[ID_TYPES]]
+    properties: NotRequired[Optional[Dict[str, Any]]]
+    timestamp: NotRequired[Optional[Union[datetime, str]]]
+    uuid: NotRequired[Optional[str]]
+    groups: NotRequired[Optional[Dict[str, str]]]
+    send_feature_flags: NotRequired[
+        Optional[Union[bool, SendFeatureFlagsOptions]]
+    ]  # Updated to support both boolean and options object
+    disable_geoip: NotRequired[
+        Optional[bool]
+    ]  # As above, optional so we can tell if the user is intentionally overriding a client setting or not
+
+
+class OptionalSetArgs(TypedDict):
+    """Optional arguments for the set method.
+
+    Args:
+        distinct_id: Unique identifier for the user to set properties on. If not set, the context
+            distinct_id is used, if available, otherwise this function does nothing. Setting
+            context-level distinct_id's is recommended.
+        properties: Dictionary of properties to set on the person
+        timestamp: When the properties were set (defaults to current time)
+        uuid: Unique identifier for this operation. If not provided, one is generated. This
+            UUID is returned, so you can correlate it with actions in your app.
+        disable_geoip: Whether to disable GeoIP lookup for this operation. Defaults to False.
+    """
+
+    distinct_id: NotRequired[Optional[ID_TYPES]]
+    properties: NotRequired[Optional[Dict[str, Any]]]
+    timestamp: NotRequired[Optional[Union[datetime, str]]]
+    uuid: NotRequired[Optional[str]]
+    disable_geoip: NotRequired[Optional[bool]]
+
+
+ExcInfo = Union[
+    Tuple[Type[BaseException], BaseException, Optional[TracebackType]],
+    Tuple[None, None, None],
+]
+
+ExceptionArg = Union[BaseException, ExcInfo]

posthog/consumer.py

@@ -107,7 +107,9 @@ class Consumer(Thread):
                 item = queue.get(block=True, timeout=self.flush_interval - elapsed)
                 item_size = len(json.dumps(item, cls=DatetimeSerializer).encode())
                 if item_size > MAX_MSG_SIZE:
-                    self.log.error("Item exceeds 900kib limit, dropping. (%s)", str(item))
+                    self.log.error(
+                        "Item exceeds 900kib limit, dropping. (%s)", str(item)
+                    )
                     continue
                 items.append(item)
                 total_size += item_size
@@ -134,7 +136,9 @@ class Consumer(Thread):
                 # retry on all other errors (eg. network)
                 return False
 
-        @backoff.on_exception(backoff.expo, Exception, max_tries=self.retries + 1, giveup=fatal_exception)
+        @backoff.on_exception(
+            backoff.expo, Exception, max_tries=self.retries + 1, giveup=fatal_exception
+        )
         def send_request():
             batch_post(
                 self.api_key,

posthog/contexts.py

@@ -0,0 +1,365 @@
+import contextvars
+from contextlib import contextmanager
+from typing import Optional, Any, Callable, Dict, TypeVar, cast, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    # To avoid circular imports
+    from posthog.client import Client
+
+
+class ContextScope:
+    def __init__(
+        self,
+        parent=None,
+        fresh: bool = False,
+        capture_exceptions: bool = True,
+        client: Optional["Client"] = None,
+    ):
+        self.client: Optional[Client] = client
+        self.parent = parent
+        self.fresh = fresh
+        self.capture_exceptions = capture_exceptions
+        self.session_id: Optional[str] = None
+        self.distinct_id: Optional[str] = None
+        self.tags: Dict[str, Any] = {}
+        self.capture_exception_code_variables: Optional[bool] = None
+        self.code_variables_mask_patterns: Optional[list] = None
+        self.code_variables_ignore_patterns: Optional[list] = None
+
+    def set_session_id(self, session_id: str):
+        self.session_id = session_id
+
+    def set_distinct_id(self, distinct_id: str):
+        self.distinct_id = distinct_id
+
+    def add_tag(self, key: str, value: Any):
+        self.tags[key] = value
+
+    def set_capture_exception_code_variables(self, enabled: bool):
+        self.capture_exception_code_variables = enabled
+
+    def set_code_variables_mask_patterns(self, mask_patterns: list):
+        self.code_variables_mask_patterns = mask_patterns
+
+    def set_code_variables_ignore_patterns(self, ignore_patterns: list):
+        self.code_variables_ignore_patterns = ignore_patterns
+
+    def get_parent(self):
+        return self.parent
+
+    def get_session_id(self) -> Optional[str]:
+        if self.session_id is not None:
+            return self.session_id
+        if self.parent is not None and not self.fresh:
+            return self.parent.get_session_id()
+        return None
+
+    def get_distinct_id(self) -> Optional[str]:
+        if self.distinct_id is not None:
+            return self.distinct_id
+        if self.parent is not None and not self.fresh:
+            return self.parent.get_distinct_id()
+        return None
+
+    def collect_tags(self) -> Dict[str, Any]:
+        tags = self.tags.copy()
+        if self.parent and not self.fresh:
+            # We want child tags to take precedence over parent tags,
+            # so we can't use a simple update here, instead collecting
+            # the parent tags and then updating with the child tags.
+            new_tags = self.parent.collect_tags()
+            tags.update(new_tags)
+        return tags
+
+    def get_capture_exception_code_variables(self) -> Optional[bool]:
+        if self.capture_exception_code_variables is not None:
+            return self.capture_exception_code_variables
+        if self.parent is not None and not self.fresh:
+            return self.parent.get_capture_exception_code_variables()
+        return None
+
+    def get_code_variables_mask_patterns(self) -> Optional[list]:
+        if self.code_variables_mask_patterns is not None:
+            return self.code_variables_mask_patterns
+        if self.parent is not None and not self.fresh:
+            return self.parent.get_code_variables_mask_patterns()
+        return None
+
+    def get_code_variables_ignore_patterns(self) -> Optional[list]:
+        if self.code_variables_ignore_patterns is not None:
+            return self.code_variables_ignore_patterns
+        if self.parent is not None and not self.fresh:
+            return self.parent.get_code_variables_ignore_patterns()
+        return None
+
+
+_context_stack: contextvars.ContextVar[Optional[ContextScope]] = contextvars.ContextVar(
+    "posthog_context_stack", default=None
+)
+
+
+def _get_current_context() -> Optional[ContextScope]:
+    return _context_stack.get()
+
+
+@contextmanager
+def new_context(
+    fresh: bool = False,
+    capture_exceptions: bool = True,
+    client: Optional["Client"] = None,
+):
+    """
+    Create a new context scope that will be active for the duration of the with block.
+    Any tags set within this scope will be isolated to this context. Any exceptions raised
+    or events captured within the context will be tagged with the context tags.
+
+    Args:
+        fresh: Whether to start with a fresh context (default: False).
+               If False, inherits tags, identity and session id's from parent context.
+               If True, starts with no state
+        capture_exceptions: Whether to capture exceptions raised within the context (default: True).
+               If True, captures exceptions and tags them with the context tags before propagating them.
+               If False, exceptions will propagate without being tagged or captured.
+        client: Optional client instance to use for capturing exceptions (default: None).
+                If provided, the client will be used to capture exceptions within the context.
+                If not provided, the default (global) client will be used. Note that the passed
+                client is only used to capture exceptions within the context - other events captured
+                within the context via `Client.capture` or `posthog.capture` will still carry the context
+                state (tags, identity, session id), but will be captured by the client directly used (or
+                the global one, in the case of `posthog.capture`)
+
+    Examples:
+        ```python
+        # Inherit parent context tags
+        with posthog.new_context():
+            posthog.tag("request_id", "123")
+            # Both this event and the exception will be tagged with the context tags
+            posthog.capture("event_name", {"property": "value"})
+            raise ValueError("Something went wrong")
+        ```
+        ```python
+        # Start with fresh context (no inherited tags)
+        with posthog.new_context(fresh=True):
+            posthog.tag("request_id", "123")
+            # Both this event and the exception will be tagged with the context tags
+            posthog.capture("event_name", {"property": "value"})
+            raise ValueError("Something went wrong")
+        ```
+
+    Category:
+        Contexts
+    """
+    from posthog import capture_exception
+
+    current_context = _get_current_context()
+    new_context = ContextScope(current_context, fresh, capture_exceptions, client)
+    _context_stack.set(new_context)
+
+    try:
+        yield
+    except Exception as e:
+        if new_context.capture_exceptions:
+            if new_context.client:
+                new_context.client.capture_exception(e)
+            else:
+                capture_exception(e)
+        raise
+    finally:
+        _context_stack.set(new_context.get_parent())
+
+
+def tag(key: str, value: Any) -> None:
+    """
+    Add a tag to the current context. All tags are added as properties to any event, including exceptions, captured
+    within the context.
+
+    Args:
+        key: The tag key
+        value: The tag value
+
+    Example:
+        ```python
+        posthog.tag("user_id", "123")
+        ```
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.add_tag(key, value)
+
+
+def get_tags() -> Dict[str, Any]:
+    """
+    Get all tags from the current context. Note, modifying
+    the returned dictionary will not affect the current context.
+
+    Returns:
+        Dict of all tags in the current context
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.collect_tags()
+    return {}
+
+
+def identify_context(distinct_id: str) -> None:
+    """
+    Identify the current context with a distinct ID, associating all events captured in this or
+    child contexts with the given distinct ID (unless identify_context is called again). This is overridden by
+    distinct id's passed directly to posthog.capture and related methods (identify, set etc). Entering a
+    fresh context will clear the context-level distinct ID. The distinct-id passed should be uniquely associated
+    with one of your users. Events captured outside of a context, or in a context with no associated distinct
+    ID, will be assigned a random UUID, and captured as "personless".
+
+    Args:
+        distinct_id: The distinct ID to associate with the current context and its children.
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.set_distinct_id(distinct_id)
+
+
+def set_context_session(session_id: str) -> None:
+    """
+    Set the session ID for the current context, associating all events captured in this or
+    child contexts with the given session ID (unless set_context_session is called again).
+    Entering a fresh context will clear the context-level session ID.
+
+    Args:
+        session_id: The session ID to associate with the current context and its children. See https://posthog.com/docs/data/sessions
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.set_session_id(session_id)
+
+
+def get_context_session_id() -> Optional[str]:
+    """
+    Get the session ID for the current context.
+
+    Returns:
+        The session ID if set, None otherwise
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.get_session_id()
+    return None
+
+
+def get_context_distinct_id() -> Optional[str]:
+    """
+    Get the distinct ID for the current context.
+
+    Returns:
+        The distinct ID if set, None otherwise
+
+    Category:
+        Contexts
+    """
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.get_distinct_id()
+    return None
+
+
+def set_capture_exception_code_variables_context(enabled: bool) -> None:
+    """
+    Set whether code variables are captured for the current context.
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.set_capture_exception_code_variables(enabled)
+
+
+def set_code_variables_mask_patterns_context(mask_patterns: list) -> None:
+    """
+    Variable names matching these patterns will be masked with *** when capturing code variables.
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.set_code_variables_mask_patterns(mask_patterns)
+
+
+def set_code_variables_ignore_patterns_context(ignore_patterns: list) -> None:
+    """
+    Variable names matching these patterns will be ignored completely when capturing code variables.
+    """
+    current_context = _get_current_context()
+    if current_context:
+        current_context.set_code_variables_ignore_patterns(ignore_patterns)
+
+
+def get_capture_exception_code_variables_context() -> Optional[bool]:
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.get_capture_exception_code_variables()
+    return None
+
+
+def get_code_variables_mask_patterns_context() -> Optional[list]:
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.get_code_variables_mask_patterns()
+    return None
+
+
+def get_code_variables_ignore_patterns_context() -> Optional[list]:
+    current_context = _get_current_context()
+    if current_context:
+        return current_context.get_code_variables_ignore_patterns()
+    return None
+
+
+F = TypeVar("F", bound=Callable[..., Any])
+
+
+def scoped(fresh: bool = False, capture_exceptions: bool = True):
+    """
+    Decorator that creates a new context for the function. Simply wraps
+    the function in a with posthog.new_context(): block.
+
+    Args:
+        fresh: Whether to start with a fresh context (default: False)
+        capture_exceptions: Whether to capture and track exceptions with posthog error tracking (default: True)
+
+    Example:
+        @posthog.scoped()
+        def process_payment(payment_id):
+            posthog.tag("payment_id", payment_id)
+            posthog.tag("payment_method", "credit_card")
+
+            # This event will be captured with tags
+            posthog.capture("payment_started")
+            # If this raises an exception, it will be captured with tags
+            # and then re-raised
+            some_risky_function()
+
+    Category:
+        Contexts
+    """
+
+    def decorator(func: F) -> F:
+        from functools import wraps
+
+        @wraps(func)
+        def wrapper(*args, **kwargs):
+            with new_context(fresh=fresh, capture_exceptions=capture_exceptions):
+                return func(*args, **kwargs)
+
+        return cast(F, wrapper)
+
+    return decorator

posthog/exception_capture.py

@@ -6,45 +6,25 @@
 import logging
 import sys
 import threading
-from enum import Enum
-from typing import TYPE_CHECKING, List, Optional
+from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
     from posthog.client import Client
 
 
-class Integrations(str, Enum):
-    Django = "django"
-
-
 class ExceptionCapture:
     # TODO: Add client side rate limiting to prevent spamming the server with exceptions
 
     log = logging.getLogger("posthog")
 
-    def __init__(self, client: "Client", integrations: Optional[List[Integrations]] = None):
+    def __init__(self, client: "Client"):
         self.client = client
         self.original_excepthook = sys.excepthook
         sys.excepthook = self.exception_handler
         threading.excepthook = self.thread_exception_handler
-        self.enabled_integrations = []
-
-        for integration in integrations or []:
-            # TODO: Maybe find a better way of enabling integrations
-            # This is very annoying currently if we had to add any configuration per integration
-            if integration == Integrations.Django:
-                try:
-                    from posthog.exception_integrations.django import DjangoIntegration
-
-                    enabled_integration = DjangoIntegration(self.exception_receiver)
-                    self.enabled_integrations.append(enabled_integration)
-                except Exception as e:
-                    self.log.exception(f"Failed to enable Django integration: {e}")
 
     def close(self):
         sys.excepthook = self.original_excepthook
-        for integration in self.enabled_integrations:
-            integration.uninstall()
 
     def exception_handler(self, exc_type, exc_value, exc_traceback):
         # don't affect default behaviour.
@@ -64,6 +44,6 @@ class ExceptionCapture:
     def capture_exception(self, exception, metadata=None):
         try:
             distinct_id = metadata.get("distinct_id") if metadata else None
-            self.client.capture_exception(exception, distinct_id)
+            self.client.capture_exception(exception, distinct_id=distinct_id)
         except Exception as e:
             self.log.exception(f"Failed to capture exception: {e}")

posthog/exception_integrations/__init__.py

@@ -1,5 +0,0 @@
-class IntegrationEnablingError(Exception):
-    """
-    The integration could not be enabled due to a user error like
-    `django` not being installed for the `DjangoIntegration`.
-    """

posthog/exception_integrations/django.py

@@ -1,93 +0,0 @@
-# Portions of this file are derived from getsentry/sentry-javascript by Software, Inc. dba Sentry
-# Licensed under the MIT License
-
-# 💖open source (under MIT License)
-
-import re
-import sys
-from typing import TYPE_CHECKING
-
-from posthog.exception_integrations import IntegrationEnablingError
-
-try:
-    from django import VERSION as DJANGO_VERSION
-    from django.core import signals
-
-except ImportError:
-    raise IntegrationEnablingError("Django not installed")
-
-
-if TYPE_CHECKING:
-    from typing import Any, Dict  # noqa: F401
-
-    from django.core.handlers.wsgi import WSGIRequest  # noqa: F401
-
-
-class DjangoIntegration:
-    # TODO: Abstract integrations one we have more and can see patterns
-    """
-    Autocapture errors from a Django application.
-    """
-
-    identifier = "django"
-
-    def __init__(self, capture_exception_fn=None):
-
-        if DJANGO_VERSION < (4, 2):
-            raise IntegrationEnablingError("Django 4.2 or newer is required.")
-
-        # TODO: Right now this seems too complicated / overkill for us, but seems like we can automatically plug in middlewares
-        # which is great for users (they don't need to do this) and everything should just work.
-        # We should consider this in the future, but for now we can just use the middleware and signals handlers.
-        # See: https://github.com/getsentry/sentry-python/blob/269d96d6e9821122fbff280e6a26956e5ed03c0b/sentry_sdk/integrations/django/__init__.py
-
-        self.capture_exception_fn = capture_exception_fn
-
-        def _got_request_exception(request=None, **kwargs):
-            # type: (WSGIRequest, **Any) -> None
-
-            extra_props = {}
-            if request is not None:
-                # get headers metadata
-                extra_props = DjangoRequestExtractor(request).extract_person_data()
-
-            self.capture_exception_fn(sys.exc_info(), extra_props)
-
-        signals.got_request_exception.connect(_got_request_exception)
-
-    def uninstall(self):
-        pass
-
-
-class DjangoRequestExtractor:
-
-    def __init__(self, request):
-        # type: (Any) -> None
-        self.request = request
-
-    def extract_person_data(self):
-        headers = self.headers()
-
-        # Extract traceparent and tracestate headers
-        traceparent = headers.get("traceparent")
-        tracestate = headers.get("tracestate")
-
-        # Extract the distinct_id from tracestate
-        distinct_id = None
-        if tracestate:
-            # TODO: Align on the format of the distinct_id in tracestate
-            # We can't have comma or equals in header values here, so maybe we should base64 encode it?
-            match = re.search(r"posthog-distinct-id=([^,]+)", tracestate)
-            if match:
-                distinct_id = match.group(1)
-
-        return {
-            "distinct_id": distinct_id,
-            "ip": headers.get("X-Forwarded-For"),
-            "user_agent": headers.get("User-Agent"),
-            "traceparent": traceparent,
-        }
-
-    def headers(self):
-        # type: () -> Dict[str, str]
-        return dict(self.request.headers)

posthog/exception_utils.py

@@ -5,12 +5,32 @@
 # 💖open source (under MIT License)
 # We want to keep payloads as similar to Sentry as possible for easy interoperability
 
+import json
 import linecache
 import os
 import re
 import sys
+import types
 from datetime import datetime
-from typing import TYPE_CHECKING
+from types import FrameType, TracebackType  # noqa: F401
+from typing import (  # noqa: F401
+    Any,
+    Dict,
+    Iterator,
+    List,
+    Literal,
+    Optional,
+    Set,
+    Tuple,
+    TypedDict,
+    TypeVar,
+    Union,
+    cast,
+    TYPE_CHECKING,
+    Pattern,
+)
+
+from posthog.args import ExcInfo, ExceptionArg  # noqa: F401
 
 try:
     # Python 3.11
@@ -22,78 +42,97 @@ except ImportError:
 
 DEFAULT_MAX_VALUE_LENGTH = 1024
 
+DEFAULT_CODE_VARIABLES_MASK_PATTERNS = [
+    r"(?i).*password.*",
+    r"(?i).*secret.*",
+    r"(?i).*passwd.*",
+    r"(?i).*pwd.*",
+    r"(?i).*api_key.*",
+    r"(?i).*apikey.*",
+    r"(?i).*auth.*",
+    r"(?i).*credentials.*",
+    r"(?i).*privatekey.*",
+    r"(?i).*private_key.*",
+    r"(?i).*token.*",
+]
 
-if TYPE_CHECKING:
+DEFAULT_CODE_VARIABLES_IGNORE_PATTERNS = [r"^__.*"]
 
-    from types import FrameType, TracebackType
-    from typing import (  # noqa: F401
-        Any,
-        Callable,
-        Dict,
-        Iterator,
-        List,
-        Literal,
-        Optional,
-        Set,
-        Tuple,
-        Type,
-        TypedDict,
-        TypeVar,
-        Union,
-        cast,
-    )
+CODE_VARIABLES_REDACTED_VALUE = "$$_posthog_redacted_based_on_masking_rules_$$"
 
-    ExcInfo = Union[
-        Tuple[Type[BaseException], BaseException, Optional[TracebackType]],
-        Tuple[None, None, None],
-    ]
-    LogLevelStr = Literal["fatal", "critical", "error", "warning", "info", "debug"]
+DEFAULT_TOTAL_VARIABLES_SIZE_LIMIT = 20 * 1024
 
-    Event = TypedDict(
-        "Event",
-        {
-            "breadcrumbs": Dict[Literal["values"], List[Dict[str, Any]]],  # TODO: We can expand on this type
-            "check_in_id": str,
-            "contexts": Dict[str, Dict[str, object]],
-            "dist": str,
-            "duration": Optional[float],
-            "environment": str,
-            "errors": List[Dict[str, Any]],  # TODO: We can expand on this type
-            "event_id": str,
-            "exception": Dict[Literal["values"], List[Dict[str, Any]]],  # TODO: We can expand on this type
-            # "extra": MutableMapping[str, object],
-            # "fingerprint": List[str],
-            "level": LogLevelStr,
-            # "logentry": Mapping[str, object],
-            "logger": str,
-            # "measurements": Dict[str, MeasurementValue],
-            "message": str,
-            "modules": Dict[str, str],
-            # "monitor_config": Mapping[str, object],
-            "monitor_slug": Optional[str],
-            "platform": Literal["python"],
-            "profile": object,  # Should be sentry_sdk.profiler.Profile, but we can't import that here due to circular imports
-            "release": str,
-            "request": Dict[str, object],
-            # "sdk": Mapping[str, object],
-            "server_name": str,
-            "spans": List[Dict[str, object]],
-            "stacktrace": Dict[str, object],  # We access this key in the code, but I am unsure whether we ever set it
-            "start_timestamp": datetime,
-            "status": Optional[str],
-            # "tags": MutableMapping[
-            #     str, str
-            # ],  # Tags must be less than 200 characters each
-            "threads": Dict[Literal["values"], List[Dict[str, Any]]],  # TODO: We can expand on this type
-            "timestamp": Optional[datetime],  # Must be set before sending the event
-            "transaction": str,
-            # "transaction_info": Mapping[str, Any],  # TODO: We can expand on this type
-            "type": Literal["check_in", "transaction"],
-            "user": Dict[str, object],
-            "_metrics_summary": Dict[str, object],
-        },
-        total=False,
-    )
+
+class VariableSizeLimiter:
+    def __init__(self, max_size=DEFAULT_TOTAL_VARIABLES_SIZE_LIMIT):
+        self.max_size = max_size
+        self.current_size = 0
+
+    def can_add(self, size):
+        return self.current_size + size <= self.max_size
+
+    def add(self, size):
+        self.current_size += size
+
+    def get_remaining_space(self):
+        return self.max_size - self.current_size
+
+
+LogLevelStr = Literal["fatal", "critical", "error", "warning", "info", "debug"]
+
+Event = TypedDict(
+    "Event",
+    {
+        "breadcrumbs": Dict[
+            Literal["values"], List[Dict[str, Any]]
+        ],  # TODO: We can expand on this type
+        "check_in_id": str,
+        "contexts": Dict[str, Dict[str, object]],
+        "dist": str,
+        "duration": Optional[float],
+        "environment": str,
+        "errors": List[Dict[str, Any]],  # TODO: We can expand on this type
+        "event_id": str,
+        "exception": Dict[
+            Literal["values"], List[Dict[str, Any]]
+        ],  # TODO: We can expand on this type
+        # "extra": MutableMapping[str, object],
+        # "fingerprint": List[str],
+        "level": LogLevelStr,
+        # "logentry": Mapping[str, object],
+        "logger": str,
+        # "measurements": Dict[str, MeasurementValue],
+        "message": str,
+        "modules": Dict[str, str],
+        # "monitor_config": Mapping[str, object],
+        "monitor_slug": Optional[str],
+        "platform": Literal["python"],
+        "profile": object,
+        "release": str,
+        "request": Dict[str, object],
+        # "sdk": Mapping[str, object],
+        "server_name": str,
+        "spans": List[Dict[str, object]],
+        "stacktrace": Dict[
+            str, object
+        ],  # We access this key in the code, but I am unsure whether we ever set it
+        "start_timestamp": datetime,
+        "status": Optional[str],
+        # "tags": MutableMapping[
+        #     str, str
+        # ],  # Tags must be less than 200 characters each
+        "threads": Dict[
+            Literal["values"], List[Dict[str, Any]]
+        ],  # TODO: We can expand on this type
+        "timestamp": Optional[datetime],  # Must be set before sending the event
+        "transaction": str,
+        # "transaction_info": Mapping[str, Any],  # TODO: We can expand on this type
+        "type": Literal["check_in", "transaction"],
+        "user": Dict[str, object],
+        "_metrics_summary": Dict[str, object],
+    },
+    total=False,
+)
 
 
 epoch = datetime(1970, 1, 1)
@@ -129,9 +168,6 @@ def event_hint_with_exc_info(exc_info=None):
 class AnnotatedValue:
     """
     Meta information for a data field in the event payload.
-    This is to tell Relay that we have tampered with the fields value.
-    See:
-    https://github.com/getsentry/relay/blob/be12cd49a0f06ea932ed9b9f93a655de5d6ad6d1/relay-general/src/types/meta.rs#L407-L423
     """
 
     __slots__ = ("value", "metadata")
@@ -273,7 +309,10 @@ def get_lines_from_file(
     upper_bound = min(lineno + 1 + context_lines, len(source))
 
     try:
-        pre_context = [strip_string(line.strip("\r\n"), max_length=max_length) for line in source[lower_bound:lineno]]
+        pre_context = [
+            strip_string(line.strip("\r\n"), max_length=max_length)
+            for line in source[lower_bound:lineno]
+        ]
         context_line = strip_string(source[lineno].strip("\r\n"), max_length=max_length)
         post_context = [
             strip_string(line.strip("\r\n"), max_length=max_length)
@@ -305,7 +344,9 @@ def get_source_context(
         loader = None
     lineno = tb_lineno - 1
     if lineno is not None and abs_path:
-        return get_lines_from_file(abs_path, lineno, max_value_length, loader=loader, module=module)
+        return get_lines_from_file(
+            abs_path, lineno, max_value_length, loader=loader, module=module
+        )
     return [], None, []
 
 
@@ -342,7 +383,9 @@ def filename_for_module(module, abs_path):
         if not base_module_path:
             return abs_path
 
-        return abs_path.split(base_module_path.rsplit(os.sep, 2)[0], 1)[-1].lstrip(os.sep)
+        return abs_path.split(base_module_path.rsplit(os.sep, 2)[0], 1)[-1].lstrip(
+            os.sep
+        )
     except Exception:
         return abs_path
 
@@ -350,12 +393,9 @@ def filename_for_module(module, abs_path):
 def serialize_frame(
     frame,
     tb_lineno=None,
-    include_local_variables=True,
-    include_source_context=True,
     max_value_length=None,
-    custom_repr=None,
 ):
-    # type: (FrameType, Optional[int], bool, bool, Optional[int], Optional[Callable[..., Optional[str]]]) -> Dict[str, Any]
+    # type: (FrameType, Optional[int], Optional[int]) -> Dict[str, Any]
     f_code = getattr(frame, "f_code", None)
     if not f_code:
         abs_path = None
@@ -380,50 +420,13 @@ def serialize_frame(
         "lineno": tb_lineno,
     }  # type: Dict[str, Any]
 
-    if include_source_context:
-        rv["pre_context"], rv["context_line"], rv["post_context"] = get_source_context(
-            frame, tb_lineno, max_value_length
-        )
-
-    if include_local_variables:
-        # TODO(nk): Sort out this current invalid import
-        # from sentry_sdk.serializer import serialize
-
-        # rv["vars"] = serialize(
-        #     dict(frame.f_locals), is_vars=True, custom_repr=custom_repr
-        # )
-        pass
+    rv["pre_context"], rv["context_line"], rv["post_context"] = get_source_context(
+        frame, tb_lineno, max_value_length
+    )
 
     return rv
 
 
-def current_stacktrace(
-    include_local_variables=True,  # type: bool
-    include_source_context=True,  # type: bool
-    max_value_length=None,  # type: Optional[int]
-):
-    # type: (...) -> Dict[str, Any]
-    __tracebackhide__ = True
-    frames = []
-
-    f = sys._getframe()  # type: Optional[FrameType]
-    while f is not None:
-        if not should_hide_frame(f):
-            frames.append(
-                serialize_frame(
-                    f,
-                    include_local_variables=include_local_variables,
-                    include_source_context=include_source_context,
-                    max_value_length=max_value_length,
-                )
-            )
-        f = f.f_back
-
-    frames.reverse()
-
-    return {"frames": frames, "type": "raw"}
-
-
 def get_errno(exc_value):
     # type: (BaseException) -> Optional[Any]
     return getattr(exc_value, "errno", None)
@@ -431,14 +434,19 @@ def get_errno(exc_value):
 
 def get_error_message(exc_value):
     # type: (Optional[BaseException]) -> str
-    return getattr(exc_value, "message", "") or getattr(exc_value, "detail", "") or safe_str(exc_value)
+    message = (
+        getattr(exc_value, "message", "")
+        or getattr(exc_value, "detail", "")
+        or exc_value
+    )
+
+    return safe_str(message)
 
 
 def single_exception_from_error_tuple(
     exc_type,  # type: Optional[type]
     exc_value,  # type: Optional[BaseException]
     tb,  # type: Optional[TracebackType]
-    client_options=None,  # type: Optional[Dict[str, Any]]
     mechanism=None,  # type: Optional[Dict[str, Any]]
     exception_id=None,  # type: Optional[int]
     parent_id=None,  # type: Optional[int]
@@ -446,13 +454,12 @@ def single_exception_from_error_tuple(
 ):
     # type: (...) -> Dict[str, Any]
     """
-    Creates a dict that goes into the events `exception.values` list and is ingestible by Sentry.
-
-    See the Exception Interface documentation for more details:
-    https://develop.sentry.dev/sdk/event-payloads/exception/
+    Creates a dict that goes into the events `exception.values` list
     """
     exception_value = {}  # type: Dict[str, Any]
-    exception_value["mechanism"] = mechanism.copy() if mechanism else {"type": "generic", "handled": True}
+    exception_value["mechanism"] = (
+        mechanism.copy() if mechanism else {"type": "generic", "handled": True}
+    )
     if exception_id is not None:
         exception_value["mechanism"]["exception_id"] = exception_id
 
@@ -462,7 +469,9 @@ def single_exception_from_error_tuple(
         errno = None
 
     if errno is not None:
-        exception_value["mechanism"].setdefault("meta", {}).setdefault("errno", {}).setdefault("number", errno)
+        exception_value["mechanism"].setdefault("meta", {}).setdefault(
+            "errno", {}
+        ).setdefault("number", errno)
 
     if source is not None:
         exception_value["mechanism"]["source"] = source
@@ -475,7 +484,9 @@ def single_exception_from_error_tuple(
     if is_root_exception and "type" not in exception_value["mechanism"]:
         exception_value["mechanism"]["type"] = "generic"
 
-    is_exception_group = BaseExceptionGroup is not None and isinstance(exc_value, BaseExceptionGroup)
+    is_exception_group = BaseExceptionGroup is not None and isinstance(
+        exc_value, BaseExceptionGroup
+    )
     if is_exception_group:
         exception_value["mechanism"]["is_exception_group"] = True
 
@@ -483,25 +494,13 @@ def single_exception_from_error_tuple(
     exception_value["type"] = get_type_name(exc_type)
     exception_value["value"] = get_error_message(exc_value)
 
-    if client_options is None:
-        include_local_variables = True
-        include_source_context = True
-        max_value_length = DEFAULT_MAX_VALUE_LENGTH  # fallback
-        custom_repr = None
-    else:
-        include_local_variables = client_options["include_local_variables"]
-        include_source_context = client_options["include_source_context"]
-        max_value_length = client_options["max_value_length"]
-        custom_repr = client_options.get("custom_repr")
+    max_value_length = DEFAULT_MAX_VALUE_LENGTH  # fallback
 
     frames = [
         serialize_frame(
             tb.tb_frame,
             tb_lineno=tb.tb_lineno,
-            include_local_variables=include_local_variables,
-            include_source_context=include_source_context,
             max_value_length=max_value_length,
-            custom_repr=custom_repr,
         )
         for tb in iter_stacks(tb)
     ]
@@ -523,7 +522,11 @@ if HAS_CHAINED_EXCEPTIONS:
         seen_exceptions = []
         seen_exception_ids = set()  # type: Set[int]
 
-        while exc_type is not None and exc_value is not None and id(exc_value) not in seen_exception_ids:
+        while (
+            exc_type is not None
+            and exc_value is not None
+            and id(exc_value) not in seen_exception_ids
+        ):
             yield exc_type, exc_value, tb
 
             # Avoid hashing random types we don't know anything
@@ -553,7 +556,6 @@ def exceptions_from_error(
     exc_type,  # type: Optional[type]
     exc_value,  # type: Optional[BaseException]
     tb,  # type: Optional[TracebackType]
-    client_options=None,  # type: Optional[Dict[str, Any]]
     mechanism=None,  # type: Optional[Dict[str, Any]]
     exception_id=0,  # type: int
     parent_id=0,  # type: int
@@ -563,16 +565,12 @@ def exceptions_from_error(
     """
     Creates the list of exceptions.
     This can include chained exceptions and exceptions from an ExceptionGroup.
-
-    See the Exception Interface documentation for more details:
-    https://develop.sentry.dev/sdk/event-payloads/exception/
     """
 
     parent = single_exception_from_error_tuple(
         exc_type=exc_type,
         exc_value=exc_value,
         tb=tb,
-        client_options=client_options,
         mechanism=mechanism,
         exception_id=exception_id,
         parent_id=parent_id,
@@ -583,18 +581,23 @@ def exceptions_from_error(
     parent_id = exception_id
     exception_id += 1
 
-    should_supress_context = hasattr(exc_value, "__suppress_context__") and exc_value.__suppress_context__  # type: ignore
+    should_supress_context = (
+        hasattr(exc_value, "__suppress_context__") and exc_value.__suppress_context__  # type: ignore
+    )
     if should_supress_context:
         # Add direct cause.
         # The field `__cause__` is set when raised with the exception (using the `from` keyword).
-        exception_has_cause = exc_value and hasattr(exc_value, "__cause__") and exc_value.__cause__ is not None
+        exception_has_cause = (
+            exc_value
+            and hasattr(exc_value, "__cause__")
+            and exc_value.__cause__ is not None
+        )
         if exception_has_cause:
             cause = exc_value.__cause__  # type: ignore
             (exception_id, child_exceptions) = exceptions_from_error(
                 exc_type=type(cause),
                 exc_value=cause,
                 tb=getattr(cause, "__traceback__", None),
-                client_options=client_options,
                 mechanism=mechanism,
                 exception_id=exception_id,
                 source="__cause__",
@@ -604,14 +607,17 @@ def exceptions_from_error(
     else:
         # Add indirect cause.
         # The field `__context__` is assigned if another exception occurs while handling the exception.
-        exception_has_content = exc_value and hasattr(exc_value, "__context__") and exc_value.__context__ is not None
+        exception_has_content = (
+            exc_value
+            and hasattr(exc_value, "__context__")
+            and exc_value.__context__ is not None
+        )
         if exception_has_content:
             context = exc_value.__context__  # type: ignore
             (exception_id, child_exceptions) = exceptions_from_error(
                 exc_type=type(context),
                 exc_value=context,
                 tb=getattr(context, "__traceback__", None),
-                client_options=client_options,
                 mechanism=mechanism,
                 exception_id=exception_id,
                 source="__context__",
@@ -626,7 +632,6 @@ def exceptions_from_error(
                 exc_type=type(e),
                 exc_value=e,
                 tb=getattr(e, "__traceback__", None),
-                client_options=client_options,
                 mechanism=mechanism,
                 exception_id=exception_id,
                 parent_id=parent_id,
@@ -639,20 +644,20 @@ def exceptions_from_error(
 
 def exceptions_from_error_tuple(
     exc_info,  # type: ExcInfo
-    client_options=None,  # type: Optional[Dict[str, Any]]
     mechanism=None,  # type: Optional[Dict[str, Any]]
 ):
     # type: (...) -> List[Dict[str, Any]]
     exc_type, exc_value, tb = exc_info
 
-    is_exception_group = BaseExceptionGroup is not None and isinstance(exc_value, BaseExceptionGroup)
+    is_exception_group = BaseExceptionGroup is not None and isinstance(
+        exc_value, BaseExceptionGroup
+    )
 
     if is_exception_group:
         (_, exceptions) = exceptions_from_error(
             exc_type=exc_type,
             exc_value=exc_value,
             tb=tb,
-            client_options=client_options,
             mechanism=mechanism,
             exception_id=0,
             parent_id=0,
@@ -661,7 +666,9 @@ def exceptions_from_error_tuple(
     else:
         exceptions = []
         for exc_type, exc_value, tb in walk_exception_chain(exc_info):
-            exceptions.append(single_exception_from_error_tuple(exc_type, exc_value, tb, client_options, mechanism))
+            exceptions.append(
+                single_exception_from_error_tuple(exc_type, exc_value, tb, mechanism)
+            )
 
     exceptions.reverse()
 
@@ -749,11 +756,42 @@ def set_in_app_in_frames(frames, in_app_exclude, in_app_include, project_root=No
     return frames
 
 
+def exception_is_already_captured(error):
+    # type: (ExceptionArg) -> bool
+    if isinstance(error, BaseException):
+        return hasattr(error, "__posthog_exception_captured")
+    # Autocaptured exceptions are passed as a tuple from our system hooks,
+    # the second item is the exception value (the first is the exception type)
+    elif isinstance(error, tuple) and len(error) > 1:
+        return error[1] is not None and hasattr(
+            error[1], "__posthog_exception_captured"
+        )
+    else:
+        return False  # type: ignore[unreachable]
+
+
+def mark_exception_as_captured(error, uuid):
+    # type: (ExceptionArg, str) -> None
+    if isinstance(error, BaseException):
+        setattr(error, "__posthog_exception_captured", True)
+        setattr(error, "__posthog_exception_uuid", uuid)
+    # Autocaptured exceptions are passed as a tuple from our system hooks,
+    # the second item is the exception value (the first is the exception type)
+    elif isinstance(error, tuple) and len(error) > 1:
+        if error[1] is not None:
+            setattr(error[1], "__posthog_exception_captured", True)
+            setattr(error[1], "__posthog_exception_uuid", uuid)
+
+
 def exc_info_from_error(error):
-    # type: (Union[BaseException, ExcInfo]) -> ExcInfo
+    # type: (ExceptionArg) -> ExcInfo
     if isinstance(error, tuple) and len(error) == 3:
         exc_type, exc_value, tb = error
     elif isinstance(error, BaseException):
+        try:
+            construct_artificial_traceback(error)
+        except Exception:
+            pass
         tb = getattr(error, "__traceback__", None)
         if tb is not None:
             exc_type = type(error)
@@ -778,21 +816,29 @@ def exc_info_from_error(error):
     return exc_info
 
 
-def event_from_exception(
-    exc_info,  # type: Union[BaseException, ExcInfo]
-    client_options=None,  # type: Optional[Dict[str, Any]]
-    mechanism=None,  # type: Optional[Dict[str, Any]]
-):
-    # type: (...) -> Tuple[Event, Dict[str, Any]]
-    exc_info = exc_info_from_error(exc_info)
-    hint = event_hint_with_exc_info(exc_info)
-    return (
-        {
-            "level": "error",
-            "exception": {"values": exceptions_from_error_tuple(exc_info, client_options, mechanism)},
-        },
-        hint,
-    )
+def construct_artificial_traceback(e):
+    # type: (BaseException) -> None
+    if getattr(e, "__traceback__", None) is not None:
+        return
+
+    depth = 0
+    frames = []
+    while True:
+        try:
+            frame = sys._getframe(depth)
+            depth += 1
+        except ValueError:
+            break
+
+        frames.append(frame)
+
+    frames.reverse()
+
+    tb = None
+    for frame in frames:
+        tb = types.TracebackType(tb, frame, frame.f_lasti, frame.f_lineno)
+
+    setattr(e, "__traceback__", tb)
 
 
 def _module_in_list(name, items):
@@ -813,7 +859,9 @@ def _module_in_list(name, items):
 def _is_external_source(abs_path):
     # type: (str) -> bool
     # check if frame is in 'site-packages' or 'dist-packages'
-    external_source = re.search(r"[\\/](?:dist|site)-packages[\\/]", abs_path) is not None
+    external_source = (
+        re.search(r"[\\/](?:dist|site)-packages[\\/]", abs_path) is not None
+    )
     return external_source
 
 
@@ -874,3 +922,157 @@ def strip_string(value, max_length=None):
             "rem": [["!limit", "x", max_length - 3, max_length]],
         },
     )
+
+
+def _compile_patterns(patterns):
+    compiled = []
+    for pattern in patterns:
+        try:
+            compiled.append(re.compile(pattern))
+        except:
+            pass
+    return compiled
+
+
+def _pattern_matches(name, patterns):
+    for pattern in patterns:
+        if pattern.search(name):
+            return True
+    return False
+
+
+def _serialize_variable_value(value, limiter, max_length=1024):
+    try:
+        if value is None:
+            result = "None"
+        elif isinstance(value, bool):
+            result = str(value)
+        elif isinstance(value, (int, float)):
+            result_size = len(str(value))
+            if not limiter.can_add(result_size):
+                return None
+            limiter.add(result_size)
+            return value
+        elif isinstance(value, str):
+            result = value
+        else:
+            result = json.dumps(value)
+
+        if len(result) > max_length:
+            result = result[: max_length - 3] + "..."
+
+        result_size = len(result)
+        if not limiter.can_add(result_size):
+            return None
+        limiter.add(result_size)
+
+        return result
+    except Exception:
+        try:
+            fallback = f"<{type(value).__name__}>"
+            fallback_size = len(fallback)
+            if not limiter.can_add(fallback_size):
+                return None
+            limiter.add(fallback_size)
+            return fallback
+        except Exception:
+            fallback = "<unserializable object>"
+            fallback_size = len(fallback)
+            if not limiter.can_add(fallback_size):
+                return None
+            limiter.add(fallback_size)
+            return fallback
+
+
+def _is_simple_type(value):
+    return isinstance(value, (type(None), bool, int, float, str))
+
+
+def serialize_code_variables(
+    frame, limiter, mask_patterns=None, ignore_patterns=None, max_length=1024
+):
+    if mask_patterns is None:
+        mask_patterns = []
+    if ignore_patterns is None:
+        ignore_patterns = []
+
+    compiled_mask = _compile_patterns(mask_patterns)
+    compiled_ignore = _compile_patterns(ignore_patterns)
+
+    try:
+        local_vars = frame.f_locals.copy()
+    except Exception:
+        return {}
+
+    simple_vars = {}
+    complex_vars = {}
+
+    for name, value in local_vars.items():
+        if _pattern_matches(name, compiled_ignore):
+            continue
+
+        if _is_simple_type(value):
+            simple_vars[name] = value
+        else:
+            complex_vars[name] = value
+
+    result = {}
+
+    all_vars = {**simple_vars, **complex_vars}
+    ordered_names = list(sorted(simple_vars.keys())) + list(sorted(complex_vars.keys()))
+
+    for name in ordered_names:
+        value = all_vars[name]
+
+        if _pattern_matches(name, compiled_mask):
+            redacted_value = CODE_VARIABLES_REDACTED_VALUE
+            redacted_size = len(redacted_value)
+            if not limiter.can_add(redacted_size):
+                break
+            limiter.add(redacted_size)
+            result[name] = redacted_value
+        else:
+            serialized = _serialize_variable_value(value, limiter, max_length)
+            if serialized is None:
+                break
+            result[name] = serialized
+
+    return result
+
+
+def try_attach_code_variables_to_frames(
+    all_exceptions, exc_info, mask_patterns, ignore_patterns
+):
+    exc_type, exc_value, traceback = exc_info
+
+    if traceback is None:
+        return
+
+    tb_frames = list(iter_stacks(traceback))
+
+    if not tb_frames:
+        return
+
+    limiter = VariableSizeLimiter()
+
+    for exception in all_exceptions:
+        stacktrace = exception.get("stacktrace")
+        if not stacktrace or "frames" not in stacktrace:
+            continue
+
+        serialized_frames = stacktrace["frames"]
+
+        for serialized_frame, tb_item in zip(serialized_frames, tb_frames):
+            if not serialized_frame.get("in_app"):
+                continue
+
+            variables = serialize_code_variables(
+                tb_item.tb_frame,
+                limiter,
+                mask_patterns=mask_patterns,
+                ignore_patterns=ignore_patterns,
+                max_length=1024,
+            )
+
+            if variables:
+                serialized_frame["code_variables"] = variables

posthog/feature_flags.py

@@ -22,6 +22,18 @@ class InconclusiveMatchError(Exception):
     pass
 
 
+class RequiresServerEvaluation(Exception):
+    """
+    Raised when feature flag evaluation requires server-side data that is not
+    available locally (e.g., static cohorts, experience continuity).
+
+    This error should propagate immediately to trigger API fallback, unlike
+    InconclusiveMatchError which allows trying other conditions.
+    """
+
+    pass
+
+
 # This function takes a distinct_id and a feature flag key and returns a float between 0 and 1.
 # Given the same distinct_id and key, it'll always return the same float. These floats are
 # uniformly distributed between 0 and 1, so if we want to show this feature to 20% of traffic
@@ -43,58 +55,251 @@ def get_matching_variant(flag, distinct_id):
 def variant_lookup_table(feature_flag):
     lookup_table = []
     value_min = 0
-    multivariates = ((feature_flag.get("filters") or {}).get("multivariate") or {}).get("variants") or []
+    multivariates = ((feature_flag.get("filters") or {}).get("multivariate") or {}).get(
+        "variants"
+    ) or []
     for variant in multivariates:
         value_max = value_min + variant["rollout_percentage"] / 100
-        lookup_table.append({"value_min": value_min, "value_max": value_max, "key": variant["key"]})
+        lookup_table.append(
+            {"value_min": value_min, "value_max": value_max, "key": variant["key"]}
+        )
         value_min = value_max
     return lookup_table
 
 
-def match_feature_flag_properties(flag, distinct_id, properties, cohort_properties=None) -> FlagValue:
+def evaluate_flag_dependency(
+    property, flags_by_key, evaluation_cache, distinct_id, properties, cohort_properties
+):
+    """
+    Evaluate a flag dependency property according to the dependency chain algorithm.
+
+    Args:
+        property: Flag property with type="flag" and dependency_chain
+        flags_by_key: Dictionary of all flags by their key
+        evaluation_cache: Cache for storing evaluation results
+        distinct_id: The distinct ID being evaluated
+        properties: Person properties for evaluation
+        cohort_properties: Cohort properties for evaluation
+
+    Returns:
+        bool: True if all dependencies in the chain evaluate to True, False otherwise
+    """
+    if flags_by_key is None or evaluation_cache is None:
+        # Cannot evaluate flag dependencies without required context
+        raise InconclusiveMatchError(
+            f"Cannot evaluate flag dependency on '{property.get('key', 'unknown')}' without flags_by_key and evaluation_cache"
+        )
+
+    # Check if dependency_chain is present - it should always be provided for flag dependencies
+    if "dependency_chain" not in property:
+        # Missing dependency_chain indicates malformed server data
+        raise InconclusiveMatchError(
+            f"Flag dependency property for '{property.get('key', 'unknown')}' is missing required 'dependency_chain' field"
+        )
+
+    dependency_chain = property["dependency_chain"]
+
+    # Handle circular dependency (empty chain means circular)
+    if len(dependency_chain) == 0:
+        log.debug(f"Circular dependency detected for flag: {property.get('key')}")
+        raise InconclusiveMatchError(
+            f"Circular dependency detected for flag '{property.get('key', 'unknown')}'"
+        )
+
+    # Evaluate all dependencies in the chain order
+    for dep_flag_key in dependency_chain:
+        if dep_flag_key not in evaluation_cache:
+            # Need to evaluate this dependency first
+            dep_flag = flags_by_key.get(dep_flag_key)
+            if not dep_flag:
+                # Missing flag dependency - cannot evaluate locally
+                evaluation_cache[dep_flag_key] = None
+                raise InconclusiveMatchError(
+                    f"Cannot evaluate flag dependency '{dep_flag_key}' - flag not found in local flags"
+                )
+            else:
+                # Check if the flag is active (same check as in client._compute_flag_locally)
+                if not dep_flag.get("active"):
+                    evaluation_cache[dep_flag_key] = False
+                else:
+                    # Recursively evaluate the dependency
+                    try:
+                        dep_result = match_feature_flag_properties(
+                            dep_flag,
+                            distinct_id,
+                            properties,
+                            cohort_properties,
+                            flags_by_key,
+                            evaluation_cache,
+                        )
+                        evaluation_cache[dep_flag_key] = dep_result
+                    except InconclusiveMatchError as e:
+                        # If we can't evaluate a dependency, store None and propagate the error
+                        evaluation_cache[dep_flag_key] = None
+                        raise InconclusiveMatchError(
+                            f"Cannot evaluate flag dependency '{dep_flag_key}': {e}"
+                        ) from e
+
+        # Check the cached result
+        cached_result = evaluation_cache[dep_flag_key]
+        if cached_result is None:
+            # Previously inconclusive - raise error again
+            raise InconclusiveMatchError(
+                f"Flag dependency '{dep_flag_key}' was previously inconclusive"
+            )
+        elif not cached_result:
+            # Definitive False result - dependency failed
+            return False
+
+    # All dependencies in the chain have been evaluated successfully
+    # Now check if the final flag value matches the expected value in the property
+    flag_key = property.get("key")
+    expected_value = property.get("value")
+    operator = property.get("operator", "exact")
+
+    if flag_key and expected_value is not None:
+        # Get the actual value of the flag we're checking
+        actual_value = evaluation_cache.get(flag_key)
+
+        if actual_value is None:
+            # Flag wasn't evaluated - this shouldn't happen if dependency chain is correct
+            raise InconclusiveMatchError(
+                f"Flag '{flag_key}' was not evaluated despite being in dependency chain"
+            )
+
+        # For flag dependencies, we need to compare the actual flag result with expected value
+        # using the flag_evaluates_to operator logic
+        if operator == "flag_evaluates_to":
+            return matches_dependency_value(expected_value, actual_value)
+        else:
+            # This should never happen, but just to be defensive.
+            raise InconclusiveMatchError(
+                f"Flag dependency property for '{property.get('key', 'unknown')}' has invalid operator '{operator}'"
+            )
+
+    # If no value check needed, return True (all dependencies passed)
+    return True
+
+
+def matches_dependency_value(expected_value, actual_value):
+    """
+    Check if the actual flag value matches the expected dependency value.
+
+    This follows the same logic as the C# MatchesDependencyValue function:
+    - String variant case: check for exact match or boolean true
+    - Boolean case: must match expected boolean value
+
+    Args:
+        expected_value: The expected value from the property
+        actual_value: The actual value returned by the flag evaluation
+
+    Returns:
+        bool: True if the values match according to flag dependency rules
+    """
+    # String variant case - check for exact match or boolean true
+    if isinstance(actual_value, str) and len(actual_value) > 0:
+        if isinstance(expected_value, bool):
+            # Any variant matches boolean true
+            return expected_value
+        elif isinstance(expected_value, str):
+            # variants are case-sensitive, hence our comparison is too
+            return actual_value == expected_value
+        else:
+            return False
+
+    # Boolean case - must match expected boolean value
+    elif isinstance(actual_value, bool) and isinstance(expected_value, bool):
+        return actual_value == expected_value
+
+    # Default case
+    return False
+
+
+def match_feature_flag_properties(
+    flag,
+    distinct_id,
+    properties,
+    cohort_properties=None,
+    flags_by_key=None,
+    evaluation_cache=None,
+) -> FlagValue:
     flag_conditions = (flag.get("filters") or {}).get("groups") or []
     is_inconclusive = False
     cohort_properties = cohort_properties or {}
     # Some filters can be explicitly set to null, which require accessing variants like so
-    flag_variants = ((flag.get("filters") or {}).get("multivariate") or {}).get("variants") or []
+    flag_variants = ((flag.get("filters") or {}).get("multivariate") or {}).get(
+        "variants"
+    ) or []
     valid_variant_keys = [variant["key"] for variant in flag_variants]
 
-    # Stable sort conditions with variant overrides to the top. This ensures that if overrides are present, they are
-    # evaluated first, and the variant override is applied to the first matching condition.
-    sorted_flag_conditions = sorted(
-        flag_conditions,
-        key=lambda condition: 0 if condition.get("variant") else 1,
-    )
-
-    for condition in sorted_flag_conditions:
+    for condition in flag_conditions:
         try:
             # if any one condition resolves to True, we can shortcircuit and return
             # the matching variant
-            if is_condition_match(flag, distinct_id, condition, properties, cohort_properties):
+            if is_condition_match(
+                flag,
+                distinct_id,
+                condition,
+                properties,
+                cohort_properties,
+                flags_by_key,
+                evaluation_cache,
+            ):
                 variant_override = condition.get("variant")
                 if variant_override and variant_override in valid_variant_keys:
                     variant = variant_override
                 else:
                     variant = get_matching_variant(flag, distinct_id)
                 return variant or True
+        except RequiresServerEvaluation:
+            # Static cohort or other missing server-side data - must fallback to API
+            raise
         except InconclusiveMatchError:
+            # Evaluation error (bad regex, invalid date, missing property, etc.)
+            # Track that we had an inconclusive match, but try other conditions
             is_inconclusive = True
 
     if is_inconclusive:
-        raise InconclusiveMatchError("Can't determine if feature flag is enabled or not with given properties")
+        raise InconclusiveMatchError(
+            "Can't determine if feature flag is enabled or not with given properties"
+        )
 
     # We can only return False when either all conditions are False, or
     # no condition was inconclusive.
     return False
 
 
-def is_condition_match(feature_flag, distinct_id, condition, properties, cohort_properties) -> bool:
+def is_condition_match(
+    feature_flag,
+    distinct_id,
+    condition,
+    properties,
+    cohort_properties,
+    flags_by_key=None,
+    evaluation_cache=None,
+) -> bool:
     rollout_percentage = condition.get("rollout_percentage")
     if len(condition.get("properties") or []) > 0:
         for prop in condition.get("properties"):
             property_type = prop.get("type")
             if property_type == "cohort":
-                matches = match_cohort(prop, properties, cohort_properties)
+                matches = match_cohort(
+                    prop,
+                    properties,
+                    cohort_properties,
+                    flags_by_key,
+                    evaluation_cache,
+                    distinct_id,
+                )
+            elif property_type == "flag":
+                matches = evaluate_flag_dependency(
+                    prop,
+                    flags_by_key,
+                    evaluation_cache,
+                    distinct_id,
+                    properties,
+                    cohort_properties,
+                )
             else:
                 matches = match_property(prop, properties)
             if not matches:
@@ -103,7 +308,9 @@ def is_condition_match(feature_flag, distinct_id, condition, properties, cohort_
         if rollout_percentage is None:
             return True
 
-    if rollout_percentage is not None and _hash(feature_flag["key"], distinct_id) > (rollout_percentage / 100):
+    if rollout_percentage is not None and _hash(feature_flag["key"], distinct_id) > (
+        rollout_percentage / 100
+    ):
         return False
 
     return True
@@ -117,7 +324,9 @@ def match_property(property, property_values) -> bool:
     value = property.get("value")
 
     if key not in property_values:
-        raise InconclusiveMatchError("can't match properties without a given property value")
+        raise InconclusiveMatchError(
+            "can't match properties without a given property value"
+        )
 
     if operator == "is_not_set":
         raise InconclusiveMatchError("can't match properties with operator is_not_set")
@@ -131,7 +340,9 @@ def match_property(property, property_values) -> bool:
 
         def compute_exact_match(value, override_value):
             if isinstance(value, list):
-                return str(override_value).casefold() in [str(val).casefold() for val in value]
+                return str(override_value).casefold() in [
+                    str(val).casefold() for val in value
+                ]
             return utils.str_iequals(value, override_value)
 
         if operator == "exact":
@@ -149,10 +360,16 @@ def match_property(property, property_values) -> bool:
         return not utils.str_icontains(override_value, value)
 
     if operator == "regex":
-        return is_valid_regex(str(value)) and re.compile(str(value)).search(str(override_value)) is not None
+        return (
+            is_valid_regex(str(value))
+            and re.compile(str(value)).search(str(override_value)) is not None
+        )
 
     if operator == "not_regex":
-        return is_valid_regex(str(value)) and re.compile(str(value)).search(str(override_value)) is None
+        return (
+            is_valid_regex(str(value))
+            and re.compile(str(value)).search(str(override_value)) is None
+        )
 
     if operator in ("gt", "gte", "lt", "lte"):
         # :TRICKY: We adjust comparison based on the override value passed in,
@@ -191,10 +408,14 @@ def match_property(property, property_values) -> bool:
                 parsed_date = parser.parse(str(value))
                 parsed_date = convert_to_datetime_aware(parsed_date)
         except Exception as e:
-            raise InconclusiveMatchError("The date set on the flag is not a valid format") from e
+            raise InconclusiveMatchError(
+                "The date set on the flag is not a valid format"
+            ) from e
 
         if not parsed_date:
-            raise InconclusiveMatchError("The date set on the flag is not a valid format")
+            raise InconclusiveMatchError(
+                "The date set on the flag is not a valid format"
+            )
 
         if isinstance(override_value, datetime.datetime):
             override_date = convert_to_datetime_aware(override_value)
@@ -218,13 +439,22 @@ def match_property(property, property_values) -> bool:
             except Exception:
                 raise InconclusiveMatchError("The date provided is not a valid format")
         else:
-            raise InconclusiveMatchError("The date provided must be a string or date object")
+            raise InconclusiveMatchError(
+                "The date provided must be a string or date object"
+            )
 
     # if we get here, we don't know how to handle the operator
     raise InconclusiveMatchError(f"Unknown operator {operator}")
 
 
-def match_cohort(property, property_values, cohort_properties) -> bool:
+def match_cohort(
+    property,
+    property_values,
+    cohort_properties,
+    flags_by_key=None,
+    evaluation_cache=None,
+    distinct_id=None,
+) -> bool:
     # Cohort properties are in the form of property groups like this:
     # {
     #     "cohort_id": {
@@ -236,13 +466,29 @@ def match_cohort(property, property_values, cohort_properties) -> bool:
     # }
     cohort_id = str(property.get("value"))
     if cohort_id not in cohort_properties:
-        raise InconclusiveMatchError("can't match cohort without a given cohort property value")
+        raise RequiresServerEvaluation(
+            f"cohort {cohort_id} not found in local cohorts - likely a static cohort that requires server evaluation"
+        )
 
     property_group = cohort_properties[cohort_id]
-    return match_property_group(property_group, property_values, cohort_properties)
+    return match_property_group(
+        property_group,
+        property_values,
+        cohort_properties,
+        flags_by_key,
+        evaluation_cache,
+        distinct_id,
+    )
 
 
-def match_property_group(property_group, property_values, cohort_properties) -> bool:
+def match_property_group(
+    property_group,
+    property_values,
+    cohort_properties,
+    flags_by_key=None,
+    evaluation_cache=None,
+    distinct_id=None,
+) -> bool:
     if not property_group:
         return True
 
@@ -259,7 +505,14 @@ def match_property_group(property_group, property_values, cohort_properties) ->
         # a nested property group
         for prop in properties:
             try:
-                matches = match_property_group(prop, property_values, cohort_properties)
+                matches = match_property_group(
+                    prop,
+                    property_values,
+                    cohort_properties,
+                    flags_by_key,
+                    evaluation_cache,
+                    distinct_id,
+                )
                 if property_group_type == "AND":
                     if not matches:
                         return False
@@ -267,12 +520,17 @@ def match_property_group(property_group, property_values, cohort_properties) ->
                     # OR group
                     if matches:
                         return True
+            except RequiresServerEvaluation:
+                # Immediately propagate - this condition requires server-side data
+                raise
             except InconclusiveMatchError as e:
                 log.debug(f"Failed to compute property {prop} locally: {e}")
                 error_matching_locally = True
 
         if error_matching_locally:
-            raise InconclusiveMatchError("Can't match cohort without a given cohort property value")
+            raise InconclusiveMatchError(
+                "Can't match cohort without a given cohort property value"
+            )
         # if we get here, all matched in AND case, or none matched in OR case
         return property_group_type == "AND"
 
@@ -280,7 +538,23 @@ def match_property_group(property_group, property_values, cohort_properties) ->
         for prop in properties:
             try:
                 if prop.get("type") == "cohort":
-                    matches = match_cohort(prop, property_values, cohort_properties)
+                    matches = match_cohort(
+                        prop,
+                        property_values,
+                        cohort_properties,
+                        flags_by_key,
+                        evaluation_cache,
+                        distinct_id,
+                    )
+                elif prop.get("type") == "flag":
+                    matches = evaluate_flag_dependency(
+                        prop,
+                        flags_by_key,
+                        evaluation_cache,
+                        distinct_id,
+                        property_values,
+                        cohort_properties,
+                    )
                 else:
                     matches = match_property(prop, property_values)
 
@@ -298,18 +572,25 @@ def match_property_group(property_group, property_values, cohort_properties) ->
                         return True
                     if not matches and negation:
                         return True
+            except RequiresServerEvaluation:
+                # Immediately propagate - this condition requires server-side data
+                raise
             except InconclusiveMatchError as e:
                 log.debug(f"Failed to compute property {prop} locally: {e}")
                 error_matching_locally = True
 
         if error_matching_locally:
-            raise InconclusiveMatchError("can't match cohort without a given cohort property value")
+            raise InconclusiveMatchError(
+                "can't match cohort without a given cohort property value"
+            )
 
         # if we get here, all matched in AND case, or none matched in OR case
         return property_group_type == "AND"
 
 
-def relative_date_parse_for_feature_flag_matching(value: str) -> Optional[datetime.datetime]:
+def relative_date_parse_for_feature_flag_matching(
+    value: str,
+) -> Optional[datetime.datetime]:
     regex = r"^-?(?P<number>[0-9]+)(?P<interval>[a-z])$"
     match = re.search(regex, value)
     parsed_dt = datetime.datetime.now(datetime.timezone.utc)

posthog/integrations/django.py

@@ -0,0 +1,319 @@
+from typing import TYPE_CHECKING, cast
+from posthog import contexts
+from posthog.client import Client
+
+try:
+    from asgiref.sync import iscoroutinefunction, markcoroutinefunction
+except ImportError:
+    # Fallback for older Django versions without asgiref
+    import asyncio
+
+    iscoroutinefunction = asyncio.iscoroutinefunction
+
+    # No-op fallback for markcoroutinefunction
+    # Older Django versions without asgiref typically don't support async middleware anyway
+    def markcoroutinefunction(func):
+        return func
+
+
+if TYPE_CHECKING:
+    from django.http import HttpRequest, HttpResponse  # noqa: F401
+    from typing import Callable, Dict, Any, Optional, Union, Awaitable  # noqa: F401
+
+
+class PosthogContextMiddleware:
+    """Middleware to automatically track Django requests.
+
+    This middleware wraps all calls with a posthog context. It attempts to extract the following from the request headers:
+    - Session ID, (extracted from `X-POSTHOG-SESSION-ID`)
+    - Distinct ID, (extracted from `X-POSTHOG-DISTINCT-ID`)
+    - Request URL as $current_url
+    - Request Method as $request_method
+
+    The context will also auto-capture exceptions and send them to PostHog, unless you disable it by setting
+    `POSTHOG_MW_CAPTURE_EXCEPTIONS` to `False` in your Django settings. The exceptions are captured using the
+    global client, unless the setting `POSTHOG_MW_CLIENT` is set to a custom client instance
+
+    The middleware behaviour is customisable through 3 additional functions:
+    - `POSTHOG_MW_EXTRA_TAGS`, which is a Callable[[HttpRequest], Dict[str, Any]] expected to return a dictionary of additional tags to be added to the context.
+    - `POSTHOG_MW_REQUEST_FILTER`, which is a Callable[[HttpRequest], bool] expected to return `False` if the request should not be tracked.
+    - `POSTHOG_MW_TAG_MAP`, which is a Callable[[Dict[str, Any]], Dict[str, Any]], which you can use to modify the tags before they're added to the context.
+
+    You can use the `POSTHOG_MW_TAG_MAP` function to remove any default tags you don't want to capture, or override them with your own values.
+
+    Context tags are automatically included as properties on all events captured within a context, including exceptions.
+    See the context documentation for more information. The extracted distinct ID and session ID, if found, are used to
+    associate all events captured in the middleware context with the same distinct ID and session as currently active on the
+    frontend. See the documentation for `set_context_session` and `identify_context` for more details.
+
+    This middleware is hybrid-capable: it supports both WSGI (sync) and ASGI (async) Django applications. The middleware
+    detects at initialization whether the next middleware in the chain is async or sync, and adapts its behavior accordingly.
+    This ensures compatibility with both pure sync and pure async middleware chains, as well as mixed chains in ASGI mode.
+    """
+
+    sync_capable = True
+    async_capable = True
+
+    def __init__(self, get_response):
+        # type: (Union[Callable[[HttpRequest], HttpResponse], Callable[[HttpRequest], Awaitable[HttpResponse]]]) -> None
+        self.get_response = get_response
+        self._is_coroutine = iscoroutinefunction(get_response)
+
+        # Mark this instance as a coroutine function if get_response is async
+        # This is required for Django to correctly detect async middleware
+        if self._is_coroutine:
+            markcoroutinefunction(self)
+
+        from django.conf import settings
+
+        if hasattr(settings, "POSTHOG_MW_EXTRA_TAGS") and callable(
+            settings.POSTHOG_MW_EXTRA_TAGS
+        ):
+            self.extra_tags = cast(
+                "Optional[Callable[[HttpRequest], Dict[str, Any]]]",
+                settings.POSTHOG_MW_EXTRA_TAGS,
+            )
+        else:
+            self.extra_tags = None
+
+        if hasattr(settings, "POSTHOG_MW_REQUEST_FILTER") and callable(
+            settings.POSTHOG_MW_REQUEST_FILTER
+        ):
+            self.request_filter = cast(
+                "Optional[Callable[[HttpRequest], bool]]",
+                settings.POSTHOG_MW_REQUEST_FILTER,
+            )
+        else:
+            self.request_filter = None
+
+        if hasattr(settings, "POSTHOG_MW_TAG_MAP") and callable(
+            settings.POSTHOG_MW_TAG_MAP
+        ):
+            self.tag_map = cast(
+                "Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]",
+                settings.POSTHOG_MW_TAG_MAP,
+            )
+        else:
+            self.tag_map = None
+
+        if hasattr(settings, "POSTHOG_MW_CAPTURE_EXCEPTIONS") and isinstance(
+            settings.POSTHOG_MW_CAPTURE_EXCEPTIONS, bool
+        ):
+            self.capture_exceptions = settings.POSTHOG_MW_CAPTURE_EXCEPTIONS
+        else:
+            self.capture_exceptions = True
+
+        if hasattr(settings, "POSTHOG_MW_CLIENT") and isinstance(
+            settings.POSTHOG_MW_CLIENT, Client
+        ):
+            self.client = cast("Optional[Client]", settings.POSTHOG_MW_CLIENT)
+        else:
+            self.client = None
+
+    def extract_tags(self, request):
+        # type: (HttpRequest) -> Dict[str, Any]
+        """Extract tags from request in sync context."""
+        user_id, user_email = self.extract_request_user(request)
+        return self._build_tags(request, user_id, user_email)
+
+    def _build_tags(self, request, user_id, user_email):
+        # type: (HttpRequest, Optional[str], Optional[str]) -> Dict[str, Any]
+        """
+        Build tags dict from request and user info.
+
+        Centralized tag extraction logic used by both sync and async paths.
+        """
+        tags = {}
+
+        # Extract session ID from X-POSTHOG-SESSION-ID header
+        session_id = request.headers.get("X-POSTHOG-SESSION-ID")
+        if session_id:
+            contexts.set_context_session(session_id)
+
+        # Extract distinct ID from X-POSTHOG-DISTINCT-ID header or request user id
+        distinct_id = request.headers.get("X-POSTHOG-DISTINCT-ID") or user_id
+        if distinct_id:
+            contexts.identify_context(distinct_id)
+
+        # Extract user email
+        if user_email:
+            tags["email"] = user_email
+
+        # Extract current URL
+        absolute_url = request.build_absolute_uri()
+        if absolute_url:
+            tags["$current_url"] = absolute_url
+
+        # Extract request method
+        if request.method:
+            tags["$request_method"] = request.method
+
+        # Extract request path
+        if request.path:
+            tags["$request_path"] = request.path
+
+        # Extract IP address
+        ip_address = request.headers.get("X-Forwarded-For")
+        if ip_address:
+            tags["$ip_address"] = ip_address
+
+        # Extract user agent
+        user_agent = request.headers.get("User-Agent")
+        if user_agent:
+            tags["$user_agent"] = user_agent
+
+        # Apply extra tags if configured
+        if self.extra_tags:
+            extra = self.extra_tags(request)
+            if extra:
+                tags.update(extra)
+
+        # Apply tag mapping if configured
+        if self.tag_map:
+            tags = self.tag_map(tags)
+
+        return tags
+
+    def extract_request_user(self, request):
+        # type: (HttpRequest) -> tuple[Optional[str], Optional[str]]
+        """Extract user ID and email from request in sync context."""
+        user = getattr(request, "user", None)
+        return self._resolve_user_details(user)
+
+    async def aextract_tags(self, request):
+        # type: (HttpRequest) -> Dict[str, Any]
+        """
+        Async version of extract_tags for use in async request handling.
+
+        Uses await request.auser() instead of request.user to avoid
+        SynchronousOnlyOperation in async context.
+
+        Follows Django's naming convention for async methods (auser, asave, etc.).
+        """
+        user_id, user_email = await self.aextract_request_user(request)
+        return self._build_tags(request, user_id, user_email)
+
+    async def aextract_request_user(self, request):
+        # type: (HttpRequest) -> tuple[Optional[str], Optional[str]]
+        """
+        Async version of extract_request_user for use in async request handling.
+
+        Uses await request.auser() instead of request.user to avoid
+        SynchronousOnlyOperation in async context.
+
+        Follows Django's naming convention for async methods (auser, asave, etc.).
+        """
+        auser = getattr(request, "auser", None)
+        if callable(auser):
+            try:
+                user = await auser()
+                return self._resolve_user_details(user)
+            except Exception:
+                # If auser() fails, return empty - don't break the request
+                # Real errors (permissions, broken auth) will be logged by Django
+                return None, None
+
+        # Fallback for test requests without auser
+        return None, None
+
+    def _resolve_user_details(self, user):
+        # type: (Any) -> tuple[Optional[str], Optional[str]]
+        """
+        Extract user ID and email from a user object.
+
+        Handles both authenticated and unauthenticated users, as well as
+        legacy Django where is_authenticated was a method.
+        """
+        user_id = None
+        email = None
+
+        if user is None:
+            return user_id, email
+
+        # Handle is_authenticated (property in modern Django, method in legacy)
+        is_authenticated = getattr(user, "is_authenticated", False)
+        if callable(is_authenticated):
+            is_authenticated = is_authenticated()
+
+        if not is_authenticated:
+            return user_id, email
+
+        # Extract user primary key
+        user_pk = getattr(user, "pk", None)
+        if user_pk is not None:
+            user_id = str(user_pk)
+
+        # Extract user email
+        user_email = getattr(user, "email", None)
+        if user_email:
+            email = str(user_email)
+
+        return user_id, email
+
+    def __call__(self, request):
+        # type: (HttpRequest) -> Union[HttpResponse, Awaitable[HttpResponse]]
+        """
+        Unified entry point for both sync and async request handling.
+
+        When sync_capable and async_capable are both True, Django passes requests
+        without conversion. This method detects the mode and routes accordingly.
+        """
+        if self._is_coroutine:
+            return self.__acall__(request)
+        else:
+            # Synchronous path
+            if self.request_filter and not self.request_filter(request):
+                return self.get_response(request)
+
+            with contexts.new_context(self.capture_exceptions, client=self.client):
+                for k, v in self.extract_tags(request).items():
+                    contexts.tag(k, v)
+
+                return self.get_response(request)
+
+    async def __acall__(self, request):
+        # type: (HttpRequest) -> Awaitable[HttpResponse]
+        """
+        Asynchronous entry point for async request handling.
+
+        This method is called when the middleware chain is async.
+        Uses aextract_tags() which calls request.auser() to avoid
+        SynchronousOnlyOperation when accessing user in async context.
+        """
+        if self.request_filter and not self.request_filter(request):
+            return await self.get_response(request)
+
+        with contexts.new_context(self.capture_exceptions, client=self.client):
+            for k, v in (await self.aextract_tags(request)).items():
+                contexts.tag(k, v)
+
+            return await self.get_response(request)
+
+    def process_exception(self, request, exception):
+        # type: (HttpRequest, Exception) -> None
+        """
+        Process exceptions from views and downstream middleware.
+
+        Django calls this WHILE still inside the context created by __call__,
+        so request tags have already been extracted and set. This method just
+        needs to capture the exception directly.
+
+        Django converts view exceptions into responses before they propagate through
+        the middleware stack, so the context manager in __call__/__acall__ never sees them.
+
+        Note: Django's process_exception is always synchronous, even for async views.
+        """
+        if self.request_filter and not self.request_filter(request):
+            return
+
+        if not self.capture_exceptions:
+            return
+
+        # Context and tags already set by __call__ or __acall__
+        # Just capture the exception
+        if self.client:
+            self.client.capture_exception(exception)
+        else:
+            from posthog import capture_exception
+
+            capture_exception(exception)

posthog/request.py

@@ -43,7 +43,12 @@ def determine_server_host(host: Optional[str]) -> str:
 
 
 def post(
-    api_key: str, host: Optional[str] = None, path=None, gzip: bool = False, timeout: int = 15, **kwargs
+    api_key: str,
+    host: Optional[str] = None,
+    path=None,
+    gzip: bool = False,
+    timeout: int = 15,
+    **kwargs,
 ) -> requests.Response:
     """Post the `kwargs` to the API"""
     log = logging.getLogger("posthog")
@@ -100,34 +105,71 @@ def _process_response(
         raise APIError(res.status_code, res.text)
 
 
-def decide(api_key: str, host: Optional[str] = None, gzip: bool = False, timeout: int = 15, **kwargs) -> Any:
+def decide(
+    api_key: str,
+    host: Optional[str] = None,
+    gzip: bool = False,
+    timeout: int = 15,
+    **kwargs,
+) -> Any:
     """Post the `kwargs to the decide API endpoint"""
     res = post(api_key, host, "/decide/?v=4", gzip, timeout, **kwargs)
     return _process_response(res, success_message="Feature flags decided successfully")
 
 
-def flags(api_key: str, host: Optional[str] = None, gzip: bool = False, timeout: int = 15, **kwargs) -> Any:
+def flags(
+    api_key: str,
+    host: Optional[str] = None,
+    gzip: bool = False,
+    timeout: int = 15,
+    **kwargs,
+) -> Any:
     """Post the `kwargs to the flags API endpoint"""
     res = post(api_key, host, "/flags/?v=2", gzip, timeout, **kwargs)
-    return _process_response(res, success_message="Feature flags evaluated successfully")
+    return _process_response(
+        res, success_message="Feature flags evaluated successfully"
+    )
 
 
-def remote_config(personal_api_key: str, host: Optional[str] = None, key: str = "", timeout: int = 15) -> Any:
+def remote_config(
+    personal_api_key: str,
+    project_api_key: str,
+    host: Optional[str] = None,
+    key: str = "",
+    timeout: int = 15,
+) -> Any:
     """Get remote config flag value from remote_config API endpoint"""
-    return get(personal_api_key, f"/api/projects/@current/feature_flags/{key}/remote_config/", host, timeout)
+    return get(
+        personal_api_key,
+        f"/api/projects/@current/feature_flags/{key}/remote_config?token={project_api_key}",
+        host,
+        timeout,
+    )
 
 
 def batch_post(
-    api_key: str, host: Optional[str] = None, gzip: bool = False, timeout: int = 15, **kwargs
+    api_key: str,
+    host: Optional[str] = None,
+    gzip: bool = False,
+    timeout: int = 15,
+    **kwargs,
 ) -> requests.Response:
     """Post the `kwargs` to the batch API endpoint for events"""
     res = post(api_key, host, "/batch/", gzip, timeout, **kwargs)
-    return _process_response(res, success_message="data uploaded successfully", return_json=False)
+    return _process_response(
+        res, success_message="data uploaded successfully", return_json=False
+    )
 
 
-def get(api_key: str, url: str, host: Optional[str] = None, timeout: Optional[int] = None) -> requests.Response:
+def get(
+    api_key: str, url: str, host: Optional[str] = None, timeout: Optional[int] = None
+) -> requests.Response:
     url = remove_trailing_slash(host or DEFAULT_HOST) + url
-    res = requests.get(url, headers={"Authorization": "Bearer %s" % api_key, "User-Agent": USER_AGENT}, timeout=timeout)
+    res = requests.get(
+        url,
+        headers={"Authorization": "Bearer %s" % api_key, "User-Agent": USER_AGENT},
+        timeout=timeout,
+    )
     return _process_response(res, success_message=f"GET {url} completed successfully")
 
 

posthog/sentry/__init__.py

@@ -1 +0,0 @@
-POSTHOG_ID_TAG = "posthog_distinct_id"

posthog/sentry/django.py

@@ -1,28 +0,0 @@
-from django.conf import settings
-from sentry_sdk import configure_scope
-
-from posthog.sentry import POSTHOG_ID_TAG
-
-GET_DISTINCT_ID = getattr(settings, "POSTHOG_DJANGO", {}).get("distinct_id")
-
-
-def get_distinct_id(request):
-    if not GET_DISTINCT_ID:
-        return None
-    try:
-        return GET_DISTINCT_ID(request)
-    except:  # noqa: E722
-        return None
-
-
-class PosthogDistinctIdMiddleware:
-    def __init__(self, get_response):
-        self.get_response = get_response
-
-    def __call__(self, request):
-        with configure_scope() as scope:
-            distinct_id = get_distinct_id(request)
-            if distinct_id:
-                scope.set_tag(POSTHOG_ID_TAG, distinct_id)
-            response = self.get_response(request)
-        return response

posthog/sentry/posthog_integration.py

@@ -1,52 +0,0 @@
-from sentry_sdk._types import MYPY
-from sentry_sdk.hub import Hub
-from sentry_sdk.integrations import Integration
-from sentry_sdk.scope import add_global_event_processor
-from sentry_sdk.utils import Dsn
-
-import posthog
-from posthog.request import DEFAULT_HOST
-from posthog.sentry import POSTHOG_ID_TAG
-
-if MYPY:
-    from typing import Optional  # noqa: F401
-
-    from sentry_sdk._types import Event, Hint  # noqa: F401
-
-
-class PostHogIntegration(Integration):
-    identifier = "posthog-python"
-    organization = None  # The Sentry organization, used to send a direct link from PostHog to Sentry
-    project_id = None  # The Sentry project id, used to send a direct link from PostHog to Sentry
-    prefix = "https://sentry.io/organizations/"  # URL of a hosted sentry instance (default: https://sentry.io/organizations/)
-
-    @staticmethod
-    def setup_once():
-        @add_global_event_processor
-        def processor(event, hint):
-            # type: (Event, Optional[Hint]) -> Optional[Event]
-            if Hub.current.get_integration(PostHogIntegration) is not None:
-                if event.get("level") != "error":
-                    return event
-
-                if event.get("tags", {}).get(POSTHOG_ID_TAG):
-                    posthog_distinct_id = event["tags"][POSTHOG_ID_TAG]
-                    event["tags"]["PostHog URL"] = f"{posthog.host or DEFAULT_HOST}/person/{posthog_distinct_id}"
-
-                    properties = {
-                        "$sentry_event_id": event["event_id"],
-                        "$sentry_exception": event["exception"],
-                    }
-
-                    if PostHogIntegration.organization:
-                        project_id = PostHogIntegration.project_id or (
-                            not not Hub.current.client.dsn and Dsn(Hub.current.client.dsn).project_id
-                        )
-                        if project_id:
-                            properties["$sentry_url"] = (
-                                f"{PostHogIntegration.prefix}{PostHogIntegration.organization}/issues/?project={project_id}&query={event['event_id']}"
-                            )
-
-                    posthog.capture(posthog_distinct_id, "$exception", properties)
-
-            return event

posthog/test/ai/gemini/test_gemini.py

@@ -11,7 +11,9 @@ try:
 except ImportError:
     GEMINI_AVAILABLE = False
 
-pytestmark = pytest.mark.skipif(not GEMINI_AVAILABLE, reason="Google Gemini package is not available")
+pytestmark = pytest.mark.skipif(
+    not GEMINI_AVAILABLE, reason="Google Gemini package is not available"
+)
 
 
 @pytest.fixture
@@ -29,6 +31,9 @@ def mock_gemini_response():
     mock_usage = MagicMock()
     mock_usage.prompt_token_count = 20
     mock_usage.candidates_token_count = 10
+    # Ensure cache and reasoning tokens are not present (not MagicMock)
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
     mock_response.usage_metadata = mock_usage
 
     mock_candidate = MagicMock()
@@ -54,7 +59,94 @@ def mock_google_genai_client():
         yield mock_client_instance
 
 
-def test_new_client_basic_generation(mock_client, mock_google_genai_client, mock_gemini_response):
+@pytest.fixture
+def mock_gemini_response_with_function_calls():
+    mock_response = MagicMock()
+
+    # Mock usage metadata
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 25
+    mock_usage.candidates_token_count = 15
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
+    mock_response.usage_metadata = mock_usage
+
+    # Mock function call
+    mock_function_call = MagicMock()
+    mock_function_call.name = "get_current_weather"
+    mock_function_call.args = {"location": "San Francisco"}
+
+    # Mock text part 1
+    mock_text_part1 = MagicMock()
+    mock_text_part1.text = "I'll check the weather for you."
+    # Make hasattr(part, "text") return True
+    type(mock_text_part1).text = mock_text_part1.text
+
+    # Mock text part 2
+    mock_text_part2 = MagicMock()
+    mock_text_part2.text = " Let me look that up."
+    type(mock_text_part2).text = mock_text_part2.text
+
+    # Mock function call part - need to ensure hasattr() works correctly
+    mock_function_part = MagicMock()
+    mock_function_part.function_call = mock_function_call
+    # Make hasattr(part, "function_call") return True
+    type(mock_function_part).function_call = mock_function_part.function_call
+    # Ensure hasattr(part, "text") returns False for the function part
+    del mock_function_part.text
+
+    # Mock content with 2 text parts and 1 function call part
+    mock_content = MagicMock()
+    mock_content.parts = [mock_text_part1, mock_text_part2, mock_function_part]
+
+    # Mock candidate
+    mock_candidate = MagicMock()
+    mock_candidate.content = mock_content
+    mock_response.candidates = [mock_candidate]
+
+    return mock_response
+
+
+@pytest.fixture
+def mock_gemini_response_function_calls_only():
+    mock_response = MagicMock()
+
+    # Mock usage metadata
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 30
+    mock_usage.candidates_token_count = 12
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
+    mock_response.usage_metadata = mock_usage
+
+    # Mock function call
+    mock_function_call = MagicMock()
+    mock_function_call.name = "get_current_weather"
+    mock_function_call.args = {"location": "New York", "unit": "fahrenheit"}
+
+    # Mock function call part (no text part) - need to ensure hasattr() works correctly
+    mock_function_part = MagicMock()
+    mock_function_part.function_call = mock_function_call
+    # Make hasattr(part, "function_call") return True
+    type(mock_function_part).function_call = mock_function_part.function_call
+    # Ensure hasattr(part, "text") returns False for the function part
+    del mock_function_part.text
+
+    # Mock content with only function call part
+    mock_content = MagicMock()
+    mock_content.parts = [mock_function_part]
+
+    # Mock candidate
+    mock_candidate = MagicMock()
+    mock_candidate.content = mock_content
+    mock_response.candidates = [mock_candidate]
+
+    return mock_response
+
+
+def test_new_client_basic_generation(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test the new Client/Models API structure"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
@@ -84,7 +176,9 @@ def test_new_client_basic_generation(mock_client, mock_google_genai_client, mock
     assert props["$ai_latency"] > 0
 
 
-def test_new_client_streaming_with_generate_content_stream(mock_client, mock_google_genai_client):
+def test_new_client_streaming_with_generate_content_stream(
+    mock_client, mock_google_genai_client
+):
     """Test the new generate_content_stream method"""
 
     def mock_streaming_response():
@@ -93,6 +187,8 @@ def test_new_client_streaming_with_generate_content_stream(mock_client, mock_goo
         mock_usage1 = MagicMock()
         mock_usage1.prompt_token_count = 10
         mock_usage1.candidates_token_count = 5
+        mock_usage1.cached_content_token_count = 0
+        mock_usage1.thoughts_token_count = 0
         mock_chunk1.usage_metadata = mock_usage1
 
         mock_chunk2 = MagicMock()
@@ -100,13 +196,17 @@ def test_new_client_streaming_with_generate_content_stream(mock_client, mock_goo
         mock_usage2 = MagicMock()
         mock_usage2.prompt_token_count = 10
         mock_usage2.candidates_token_count = 10
+        mock_usage2.cached_content_token_count = 0
+        mock_usage2.thoughts_token_count = 0
         mock_chunk2.usage_metadata = mock_usage2
 
         yield mock_chunk1
         yield mock_chunk2
 
     # Mock the generate_content_stream method
-    mock_google_genai_client.models.generate_content_stream.return_value = mock_streaming_response()
+    mock_google_genai_client.models.generate_content_stream.return_value = (
+        mock_streaming_response()
+    )
 
     client = Client(api_key="test-key", posthog_client=mock_client)
 
@@ -137,6 +237,91 @@ def test_new_client_streaming_with_generate_content_stream(mock_client, mock_goo
     assert isinstance(props["$ai_latency"], float)
 
 
+def test_new_client_streaming_with_tools(mock_client, mock_google_genai_client):
+    """Test that tools are captured in streaming mode"""
+
+    def mock_streaming_response():
+        mock_chunk1 = MagicMock()
+        mock_chunk1.text = "I'll check "
+        mock_usage1 = MagicMock()
+        mock_usage1.prompt_token_count = 15
+        mock_usage1.candidates_token_count = 5
+        mock_usage1.cached_content_token_count = 0
+        mock_usage1.thoughts_token_count = 0
+        mock_chunk1.usage_metadata = mock_usage1
+
+        mock_chunk2 = MagicMock()
+        mock_chunk2.text = "the weather"
+        mock_usage2 = MagicMock()
+        mock_usage2.prompt_token_count = 15
+        mock_usage2.candidates_token_count = 10
+        mock_usage2.cached_content_token_count = 0
+        mock_usage2.thoughts_token_count = 0
+        mock_chunk2.usage_metadata = mock_usage2
+
+        yield mock_chunk1
+        yield mock_chunk2
+
+    # Mock the generate_content_stream method
+    mock_google_genai_client.models.generate_content_stream.return_value = (
+        mock_streaming_response()
+    )
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    # Create mock tools configuration
+    mock_tool = MagicMock()
+    mock_tool.function_declarations = [
+        MagicMock(
+            name="get_current_weather",
+            description="Gets the current weather for a given location.",
+            parameters=MagicMock(
+                type="OBJECT",
+                properties={
+                    "location": MagicMock(
+                        type="STRING",
+                        description="The city and state, e.g. San Francisco, CA",
+                    )
+                },
+                required=["location"],
+            ),
+        )
+    ]
+
+    mock_config = MagicMock()
+    mock_config.tools = [mock_tool]
+
+    response = client.models.generate_content_stream(
+        model="gemini-2.0-flash",
+        contents=["What's the weather in SF?"],
+        config=mock_config,
+        posthog_distinct_id="test-id",
+        posthog_properties={"feature": "streaming_with_tools"},
+    )
+
+    chunks = list(response)
+    assert len(chunks) == 2
+    assert chunks[0].text == "I'll check "
+    assert chunks[1].text == "the weather"
+
+    # Check that the streaming event was captured with tools
+    assert mock_client.capture.call_count == 1
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    assert call_args["distinct_id"] == "test-id"
+    assert call_args["event"] == "$ai_generation"
+    assert props["$ai_provider"] == "gemini"
+    assert props["$ai_model"] == "gemini-2.0-flash"
+    assert props["$ai_input_tokens"] == 15
+    assert props["$ai_output_tokens"] == 10
+    assert props["feature"] == "streaming_with_tools"
+    assert isinstance(props["$ai_latency"], float)
+
+    # Verify that tools are captured in the $ai_tools property in streaming mode
+    assert props["$ai_tools"] == [mock_tool]
+
+
 def test_new_client_groups(mock_client, mock_google_genai_client, mock_gemini_response):
     """Test groups functionality with new Client API"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
@@ -154,7 +339,9 @@ def test_new_client_groups(mock_client, mock_google_genai_client, mock_gemini_re
     assert call_args["groups"] == {"company": "company_123"}
 
 
-def test_new_client_privacy_mode_local(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_privacy_mode_local(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test local privacy mode with new Client API"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
@@ -173,7 +360,9 @@ def test_new_client_privacy_mode_local(mock_client, mock_google_genai_client, mo
     assert props["$ai_output_choices"] is None
 
 
-def test_new_client_privacy_mode_global(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_privacy_mode_global(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test global privacy mode with new Client API"""
     mock_client.privacy_mode = True
 
@@ -193,29 +382,57 @@ def test_new_client_privacy_mode_global(mock_client, mock_google_genai_client, m
     assert props["$ai_output_choices"] is None
 
 
-def test_new_client_different_input_formats(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_different_input_formats(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test different input formats with new Client API"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
     client = Client(api_key="test-key", posthog_client=mock_client)
 
     # Test string input
-    client.models.generate_content(model="gemini-2.0-flash", contents="Hello", posthog_distinct_id="test-id")
+    client.models.generate_content(
+        model="gemini-2.0-flash", contents="Hello", posthog_distinct_id="test-id"
+    )
     call_args = mock_client.capture.call_args[1]
     props = call_args["properties"]
     assert props["$ai_input"] == [{"role": "user", "content": "Hello"}]
 
-    # Test list input
+    # Test Gemini-specific format with parts array (like in the screenshot)
+    mock_client.reset_mock()
+    client.models.generate_content(
+        model="gemini-2.0-flash",
+        contents=[{"role": "user", "parts": [{"text": "hey"}]}],
+        posthog_distinct_id="test-id",
+    )
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+    assert props["$ai_input"] == [{"role": "user", "content": "hey"}]
+
+    # Test multiple parts in the parts array
+    mock_client.reset_mock()
+    client.models.generate_content(
+        model="gemini-2.0-flash",
+        contents=[{"role": "user", "parts": [{"text": "Hello "}, {"text": "world"}]}],
+        posthog_distinct_id="test-id",
+    )
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+    assert props["$ai_input"] == [{"role": "user", "content": "Hello world"}]
+
+    # Test list input with string
     mock_client.capture.reset_mock()
-    mock_part = MagicMock()
-    mock_part.text = "List item"
-    client.models.generate_content(model="gemini-2.0-flash", contents=[mock_part], posthog_distinct_id="test-id")
+    client.models.generate_content(
+        model="gemini-2.0-flash", contents=["List item"], posthog_distinct_id="test-id"
+    )
     call_args = mock_client.capture.call_args[1]
     props = call_args["properties"]
     assert props["$ai_input"] == [{"role": "user", "content": "List item"}]
 
 
-def test_new_client_model_parameters(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_model_parameters(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test model parameters with new Client API"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
@@ -235,7 +452,9 @@ def test_new_client_model_parameters(mock_client, mock_google_genai_client, mock
     assert props["$ai_model_parameters"]["max_tokens"] == 100
 
 
-def test_new_client_default_settings(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_default_settings(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test client with default PostHog settings"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
@@ -259,7 +478,9 @@ def test_new_client_default_settings(mock_client, mock_google_genai_client, mock
     assert props["team"] == "ai"
 
 
-def test_new_client_override_defaults(mock_client, mock_google_genai_client, mock_gemini_response):
+def test_new_client_override_defaults(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
     """Test overriding client defaults per call"""
     mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
 
@@ -294,3 +515,584 @@ def test_new_client_override_defaults(mock_client, mock_google_genai_client, moc
     assert props["team"] == "ai"  # from defaults
     assert props["feature"] == "chat"  # from call
     assert props["urgent"] is True  # from call
+
+
+def test_vertex_ai_parameters_passed_through(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
+    """Test that Vertex AI parameters are properly passed to genai.Client"""
+    mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
+
+    # Mock credentials object
+    mock_credentials = MagicMock()
+    mock_debug_config = MagicMock()
+    mock_http_options = MagicMock()
+
+    # Create client with Vertex AI parameters
+    Client(
+        vertexai=True,
+        credentials=mock_credentials,
+        project="test-project",
+        location="us-central1",
+        debug_config=mock_debug_config,
+        http_options=mock_http_options,
+        posthog_client=mock_client,
+    )
+
+    # Verify genai.Client was called with correct parameters
+    google_genai.Client.assert_called_once_with(
+        vertexai=True,
+        credentials=mock_credentials,
+        project="test-project",
+        location="us-central1",
+        debug_config=mock_debug_config,
+        http_options=mock_http_options,
+    )
+
+
+def test_api_key_mode(mock_client, mock_google_genai_client):
+    """Test API key authentication mode"""
+
+    # Create client with just API key (traditional mode)
+    Client(
+        api_key="test-api-key",
+        posthog_client=mock_client,
+    )
+
+    # Verify genai.Client was called with only api_key
+    google_genai.Client.assert_called_once_with(api_key="test-api-key")
+
+
+def test_vertex_ai_mode_with_optional_api_key(
+    mock_client, mock_google_genai_client, mock_gemini_response
+):
+    """Test Vertex AI mode with optional API key"""
+    mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
+
+    mock_credentials = MagicMock()
+
+    # Create client with Vertex AI + API key
+    Client(
+        vertexai=True,
+        api_key="test-api-key",
+        credentials=mock_credentials,
+        project="test-project",
+        posthog_client=mock_client,
+    )
+
+    # Verify genai.Client was called with both Vertex AI params and API key
+    google_genai.Client.assert_called_once_with(
+        vertexai=True,
+        api_key="test-api-key",
+        credentials=mock_credentials,
+        project="test-project",
+    )
+
+
+def test_tool_use_response(mock_client, mock_google_genai_client, mock_gemini_response):
+    """Test that tools defined in config are captured in $ai_tools property"""
+    mock_google_genai_client.models.generate_content.return_value = mock_gemini_response
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    # Create mock tools configuration
+    mock_tool = MagicMock()
+    mock_tool.function_declarations = [
+        MagicMock(
+            name="get_current_weather",
+            description="Gets the current weather for a given location.",
+            parameters=MagicMock(
+                type="OBJECT",
+                properties={
+                    "location": MagicMock(
+                        type="STRING",
+                        description="The city and state, e.g. San Francisco, CA",
+                    )
+                },
+                required=["location"],
+            ),
+        )
+    ]
+
+    mock_config = MagicMock()
+    mock_config.tools = [mock_tool]
+    # Explicitly specify this config doesn't have system_instruction
+    del mock_config.system_instruction
+
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents=["hey"],
+        config=mock_config,
+        posthog_distinct_id="test-id",
+        posthog_properties={"foo": "bar"},
+    )
+
+    assert response == mock_gemini_response
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    assert call_args["distinct_id"] == "test-id"
+    assert call_args["event"] == "$ai_generation"
+    assert props["$ai_provider"] == "gemini"
+    assert props["$ai_model"] == "gemini-2.5-flash"
+    assert props["$ai_input"] == [{"role": "user", "content": "hey"}]
+    assert props["$ai_output_choices"] == [
+        {
+            "role": "assistant",
+            "content": [{"type": "text", "text": "Test response from Gemini"}],
+        }
+    ]
+    assert props["$ai_input_tokens"] == 20
+    assert props["$ai_output_tokens"] == 10
+    assert props["$ai_http_status"] == 200
+    assert props["foo"] == "bar"
+    assert isinstance(props["$ai_latency"], float)
+    # Verify that tools are captured in the $ai_tools property
+    assert props["$ai_tools"] == [mock_tool]
+
+
+def test_function_calls_in_output_choices(
+    mock_client, mock_google_genai_client, mock_gemini_response_with_function_calls
+):
+    """Test that function calls are properly included in $ai_output_choices"""
+    mock_google_genai_client.models.generate_content.return_value = (
+        mock_gemini_response_with_function_calls
+    )
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents=["What's the weather in San Francisco?"],
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_gemini_response_with_function_calls
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    assert call_args["distinct_id"] == "test-id"
+    assert call_args["event"] == "$ai_generation"
+    assert props["$ai_provider"] == "gemini"
+    assert props["$ai_model"] == "gemini-2.5-flash"
+    assert props["$ai_output_choices"] == [
+        {
+            "role": "assistant",
+            "content": [
+                {"type": "text", "text": "I'll check the weather for you."},
+                {"type": "text", "text": " Let me look that up."},
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "get_current_weather",
+                        "arguments": {"location": "San Francisco"},
+                    },
+                },
+            ],
+        }
+    ]
+
+    # Check token usage
+    assert props["$ai_input_tokens"] == 25
+    assert props["$ai_output_tokens"] == 15
+    assert props["$ai_http_status"] == 200
+
+
+def test_function_calls_only_no_content(
+    mock_client, mock_google_genai_client, mock_gemini_response_function_calls_only
+):
+    """Test function calls without text content in $ai_output_choices"""
+    mock_google_genai_client.models.generate_content.return_value = (
+        mock_gemini_response_function_calls_only
+    )
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents=["Get weather for New York"],
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_gemini_response_function_calls_only
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    assert call_args["distinct_id"] == "test-id"
+    assert call_args["event"] == "$ai_generation"
+    assert props["$ai_provider"] == "gemini"
+    assert props["$ai_model"] == "gemini-2.5-flash"
+    assert props["$ai_output_choices"] == [
+        {
+            "role": "assistant",
+            "content": [
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "get_current_weather",
+                        "arguments": {"location": "New York", "unit": "fahrenheit"},
+                    },
+                }
+            ],
+        }
+    ]
+
+    # Check token usage
+    assert props["$ai_input_tokens"] == 30
+    assert props["$ai_output_tokens"] == 12
+    assert props["$ai_http_status"] == 200
+
+
+def test_cache_and_reasoning_tokens(mock_client, mock_google_genai_client):
+    """Test that cache and reasoning tokens are properly extracted"""
+    # Create a mock response with cache and reasoning tokens
+    mock_response = MagicMock()
+    mock_response.text = "Test response with cache"
+
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 100
+    mock_usage.candidates_token_count = 50
+    mock_usage.cached_content_token_count = 30  # Cache tokens
+    mock_usage.thoughts_token_count = 10  # Reasoning tokens
+    mock_response.usage_metadata = mock_usage
+
+    # Mock candidates
+    mock_candidate = MagicMock()
+    mock_candidate.text = "Test response with cache"
+    mock_response.candidates = [mock_candidate]
+
+    mock_google_genai_client.models.generate_content.return_value = mock_response
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content(
+        model="gemini-2.5-pro",
+        contents="Test with cache",
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_response
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Check that all token types are present
+    assert props["$ai_input_tokens"] == 100
+    assert props["$ai_output_tokens"] == 50
+    assert props["$ai_cache_read_input_tokens"] == 30
+    assert props["$ai_reasoning_tokens"] == 10
+
+
+def test_streaming_cache_and_reasoning_tokens(mock_client, mock_google_genai_client):
+    """Test that cache and reasoning tokens are properly extracted in streaming"""
+    # Create mock chunks with cache and reasoning tokens
+    chunk1 = MagicMock()
+    chunk1.text = "Hello "
+    chunk1_usage = MagicMock()
+    chunk1_usage.prompt_token_count = 100
+    chunk1_usage.candidates_token_count = 5
+    chunk1_usage.cached_content_token_count = 30  # Cache tokens
+    chunk1_usage.thoughts_token_count = 0
+    chunk1.usage_metadata = chunk1_usage
+
+    chunk2 = MagicMock()
+    chunk2.text = "world!"
+    chunk2_usage = MagicMock()
+    chunk2_usage.prompt_token_count = 100
+    chunk2_usage.candidates_token_count = 10
+    chunk2_usage.cached_content_token_count = 30  # Same cache tokens
+    chunk2_usage.thoughts_token_count = 5  # Reasoning tokens
+    chunk2.usage_metadata = chunk2_usage
+
+    mock_stream = iter([chunk1, chunk2])
+    mock_google_genai_client.models.generate_content_stream.return_value = mock_stream
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content_stream(
+        model="gemini-2.5-pro",
+        contents="Test streaming with cache",
+        posthog_distinct_id="test-id",
+    )
+
+    # Consume the stream
+    result = list(response)
+    assert len(result) == 2
+
+    # Check PostHog capture was called
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Check that all token types are present (should use final chunk's usage)
+    assert props["$ai_input_tokens"] == 100
+    assert props["$ai_output_tokens"] == 10
+    assert props["$ai_cache_read_input_tokens"] == 30
+    assert props["$ai_reasoning_tokens"] == 5
+
+
+def test_web_search_grounding(mock_client, mock_google_genai_client):
+    """Test web search detection via grounding_metadata."""
+
+    # Create mock response with grounding metadata
+    mock_response = MagicMock()
+
+    # Mock usage metadata
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 60
+    mock_usage.candidates_token_count = 40
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
+    mock_response.usage_metadata = mock_usage
+
+    # Mock grounding metadata
+    mock_grounding_chunk = MagicMock()
+    mock_grounding_chunk.uri = "https://example.com"
+
+    mock_grounding_metadata = MagicMock()
+    mock_grounding_metadata.grounding_chunks = [mock_grounding_chunk]
+
+    # Mock text part
+    mock_text_part = MagicMock()
+    mock_text_part.text = "According to search results..."
+    type(mock_text_part).text = mock_text_part.text
+
+    # Mock content with parts
+    mock_content = MagicMock()
+    mock_content.parts = [mock_text_part]
+
+    # Mock candidate with grounding metadata
+    mock_candidate = MagicMock()
+    mock_candidate.content = mock_content
+    mock_candidate.grounding_metadata = mock_grounding_metadata
+    type(mock_candidate).grounding_metadata = mock_candidate.grounding_metadata
+
+    mock_response.candidates = [mock_candidate]
+    mock_response.text = "According to search results..."
+
+    # Mock the generate_content method
+    mock_google_genai_client.models.generate_content.return_value = mock_response
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents="What's the latest news?",
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_response
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Verify web search count is detected (binary for grounding)
+    assert props["$ai_web_search_count"] == 1
+    assert props["$ai_input_tokens"] == 60
+    assert props["$ai_output_tokens"] == 40
+
+
+def test_streaming_with_web_search(mock_client, mock_google_genai_client):
+    """Test that web search count is properly captured in streaming mode."""
+
+    def mock_streaming_response():
+        # Create chunk 1 with grounding metadata
+        mock_chunk1 = MagicMock()
+        mock_chunk1.text = "According to "
+
+        mock_usage1 = MagicMock()
+        mock_usage1.prompt_token_count = 30
+        mock_usage1.candidates_token_count = 5
+        mock_usage1.cached_content_token_count = 0
+        mock_usage1.thoughts_token_count = 0
+        mock_chunk1.usage_metadata = mock_usage1
+
+        # Add grounding metadata to first chunk
+        mock_grounding_chunk = MagicMock()
+        mock_grounding_chunk.uri = "https://example.com"
+
+        mock_grounding_metadata = MagicMock()
+        mock_grounding_metadata.grounding_chunks = [mock_grounding_chunk]
+
+        mock_candidate1 = MagicMock()
+        mock_candidate1.grounding_metadata = mock_grounding_metadata
+        type(mock_candidate1).grounding_metadata = mock_candidate1.grounding_metadata
+
+        mock_chunk1.candidates = [mock_candidate1]
+
+        # Create chunk 2
+        mock_chunk2 = MagicMock()
+        mock_chunk2.text = "search results..."
+
+        mock_usage2 = MagicMock()
+        mock_usage2.prompt_token_count = 30
+        mock_usage2.candidates_token_count = 15
+        mock_usage2.cached_content_token_count = 0
+        mock_usage2.thoughts_token_count = 0
+        mock_chunk2.usage_metadata = mock_usage2
+
+        mock_candidate2 = MagicMock()
+        mock_chunk2.candidates = [mock_candidate2]
+
+        yield mock_chunk1
+        yield mock_chunk2
+
+    # Mock the generate_content_stream method
+    mock_google_genai_client.models.generate_content_stream.return_value = (
+        mock_streaming_response()
+    )
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content_stream(
+        model="gemini-2.5-flash",
+        contents="What's the latest news?",
+        posthog_distinct_id="test-id",
+    )
+
+    chunks = list(response)
+    assert len(chunks) == 2
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Verify web search count is detected (binary for grounding)
+    assert props["$ai_web_search_count"] == 1
+    assert props["$ai_input_tokens"] == 30
+    assert props["$ai_output_tokens"] == 15
+
+
+def test_empty_grounding_metadata_no_web_search(mock_client, mock_google_genai_client):
+    """Test that empty grounding_metadata (all null fields) does not count as web search."""
+
+    # Create mock response with empty grounding metadata (all null fields)
+    mock_response = MagicMock()
+
+    # Mock usage metadata
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 10
+    mock_usage.candidates_token_count = 10
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
+    mock_response.usage_metadata = mock_usage
+
+    # Mock empty grounding metadata (all fields are None)
+    mock_grounding_metadata = MagicMock()
+    mock_grounding_metadata.web_search_queries = None
+    mock_grounding_metadata.grounding_chunks = None
+    mock_grounding_metadata.grounding_supports = None
+    mock_grounding_metadata.retrieval_metadata = None
+    mock_grounding_metadata.retrieval_queries = None
+    mock_grounding_metadata.search_entry_point = None
+
+    # Mock text part
+    mock_text_part = MagicMock()
+    mock_text_part.text = "Hey there! How can I help you today?"
+    type(mock_text_part).text = mock_text_part.text
+
+    # Mock content with parts
+    mock_content = MagicMock()
+    mock_content.parts = [mock_text_part]
+
+    # Mock candidate with empty grounding metadata
+    mock_candidate = MagicMock()
+    mock_candidate.content = mock_content
+    mock_candidate.grounding_metadata = mock_grounding_metadata
+    type(mock_candidate).grounding_metadata = mock_candidate.grounding_metadata
+
+    mock_response.candidates = [mock_candidate]
+    mock_response.text = "Hey there! How can I help you today?"
+
+    # Mock the generate_content method
+    mock_google_genai_client.models.generate_content.return_value = mock_response
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents="Hello",
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_response
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Verify web search count is 0 (not present in properties when 0)
+    assert "$ai_web_search_count" not in props
+    assert props["$ai_input_tokens"] == 10
+    assert props["$ai_output_tokens"] == 10
+
+
+def test_empty_array_grounding_metadata_no_web_search(
+    mock_client, mock_google_genai_client
+):
+    """Test that grounding_metadata with empty arrays does not count as web search."""
+
+    # Create mock response with grounding metadata having empty arrays
+    mock_response = MagicMock()
+
+    # Mock usage metadata
+    mock_usage = MagicMock()
+    mock_usage.prompt_token_count = 15
+    mock_usage.candidates_token_count = 12
+    mock_usage.cached_content_token_count = 0
+    mock_usage.thoughts_token_count = 0
+    mock_response.usage_metadata = mock_usage
+
+    # Mock grounding metadata with empty arrays
+    mock_grounding_metadata = MagicMock()
+    mock_grounding_metadata.web_search_queries = []
+    mock_grounding_metadata.grounding_chunks = []
+    mock_grounding_metadata.grounding_supports = []
+
+    # Mock text part
+    mock_text_part = MagicMock()
+    mock_text_part.text = "I can help with that."
+    type(mock_text_part).text = mock_text_part.text
+
+    # Mock content with parts
+    mock_content = MagicMock()
+    mock_content.parts = [mock_text_part]
+
+    # Mock candidate with grounding metadata containing empty arrays
+    mock_candidate = MagicMock()
+    mock_candidate.content = mock_content
+    mock_candidate.grounding_metadata = mock_grounding_metadata
+    type(mock_candidate).grounding_metadata = mock_candidate.grounding_metadata
+
+    mock_response.candidates = [mock_candidate]
+    mock_response.text = "I can help with that."
+
+    # Mock the generate_content method
+    mock_google_genai_client.models.generate_content.return_value = mock_response
+
+    client = Client(api_key="test-key", posthog_client=mock_client)
+
+    response = client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents="What can you do?",
+        posthog_distinct_id="test-id",
+    )
+
+    assert response == mock_response
+    assert mock_client.capture.call_count == 1
+
+    call_args = mock_client.capture.call_args[1]
+    props = call_args["properties"]
+
+    # Verify web search count is 0 (not present in properties when 0)
+    assert "$ai_web_search_count" not in props
+    assert props["$ai_input_tokens"] == 15
+    assert props["$ai_output_tokens"] == 12

posthog/test/ai/test_sanitization.py

@@ -0,0 +1,335 @@
+import unittest
+
+from posthog.ai.sanitization import (
+    redact_base64_data_url,
+    sanitize_openai,
+    sanitize_openai_response,
+    sanitize_anthropic,
+    sanitize_gemini,
+    sanitize_langchain,
+    is_base64_data_url,
+    is_raw_base64,
+    REDACTED_IMAGE_PLACEHOLDER,
+)
+
+
+class TestSanitization(unittest.TestCase):
+    def setUp(self):
+        self.sample_base64_image = "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQ..."
+        self.sample_base64_png = "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAUA..."
+        self.regular_url = "https://example.com/image.jpg"
+        self.raw_base64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUl=="
+
+    def test_is_base64_data_url(self):
+        self.assertTrue(is_base64_data_url(self.sample_base64_image))
+        self.assertTrue(is_base64_data_url(self.sample_base64_png))
+        self.assertFalse(is_base64_data_url(self.regular_url))
+        self.assertFalse(is_base64_data_url("regular text"))
+
+    def test_is_raw_base64(self):
+        self.assertTrue(is_raw_base64(self.raw_base64))
+        self.assertFalse(is_raw_base64("short"))
+        self.assertFalse(is_raw_base64(self.regular_url))
+        self.assertFalse(is_raw_base64("/path/to/file"))
+
+    def test_redact_base64_data_url(self):
+        self.assertEqual(
+            redact_base64_data_url(self.sample_base64_image), REDACTED_IMAGE_PLACEHOLDER
+        )
+        self.assertEqual(
+            redact_base64_data_url(self.sample_base64_png), REDACTED_IMAGE_PLACEHOLDER
+        )
+        self.assertEqual(redact_base64_data_url(self.regular_url), self.regular_url)
+        self.assertEqual(redact_base64_data_url(None), None)
+        self.assertEqual(redact_base64_data_url(123), 123)
+
+    def test_sanitize_openai(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": "What is in this image?"},
+                    {
+                        "type": "image_url",
+                        "image_url": {
+                            "url": self.sample_base64_image,
+                            "detail": "high",
+                        },
+                    },
+                ],
+            }
+        ]
+
+        result = sanitize_openai(input_data)
+
+        self.assertEqual(result[0]["content"][0]["text"], "What is in this image?")
+        self.assertEqual(
+            result[0]["content"][1]["image_url"]["url"], REDACTED_IMAGE_PLACEHOLDER
+        )
+        self.assertEqual(result[0]["content"][1]["image_url"]["detail"], "high")
+
+    def test_sanitize_openai_preserves_regular_urls(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": self.regular_url},
+                    }
+                ],
+            }
+        ]
+
+        result = sanitize_openai(input_data)
+        self.assertEqual(result[0]["content"][0]["image_url"]["url"], self.regular_url)
+
+    def test_sanitize_openai_response(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "input_image",
+                        "image_url": self.sample_base64_image,
+                    }
+                ],
+            }
+        ]
+
+        result = sanitize_openai_response(input_data)
+        self.assertEqual(
+            result[0]["content"][0]["image_url"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+    def test_sanitize_anthropic(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": "What is in this image?"},
+                    {
+                        "type": "image",
+                        "source": {
+                            "type": "base64",
+                            "media_type": "image/jpeg",
+                            "data": "base64data",
+                        },
+                    },
+                ],
+            }
+        ]
+
+        result = sanitize_anthropic(input_data)
+
+        self.assertEqual(result[0]["content"][0]["text"], "What is in this image?")
+        self.assertEqual(
+            result[0]["content"][1]["source"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+        self.assertEqual(result[0]["content"][1]["source"]["type"], "base64")
+        self.assertEqual(result[0]["content"][1]["source"]["media_type"], "image/jpeg")
+
+    def test_sanitize_gemini(self):
+        input_data = [
+            {
+                "parts": [
+                    {"text": "What is in this image?"},
+                    {
+                        "inline_data": {
+                            "mime_type": "image/jpeg",
+                            "data": "base64data",
+                        }
+                    },
+                ]
+            }
+        ]
+
+        result = sanitize_gemini(input_data)
+
+        self.assertEqual(result[0]["parts"][0]["text"], "What is in this image?")
+        self.assertEqual(
+            result[0]["parts"][1]["inline_data"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+        self.assertEqual(
+            result[0]["parts"][1]["inline_data"]["mime_type"], "image/jpeg"
+        )
+
+    def test_sanitize_langchain_openai_style(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": self.sample_base64_image},
+                    }
+                ],
+            }
+        ]
+
+        result = sanitize_langchain(input_data)
+        self.assertEqual(
+            result[0]["content"][0]["image_url"]["url"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+    def test_sanitize_langchain_anthropic_style(self):
+        input_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image",
+                        "source": {"data": "base64data"},
+                    }
+                ],
+            }
+        ]
+
+        result = sanitize_langchain(input_data)
+        self.assertEqual(
+            result[0]["content"][0]["source"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+    def test_sanitize_with_data_url_format(self):
+        # Test that data URLs are properly detected and redacted across providers
+        data_url = "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD"
+
+        # OpenAI format
+        openai_data = [
+            {
+                "role": "user",
+                "content": [{"type": "image_url", "image_url": {"url": data_url}}],
+            }
+        ]
+        result = sanitize_openai(openai_data)
+        self.assertEqual(
+            result[0]["content"][0]["image_url"]["url"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+        # Anthropic format
+        anthropic_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image",
+                        "source": {
+                            "type": "base64",
+                            "media_type": "image/jpeg",
+                            "data": data_url,
+                        },
+                    }
+                ],
+            }
+        ]
+        result = sanitize_anthropic(anthropic_data)
+        self.assertEqual(
+            result[0]["content"][0]["source"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+        # LangChain format
+        langchain_data = [
+            {"role": "user", "content": [{"type": "image", "data": data_url}]}
+        ]
+        result = sanitize_langchain(langchain_data)
+        self.assertEqual(result[0]["content"][0]["data"], REDACTED_IMAGE_PLACEHOLDER)
+
+    def test_sanitize_with_raw_base64(self):
+        # Test that raw base64 strings (without data URL prefix) are detected
+        raw_base64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUl=="
+
+        # Test with Anthropic format
+        anthropic_data = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image",
+                        "source": {
+                            "type": "base64",
+                            "media_type": "image/png",
+                            "data": raw_base64,
+                        },
+                    }
+                ],
+            }
+        ]
+        result = sanitize_anthropic(anthropic_data)
+        self.assertEqual(
+            result[0]["content"][0]["source"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+        # Test with Gemini format
+        gemini_data = [
+            {"parts": [{"inline_data": {"mime_type": "image/png", "data": raw_base64}}]}
+        ]
+        result = sanitize_gemini(gemini_data)
+        self.assertEqual(
+            result[0]["parts"][0]["inline_data"]["data"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+    def test_sanitize_preserves_regular_content(self):
+        # Ensure non-base64 content is preserved across all providers
+        regular_url = "https://example.com/image.jpg"
+        text_content = "What do you see?"
+
+        # OpenAI
+        openai_data = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": text_content},
+                    {"type": "image_url", "image_url": {"url": regular_url}},
+                ],
+            }
+        ]
+        result = sanitize_openai(openai_data)
+        self.assertEqual(result[0]["content"][0]["text"], text_content)
+        self.assertEqual(result[0]["content"][1]["image_url"]["url"], regular_url)
+
+        # Anthropic
+        anthropic_data = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": text_content},
+                    {"type": "image", "source": {"type": "url", "url": regular_url}},
+                ],
+            }
+        ]
+        result = sanitize_anthropic(anthropic_data)
+        self.assertEqual(result[0]["content"][0]["text"], text_content)
+        # URL-based images should remain unchanged
+        self.assertEqual(result[0]["content"][1]["source"]["url"], regular_url)
+
+    def test_sanitize_handles_non_dict_content(self):
+        input_data = [{"role": "user", "content": "Just text"}]
+
+        result = sanitize_openai(input_data)
+        self.assertEqual(result, input_data)
+
+    def test_sanitize_handles_none_input(self):
+        self.assertIsNone(sanitize_openai(None))
+        self.assertIsNone(sanitize_anthropic(None))
+        self.assertIsNone(sanitize_gemini(None))
+        self.assertIsNone(sanitize_langchain(None))
+
+    def test_sanitize_handles_single_message(self):
+        input_data = {
+            "role": "user",
+            "content": [
+                {
+                    "type": "image_url",
+                    "image_url": {"url": self.sample_base64_image},
+                }
+            ],
+        }
+
+        result = sanitize_openai(input_data)
+        self.assertEqual(
+            result["content"][0]["image_url"]["url"], REDACTED_IMAGE_PLACEHOLDER
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()

posthog/test/ai/test_system_prompts.py

@@ -0,0 +1,354 @@
+"""
+Tests for system prompt capture across all LLM providers.
+
+This test suite ensures that system prompts are correctly captured in analytics
+regardless of how they're passed to the providers:
+- As first message in messages/contents array (standard format)
+- As separate system parameter (Anthropic, OpenAI)
+- As instructions parameter (OpenAI Responses API)
+- As system_instruction parameter (Gemini)
+"""
+
+import time
+import unittest
+from unittest.mock import patch, MagicMock
+
+
+class TestSystemPromptCapture(unittest.TestCase):
+    """Test system prompt capture for all providers."""
+
+    def setUp(self):
+        super().setUp()
+        self.test_system_prompt = "You are a helpful AI assistant."
+        self.test_user_message = "Hello, how are you?"
+        self.test_response = "I'm doing well, thank you!"
+
+        # Create mock PostHog client
+        self.client = MagicMock()
+        self.client.privacy_mode = False
+
+    def _assert_system_prompt_captured(self, captured_input):
+        """Helper to assert system prompt is correctly captured."""
+        self.assertEqual(
+            len(captured_input), 2, "Should have 2 messages (system + user)"
+        )
+        self.assertEqual(
+            captured_input[0]["role"], "system", "First message should be system"
+        )
+        self.assertEqual(
+            captured_input[0]["content"],
+            self.test_system_prompt,
+            "System content should match",
+        )
+        self.assertEqual(
+            captured_input[1]["role"], "user", "Second message should be user"
+        )
+        self.assertEqual(
+            captured_input[1]["content"],
+            self.test_user_message,
+            "User content should match",
+        )
+
+    # OpenAI Tests
+    def test_openai_messages_array_system_prompt(self):
+        """Test OpenAI with system prompt in messages array."""
+        try:
+            from posthog.ai.openai import OpenAI
+            from openai.types.chat import ChatCompletion, ChatCompletionMessage
+            from openai.types.chat.chat_completion import Choice
+            from openai.types.completion_usage import CompletionUsage
+        except ImportError:
+            self.skipTest("OpenAI package not available")
+
+        mock_response = ChatCompletion(
+            id="test",
+            model="gpt-4",
+            object="chat.completion",
+            created=int(time.time()),
+            choices=[
+                Choice(
+                    finish_reason="stop",
+                    index=0,
+                    message=ChatCompletionMessage(
+                        content=self.test_response, role="assistant"
+                    ),
+                )
+            ],
+            usage=CompletionUsage(
+                completion_tokens=10, prompt_tokens=20, total_tokens=30
+            ),
+        )
+
+        with patch(
+            "openai.resources.chat.completions.Completions.create",
+            return_value=mock_response,
+        ):
+            client = OpenAI(posthog_client=self.client, api_key="test")
+
+            messages = [
+                {"role": "system", "content": self.test_system_prompt},
+                {"role": "user", "content": self.test_user_message},
+            ]
+
+            client.chat.completions.create(
+                model="gpt-4", messages=messages, posthog_distinct_id="test-user"
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    def test_openai_separate_system_parameter(self):
+        """Test OpenAI with system prompt as separate parameter."""
+        try:
+            from posthog.ai.openai import OpenAI
+            from openai.types.chat import ChatCompletion, ChatCompletionMessage
+            from openai.types.chat.chat_completion import Choice
+            from openai.types.completion_usage import CompletionUsage
+        except ImportError:
+            self.skipTest("OpenAI package not available")
+
+        mock_response = ChatCompletion(
+            id="test",
+            model="gpt-4",
+            object="chat.completion",
+            created=int(time.time()),
+            choices=[
+                Choice(
+                    finish_reason="stop",
+                    index=0,
+                    message=ChatCompletionMessage(
+                        content=self.test_response, role="assistant"
+                    ),
+                )
+            ],
+            usage=CompletionUsage(
+                completion_tokens=10, prompt_tokens=20, total_tokens=30
+            ),
+        )
+
+        with patch(
+            "openai.resources.chat.completions.Completions.create",
+            return_value=mock_response,
+        ):
+            client = OpenAI(posthog_client=self.client, api_key="test")
+
+            messages = [{"role": "user", "content": self.test_user_message}]
+
+            client.chat.completions.create(
+                model="gpt-4",
+                messages=messages,
+                system=self.test_system_prompt,
+                posthog_distinct_id="test-user",
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    def test_openai_streaming_system_parameter(self):
+        """Test OpenAI streaming with system parameter."""
+        try:
+            from posthog.ai.openai import OpenAI
+            from openai.types.chat.chat_completion_chunk import ChatCompletionChunk
+            from openai.types.chat.chat_completion_chunk import Choice as ChoiceChunk
+            from openai.types.chat.chat_completion_chunk import ChoiceDelta
+            from openai.types.completion_usage import CompletionUsage
+        except ImportError:
+            self.skipTest("OpenAI package not available")
+
+        chunk1 = ChatCompletionChunk(
+            id="test",
+            model="gpt-4",
+            object="chat.completion.chunk",
+            created=int(time.time()),
+            choices=[
+                ChoiceChunk(
+                    finish_reason=None,
+                    index=0,
+                    delta=ChoiceDelta(content="Hello", role="assistant"),
+                )
+            ],
+        )
+
+        chunk2 = ChatCompletionChunk(
+            id="test",
+            model="gpt-4",
+            object="chat.completion.chunk",
+            created=int(time.time()),
+            choices=[
+                ChoiceChunk(
+                    finish_reason="stop",
+                    index=0,
+                    delta=ChoiceDelta(content=" there!", role=None),
+                )
+            ],
+            usage=CompletionUsage(
+                completion_tokens=10, prompt_tokens=20, total_tokens=30
+            ),
+        )
+
+        with patch(
+            "openai.resources.chat.completions.Completions.create",
+            return_value=[chunk1, chunk2],
+        ):
+            client = OpenAI(posthog_client=self.client, api_key="test")
+
+            messages = [{"role": "user", "content": self.test_user_message}]
+
+            response_generator = client.chat.completions.create(
+                model="gpt-4",
+                messages=messages,
+                system=self.test_system_prompt,
+                stream=True,
+                posthog_distinct_id="test-user",
+            )
+
+            list(response_generator)  # Consume generator
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    # Anthropic Tests
+    def test_anthropic_messages_array_system_prompt(self):
+        """Test Anthropic with system prompt in messages array."""
+        try:
+            from posthog.ai.anthropic import Anthropic
+        except ImportError:
+            self.skipTest("Anthropic package not available")
+
+        with patch("anthropic.resources.messages.Messages.create") as mock_create:
+            mock_response = MagicMock()
+            mock_response.usage.input_tokens = 20
+            mock_response.usage.output_tokens = 10
+            mock_response.usage.cache_read_input_tokens = None
+            mock_response.usage.cache_creation_input_tokens = None
+            mock_create.return_value = mock_response
+
+            client = Anthropic(posthog_client=self.client, api_key="test")
+
+            messages = [
+                {"role": "system", "content": self.test_system_prompt},
+                {"role": "user", "content": self.test_user_message},
+            ]
+
+            client.messages.create(
+                model="claude-3-5-sonnet-20241022",
+                messages=messages,
+                posthog_distinct_id="test-user",
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    def test_anthropic_separate_system_parameter(self):
+        """Test Anthropic with system prompt as separate parameter."""
+        try:
+            from posthog.ai.anthropic import Anthropic
+        except ImportError:
+            self.skipTest("Anthropic package not available")
+
+        with patch("anthropic.resources.messages.Messages.create") as mock_create:
+            mock_response = MagicMock()
+            mock_response.usage.input_tokens = 20
+            mock_response.usage.output_tokens = 10
+            mock_response.usage.cache_read_input_tokens = None
+            mock_response.usage.cache_creation_input_tokens = None
+            mock_create.return_value = mock_response
+
+            client = Anthropic(posthog_client=self.client, api_key="test")
+
+            messages = [{"role": "user", "content": self.test_user_message}]
+
+            client.messages.create(
+                model="claude-3-5-sonnet-20241022",
+                messages=messages,
+                system=self.test_system_prompt,
+                posthog_distinct_id="test-user",
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    # Gemini Tests
+    def test_gemini_contents_array_system_prompt(self):
+        """Test Gemini with system prompt in contents array."""
+        try:
+            from posthog.ai.gemini import Client
+        except ImportError:
+            self.skipTest("Gemini package not available")
+
+        with patch("google.genai.Client") as mock_genai_class:
+            mock_response = MagicMock()
+            mock_response.candidates = [MagicMock()]
+            mock_response.candidates[0].content.parts = [MagicMock()]
+            mock_response.candidates[0].content.parts[0].text = self.test_response
+            mock_response.usage_metadata.prompt_token_count = 20
+            mock_response.usage_metadata.candidates_token_count = 10
+            mock_response.usage_metadata.cached_content_token_count = None
+            mock_response.usage_metadata.thoughts_token_count = None
+
+            mock_client_instance = MagicMock()
+            mock_models_instance = MagicMock()
+            mock_models_instance.generate_content.return_value = mock_response
+            mock_client_instance.models = mock_models_instance
+            mock_genai_class.return_value = mock_client_instance
+
+            client = Client(posthog_client=self.client, api_key="test")
+
+            contents = [
+                {"role": "system", "content": self.test_system_prompt},
+                {"role": "user", "content": self.test_user_message},
+            ]
+
+            client.models.generate_content(
+                model="gemini-2.0-flash",
+                contents=contents,
+                posthog_distinct_id="test-user",
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])
+
+    def test_gemini_system_instruction_parameter(self):
+        """Test Gemini with system_instruction in config parameter."""
+        try:
+            from posthog.ai.gemini import Client
+        except ImportError:
+            self.skipTest("Gemini package not available")
+
+        with patch("google.genai.Client") as mock_genai_class:
+            mock_response = MagicMock()
+            mock_response.candidates = [MagicMock()]
+            mock_response.candidates[0].content.parts = [MagicMock()]
+            mock_response.candidates[0].content.parts[0].text = self.test_response
+            mock_response.usage_metadata.prompt_token_count = 20
+            mock_response.usage_metadata.candidates_token_count = 10
+            mock_response.usage_metadata.cached_content_token_count = None
+            mock_response.usage_metadata.thoughts_token_count = None
+
+            mock_client_instance = MagicMock()
+            mock_models_instance = MagicMock()
+            mock_models_instance.generate_content.return_value = mock_response
+            mock_client_instance.models = mock_models_instance
+            mock_genai_class.return_value = mock_client_instance
+
+            client = Client(posthog_client=self.client, api_key="test")
+
+            contents = [{"role": "user", "content": self.test_user_message}]
+            config = {"system_instruction": self.test_system_prompt}
+
+            client.models.generate_content(
+                model="gemini-2.0-flash",
+                contents=contents,
+                config=config,
+                posthog_distinct_id="test-user",
+            )
+
+            self.assertEqual(len(self.client.capture.call_args_list), 1)
+            properties = self.client.capture.call_args_list[0][1]["properties"]
+            self._assert_system_prompt_captured(properties["$ai_input"])

posthog/test/exception_integrations/__init__.py

@@ -1,3 +0,0 @@
-import pytest
-
-pytest.importorskip("django")

posthog/test/exception_integrations/test_django.py

@@ -1,68 +0,0 @@
-from posthog.exception_integrations.django import DjangoRequestExtractor
-
-DEFAULT_USER_AGENT = (
-    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3"
-)
-
-
-def mock_request_factory(override_headers):
-    class Request:
-        META = {}
-        # TRICKY: Actual django request dict object has case insensitive matching, and strips http from the names
-        headers = {
-            "User-Agent": DEFAULT_USER_AGENT,
-            "Referrer": "http://example.com",
-            "X-Forwarded-For": "193.4.5.12",
-            **(override_headers or {}),
-        }
-
-    return Request()
-
-
-def test_request_extractor_with_no_trace():
-    request = mock_request_factory(None)
-    extractor = DjangoRequestExtractor(request)
-    assert extractor.extract_person_data() == {
-        "ip": "193.4.5.12",
-        "user_agent": DEFAULT_USER_AGENT,
-        "traceparent": None,
-        "distinct_id": None,
-    }
-
-
-def test_request_extractor_with_trace():
-    request = mock_request_factory({"traceparent": "00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01"})
-    extractor = DjangoRequestExtractor(request)
-    assert extractor.extract_person_data() == {
-        "ip": "193.4.5.12",
-        "user_agent": DEFAULT_USER_AGENT,
-        "traceparent": "00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01",
-        "distinct_id": None,
-    }
-
-
-def test_request_extractor_with_tracestate():
-    request = mock_request_factory(
-        {
-            "traceparent": "00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01",
-            "tracestate": "posthog-distinct-id=1234",
-        }
-    )
-    extractor = DjangoRequestExtractor(request)
-    assert extractor.extract_person_data() == {
-        "ip": "193.4.5.12",
-        "user_agent": DEFAULT_USER_AGENT,
-        "traceparent": "00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01",
-        "distinct_id": "1234",
-    }
-
-
-def test_request_extractor_with_complicated_tracestate():
-    request = mock_request_factory({"tracestate": "posthog-distinct-id=alohaMountainsXUYZ,rojo=00f067aa0ba902b7"})
-    extractor = DjangoRequestExtractor(request)
-    assert extractor.extract_person_data() == {
-        "ip": "193.4.5.12",
-        "user_agent": DEFAULT_USER_AGENT,
-        "traceparent": None,
-        "distinct_id": "alohaMountainsXUYZ",
-    }

posthog/test/integrations/test_middleware.py

@@ -0,0 +1,771 @@
+from posthog.contexts import (
+    new_context,
+    get_context_session_id,
+    get_context_distinct_id,
+)
+import unittest
+from unittest.mock import Mock, patch
+import asyncio
+
+# Configure Django settings before importing middleware
+import django
+from django.conf import settings
+
+if not settings.configured:
+    settings.configure(
+        DEBUG=True,
+        SECRET_KEY="test-secret-key",
+        INSTALLED_APPS=[],
+        MIDDLEWARE=[],
+    )
+    django.setup()
+
+from posthog.integrations.django import PosthogContextMiddleware
+
+
+class MockRequest:
+    """Mock Django HttpRequest object"""
+
+    def __init__(
+        self,
+        headers=None,
+        method="GET",
+        path="/test",
+        host="example.com",
+        is_secure=False,
+    ):
+        self.headers = headers or {}
+        self.method = method
+        self.path = path
+        self._host = host
+        self._is_secure = is_secure
+
+    def build_absolute_uri(self):
+        scheme = "https" if self._is_secure else "http"
+        return f"{scheme}://{self._host}{self.path}"
+
+
+class TestPosthogContextMiddleware(unittest.TestCase):
+    def create_middleware(
+        self,
+        extra_tags=None,
+        request_filter=None,
+        tag_map=None,
+        capture_exceptions=True,
+        get_response=None,
+    ):
+        """Helper to create middleware instance with mock Django settings"""
+        if get_response is None:
+            get_response = Mock()
+
+        with patch("django.conf.settings") as mock_settings:
+            # Configure mock settings
+            mock_settings.POSTHOG_MW_EXTRA_TAGS = extra_tags
+            mock_settings.POSTHOG_MW_REQUEST_FILTER = request_filter
+            mock_settings.POSTHOG_MW_TAG_MAP = tag_map
+            mock_settings.POSTHOG_MW_CAPTURE_EXCEPTIONS = capture_exceptions
+            mock_settings.POSTHOG_MW_CLIENT = None
+
+            # Make hasattr work correctly
+            def mock_hasattr(obj, name):
+                return name in [
+                    "POSTHOG_MW_EXTRA_TAGS",
+                    "POSTHOG_MW_REQUEST_FILTER",
+                    "POSTHOG_MW_TAG_MAP",
+                    "POSTHOG_MW_CAPTURE_EXCEPTIONS",
+                    "POSTHOG_MW_CLIENT",
+                ]
+
+            with patch("builtins.hasattr", side_effect=mock_hasattr):
+                middleware = PosthogContextMiddleware(get_response)
+
+        return middleware
+
+    def test_extract_tags_basic(self):
+        with new_context():
+            """Test basic tag extraction from request"""
+            middleware = self.create_middleware()
+            request = MockRequest(
+                headers={
+                    "X-POSTHOG-SESSION-ID": "session-123",
+                    "X-POSTHOG-DISTINCT-ID": "user-456",
+                },
+                method="POST",
+                path="/api/test",
+                host="example.com",
+                is_secure=True,
+            )
+
+            tags = middleware.extract_tags(request)
+
+            self.assertEqual(get_context_session_id(), "session-123")
+            self.assertEqual(get_context_distinct_id(), "user-456")
+            self.assertEqual(tags["$current_url"], "https://example.com/api/test")
+            self.assertEqual(tags["$request_method"], "POST")
+
+    def test_extract_tags_missing_headers(self):
+        """Test tag extraction when PostHog headers are missing"""
+
+        with new_context():
+            middleware = self.create_middleware()
+            request = MockRequest(headers={}, method="GET", path="/home")
+
+            tags = middleware.extract_tags(request)
+
+            self.assertIsNone(get_context_session_id())
+            self.assertIsNone(get_context_distinct_id())
+            self.assertEqual(tags["$current_url"], "http://example.com/home")
+            self.assertEqual(tags["$request_method"], "GET")
+
+    def test_extract_tags_partial_headers(self):
+        """Test tag extraction with only some PostHog headers present"""
+
+        with new_context():
+            middleware = self.create_middleware()
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "session-only"}, method="PUT"
+            )
+
+            tags = middleware.extract_tags(request)
+
+            self.assertEqual(get_context_session_id(), "session-only")
+            self.assertIsNone(get_context_distinct_id())
+            self.assertEqual(tags["$request_method"], "PUT")
+
+    def test_extract_tags_with_extra_tags(self):
+        """Test tag extraction with extra_tags function"""
+
+        def extra_tags_func(request):
+            return {"custom_tag": "custom_value", "user_id": "789"}
+
+        with new_context():
+            middleware = self.create_middleware(extra_tags=extra_tags_func)
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "session-123"}, method="GET"
+            )
+
+            tags = middleware.extract_tags(request)
+
+            self.assertEqual(get_context_session_id(), "session-123")
+            self.assertEqual(tags["custom_tag"], "custom_value")
+            self.assertEqual(tags["user_id"], "789")
+
+    def test_extract_tags_with_tag_map(self):
+        """Test tag extraction with tag_map function"""
+
+        def extra_tags_func(request):
+            return {"custom_tag": "custom_value", "user_id": "789"}
+
+        def tag_map_func(tags):
+            if "custom_tag" in tags:
+                tags["mapped_custom_tag"] = f"mapped_{tags['custom_tag']}"
+                del tags["custom_tag"]
+            return tags
+
+        with new_context():
+            middleware = self.create_middleware(
+                tag_map=tag_map_func, extra_tags=extra_tags_func
+            )
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "session-123"}, method="GET"
+            )
+
+            tags = middleware.extract_tags(request)
+
+            self.assertEqual(tags["mapped_custom_tag"], "mapped_custom_value")
+
+    def test_extract_tags_extra_tags_returns_none(self):
+        """Test tag extraction when extra_tags returns None"""
+
+        def extra_tags_func(request):
+            return None
+
+        middleware = self.create_middleware(extra_tags=extra_tags_func)
+        request = MockRequest(method="GET")
+
+        tags = middleware.extract_tags(request)
+
+        self.assertEqual(tags["$request_method"], "GET")
+        # Should not crash when extra_tags returns None
+
+    def test_extract_tags_extra_tags_returns_empty_dict(self):
+        """Test tag extraction when extra_tags returns empty dict"""
+
+        def extra_tags_func(request):
+            return {}
+
+        middleware = self.create_middleware(extra_tags=extra_tags_func)
+        request = MockRequest(method="PATCH")
+
+        tags = middleware.extract_tags(request)
+
+        self.assertEqual(tags["$request_method"], "PATCH")
+
+    def test_process_exception_called_during_view_exception(self):
+        """
+        Unit test verifying process_exception captures exceptions per Django's contract.
+
+        Since this is a library test (no Django runtime), we simulate how Django
+        would invoke our middleware in production:
+        1. Middleware.__call__ creates context with request tags
+        2. View raises exception inside get_response
+        3. Django's BaseHandler catches it, calls process_exception, returns error response
+        4. Exception never propagates to middleware's context manager
+
+        We manually call process_exception to simulate Django's behavior - this is
+        the only way to test the hook without a full Django integration test.
+        """
+        mock_client = Mock()
+        view_exception = ValueError("View raised this error")
+        error_response = Mock(status_code=500)
+
+        def mock_get_response(request):
+            # Simulate Django's exception handling: catches view exception,
+            # calls process_exception hook if it exists, returns error response
+            if hasattr(middleware, "process_exception"):
+                middleware.process_exception(request, view_exception)
+            return error_response
+
+        middleware = self.create_middleware(get_response=mock_get_response)
+        middleware.client = mock_client
+
+        request = MockRequest(
+            headers={"X-POSTHOG-DISTINCT-ID": "test-user"},
+            method="POST",
+            path="/api/endpoint",
+        )
+        response = middleware(request)
+
+        self.assertEqual(response.status_code, 500)
+        mock_client.capture_exception.assert_called_once_with(view_exception)
+
+    def test_process_exception_respects_capture_exceptions_false(self):
+        """Verify process_exception respects capture_exceptions=False setting"""
+        mock_client = Mock()
+        view_exception = ValueError("Should not be captured")
+
+        def mock_get_response(request):
+            if hasattr(middleware, "process_exception"):
+                middleware.process_exception(request, view_exception)
+            return Mock(status_code=500)
+
+        middleware = self.create_middleware(
+            capture_exceptions=False, get_response=mock_get_response
+        )
+        middleware.client = mock_client
+
+        request = MockRequest()
+        middleware(request)
+
+        mock_client.capture_exception.assert_not_called()
+
+    def test_process_exception_respects_request_filter(self):
+        """Verify process_exception respects request_filter setting"""
+        mock_client = Mock()
+        view_exception = ValueError("Should be filtered")
+
+        def mock_get_response(request):
+            if hasattr(middleware, "process_exception"):
+                middleware.process_exception(request, view_exception)
+            return Mock(status_code=500)
+
+        middleware = self.create_middleware(
+            request_filter=lambda req: False,
+            capture_exceptions=True,
+            get_response=mock_get_response,
+        )
+        middleware.client = mock_client
+
+        request = MockRequest()
+        middleware(request)
+
+        mock_client.capture_exception.assert_not_called()
+
+
+class TestPosthogContextMiddlewareSync(unittest.TestCase):
+    """Test synchronous middleware behavior"""
+
+    def test_sync_middleware_call(self):
+        """Test that sync middleware correctly processes requests"""
+        mock_response = Mock()
+        get_response = Mock(return_value=mock_response)
+
+        # Create middleware with sync get_response
+        middleware = PosthogContextMiddleware(get_response)
+
+        # Verify sync mode detected
+        self.assertFalse(middleware._is_coroutine)
+
+        request = MockRequest(
+            headers={"X-POSTHOG-SESSION-ID": "test-session"},
+            method="GET",
+            path="/test",
+        )
+
+        with new_context():
+            response = middleware(request)
+
+            # Verify response returned
+            self.assertEqual(response, mock_response)
+            get_response.assert_called_once_with(request)
+
+    def test_sync_middleware_with_filter(self):
+        """Test sync middleware respects request filter"""
+        mock_response = Mock()
+        get_response = Mock(return_value=mock_response)
+
+        # Create middleware with request filter that filters all requests
+        request_filter = lambda req: False
+        middleware = PosthogContextMiddleware.__new__(PosthogContextMiddleware)
+        middleware.get_response = get_response
+        middleware._is_coroutine = False
+        middleware.request_filter = request_filter
+        middleware.capture_exceptions = True
+        middleware.client = None
+
+        request = MockRequest()
+
+        # Should skip context creation and return response directly
+        response = middleware(request)
+        self.assertEqual(response, mock_response)
+        get_response.assert_called_once_with(request)
+
+    def test_view_exceptions_only_captured_via_process_exception(self):
+        """
+        Demonstrates that process_exception is required to capture view exceptions.
+
+        In production Django, view exceptions don't propagate to middleware's context
+        manager because Django's BaseHandler catches them first and converts them to
+        error responses. Django provides the exception via process_exception hook instead.
+
+        This unit test proves:
+        1. Context manager in __call__ never sees view exceptions (Django intercepts)
+        2. Only process_exception can capture them
+        3. Without process_exception, exceptions are silently lost (v6.7.5 regression)
+
+        We manually call process_exception to verify the hook works - in production,
+        Django's BaseHandler would call it when a view raises.
+        """
+        mock_client = Mock()
+        get_response = Mock(return_value=Mock(status_code=500))
+
+        middleware = PosthogContextMiddleware(get_response)
+        middleware.client = mock_client
+
+        def get_response_simulating_django(request):
+            # Simulates Django behavior: view exception converted to error response,
+            # never propagates to middleware's context manager
+            return Mock(status_code=500)
+
+        middleware._sync_get_response = get_response_simulating_django
+
+        request = MockRequest()
+
+        response = middleware(request)
+        self.assertEqual(response.status_code, 500)
+
+        # Context manager didn't capture anything - exception was intercepted by Django
+        mock_client.capture_exception.assert_not_called()
+
+        # Verify process_exception hook exists and captures exceptions when called
+        if hasattr(middleware, "process_exception"):
+            exception = ValueError("View error")
+            middleware.process_exception(request, exception)
+            mock_client.capture_exception.assert_called_once_with(exception)
+        else:
+            self.fail(
+                "process_exception missing - view exceptions will not be captured!"
+            )
+
+
+class TestPosthogContextMiddlewareAsync(unittest.TestCase):
+    """Test asynchronous middleware behavior"""
+
+    def test_async_middleware_detection(self):
+        """Test that async get_response is correctly detected"""
+
+        async def async_get_response(request):
+            return Mock()
+
+        middleware = PosthogContextMiddleware(async_get_response)
+
+        # Verify async mode detected
+        self.assertTrue(middleware._is_coroutine)
+
+    def test_async_middleware_call(self):
+        """Test that async middleware correctly processes requests"""
+
+        async def run_test():
+            mock_response = Mock()
+
+            async def async_get_response(request):
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "async-session"},
+                method="POST",
+                path="/async-test",
+            )
+
+            with new_context():
+                # Call should return the coroutine from __acall__
+                result = middleware(request)
+
+                # Verify it's a coroutine
+                self.assertTrue(asyncio.iscoroutine(result))
+
+                # Await the result
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_with_filter(self):
+        """Test async middleware respects request filter"""
+
+        async def run_test():
+            mock_response = Mock()
+
+            async def async_get_response(request):
+                return mock_response
+
+            # Properly initialize middleware
+            middleware = PosthogContextMiddleware(async_get_response)
+            # Override request filter after initialization
+            middleware.request_filter = lambda req: False
+
+            request = MockRequest()
+
+            # Should skip context creation and return response directly
+            result = middleware(request)
+            response = await result
+            self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_context_propagation(self):
+        """Test that async middleware properly propagates context"""
+
+        async def run_test():
+            mock_response = Mock()
+
+            async def async_get_response(request):
+                # Verify context is available during async processing
+                session_id = get_context_session_id()
+                self.assertEqual(session_id, "async-session-123")
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "async-session-123"},
+                method="GET",
+            )
+
+            with new_context():
+                result = middleware(request)
+                await result
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_exception_capture(self):
+        """Test that async middleware captures exceptions during request processing"""
+
+        async def run_test():
+            mock_client = Mock()
+
+            # Make async_get_response raise an exception
+            async def raise_exception(request):
+                raise ValueError("Async test exception")
+
+            # Properly initialize middleware
+            middleware = PosthogContextMiddleware(raise_exception)
+            middleware.client = mock_client  # Override with mock client
+
+            request = MockRequest()
+
+            # Should capture exception and re-raise
+            with self.assertRaises(ValueError):
+                result = middleware(request)
+                await result
+
+            # Verify exception was captured by middleware
+            mock_client.capture_exception.assert_called_once()
+            captured_exception = mock_client.capture_exception.call_args[0][0]
+            self.assertIsInstance(captured_exception, ValueError)
+            self.assertEqual(str(captured_exception), "Async test exception")
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_with_authenticated_user(self):
+        """
+        Test that async middleware correctly extracts user info in async context.
+
+        Django's request.user is a SimpleLazyObject that defers DB access.
+        In async context, accessing it directly raises SynchronousOnlyOperation.
+        The middleware should use request.auser() instead.
+
+        This tests the fix for issue #355.
+        """
+
+        async def run_test():
+            mock_response = Mock()
+            mock_user = Mock()
+            mock_user.is_authenticated = True
+            mock_user.pk = 123
+            mock_user.email = "test@example.com"
+
+            async def async_get_response(request):
+                # Verify user info was extracted and set as distinct_id
+                distinct_id = get_context_distinct_id()
+                self.assertEqual(distinct_id, "123")
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.client = Mock()
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "test-session"}, method="GET"
+            )
+
+            # Mock auser() to return authenticated user
+            async def mock_auser():
+                return mock_user
+
+            request.auser = mock_auser
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_with_unauthenticated_user(self):
+        """
+        Test that async middleware handles unauthenticated users correctly.
+        """
+
+        async def run_test():
+            mock_response = Mock()
+            mock_user = Mock()
+            mock_user.is_authenticated = False  # Not authenticated
+
+            async def async_get_response(request):
+                # Verify no distinct_id was set (no user)
+                distinct_id = get_context_distinct_id()
+                self.assertIsNone(distinct_id)
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.client = Mock()
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "test-session"}, method="GET"
+            )
+
+            async def mock_auser():
+                return mock_user
+
+            request.auser = mock_auser
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_without_user_attribute(self):
+        """
+        Test that async middleware handles requests without user attribute (no auth middleware).
+        """
+
+        async def run_test():
+            mock_response = Mock()
+
+            async def async_get_response(request):
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.client = Mock()
+
+            # Request without auser method (no auth middleware)
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "test-session"}, method="GET"
+            )
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_with_extra_tags(self):
+        """
+        Test that async middleware works with extra_tags callback.
+        """
+
+        async def run_test():
+            mock_response = Mock()
+
+            def extra_tags_callback(request):
+                # Simple sync callback - should work
+                return {"custom_tag": "custom_value"}
+
+            async def async_get_response(request):
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.extra_tags = extra_tags_callback
+            middleware.client = Mock()
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "test-session"}, method="GET"
+            )
+
+            # Mock auser for no user
+            async def mock_auser():
+                return None
+
+            request.auser = mock_auser
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_with_tag_map(self):
+        """
+        Test that async middleware works with tag_map callback.
+        """
+
+        async def run_test():
+            mock_response = Mock()
+
+            def tag_map_callback(tags):
+                # Simple sync callback - should work
+                tags["mapped"] = "yes"
+                return tags
+
+            async def async_get_response(request):
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.tag_map = tag_map_callback
+            middleware.client = Mock()
+
+            request = MockRequest(
+                headers={"X-POSTHOG-SESSION-ID": "test-session"}, method="GET"
+            )
+
+            # Mock auser for no user
+            async def mock_auser():
+                return None
+
+            request.auser = mock_auser
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+    def test_async_middleware_user_extraction_with_all_headers(self):
+        """
+        Test async middleware extracts all request info correctly.
+        """
+
+        async def run_test():
+            mock_response = Mock()
+            mock_user = Mock()
+            mock_user.is_authenticated = True
+            mock_user.pk = 456
+            mock_user.email = "async@test.com"
+
+            async def async_get_response(request):
+                # Verify all context was set correctly
+                distinct_id = get_context_distinct_id()
+                session_id = get_context_session_id()
+                self.assertEqual(distinct_id, "456")
+                self.assertEqual(session_id, "async-sess-123")
+                return mock_response
+
+            middleware = PosthogContextMiddleware(async_get_response)
+            middleware.client = Mock()
+
+            request = MockRequest(
+                headers={
+                    "X-POSTHOG-SESSION-ID": "async-sess-123",
+                    "X-Forwarded-For": "192.168.1.1",
+                    "User-Agent": "TestAgent/1.0",
+                },
+                method="POST",
+                path="/api/test",
+            )
+
+            async def mock_auser():
+                return mock_user
+
+            request.auser = mock_auser
+
+            with new_context():
+                result = middleware(request)
+                response = await result
+                self.assertEqual(response, mock_response)
+
+        asyncio.run(run_test())
+
+
+class TestPosthogContextMiddlewareHybrid(unittest.TestCase):
+    """Test hybrid middleware behavior with mixed sync/async chains"""
+
+    def test_hybrid_flags_set(self):
+        """Test that both capability flags are set"""
+        self.assertTrue(PosthogContextMiddleware.sync_capable)
+        self.assertTrue(PosthogContextMiddleware.async_capable)
+
+    def test_sync_to_async_routing(self):
+        """Test that __call__ routes to __acall__ when async"""
+
+        async def run_test():
+            async def async_get_response(request):
+                return Mock()
+
+            middleware = PosthogContextMiddleware(async_get_response)
+
+            # Verify routing happens
+            request = MockRequest()
+            result = middleware(request)
+
+            # Should be a coroutine from __acall__
+            self.assertTrue(asyncio.iscoroutine(result))
+            await result  # Clean up
+
+        asyncio.run(run_test())
+
+    def test_sync_path_direct_return(self):
+        """Test that sync path returns directly without coroutine"""
+        mock_response = Mock()
+
+        def sync_get_response(request):
+            return mock_response
+
+        middleware = PosthogContextMiddleware(sync_get_response)
+
+        request = MockRequest()
+        result = middleware(request)
+
+        # Should NOT be a coroutine
+        self.assertFalse(asyncio.iscoroutine(result))
+        self.assertEqual(result, mock_response)
+
+
+if __name__ == "__main__":
+    unittest.main()

posthog/test/test_before_send.py

@@ -0,0 +1,218 @@
+import unittest
+
+import mock
+
+from posthog.client import Client
+from posthog.test.test_utils import FAKE_TEST_API_KEY
+
+
+class TestClient(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        # This ensures no real HTTP POST requests are made
+        cls.client_post_patcher = mock.patch("posthog.client.batch_post")
+        cls.consumer_post_patcher = mock.patch("posthog.consumer.batch_post")
+        cls.client_post_patcher.start()
+        cls.consumer_post_patcher.start()
+
+    @classmethod
+    def tearDownClass(cls):
+        cls.client_post_patcher.stop()
+        cls.consumer_post_patcher.stop()
+
+    def set_fail(self, e, batch):
+        """Mark the failure handler"""
+        print("FAIL", e, batch)  # noqa: T201
+        self.failed = True
+
+    def setUp(self):
+        self.failed = False
+        self.client = Client(FAKE_TEST_API_KEY, on_error=self.set_fail)
+
+    def test_before_send_callback_modifies_event(self):
+        """Test that before_send callback can modify events."""
+        processed_events = []
+
+        def my_before_send(event):
+            processed_events.append(event.copy())
+            if "properties" not in event:
+                event["properties"] = {}
+            event["properties"]["processed_by_before_send"] = True
+            return event
+
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=my_before_send,
+                sync_mode=True,
+            )
+            msg_uuid = client.capture(
+                "test_event", distinct_id="user1", properties={"original": "value"}
+            )
+
+            self.assertIsNotNone(msg_uuid)
+
+            # Get the enqueued message from the mock
+            mock_post.assert_called_once()
+            batch_data = mock_post.call_args[1]["batch"]
+            enqueued_msg = batch_data[0]
+
+            self.assertEqual(
+                enqueued_msg["properties"]["processed_by_before_send"], True
+            )
+            self.assertEqual(enqueued_msg["properties"]["original"], "value")
+            self.assertEqual(len(processed_events), 1)
+            self.assertEqual(processed_events[0]["event"], "test_event")
+
+    def test_before_send_callback_drops_event(self):
+        """Test that before_send callback can drop events by returning None."""
+
+        def drop_test_events(event):
+            if event.get("event") == "test_drop_me":
+                return None
+            return event
+
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=drop_test_events,
+                sync_mode=True,
+            )
+
+            # Event should be dropped
+            msg_uuid = client.capture("test_drop_me", distinct_id="user1")
+            self.assertIsNone(msg_uuid)
+
+            # Event should go through
+            msg_uuid = client.capture("keep_me", distinct_id="user1")
+            self.assertIsNotNone(msg_uuid)
+
+            # Check the enqueued message
+            mock_post.assert_called_once()
+            batch_data = mock_post.call_args[1]["batch"]
+            enqueued_msg = batch_data[0]
+            self.assertEqual(enqueued_msg["event"], "keep_me")
+
+    def test_before_send_callback_handles_exceptions(self):
+        """Test that exceptions in before_send don't crash the client."""
+
+        def buggy_before_send(event):
+            raise ValueError("Oops!")
+
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=buggy_before_send,
+                sync_mode=True,
+            )
+            msg_uuid = client.capture("robust_event", distinct_id="user1")
+
+            # Event should still be sent despite the exception
+            self.assertIsNotNone(msg_uuid)
+
+            # Check the enqueued message
+            mock_post.assert_called_once()
+            batch_data = mock_post.call_args[1]["batch"]
+            enqueued_msg = batch_data[0]
+            self.assertEqual(enqueued_msg["event"], "robust_event")
+
+    def test_before_send_callback_works_with_all_event_types(self):
+        """Test that before_send works with capture, set, etc."""
+
+        def add_marker(event):
+            if "properties" not in event:
+                event["properties"] = {}
+            event["properties"]["marked"] = True
+            return event
+
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=add_marker,
+                sync_mode=True,
+            )
+
+            # Test capture
+            msg_uuid = client.capture("event", distinct_id="user1")
+            self.assertIsNotNone(msg_uuid)
+
+            # Test set
+            msg_uuid = client.set(distinct_id="user1", properties={"prop": "value"})
+            self.assertIsNotNone(msg_uuid)
+
+            # Check all events were marked
+            self.assertEqual(mock_post.call_count, 2)
+            for call in mock_post.call_args_list:
+                batch_data = call[1]["batch"]
+                enqueued_msg = batch_data[0]
+                self.assertTrue(enqueued_msg["properties"]["marked"])
+
+    def test_before_send_callback_disabled_when_none(self):
+        """Test that client works normally when before_send is None."""
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=None,
+                sync_mode=True,
+            )
+            msg_uuid = client.capture("normal_event", distinct_id="user1")
+            self.assertIsNotNone(msg_uuid)
+
+            # Check the event was sent normally
+            mock_post.assert_called_once()
+            batch_data = mock_post.call_args[1]["batch"]
+            enqueued_msg = batch_data[0]
+            self.assertEqual(enqueued_msg["event"], "normal_event")
+
+    def test_before_send_callback_pii_scrubbing_example(self):
+        """Test a realistic PII scrubbing use case."""
+
+        def scrub_pii(event):
+            properties = event.get("properties", {})
+
+            # Mask email but keep domain
+            if "email" in properties:
+                email = properties["email"]
+                if "@" in email:
+                    domain = email.split("@")[1]
+                    properties["email"] = f"***@{domain}"
+                else:
+                    properties["email"] = "***"
+
+            # Remove credit card
+            properties.pop("credit_card", None)
+
+            return event
+
+        with mock.patch("posthog.client.batch_post") as mock_post:
+            client = Client(
+                FAKE_TEST_API_KEY,
+                on_error=self.set_fail,
+                before_send=scrub_pii,
+                sync_mode=True,
+            )
+            msg_uuid = client.capture(
+                "form_submit",
+                distinct_id="user1",
+                properties={
+                    "email": "user@example.com",
+                    "credit_card": "1234-5678-9012-3456",
+                    "form_name": "contact",
+                },
+            )
+
+            self.assertIsNotNone(msg_uuid)
+
+            # Check the enqueued message was scrubbed
+            mock_post.assert_called_once()
+            batch_data = mock_post.call_args[1]["batch"]
+            enqueued_msg = batch_data[0]
+
+            self.assertEqual(enqueued_msg["properties"]["email"], "***@example.com")
+            self.assertNotIn("credit_card", enqueued_msg["properties"])
+            self.assertEqual(enqueued_msg["properties"]["form_name"], "contact")

posthog/test/test_consumer.py

@@ -58,7 +58,11 @@ class TestConsumer(unittest.TestCase):
         with mock.patch("posthog.consumer.batch_post") as mock_post:
             consumer.start()
             for i in range(0, 3):
-                track = {"type": "track", "event": "python event %d" % i, "distinct_id": "distinct_id"}
+                track = {
+                    "type": "track",
+                    "event": "python event %d" % i,
+                    "distinct_id": "distinct_id",
+                }
                 q.put(track)
                 time.sleep(flush_interval * 1.1)
             self.assertEqual(mock_post.call_count, 3)
@@ -69,11 +73,17 @@ class TestConsumer(unittest.TestCase):
         q = Queue()
         flush_interval = 0.5
         flush_at = 10
-        consumer = Consumer(q, TEST_API_KEY, flush_at=flush_at, flush_interval=flush_interval)
+        consumer = Consumer(
+            q, TEST_API_KEY, flush_at=flush_at, flush_interval=flush_interval
+        )
         with mock.patch("posthog.consumer.batch_post") as mock_post:
             consumer.start()
             for i in range(0, flush_at * 2):
-                track = {"type": "track", "event": "python event %d" % i, "distinct_id": "distinct_id"}
+                track = {
+                    "type": "track",
+                    "event": "python event %d" % i,
+                    "distinct_id": "distinct_id",
+                }
                 q.put(track)
             time.sleep(flush_interval * 1.1)
             self.assertEqual(mock_post.call_count, 2)
@@ -91,8 +101,14 @@ class TestConsumer(unittest.TestCase):
 
         mock_post.call_count = 0
 
-        with mock.patch("posthog.consumer.batch_post", mock.Mock(side_effect=mock_post)):
-            track = {"type": "track", "event": "python event", "distinct_id": "distinct_id"}
+        with mock.patch(
+            "posthog.consumer.batch_post", mock.Mock(side_effect=mock_post)
+        ):
+            track = {
+                "type": "track",
+                "event": "python event",
+                "distinct_id": "distinct_id",
+            }
             # request() should succeed if the number of exceptions raised is
             # less than the retries paramater.
             if exception_count <= consumer.retries:
@@ -107,7 +123,8 @@ class TestConsumer(unittest.TestCase):
                     self.assertEqual(exc, expected_exception)
                 else:
                     self.fail(
-                        "request() should raise an exception if still failing after %d retries" % consumer.retries
+                        "request() should raise an exception if still failing after %d retries"
+                        % consumer.retries
                     )
 
     def test_request_retry(self):
@@ -148,7 +165,12 @@ class TestConsumer(unittest.TestCase):
         properties = {}
         for n in range(0, 500):
             properties[str(n)] = "one_long_property_value_to_build_a_big_event"
-        track = {"type": "track", "event": "python event", "distinct_id": "distinct_id", "properties": properties}
+        track = {
+            "type": "track",
+            "event": "python event",
+            "distinct_id": "distinct_id",
+            "properties": properties,
+        }
         msg_size = len(json.dumps(track).encode())
         # Let's capture 8MB of data to trigger two batches
         n_msgs = int(8_000_000 / msg_size)
@@ -158,10 +180,15 @@ class TestConsumer(unittest.TestCase):
             res.status_code = 200
             request_size = len(data.encode())
             # Batches close after the first message bringing it bigger than BATCH_SIZE_LIMIT, let's add 10% of margin
-            self.assertTrue(request_size < (5 * 1024 * 1024) * 1.1, "batch size (%d) higher than limit" % request_size)
+            self.assertTrue(
+                request_size < (5 * 1024 * 1024) * 1.1,
+                "batch size (%d) higher than limit" % request_size,
+            )
             return res
 
-        with mock.patch("posthog.request._session.post", side_effect=mock_post_fn) as mock_post:
+        with mock.patch(
+            "posthog.request._session.post", side_effect=mock_post_fn
+        ) as mock_post:
             consumer.start()
             for _ in range(0, n_msgs + 2):
                 q.put(track)

posthog/test/test_contexts.py

@@ -0,0 +1,207 @@
+import unittest
+from unittest.mock import patch
+
+from posthog.contexts import (
+    get_tags,
+    new_context,
+    scoped,
+    tag,
+    identify_context,
+    set_context_session,
+    get_context_session_id,
+    get_context_distinct_id,
+)
+
+
+class TestContexts(unittest.TestCase):
+    def test_tag_and_get_tags(self):
+        with new_context(fresh=True):
+            tag("key1", "value1")
+            tag("key2", 2)
+
+            tags = get_tags()
+            assert tags["key1"] == "value1"
+            assert tags["key2"] == 2
+
+    def test_new_context_isolation(self):
+        with new_context(fresh=True):
+            # Set tag in outer context
+            tag("outer", "value")
+
+            with new_context(fresh=True):
+                # Inner context should start empty
+                assert get_tags() == {}
+
+                # Set tag in inner context
+                tag("inner", "value")
+                assert get_tags()["inner"] == "value"
+
+                # Outer tag should not be visible
+                self.assertNotIn("outer", get_tags())
+
+            with new_context(fresh=False):
+                # Inner context should inherit outer tag
+                assert get_tags() == {"outer": "value"}
+
+            # After exiting context, inner tag should be gone
+            self.assertNotIn("inner", get_tags())
+
+            # Outer tag should still be there
+            assert get_tags()["outer"] == "value"
+
+    def test_nested_contexts(self):
+        with new_context(fresh=True):
+            tag("level1", "value1")
+
+            with new_context(fresh=True):
+                tag("level2", "value2")
+
+                with new_context(fresh=True):
+                    tag("level3", "value3")
+                    assert get_tags() == {"level3": "value3"}
+
+                # Back to level 2
+                assert get_tags() == {"level2": "value2"}
+
+            # Back to level 1
+            assert get_tags() == {"level1": "value1"}
+
+    @patch("posthog.capture_exception")
+    def test_scoped_decorator_success(self, mock_capture):
+        @scoped()
+        def successful_function(x, y):
+            tag("x", x)
+            tag("y", y)
+            return x + y
+
+        result = successful_function(1, 2)
+
+        # Function should execute normally
+        assert result == 3
+
+        # No exception should be captured
+        mock_capture.assert_not_called()
+
+        # Context should be cleared after function execution
+        assert get_tags() == {}
+
+    @patch("posthog.capture_exception")
+    def test_scoped_decorator_exception(self, mock_capture):
+        test_exception = ValueError("Test exception")
+
+        def check_context_on_capture(exception, **kwargs):
+            # Assert tags are available when capture_exception is called
+            current_tags = get_tags()
+            assert current_tags.get("important_context") == "value"
+
+        mock_capture.side_effect = check_context_on_capture
+
+        @scoped()
+        def failing_function():
+            tag("important_context", "value")
+            raise test_exception
+
+        # Function should raise the exception
+        with self.assertRaises(ValueError):
+            failing_function()
+
+        # Verify capture_exception was called
+        mock_capture.assert_called_once_with(test_exception)
+
+        # Context should be cleared after function execution
+        assert get_tags() == {}
+
+    @patch("posthog.capture_exception")
+    def test_new_context_exception_handling(self, mock_capture):
+        test_exception = RuntimeError("Context exception")
+
+        def check_context_on_capture(exception, **kwargs):
+            # Assert inner context tags are available when capture_exception is called
+            current_tags = get_tags()
+            assert current_tags.get("inner_context") == "inner_value"
+
+        mock_capture.side_effect = check_context_on_capture
+
+        # Set up outer context
+        with new_context():
+            tag("outer_context", "outer_value")
+
+            try:
+                with new_context():
+                    tag("inner_context", "inner_value")
+                    raise test_exception
+            except RuntimeError:
+                pass  # Expected exception
+
+            # Outer context should still be intact
+            assert get_tags()["outer_context"] == "outer_value"
+
+        # Verify capture_exception was called
+        mock_capture.assert_called_once_with(test_exception)
+
+    def test_identify_context(self):
+        with new_context(fresh=True):
+            # Initially no distinct ID
+            assert get_context_distinct_id() is None
+
+            # Set distinct ID
+            identify_context("user123")
+            assert get_context_distinct_id() == "user123"
+
+    def test_set_context_session(self):
+        with new_context(fresh=True):
+            # Initially no session ID
+            assert get_context_session_id() is None
+
+            # Set session ID
+            set_context_session("session456")
+            assert get_context_session_id() == "session456"
+
+    def test_context_inheritance_fresh_context(self):
+        with new_context(fresh=True):
+            identify_context("user123")
+            set_context_session("session456")
+
+            with new_context(fresh=True):
+                # Fresh context should not inherit
+                assert get_context_distinct_id() is None
+                assert get_context_session_id() is None
+
+            # Original context should still have values
+            assert get_context_distinct_id() == "user123"
+            assert get_context_session_id() == "session456"
+
+    def test_context_inheritance_non_fresh_context(self):
+        with new_context(fresh=True):
+            identify_context("user123")
+            set_context_session("session456")
+
+            with new_context(fresh=False):
+                # Non-fresh context should inherit
+                assert get_context_distinct_id() == "user123"
+                assert get_context_session_id() == "session456"
+
+                # Override in child context
+                identify_context("user789")
+                set_context_session("session999")
+                assert get_context_distinct_id() == "user789"
+                assert get_context_session_id() == "session999"
+
+            # Original context should still have original values
+            assert get_context_distinct_id() == "user123"
+            assert get_context_session_id() == "session456"
+
+    def test_scoped_decorator_with_context_ids(self):
+        @scoped()
+        def function_with_context():
+            identify_context("user456")
+            set_context_session("session789")
+            return get_context_distinct_id(), get_context_session_id()
+
+        distinct_id, session_id = function_with_context()
+        assert distinct_id == "user456"
+        assert session_id == "session789"
+
+        # Context should be cleared after function execution
+        assert get_context_distinct_id() is None
+        assert get_context_session_id() is None

posthog/test/test_exception_capture.py

@@ -34,17 +34,51 @@ def test_excepthook(tmpdir):
     )
 
 
-def test_trying_to_use_django_integration(tmpdir):
+def test_code_variables_capture(tmpdir):
     app = tmpdir.join("app.py")
     app.write(
         dedent(
             """
-    from posthog import Posthog, Integrations
-    posthog = Posthog('phc_x', host='https://eu.i.posthog.com', enable_exception_autocapture=True, exception_autocapture_integrations=[Integrations.Django], debug=True, on_error=lambda e, batch: print('error handling batch: ', e, batch))
-
-    # frame_value = "LOL"
-
-    1/0
+    import os
+    from posthog import Posthog
+    
+    class UnserializableObject:
+        pass
+    
+    posthog = Posthog(
+        'phc_x', 
+        host='https://eu.i.posthog.com', 
+        debug=True, 
+        enable_exception_autocapture=True,
+        capture_exception_code_variables=True,
+        project_root=os.path.dirname(os.path.abspath(__file__))
+    )
+    
+    def trigger_error():
+        my_string = "hello world"
+        my_number = 42
+        my_bool = True
+        my_dict = {"name": "test", "value": 123}
+        my_obj = UnserializableObject()
+        my_password = "secret123"  # Should be masked by default
+        __should_be_ignored = "hidden"  # Should be ignored by default
+        
+        1/0  # Trigger exception
+    
+    def intermediate_function():
+        request_id = "abc-123"
+        user_count = 100
+        is_active = True
+        
+        trigger_error()
+    
+    def process_data():
+        batch_size = 50
+        retry_count = 3
+        
+        intermediate_function()
+    
+    process_data()
     """
         )
     )
@@ -55,9 +89,246 @@ def test_trying_to_use_django_integration(tmpdir):
     output = excinfo.value.output
 
     assert b"ZeroDivisionError" in output
-    assert b"LOL" in output
-    assert b"DEBUG:posthog:data uploaded successfully" in output
-    assert (
-        b'"$exception_list": [{"mechanism": {"type": "generic", "handled": true}, "module": null, "type": "ZeroDivisionError", "value": "division by zero", "stacktrace": {"frames": [{"platform": "python", "filename": "app.py", "abs_path"'
-        in output
+    assert b"code_variables" in output
+
+    # Variables from trigger_error frame
+    assert b"'my_string': 'hello world'" in output
+    assert b"'my_number': 42" in output
+    assert b"'my_bool': 'True'" in output
+    assert b'"my_dict": "{\\"name\\": \\"test\\", \\"value\\": 123}"' in output
+    assert b'"my_obj": "<UnserializableObject>"' in output
+    assert b"'my_password': '$$_posthog_redacted_based_on_masking_rules_$$'" in output
+    assert b"'__should_be_ignored':" not in output
+
+    # Variables from intermediate_function frame
+    assert b"'request_id': 'abc-123'" in output
+    assert b"'user_count': 100" in output
+    assert b"'is_active': 'True'" in output
+
+    # Variables from process_data frame
+    assert b"'batch_size': 50" in output
+    assert b"'retry_count': 3" in output
+
+
+def test_code_variables_context_override(tmpdir):
+    app = tmpdir.join("app.py")
+    app.write(
+        dedent(
+            """
+    import os
+    import posthog
+    from posthog import Posthog
+    
+    posthog_client = Posthog(
+        'phc_x', 
+        host='https://eu.i.posthog.com', 
+        debug=True, 
+        enable_exception_autocapture=True,
+        capture_exception_code_variables=False,
+        project_root=os.path.dirname(os.path.abspath(__file__))
     )
+    
+    def process_data():
+        bank = "should_be_masked"
+        __dunder_var = "should_be_visible"
+        
+        1/0
+    
+    with posthog.new_context(client=posthog_client):
+        posthog.set_capture_exception_code_variables_context(True)
+        posthog.set_code_variables_mask_patterns_context([r"(?i).*bank.*"])
+        posthog.set_code_variables_ignore_patterns_context([])
+        
+        process_data()
+    """
+        )
+    )
+
+    with pytest.raises(subprocess.CalledProcessError) as excinfo:
+        subprocess.check_output([sys.executable, str(app)], stderr=subprocess.STDOUT)
+
+    output = excinfo.value.output
+
+    assert b"ZeroDivisionError" in output
+    assert b"code_variables" in output
+    assert b"'bank': '$$_posthog_redacted_based_on_masking_rules_$$'" in output
+    assert b"'__dunder_var': 'should_be_visible'" in output
+
+
+def test_code_variables_size_limiter(tmpdir):
+    app = tmpdir.join("app.py")
+    app.write(
+        dedent(
+            """
+    import os
+    from posthog import Posthog
+    
+    posthog = Posthog(
+        'phc_x', 
+        host='https://eu.i.posthog.com', 
+        debug=True, 
+        enable_exception_autocapture=True,
+        capture_exception_code_variables=True,
+        project_root=os.path.dirname(os.path.abspath(__file__))
+    )
+    
+    def trigger_error():
+        var_a = "a" * 2000
+        var_b = "b" * 2000
+        var_c = "c" * 2000
+        var_d = "d" * 2000
+        var_e = "e" * 2000
+        var_f = "f" * 2000
+        var_g = "g" * 2000
+        
+        1/0
+    
+    def intermediate_function():
+        var_h = "h" * 2000
+        var_i = "i" * 2000
+        var_j = "j" * 2000
+        var_k = "k" * 2000
+        var_l = "l" * 2000
+        var_m = "m" * 2000
+        var_n = "n" * 2000
+        
+        trigger_error()
+    
+    def process_data():
+        var_o = "o" * 2000
+        var_p = "p" * 2000
+        var_q = "q" * 2000
+        var_r = "r" * 2000
+        var_s = "s" * 2000
+        var_t = "t" * 2000
+        var_u = "u" * 2000
+        
+        intermediate_function()
+    
+    process_data()
+    """
+        )
+    )
+
+    with pytest.raises(subprocess.CalledProcessError) as excinfo:
+        subprocess.check_output([sys.executable, str(app)], stderr=subprocess.STDOUT)
+
+    output = excinfo.value.output.decode("utf-8")
+
+    assert "ZeroDivisionError" in output
+    assert "code_variables" in output
+
+    captured_vars = []
+    for var_name in [
+        "var_a",
+        "var_b",
+        "var_c",
+        "var_d",
+        "var_e",
+        "var_f",
+        "var_g",
+        "var_h",
+        "var_i",
+        "var_j",
+        "var_k",
+        "var_l",
+        "var_m",
+        "var_n",
+        "var_o",
+        "var_p",
+        "var_q",
+        "var_r",
+        "var_s",
+        "var_t",
+        "var_u",
+    ]:
+        if f"'{var_name}'" in output:
+            captured_vars.append(var_name)
+
+    assert len(captured_vars) > 0
+    assert len(captured_vars) < 21
+
+
+def test_code_variables_disabled_capture(tmpdir):
+    app = tmpdir.join("app.py")
+    app.write(
+        dedent(
+            """
+    import os
+    from posthog import Posthog
+    
+    posthog = Posthog(
+        'phc_x', 
+        host='https://eu.i.posthog.com', 
+        debug=True, 
+        enable_exception_autocapture=True,
+        capture_exception_code_variables=False,
+        project_root=os.path.dirname(os.path.abspath(__file__))
+    )
+    
+    def trigger_error():
+        my_string = "hello world"
+        my_number = 42
+        my_bool = True
+        
+        1/0
+    
+    trigger_error()
+    """
+        )
+    )
+
+    with pytest.raises(subprocess.CalledProcessError) as excinfo:
+        subprocess.check_output([sys.executable, str(app)], stderr=subprocess.STDOUT)
+
+    output = excinfo.value.output.decode("utf-8")
+
+    assert "ZeroDivisionError" in output
+    assert "'code_variables':" not in output
+    assert '"code_variables":' not in output
+    assert "'my_string'" not in output
+    assert "'my_number'" not in output
+
+
+def test_code_variables_enabled_then_disabled_in_context(tmpdir):
+    app = tmpdir.join("app.py")
+    app.write(
+        dedent(
+            """
+    import os
+    import posthog
+    from posthog import Posthog
+    
+    posthog_client = Posthog(
+        'phc_x', 
+        host='https://eu.i.posthog.com', 
+        debug=True, 
+        enable_exception_autocapture=True,
+        capture_exception_code_variables=True,
+        project_root=os.path.dirname(os.path.abspath(__file__))
+    )
+    
+    def process_data():
+        my_var = "should not be captured"
+        important_value = 123
+        
+        1/0
+    
+    with posthog.new_context(client=posthog_client):
+        posthog.set_capture_exception_code_variables_context(False)
+        
+        process_data()
+    """
+        )
+    )
+
+    with pytest.raises(subprocess.CalledProcessError) as excinfo:
+        subprocess.check_output([sys.executable, str(app)], stderr=subprocess.STDOUT)
+
+    output = excinfo.value.output.decode("utf-8")
+
+    assert "ZeroDivisionError" in output
+    assert "'code_variables':" not in output
+    assert '"code_variables":' not in output
+    assert "'my_var'" not in output
+    assert "'important_value'" not in output

posthog/test/test_feature_flag.py

@@ -10,8 +10,17 @@ class TestFeatureFlag(unittest.TestCase):
             "key": "test-flag",
             "enabled": True,
             "variant": "test-variant",
-            "reason": {"code": "matched_condition", "condition_index": 0, "description": "Matched condition set 1"},
-            "metadata": {"id": 1, "payload": '{"some": "json"}', "version": 2, "description": "test-description"},
+            "reason": {
+                "code": "matched_condition",
+                "condition_index": 0,
+                "description": "Matched condition set 1",
+            },
+            "metadata": {
+                "id": 1,
+                "payload": '{"some": "json"}',
+                "version": 2,
+                "description": "test-description",
+            },
         }
 
         flag = FeatureFlag.from_json(resp)
@@ -20,10 +29,21 @@ class TestFeatureFlag(unittest.TestCase):
         self.assertEqual(flag.variant, "test-variant")
         self.assertEqual(flag.get_value(), "test-variant")
         self.assertEqual(
-            flag.reason, FlagReason(code="matched_condition", condition_index=0, description="Matched condition set 1")
+            flag.reason,
+            FlagReason(
+                code="matched_condition",
+                condition_index=0,
+                description="Matched condition set 1",
+            ),
         )
         self.assertEqual(
-            flag.metadata, FlagMetadata(id=1, payload='{"some": "json"}', version=2, description="test-description")
+            flag.metadata,
+            FlagMetadata(
+                id=1,
+                payload='{"some": "json"}',
+                version=2,
+                description="test-description",
+            ),
         )
 
     def test_feature_flag_from_json_minimal(self):
@@ -44,7 +64,11 @@ class TestFeatureFlag(unittest.TestCase):
             "key": "test-flag",
             "enabled": True,
             "variant": "test-variant",
-            "reason": {"code": "matched_condition", "condition_index": 0, "description": "Matched condition set 1"},
+            "reason": {
+                "code": "matched_condition",
+                "condition_index": 0,
+                "description": "Matched condition set 1",
+            },
         }
 
         flag = FeatureFlag.from_json(resp)
@@ -53,13 +77,22 @@ class TestFeatureFlag(unittest.TestCase):
         self.assertEqual(flag.variant, "test-variant")
         self.assertEqual(flag.get_value(), "test-variant")
         self.assertEqual(
-            flag.reason, FlagReason(code="matched_condition", condition_index=0, description="Matched condition set 1")
+            flag.reason,
+            FlagReason(
+                code="matched_condition",
+                condition_index=0,
+                description="Matched condition set 1",
+            ),
         )
         self.assertEqual(flag.metadata, LegacyFlagMetadata(payload=None))
 
     def test_flag_reason_from_json(self):
         # Test with complete data
-        resp = {"code": "user_in_segment", "condition_index": 1, "description": "User is in segment 'beta_users'"}
+        resp = {
+            "code": "user_in_segment",
+            "condition_index": 1,
+            "description": "User is in segment 'beta_users'",
+        }
         reason = FlagReason.from_json(resp)
         self.assertEqual(reason.code, "user_in_segment")
         self.assertEqual(reason.condition_index, 1)
@@ -77,7 +110,12 @@ class TestFeatureFlag(unittest.TestCase):
 
     def test_flag_metadata_from_json(self):
         # Test with complete data
-        resp = {"id": 123, "payload": {"key": "value"}, "version": 1, "description": "Test flag"}
+        resp = {
+            "id": 123,
+            "payload": {"key": "value"},
+            "version": 1,
+            "description": "Test flag",
+        }
         metadata = FlagMetadata.from_json(resp)
         self.assertEqual(metadata.id, 123)
         self.assertEqual(metadata.payload, {"key": "value"})
@@ -106,7 +144,12 @@ class TestFeatureFlag(unittest.TestCase):
                 "condition_index": 1,
                 "description": "User is in segment 'beta_users'",
             },
-            "metadata": {"id": 123, "payload": {"key": "value"}, "version": 1, "description": "Test flag"},
+            "metadata": {
+                "id": 123,
+                "payload": {"key": "value"},
+                "version": 1,
+                "description": "Test flag",
+            },
         }
         flag = FeatureFlag.from_json(resp)
         self.assertEqual(flag.key, "test-flag")
@@ -131,7 +174,11 @@ class TestFeatureFlag(unittest.TestCase):
 
     def test_feature_flag_from_json_with_reason(self):
         # Test with reason but no metadata
-        resp = {"key": "test-flag", "enabled": True, "reason": {"code": "user_in_segment"}}
+        resp = {
+            "key": "test-flag",
+            "enabled": True,
+            "reason": {"code": "user_in_segment"},
+        }
         flag = FeatureFlag.from_json(resp)
         self.assertEqual(flag.key, "test-flag")
         self.assertTrue(flag.enabled)

posthog/test/test_feature_flag_result.py

@@ -9,7 +9,9 @@ from posthog.types import FeatureFlag, FeatureFlagResult, FlagMetadata, FlagReas
 
 class TestFeatureFlagResult(unittest.TestCase):
     def test_from_bool_value_and_payload(self):
-        result = FeatureFlagResult.from_value_and_payload("test-flag", True, "[1, 2, 3]")
+        result = FeatureFlagResult.from_value_and_payload(
+            "test-flag", True, "[1, 2, 3]"
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, True)
@@ -17,7 +19,9 @@ class TestFeatureFlagResult(unittest.TestCase):
         self.assertEqual(result.payload, [1, 2, 3])
 
     def test_from_false_value_and_payload(self):
-        result = FeatureFlagResult.from_value_and_payload("test-flag", False, '{"some": "value"}')
+        result = FeatureFlagResult.from_value_and_payload(
+            "test-flag", False, '{"some": "value"}'
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, False)
@@ -25,7 +29,9 @@ class TestFeatureFlagResult(unittest.TestCase):
         self.assertEqual(result.payload, {"some": "value"})
 
     def test_from_variant_value_and_payload(self):
-        result = FeatureFlagResult.from_value_and_payload("test-flag", "control", "true")
+        result = FeatureFlagResult.from_value_and_payload(
+            "test-flag", "control", "true"
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, True)
@@ -33,7 +39,9 @@ class TestFeatureFlagResult(unittest.TestCase):
         self.assertEqual(result.payload, True)
 
     def test_from_none_value_and_payload(self):
-        result = FeatureFlagResult.from_value_and_payload("test-flag", None, '{"some": "value"}')
+        result = FeatureFlagResult.from_value_and_payload(
+            "test-flag", None, '{"some": "value"}'
+        )
         self.assertIsNone(result)
 
     def test_from_boolean_flag_details(self):
@@ -41,8 +49,12 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant=None,
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload='"Some string"'),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload='"Some string"'
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
         result = FeatureFlagResult.from_flag_details(flag_details)
@@ -57,11 +69,17 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant=None,
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload='"Some string"'),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload='"Some string"'
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
-        result = FeatureFlagResult.from_flag_details(flag_details, override_match_value="control")
+        result = FeatureFlagResult.from_flag_details(
+            flag_details, override_match_value="control"
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, True)
@@ -73,11 +91,17 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant="control",
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload='{"some": "value"}'),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload='{"some": "value"}'
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
-        result = FeatureFlagResult.from_flag_details(flag_details, override_match_value=True)
+        result = FeatureFlagResult.from_flag_details(
+            flag_details, override_match_value=True
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, True)
@@ -89,11 +113,17 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant="control",
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload='{"some": "value"}'),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload='{"some": "value"}'
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
-        result = FeatureFlagResult.from_flag_details(flag_details, override_match_value=False)
+        result = FeatureFlagResult.from_flag_details(
+            flag_details, override_match_value=False
+        )
 
         self.assertEqual(result.key, "test-flag")
         self.assertEqual(result.enabled, False)
@@ -105,8 +135,12 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant="control",
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload='{"some": "value"}'),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload='{"some": "value"}'
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
         result = FeatureFlagResult.from_flag_details(flag_details)
@@ -126,8 +160,12 @@ class TestFeatureFlagResult(unittest.TestCase):
             key="test-flag",
             enabled=True,
             variant=None,
-            metadata=FlagMetadata(id=1, version=1, description="test-flag", payload=None),
-            reason=FlagReason(code="test-reason", description="test-reason", condition_index=0),
+            metadata=FlagMetadata(
+                id=1, version=1, description="test-flag", payload=None
+            ),
+            reason=FlagReason(
+                code="test-reason", description="test-reason", condition_index=0
+            ),
         )
 
         result = FeatureFlagResult.from_flag_details(flag_details)
@@ -191,9 +229,9 @@ class TestGetFeatureFlagResult(unittest.TestCase):
         self.assertEqual(flag_result.variant, None)
         self.assertEqual(flag_result.payload, 300)
         patch_capture.assert_called_with(
-            "some-distinct-id",
             "$feature_flag_called",
-            {
+            distinct_id="some-distinct-id",
+            properties={
                 "$feature_flag": "person-flag",
                 "$feature_flag_response": True,
                 "locally_evaluated": True,
@@ -245,9 +283,9 @@ class TestGetFeatureFlagResult(unittest.TestCase):
         self.assertEqual(flag_result.payload, {"some": "value"})
 
         patch_capture.assert_called_with(
-            "distinct_id",
             "$feature_flag_called",
-            {
+            distinct_id="distinct_id",
+            properties={
                 "$feature_flag": "person-flag",
                 "$feature_flag_response": "variant-1",
                 "locally_evaluated": True,
@@ -267,9 +305,9 @@ class TestGetFeatureFlagResult(unittest.TestCase):
         self.assertIsNone(another_flag_result.payload)
 
         patch_capture.assert_called_with(
-            "another-distinct-id",
             "$feature_flag_called",
-            {
+            distinct_id="another-distinct-id",
+            properties={
                 "$feature_flag": "person-flag",
                 "$feature_flag_response": "variant-2",
                 "locally_evaluated": True,
@@ -300,14 +338,16 @@ class TestGetFeatureFlagResult(unittest.TestCase):
             },
         }
 
-        flag_result = self.client.get_feature_flag_result("person-flag", "some-distinct-id")
+        flag_result = self.client.get_feature_flag_result(
+            "person-flag", "some-distinct-id"
+        )
         self.assertEqual(flag_result.enabled, True)
         self.assertEqual(flag_result.variant, None)
         self.assertEqual(flag_result.payload, 300)
         patch_capture.assert_called_with(
-            "some-distinct-id",
             "$feature_flag_called",
-            {
+            distinct_id="some-distinct-id",
+            properties={
                 "$feature_flag": "person-flag",
                 "$feature_flag_response": True,
                 "locally_evaluated": False,
@@ -348,9 +388,9 @@ class TestGetFeatureFlagResult(unittest.TestCase):
         self.assertEqual(flag_result.get_value(), "variant-1")
         self.assertEqual(flag_result.payload, [1, 2, 3])
         patch_capture.assert_called_with(
-            "distinct_id",
             "$feature_flag_called",
-            {
+            distinct_id="distinct_id",
+            properties={
                 "$feature_flag": "person-flag",
                 "$feature_flag_response": "variant-1",
                 "locally_evaluated": False,
@@ -385,13 +425,15 @@ class TestGetFeatureFlagResult(unittest.TestCase):
             },
         }
 
-        flag_result = self.client.get_feature_flag_result("no-person-flag", "some-distinct-id")
+        flag_result = self.client.get_feature_flag_result(
+            "no-person-flag", "some-distinct-id"
+        )
 
         self.assertIsNone(flag_result)
         patch_capture.assert_called_with(
-            "some-distinct-id",
             "$feature_flag_called",
-            {
+            distinct_id="some-distinct-id",
+            properties={
                 "$feature_flag": "no-person-flag",
                 "$feature_flag_response": None,
                 "locally_evaluated": False,

posthog/test/test_module.py

@@ -7,31 +7,19 @@ class TestModule(unittest.TestCase):
     posthog = None
 
     def _assert_enqueue_result(self, result):
-        self.assertEqual(type(result[0]), bool)
-        self.assertEqual(type(result[1]), dict)
+        self.assertEqual(type(result[0]), str)
 
     def failed(self):
         self.failed = True
 
     def setUp(self):
         self.failed = False
-        self.posthog = Posthog("testsecret", host="http://localhost:8000", on_error=self.failed)
-
-    def test_no_api_key(self):
-        self.posthog.api_key = None
-        self.assertRaises(Exception, self.posthog.capture)
-
-    def test_no_host(self):
-        self.posthog.host = None
-        self.assertRaises(Exception, self.posthog.capture)
+        self.posthog = Posthog(
+            "testsecret", host="http://localhost:8000", on_error=self.failed
+        )
 
     def test_track(self):
-        res = self.posthog.capture("distinct_id", "python module event")
-        self._assert_enqueue_result(res)
-        self.posthog.flush()
-
-    def test_identify(self):
-        res = self.posthog.identify("distinct_id", {"email": "user@email.com"})
+        res = self.posthog.capture("python module event", distinct_id="distinct_id")
         self._assert_enqueue_result(res)
         self.posthog.flush()
 
@@ -40,9 +28,5 @@ class TestModule(unittest.TestCase):
         self._assert_enqueue_result(res)
         self.posthog.flush()
 
-    def test_page(self):
-        self.posthog.page("distinct_id", "https://posthog.com/contact")
-        self.posthog.flush()
-
     def test_flush(self):
         self.posthog.flush()

posthog/test/test_request.py

@@ -6,20 +6,35 @@ import mock
 import pytest
 import requests
 
-from posthog.request import DatetimeSerializer, QuotaLimitError, batch_post, decide, determine_server_host
+from posthog.request import (
+    DatetimeSerializer,
+    QuotaLimitError,
+    batch_post,
+    decide,
+    determine_server_host,
+)
 from posthog.test.test_utils import TEST_API_KEY
 
 
 class TestRequests(unittest.TestCase):
     def test_valid_request(self):
-        res = batch_post(TEST_API_KEY, batch=[{"distinct_id": "distinct_id", "event": "python event", "type": "track"}])
+        res = batch_post(
+            TEST_API_KEY,
+            batch=[
+                {"distinct_id": "distinct_id", "event": "python event", "type": "track"}
+            ],
+        )
         self.assertEqual(res.status_code, 200)
 
     def test_invalid_request_error(self):
-        self.assertRaises(Exception, batch_post, "testsecret", "https://t.posthog.com", False, "[{]")
+        self.assertRaises(
+            Exception, batch_post, "testsecret", "https://t.posthog.com", False, "[{]"
+        )
 
     def test_invalid_host(self):
-        self.assertRaises(Exception, batch_post, "testsecret", "t.posthog.com/", batch=[])
+        self.assertRaises(
+            Exception, batch_post, "testsecret", "t.posthog.com/", batch=[]
+        )
 
     def test_datetime_serialization(self):
         data = {"created": datetime(2012, 3, 4, 5, 6, 7, 891011)}
@@ -35,14 +50,26 @@ class TestRequests(unittest.TestCase):
 
     def test_should_not_timeout(self):
         res = batch_post(
-            TEST_API_KEY, batch=[{"distinct_id": "distinct_id", "event": "python event", "type": "track"}], timeout=15
+            TEST_API_KEY,
+            batch=[
+                {"distinct_id": "distinct_id", "event": "python event", "type": "track"}
+            ],
+            timeout=15,
         )
         self.assertEqual(res.status_code, 200)
 
     def test_should_timeout(self):
         with self.assertRaises(requests.ReadTimeout):
             batch_post(
-                "key", batch=[{"distinct_id": "distinct_id", "event": "python event", "type": "track"}], timeout=0.0001
+                "key",
+                batch=[
+                    {
+                        "distinct_id": "distinct_id",
+                        "event": "python event",
+                        "type": "track",
+                    }
+                ],
+                timeout=0.0001,
             )
 
     def test_quota_limited_response(self):
@@ -68,7 +95,11 @@ class TestRequests(unittest.TestCase):
         mock_response = requests.Response()
         mock_response.status_code = 200
         mock_response._content = json.dumps(
-            {"featureFlags": {"flag1": True}, "featureFlagPayloads": {}, "errorsWhileComputingFlags": False}
+            {
+                "featureFlags": {"flag1": True},
+                "featureFlagPayloads": {},
+                "errorsWhileComputingFlags": False,
+            }
         ).encode("utf-8")
 
         with mock.patch("posthog.request._session.post", return_value=mock_response):

posthog/test/test_size_limited_dict.py

@@ -0,0 +1,24 @@
+import unittest
+
+from parameterized import parameterized
+
+from posthog import utils
+
+
+class TestSizeLimitedDict(unittest.TestCase):
+    @parameterized.expand([(10, 100), (5, 20), (20, 200)])
+    def test_size_limited_dict(self, size: int, iterations: int) -> None:
+        values = utils.SizeLimitedDict(size, lambda _: -1)
+
+        for i in range(iterations):
+            values[i] = i
+
+            assert values[i] == i
+            assert len(values) == i % size + 1
+
+            if i % size == 0:
+                # old numbers should've been removed
+                self.assertIsNone(values.get(i - 1))
+                self.assertIsNone(values.get(i - 3))
+                self.assertIsNone(values.get(i - 5))
+                self.assertIsNone(values.get(i - 9))

posthog/test/test_types.py

@@ -22,9 +22,16 @@ class TestTypes(unittest.TestCase):
                     enabled=True,
                     variant="test-variant",
                     reason=FlagReason(
-                        code="matched_condition", condition_index=0, description="Matched condition set 1"
+                        code="matched_condition",
+                        condition_index=0,
+                        description="Matched condition set 1",
+                    ),
+                    metadata=FlagMetadata(
+                        id=1,
+                        payload='{"some": "json"}',
+                        version=2,
+                        description="test-description",
                     ),
-                    metadata=FlagMetadata(id=1, payload='{"some": "json"}', version=2, description="test-description"),
                 )
             },
             "errorsWhileComputingFlags": has_errors,
@@ -39,10 +46,21 @@ class TestTypes(unittest.TestCase):
         self.assertEqual(flag.variant, "test-variant")
         self.assertEqual(flag.get_value(), "test-variant")
         self.assertEqual(
-            flag.reason, FlagReason(code="matched_condition", condition_index=0, description="Matched condition set 1")
+            flag.reason,
+            FlagReason(
+                code="matched_condition",
+                condition_index=0,
+                description="Matched condition set 1",
+            ),
         )
         self.assertEqual(
-            flag.metadata, FlagMetadata(id=1, payload='{"some": "json"}', version=2, description="test-description")
+            flag.metadata,
+            FlagMetadata(
+                id=1,
+                payload='{"some": "json"}',
+                version=2,
+                description="test-description",
+            ),
         )
         self.assertEqual(result["errorsWhileComputingFlags"], has_errors)
         self.assertEqual(result["requestId"], "test-id")
@@ -64,7 +82,9 @@ class TestTypes(unittest.TestCase):
         self.assertEqual(flag.variant, "test-variant")
         self.assertEqual(flag.get_value(), "test-variant")
         self.assertIsNone(flag.reason)
-        self.assertEqual(flag.metadata, LegacyFlagMetadata(payload='{"some": "json-payload"}'))
+        self.assertEqual(
+            flag.metadata, LegacyFlagMetadata(payload='{"some": "json-payload"}')
+        )
         self.assertFalse(result["errorsWhileComputingFlags"])
         self.assertEqual(result["requestId"], "test-id")
         # Verify legacy fields are removed
@@ -99,18 +119,29 @@ class TestTypes(unittest.TestCase):
                     enabled=True,
                     variant="test-variant",
                     reason=FlagReason(
-                        code="matched_condition", condition_index=0, description="Matched condition set 1"
+                        code="matched_condition",
+                        condition_index=0,
+                        description="Matched condition set 1",
+                    ),
+                    metadata=FlagMetadata(
+                        id=1,
+                        payload='{"some": "json"}',
+                        version=2,
+                        description="test-description",
                     ),
-                    metadata=FlagMetadata(id=1, payload='{"some": "json"}', version=2, description="test-description"),
                 ),
                 "my-boolean-flag": FeatureFlag(
                     key="my-boolean-flag",
                     enabled=True,
                     variant=None,
                     reason=FlagReason(
-                        code="matched_condition", condition_index=0, description="Matched condition set 1"
+                        code="matched_condition",
+                        condition_index=0,
+                        description="Matched condition set 1",
+                    ),
+                    metadata=FlagMetadata(
+                        id=1, payload=None, version=2, description="test-description"
                     ),
-                    metadata=FlagMetadata(id=1, payload=None, version=2, description="test-description"),
                 ),
                 "disabled-flag": FeatureFlag(
                     key="disabled-flag",
@@ -129,7 +160,9 @@ class TestTypes(unittest.TestCase):
         self.assertEqual(result["featureFlags"]["my-variant-flag"], "test-variant")
         self.assertEqual(result["featureFlags"]["my-boolean-flag"], True)
         self.assertEqual(result["featureFlags"]["disabled-flag"], False)
-        self.assertEqual(result["featureFlagPayloads"]["my-variant-flag"], '{"some": "json"}')
+        self.assertEqual(
+            result["featureFlagPayloads"]["my-variant-flag"], '{"some": "json"}'
+        )
         self.assertNotIn("my-boolean-flag", result["featureFlagPayloads"])
         self.assertNotIn("disabled-flag", result["featureFlagPayloads"])
 

posthog/test/test_utils.py

@@ -1,3 +1,4 @@
+import time
 import unittest
 from dataclasses import dataclass
 from datetime import date, datetime, timedelta
@@ -7,27 +8,41 @@ from uuid import UUID
 
 import six
 from dateutil.tz import tzutc
+from parameterized import parameterized
 from pydantic import BaseModel
 from pydantic.v1 import BaseModel as BaseModelV1
 
 from posthog import utils
+from posthog.types import FeatureFlagResult
 
 TEST_API_KEY = "kOOlRy2QlMY9jHZQv0bKz0FZyazBUoY8Arj0lFVNjs4"
 FAKE_TEST_API_KEY = "random_key"
 
 
 class TestUtils(unittest.TestCase):
+    @parameterized.expand(
+        [
+            ("naive datetime should be naive", True),
+            ("timezone-aware datetime should not be naive", False),
+        ]
+    )
+    def test_is_naive(self, _name: str, expected_naive: bool):
+        if expected_naive:
+            dt = datetime.now()  # naive datetime
+        else:
+            dt = datetime.now(tz=tzutc())  # timezone-aware datetime
+
+        assert utils.is_naive(dt) is expected_naive
+
     def test_timezone_utils(self):
         now = datetime.now()
         utcnow = datetime.now(tz=tzutc())
-        self.assertTrue(utils.is_naive(now))
-        self.assertFalse(utils.is_naive(utcnow))
 
         fixed = utils.guess_timezone(now)
-        self.assertFalse(utils.is_naive(fixed))
+        assert utils.is_naive(fixed) is False
 
         shouldnt_be_edited = utils.guess_timezone(utcnow)
-        self.assertEqual(utcnow, shouldnt_be_edited)
+        assert utcnow == shouldnt_be_edited
 
     def test_clean(self):
         simple = {
@@ -54,12 +69,12 @@ class TestUtils(unittest.TestCase):
         pre_clean_keys = combined.keys()
 
         utils.clean(combined)
-        self.assertEqual(combined.keys(), pre_clean_keys)
+        assert combined.keys() == pre_clean_keys
 
         # test UUID separately, as the UUID object doesn't equal its string representation according to Python
-        self.assertEqual(
-            utils.clean(UUID("12345678123456781234567812345678")),
-            "12345678-1234-5678-1234-567812345678",
+        assert (
+            utils.clean(UUID("12345678123456781234567812345678"))
+            == "12345678-1234-5678-1234-567812345678"
         )
 
     def test_clean_with_dates(self):
@@ -67,26 +82,27 @@ class TestUtils(unittest.TestCase):
             "birthdate": date(1980, 1, 1),
             "registration": datetime.now(tz=tzutc()),
         }
-        self.assertEqual(dict_with_dates, utils.clean(dict_with_dates))
+        assert dict_with_dates == utils.clean(dict_with_dates)
 
     def test_bytes(self):
-        if six.PY3:
-            item = bytes(10)
-        else:
-            item = bytearray(10)
-
+        item = bytes(10)
         utils.clean(item)
+        assert utils.clean(item) == "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
 
     def test_clean_fn(self):
         cleaned = utils.clean({"fn": lambda x: x, "number": 4})
-        self.assertEqual(cleaned["number"], 4)
-        # TODO: fixme, different behavior on python 2 and 3
-        if "fn" in cleaned:
-            self.assertEqual(cleaned["fn"], None)
+        assert cleaned == {"fn": None, "number": 4}
 
-    def test_remove_slash(self):
-        self.assertEqual("http://posthog.io", utils.remove_trailing_slash("http://posthog.io/"))
-        self.assertEqual("http://posthog.io", utils.remove_trailing_slash("http://posthog.io"))
+    @parameterized.expand(
+        [
+            ("http://posthog.io/", "http://posthog.io"),
+            ("http://posthog.io", "http://posthog.io"),
+            ("https://example.com/path/", "https://example.com/path"),
+            ("https://example.com/path", "https://example.com/path"),
+        ]
+    )
+    def test_remove_slash(self, input_url, expected_url):
+        assert expected_url == utils.remove_trailing_slash(input_url)
 
     def test_clean_pydantic(self):
         class ModelV2(BaseModel):
@@ -101,19 +117,26 @@ class TestUtils(unittest.TestCase):
         class NestedModel(BaseModel):
             foo: ModelV2
 
-        self.assertEqual(utils.clean(ModelV2(foo="1", bar=2)), {"foo": "1", "bar": 2, "baz": None})
-        self.assertEqual(utils.clean(ModelV1(foo=1, bar="2")), {"foo": 1, "bar": "2"})
-        self.assertEqual(
-            utils.clean(NestedModel(foo=ModelV2(foo="1", bar=2, baz="3"))),
-            {"foo": {"foo": "1", "bar": 2, "baz": "3"}},
-        )
+        assert utils.clean(ModelV2(foo="1", bar=2)) == {
+            "foo": "1",
+            "bar": 2,
+            "baz": None,
+        }
+        assert utils.clean(ModelV1(foo=1, bar="2")) == {"foo": 1, "bar": "2"}
+        assert utils.clean(NestedModel(foo=ModelV2(foo="1", bar=2, baz="3"))) == {
+            "foo": {"foo": "1", "bar": 2, "baz": "3"}
+        }
 
+    def test_clean_pydantic_like_class(self) -> None:
         class Dummy:
-            def model_dump(self, required_param):
-                pass
+            def model_dump(self, required_param: str) -> dict:
+                return {}
 
-        # Skips a class with a defined non-Pydantic `model_dump` method.
-        self.assertEqual(utils.clean({"test": Dummy()}), {})
+        # previously python 2 code would cause an error while cleaning,
+        # and this entire object would be None, and we would log an error
+        # let's allow ourselves to clean `Dummy` as None,
+        # without blatting the `test` key
+        assert utils.clean({"test": Dummy()}) == {"test": None}
 
     def test_clean_dataclass(self):
         @dataclass
@@ -130,47 +153,146 @@ class TestUtils(unittest.TestCase):
             bar: int
             nested: InnerDataClass
 
-        self.assertEqual(
-            utils.clean(
-                TestDataClass(
-                    foo="1",
-                    bar=2,
-                    nested=InnerDataClass(
-                        inner_foo="3",
-                        inner_bar=4,
-                        inner_uuid=UUID("12345678123456781234567812345678"),
-                        inner_date=datetime(2025, 1, 1),
-                    ),
-                )
-            ),
-            {
-                "foo": "1",
-                "bar": 2,
-                "nested": {
-                    "inner_foo": "3",
-                    "inner_bar": 4,
-                    "inner_uuid": "12345678-1234-5678-1234-567812345678",
-                    "inner_date": datetime(2025, 1, 1),
-                    "inner_optional": None,
-                },
+        assert utils.clean(
+            TestDataClass(
+                foo="1",
+                bar=2,
+                nested=InnerDataClass(
+                    inner_foo="3",
+                    inner_bar=4,
+                    inner_uuid=UUID("12345678123456781234567812345678"),
+                    inner_date=datetime(2025, 1, 1),
+                ),
+            )
+        ) == {
+            "foo": "1",
+            "bar": 2,
+            "nested": {
+                "inner_foo": "3",
+                "inner_bar": 4,
+                "inner_uuid": "12345678-1234-5678-1234-567812345678",
+                "inner_date": datetime(2025, 1, 1),
+                "inner_optional": None,
             },
+        }
+
+
+class TestFlagCache(unittest.TestCase):
+    def setUp(self):
+        self.cache = utils.FlagCache(max_size=3, default_ttl=1)
+        self.flag_result = FeatureFlagResult.from_value_and_payload(
+            "test-flag", True, None
         )
 
+    def test_cache_basic_operations(self):
+        distinct_id = "user123"
+        flag_key = "test-flag"
+        flag_version = 1
 
-class TestSizeLimitedDict(unittest.TestCase):
-    def test_size_limited_dict(self):
-        size = 10
-        values = utils.SizeLimitedDict(size, lambda _: -1)
+        # Test cache miss
+        result = self.cache.get_cached_flag(distinct_id, flag_key, flag_version)
+        assert result is None
 
-        for i in range(100):
-            values[i] = i
+        # Test cache set and hit
+        self.cache.set_cached_flag(
+            distinct_id, flag_key, self.flag_result, flag_version
+        )
+        result = self.cache.get_cached_flag(distinct_id, flag_key, flag_version)
+        assert result is not None
+        assert result.get_value()
 
-            self.assertEqual(values[i], i)
-            self.assertEqual(len(values), i % size + 1)
+    def test_cache_ttl_expiration(self):
+        distinct_id = "user123"
+        flag_key = "test-flag"
+        flag_version = 1
 
-            if i % size == 0:
-                # old numbers should've been removed
-                self.assertIsNone(values.get(i - 1))
-                self.assertIsNone(values.get(i - 3))
-                self.assertIsNone(values.get(i - 5))
-                self.assertIsNone(values.get(i - 9))
+        # Set flag in cache
+        self.cache.set_cached_flag(
+            distinct_id, flag_key, self.flag_result, flag_version
+        )
+
+        # Should be available immediately
+        result = self.cache.get_cached_flag(distinct_id, flag_key, flag_version)
+        assert result is not None
+
+        # Wait for TTL to expire (1 second + buffer)
+        time.sleep(1.1)
+
+        # Should be expired
+        result = self.cache.get_cached_flag(distinct_id, flag_key, flag_version)
+        assert result is None
+
+    def test_cache_version_invalidation(self):
+        distinct_id = "user123"
+        flag_key = "test-flag"
+        old_version = 1
+        new_version = 2
+
+        # Set flag with old version
+        self.cache.set_cached_flag(distinct_id, flag_key, self.flag_result, old_version)
+
+        # Should hit with old version
+        result = self.cache.get_cached_flag(distinct_id, flag_key, old_version)
+        assert result is not None
+
+        # Should miss with new version
+        result = self.cache.get_cached_flag(distinct_id, flag_key, new_version)
+        assert result is None
+
+        # Invalidate old version
+        self.cache.invalidate_version(old_version)
+
+        # Should miss even with old version after invalidation
+        result = self.cache.get_cached_flag(distinct_id, flag_key, old_version)
+        assert result is None
+
+    def test_stale_cache_functionality(self):
+        distinct_id = "user123"
+        flag_key = "test-flag"
+        flag_version = 1
+
+        # Set flag in cache
+        self.cache.set_cached_flag(
+            distinct_id, flag_key, self.flag_result, flag_version
+        )
+
+        # Wait for TTL to expire
+        time.sleep(1.1)
+
+        # Should not get fresh cache
+        result = self.cache.get_cached_flag(distinct_id, flag_key, flag_version)
+        assert result is None
+
+        # Should get stale cache (within 1 hour default)
+        stale_result = self.cache.get_stale_cached_flag(distinct_id, flag_key)
+        assert stale_result is not None
+        assert stale_result.get_value()
+
+    def test_lru_eviction(self):
+        # Cache has max_size=3, so adding 4 users should evict the LRU one
+        flag_version = 1
+
+        # Add 3 users
+        for i in range(3):
+            user_id = f"user{i}"
+            self.cache.set_cached_flag(
+                user_id, "test-flag", self.flag_result, flag_version
+            )
+
+        # Access user0 to make it recently used
+        self.cache.get_cached_flag("user0", "test-flag", flag_version)
+
+        # Add 4th user, should evict user1 (least recently used)
+        self.cache.set_cached_flag("user3", "test-flag", self.flag_result, flag_version)
+
+        # user0 should still be there (was recently accessed)
+        result = self.cache.get_cached_flag("user0", "test-flag", flag_version)
+        assert result is not None
+
+        # user2 should still be there (was recently added)
+        result = self.cache.get_cached_flag("user2", "test-flag", flag_version)
+        assert result is not None
+
+        # user3 should be there (just added)
+        result = self.cache.get_cached_flag("user3", "test-flag", flag_version)
+        assert result is not None

posthog/types.py

@@ -1,9 +1,34 @@
 import json
 from dataclasses import dataclass
-from typing import Any, List, Optional, TypedDict, Union, cast
+from typing import Any, Callable, List, Optional, TypedDict, Union, cast
 
 FlagValue = Union[bool, str]
 
+# Type alias for the before_send callback function
+# Takes an event dictionary and returns the modified event or None to drop it
+BeforeSendCallback = Callable[[dict[str, Any]], Optional[dict[str, Any]]]
+
+
+# Type alias for the send_feature_flags parameter
+class SendFeatureFlagsOptions(TypedDict, total=False):
+    """Options for sending feature flags with capture events.
+
+    Args:
+        only_evaluate_locally: Whether to only use local evaluation for feature flags.
+            If True, only flags that can be evaluated locally will be included.
+            If False, remote evaluation via /flags API will be used when needed.
+        person_properties: Properties to use for feature flag evaluation specific to this event.
+            These properties will be merged with any existing person properties.
+        group_properties: Group properties to use for feature flag evaluation specific to this event.
+            Format: { group_type_name: { group_properties } }
+    """
+
+    should_send: bool
+    only_evaluate_locally: Optional[bool]
+    person_properties: Optional[dict[str, Any]]
+    group_properties: Optional[dict[str, dict[str, Any]]]
+    flag_keys_filter: Optional[list[str]]
+
 
 @dataclass(frozen=True)
 class FlagReason:
@@ -78,7 +103,9 @@ class FeatureFlag:
         )
 
     @classmethod
-    def from_value_and_payload(cls, key: str, value: FlagValue, payload: Any) -> "FeatureFlag":
+    def from_value_and_payload(
+        cls, key: str, value: FlagValue, payload: Any
+    ) -> "FeatureFlag":
         enabled, variant = (True, value) if isinstance(value, str) else (value, None)
         return cls(
             key=key,
@@ -86,7 +113,7 @@ class FeatureFlag:
             variant=variant,
             reason=None,
             metadata=LegacyFlagMetadata(
-                payload=payload if payload else None,
+                payload=payload,
             ),
         )
 
@@ -154,13 +181,17 @@ class FeatureFlagResult:
             key=key,
             enabled=enabled,
             variant=variant,
-            payload=json.loads(payload) if isinstance(payload, str) else payload,
+            payload=json.loads(payload)
+            if isinstance(payload, str) and payload
+            else payload,
             reason=None,
         )
 
     @classmethod
     def from_flag_details(
-        cls, details: Union[FeatureFlag, None], override_match_value: Optional[FlagValue] = None
+        cls,
+        details: Union[FeatureFlag, None],
+        override_match_value: Optional[FlagValue] = None,
     ) -> "FeatureFlagResult | None":
         """
         Create a FeatureFlagResult from a FeatureFlag object.
@@ -179,7 +210,9 @@ class FeatureFlagResult:
 
         if override_match_value is not None:
             enabled, variant = (
-                (True, override_match_value) if isinstance(override_match_value, str) else (override_match_value, None)
+                (True, override_match_value)
+                if isinstance(override_match_value, str)
+                else (override_match_value, None)
             )
         else:
             enabled, variant = (details.enabled, details.variant)
@@ -191,6 +224,7 @@ class FeatureFlagResult:
             payload=(
                 json.loads(details.metadata.payload)
                 if isinstance(details.metadata.payload, str)
+                and details.metadata.payload
                 else details.metadata.payload
             ),
             reason=details.reason.description if details.reason else None,
@@ -226,7 +260,9 @@ def normalize_flags_response(resp: Any) -> FlagsResponse:
         # look at each key in featureFlags and create a FeatureFlag object
         flags = {}
         for key, value in featureFlags.items():
-            flags[key] = FeatureFlag.from_value_and_payload(key, value, featureFlagPayloads.get(key, None))
+            flags[key] = FeatureFlag.from_value_and_payload(
+                key, value, featureFlagPayloads.get(key, None)
+            )
         resp["flags"] = flags
     return cast(FlagsResponse, resp)
 
@@ -252,7 +288,11 @@ def to_values(response: FlagsResponse) -> Optional[dict[str, FlagValue]]:
         return None
 
     flags = response.get("flags", {})
-    return {key: value.get_value() for key, value in flags.items() if isinstance(value, FeatureFlag)}
+    return {
+        key: value.get_value()
+        for key, value in flags.items()
+        if isinstance(value, FeatureFlag)
+    }
 
 
 def to_payloads(response: FlagsResponse) -> Optional[dict[str, str]]:
@@ -262,5 +302,7 @@ def to_payloads(response: FlagsResponse) -> Optional[dict[str, str]]:
     return {
         key: value.metadata.payload
         for key, value in response.get("flags", {}).items()
-        if isinstance(value, FeatureFlag) and value.enabled and value.metadata.payload
+        if isinstance(value, FeatureFlag)
+        and value.enabled
+        and value.metadata.payload is not None
     }

posthog/utils.py

@@ -1,11 +1,17 @@
+import json
 import logging
 import numbers
 import re
+import time
 from collections import defaultdict
 from dataclasses import asdict, is_dataclass
 from datetime import date, datetime, timezone
 from decimal import Decimal
+from typing import Any, Optional
 from uuid import UUID
+import sys
+import platform
+import distro  # For Linux OS detection
 
 import six
 from dateutil.tz import tzlocal, tzutc
@@ -52,7 +58,9 @@ def clean(item):
         return float(item)
     if isinstance(item, UUID):
         return str(item)
-    if isinstance(item, (six.string_types, bool, numbers.Number, datetime, date, type(None))):
+    if isinstance(
+        item, (six.string_types, bool, numbers.Number, datetime, date, type(None))
+    ):
         return item
     if isinstance(item, (set, list, tuple)):
         return _clean_list(item)
@@ -99,14 +107,35 @@ def _clean_dataclass(dataclass_):
     return data
 
 
-def _coerce_unicode(cmplx):
+def _coerce_unicode(cmplx: Any) -> Optional[str]:
+    """
+    In theory, this method is only called
+    after many isinstance checks are carried out in `utils.clean`.
+    When we supported Python 2 it was safe to call `decode` on a `str`
+    but in Python 3 that will throw.
+    So, we check if the input is bytes and only call `decode` in that case.
+
+    Previously we would always call `decode` on the input
+    That would throw an error.
+    Then we would call `decode` on the stringified error
+    That would throw an error.
+    And then we would return `None`
+
+    To avoid a breaking change, we can maintain the behavior
+    that anything which did not have `decode` in Python 2
+    returns None.
+    """
+    item = None
     try:
-        item = cmplx.decode("utf-8", "strict")
-    except AttributeError as exception:
-        item = ":".join(exception)
-        item.decode("utf-8", "strict")
+        if isinstance(cmplx, bytes):
+            item = cmplx.decode("utf-8", "strict")
+        elif isinstance(cmplx, str):
+            item = cmplx
+    except Exception as exception:
+        item = ":".join(map(str, exception.args))
         log.warning("Error decoding: %s", item)
         return None
+
     return item
 
 
@@ -130,6 +159,266 @@ class SizeLimitedDict(defaultdict):
         super().__setitem__(key, value)
 
 
+class FlagCacheEntry:
+    def __init__(self, flag_result, flag_definition_version, timestamp=None):
+        self.flag_result = flag_result
+        self.flag_definition_version = flag_definition_version
+        self.timestamp = timestamp or time.time()
+
+    def is_valid(self, current_time, ttl, current_flag_version):
+        time_valid = (current_time - self.timestamp) < ttl
+        version_valid = self.flag_definition_version == current_flag_version
+        return time_valid and version_valid
+
+    def is_stale_but_usable(self, current_time, max_stale_age=3600):
+        return (current_time - self.timestamp) < max_stale_age
+
+
+class FlagCache:
+    def __init__(self, max_size=10000, default_ttl=300):
+        self.cache = {}  # distinct_id -> {flag_key: FlagCacheEntry}
+        self.access_times = {}  # distinct_id -> last_access_time
+        self.max_size = max_size
+        self.default_ttl = default_ttl
+
+    def get_cached_flag(self, distinct_id, flag_key, current_flag_version):
+        current_time = time.time()
+
+        if distinct_id not in self.cache:
+            return None
+
+        user_flags = self.cache[distinct_id]
+        if flag_key not in user_flags:
+            return None
+
+        entry = user_flags[flag_key]
+        if entry.is_valid(current_time, self.default_ttl, current_flag_version):
+            self.access_times[distinct_id] = current_time
+            return entry.flag_result
+
+        return None
+
+    def get_stale_cached_flag(self, distinct_id, flag_key, max_stale_age=3600):
+        current_time = time.time()
+
+        if distinct_id not in self.cache:
+            return None
+
+        user_flags = self.cache[distinct_id]
+        if flag_key not in user_flags:
+            return None
+
+        entry = user_flags[flag_key]
+        if entry.is_stale_but_usable(current_time, max_stale_age):
+            return entry.flag_result
+
+        return None
+
+    def set_cached_flag(
+        self, distinct_id, flag_key, flag_result, flag_definition_version
+    ):
+        current_time = time.time()
+
+        # Evict LRU users if we're at capacity
+        if distinct_id not in self.cache and len(self.cache) >= self.max_size:
+            self._evict_lru()
+
+        # Initialize user cache if needed
+        if distinct_id not in self.cache:
+            self.cache[distinct_id] = {}
+
+        # Store the flag result
+        self.cache[distinct_id][flag_key] = FlagCacheEntry(
+            flag_result, flag_definition_version, current_time
+        )
+        self.access_times[distinct_id] = current_time
+
+    def invalidate_version(self, old_version):
+        users_to_remove = []
+
+        for distinct_id, user_flags in self.cache.items():
+            flags_to_remove = []
+            for flag_key, entry in user_flags.items():
+                if entry.flag_definition_version == old_version:
+                    flags_to_remove.append(flag_key)
+
+            # Remove invalidated flags
+            for flag_key in flags_to_remove:
+                del user_flags[flag_key]
+
+            # Remove user entirely if no flags remain
+            if not user_flags:
+                users_to_remove.append(distinct_id)
+
+        # Clean up empty users
+        for distinct_id in users_to_remove:
+            del self.cache[distinct_id]
+            if distinct_id in self.access_times:
+                del self.access_times[distinct_id]
+
+    def _evict_lru(self):
+        if not self.access_times:
+            return
+
+        # Remove 20% of least recently used entries
+        sorted_users = sorted(self.access_times.items(), key=lambda x: x[1])
+        to_remove = max(1, len(sorted_users) // 5)
+
+        for distinct_id, _ in sorted_users[:to_remove]:
+            if distinct_id in self.cache:
+                del self.cache[distinct_id]
+            if distinct_id in self.access_times:
+                del self.access_times[distinct_id]
+
+    def clear(self):
+        self.cache.clear()
+        self.access_times.clear()
+
+
+class RedisFlagCache:
+    def __init__(
+        self, redis_client, default_ttl=300, stale_ttl=3600, key_prefix="posthog:flags:"
+    ):
+        self.redis = redis_client
+        self.default_ttl = default_ttl
+        self.stale_ttl = stale_ttl
+        self.key_prefix = key_prefix
+        self.version_key = f"{key_prefix}version"
+
+    def _get_cache_key(self, distinct_id, flag_key):
+        return f"{self.key_prefix}{distinct_id}:{flag_key}"
+
+    def _serialize_entry(self, flag_result, flag_definition_version, timestamp=None):
+        if timestamp is None:
+            timestamp = time.time()
+
+        # Use clean to make flag_result JSON-serializable for cross-platform compatibility
+        serialized_result = clean(flag_result)
+
+        entry = {
+            "flag_result": serialized_result,
+            "flag_version": flag_definition_version,
+            "timestamp": timestamp,
+        }
+        return json.dumps(entry)
+
+    def _deserialize_entry(self, data):
+        try:
+            entry = json.loads(data)
+            flag_result = entry["flag_result"]
+            return FlagCacheEntry(
+                flag_result=flag_result,
+                flag_definition_version=entry["flag_version"],
+                timestamp=entry["timestamp"],
+            )
+        except (json.JSONDecodeError, KeyError, ValueError):
+            # If deserialization fails, treat as cache miss
+            return None
+
+    def get_cached_flag(self, distinct_id, flag_key, current_flag_version):
+        try:
+            cache_key = self._get_cache_key(distinct_id, flag_key)
+            data = self.redis.get(cache_key)
+
+            if data:
+                entry = self._deserialize_entry(data)
+                if entry and entry.is_valid(
+                    time.time(), self.default_ttl, current_flag_version
+                ):
+                    return entry.flag_result
+
+            return None
+        except Exception:
+            # Redis error - return None to fall back to normal evaluation
+            return None
+
+    def get_stale_cached_flag(self, distinct_id, flag_key, max_stale_age=None):
+        try:
+            if max_stale_age is None:
+                max_stale_age = self.stale_ttl
+
+            cache_key = self._get_cache_key(distinct_id, flag_key)
+            data = self.redis.get(cache_key)
+
+            if data:
+                entry = self._deserialize_entry(data)
+                if entry and entry.is_stale_but_usable(time.time(), max_stale_age):
+                    return entry.flag_result
+
+            return None
+        except Exception:
+            # Redis error - return None
+            return None
+
+    def set_cached_flag(
+        self, distinct_id, flag_key, flag_result, flag_definition_version
+    ):
+        try:
+            cache_key = self._get_cache_key(distinct_id, flag_key)
+            serialized_entry = self._serialize_entry(
+                flag_result, flag_definition_version
+            )
+
+            # Set with TTL for automatic cleanup (use stale_ttl for total lifetime)
+            self.redis.setex(cache_key, self.stale_ttl, serialized_entry)
+
+            # Update the current version
+            self.redis.set(self.version_key, flag_definition_version)
+
+        except Exception:
+            # Redis error - silently fail, don't break flag evaluation
+            pass
+
+    def invalidate_version(self, old_version):
+        try:
+            # For Redis, we use a simple approach: scan for keys with old version
+            # and delete them. This could be expensive with many keys, but it's
+            # necessary for correctness.
+
+            cursor = 0
+            pattern = f"{self.key_prefix}*"
+
+            while True:
+                cursor, keys = self.redis.scan(cursor, match=pattern, count=100)
+
+                for key in keys:
+                    if key.decode() == self.version_key:
+                        continue
+
+                    try:
+                        data = self.redis.get(key)
+                        if data:
+                            entry_dict = json.loads(data)
+                            if entry_dict.get("flag_version") == old_version:
+                                self.redis.delete(key)
+                    except (json.JSONDecodeError, KeyError):
+                        # If we can't parse the entry, delete it to be safe
+                        self.redis.delete(key)
+
+                if cursor == 0:
+                    break
+
+        except Exception:
+            # Redis error - silently fail
+            pass
+
+    def clear(self):
+        try:
+            # Delete all keys matching our pattern
+            cursor = 0
+            pattern = f"{self.key_prefix}*"
+
+            while True:
+                cursor, keys = self.redis.scan(cursor, match=pattern, count=100)
+                if keys:
+                    self.redis.delete(*keys)
+                if cursor == 0:
+                    break
+        except Exception:
+            # Redis error - silently fail
+            pass
+
+
 def convert_to_datetime_aware(date_obj):
     if date_obj.tzinfo is None:
         date_obj = date_obj.replace(tzinfo=timezone.utc)
@@ -174,3 +463,57 @@ def str_iequals(value, comparand):
         False
     """
     return str(value).casefold() == str(comparand).casefold()
+
+
+def get_os_info():
+    """
+    Returns standardized OS name and version information.
+    Similar to how user agent parsing works in JS.
+    """
+    os_name = ""
+    os_version = ""
+
+    platform_name = sys.platform
+
+    if platform_name.startswith("win"):
+        os_name = "Windows"
+        if hasattr(platform, "win32_ver"):
+            win_version = platform.win32_ver()[0]
+            if win_version:
+                os_version = win_version
+
+    elif platform_name == "darwin":
+        os_name = "Mac OS X"
+        if hasattr(platform, "mac_ver"):
+            mac_version = platform.mac_ver()[0]
+            if mac_version:
+                os_version = mac_version
+
+    elif platform_name.startswith("linux"):
+        os_name = "Linux"
+        linux_info = distro.info()
+        if linux_info["version"]:
+            os_version = linux_info["version"]
+
+    elif platform_name.startswith("freebsd"):
+        os_name = "FreeBSD"
+        if hasattr(platform, "release"):
+            os_version = platform.release()
+
+    else:
+        os_name = platform_name
+        if hasattr(platform, "release"):
+            os_version = platform.release()
+
+    return os_name, os_version
+
+
+def system_context() -> dict[str, Any]:
+    os_name, os_version = get_os_info()
+
+    return {
+        "$python_runtime": platform.python_implementation(),
+        "$python_version": "%s.%s.%s" % (sys.version_info[:3]),
+        "$os": os_name,
+        "$os_version": os_version,
+    }

posthog/version.py

@@ -1,4 +1,4 @@
-VERSION = "4.2.0"
+VERSION = "6.9.0"
 
 if __name__ == "__main__":
     print(VERSION, end="")  # noqa: T201

pyproject.toml

@@ -1,10 +1,100 @@
-[tool.black]
-line-length = 120
+[build-system]
+requires = ["setuptools>=61.0", "wheel"]
+build-backend = "setuptools.build_meta"
 
-[tool.isort]
-multi_line_output = 3
-include_trailing_comma = true
-force_grid_wrap = 8
-ensure_newline_before_comments = true
-line_length = 120
-virtual_env = "env"
+[project]
+name = "posthog"
+dynamic = ["version"]
+description = "Integrate PostHog into any python application."
+authors = [{ name = "PostHog", email = "hey@posthog.com" }]
+maintainers = [{ name = "PostHog", email = "hey@posthog.com" }]
+license = { text = "MIT" }
+readme = "README.md"
+requires-python = ">=3.9"
+classifiers = [
+    "Development Status :: 5 - Production/Stable",
+    "Intended Audience :: Developers",
+    "Operating System :: OS Independent",
+    "License :: OSI Approved :: MIT License",
+    "Programming Language :: Python",
+    "Programming Language :: Python :: 3.9",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+]
+dependencies = [
+    "requests>=2.7,<3.0",
+    "six>=1.5",
+    "python-dateutil>=2.2",
+    "backoff>=1.10.0",
+    "distro>=1.5.0",
+    "typing-extensions>=4.2.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/posthog/posthog-python"
+Repository = "https://github.com/posthog/posthog-python"
+
+[project.optional-dependencies]
+langchain = ["langchain>=0.2.0"]
+dev = [
+    "django-stubs",
+    "lxml",
+    "mypy",
+    "mypy-baseline",
+    "types-mock",
+    "types-python-dateutil",
+    "types-requests",
+    "types-setuptools",
+    "types-six",
+    "pre-commit",
+    "pydantic",
+    "ruff",
+    "setuptools",
+    "packaging",
+    "wheel",
+    "twine",
+    "tomli",
+    "tomli_w",
+]
+test = [
+    "mock>=2.0.0",
+    "freezegun==1.5.1",
+    "coverage",
+    "pytest",
+    "pytest-timeout",
+    "pytest-asyncio",
+    "django",
+    "openai",
+    "anthropic",
+    "langgraph>=0.4.8",
+    "langchain-core>=0.3.65",
+    "langchain-community>=0.3.25",
+    "langchain-openai>=0.3.22",
+    "langchain-anthropic>=0.3.15",
+    "google-genai",
+    "pydantic",
+    "parameterized>=0.8.1",
+]
+
+[tool.setuptools]
+packages = [
+    "posthog",
+    "posthog.ai",
+    "posthog.ai.langchain",
+    "posthog.ai.openai",
+    "posthog.ai.anthropic",
+    "posthog.ai.gemini",
+    "posthog.test",
+    "posthog.integrations",
+]
+
+[tool.setuptools.dynamic]
+version = { attr = "posthog.version.VERSION" }
+
+[tool.pytest.ini_options]
+asyncio_mode = "auto"
+asyncio_default_fixture_loop_scope = "function"
+testpaths = ["posthog/test"]
+norecursedirs = ["integration_tests"]