test(documents): skip a format when its wheel is absent (portable unit gate)

pytest tests-unit runs in the GitHub-mirror unit workflow across an
ubuntu/windows/macos matrix. PDF (built-in fallback) and Markdown are stdlib and
always render; DOCX/XLSX/PPTX now skip gracefully when python-docx/openpyxl/
python-pptx isn't importable in that env instead of erroring. The canonical gate
(hanzo.yml installs requirements.txt) still exercises every format for real.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

tests-unit/documents_test/test_documents.py

@@ -13,6 +13,7 @@ Covers the four concerns end to end:
 Runs without a GPU, without torch, and without the WeasyPrint native stack (the built-in
 PDF fallback produces a valid %PDF), so it is safe for the dependency-light CI unit gate.
 """
+import importlib.util
 import io
 import json
 import sys
@@ -21,6 +22,16 @@ from pathlib import Path
 
 import pytest
 
+# PDF (built-in fallback) and Markdown are pure-stdlib and always render. DOCX/XLSX/PPTX
+# need their wheel; when it is absent (a bare env) we SKIP that format instead of failing —
+# the canonical gate (requirements.txt installed) exercises every format for real.
+_FMT_LIB = {"docx": "docx", "xlsx": "openpyxl", "pptx": "pptx"}
+
+
+def _fmt_available(fmt: str) -> bool:
+    lib = _FMT_LIB.get(fmt)
+    return lib is None or importlib.util.find_spec(lib) is not None
+
 sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
 
 from aiohttp import web
@@ -105,6 +116,8 @@ def test_public_catalog_is_json_serializable_without_build_fn():
 @pytest.mark.parametrize("tmpl", D.CATALOG, ids=[t["id"] for t in D.CATALOG])
 def test_every_template_renders_every_format(tmpl):
     for fmt in tmpl["formats"]:
+        if not _fmt_available(fmt):
+            continue  # wheel not installed here; the canonical gate covers it
         data, title = D.render_document(tmpl, tmpl["sample"], fmt)
         assert isinstance(data, (bytes, bytearray)) and data, f"{tmpl['id']}/{fmt} produced no bytes"
         assert _valid(fmt, data), f"{tmpl['id']}/{fmt} invalid magic bytes"
@@ -310,8 +323,10 @@ def test_generate_from_fields_then_download_pdf_and_docx(client):
     assert doc["id"] and doc["title"].startswith("Invoice")
     assert set(doc["download"]) == set(inv["formats"])
 
-    for fmt, want_ct in (("pdf", "application/pdf"),
-                         ("docx", "application/vnd.openxmlformats-officedocument.wordprocessingml.document")):
+    fmts = [("pdf", "application/pdf")]
+    if _fmt_available("docx"):
+        fmts.append(("docx", "application/vnd.openxmlformats-officedocument.wordprocessingml.document"))
+    for fmt, want_ct in fmts:
         dl = client.get(doc["download"][fmt])
         assert dl.status == 200
         assert dl.headers["Content-Type"] == want_ct
@@ -322,6 +337,8 @@ def test_generate_from_fields_then_download_pdf_and_docx(client):
 
 def test_generate_each_archetype_native_format(client):
     for tid, fmt in (("resume", "pdf"), ("budget", "xlsx"), ("pitch_deck", "pptx")):
+        if not _fmt_available(fmt):
+            continue
         t = D.get_template(tid)
         doc = _gen(client, tid, fields=t["sample"]).json()
         dl = client.get(doc["download"][fmt])