studio: Dedupe — one-click removal of exact-duplicate images (recoverable, ⌘Z)

The catalog seeder emits byte-identical copies (found 94 dups in one org's 268
visible assets). POST /v1/library/dedup hashes visible assets and soft-deletes
exact-content duplicates — keeps one per cluster, distinct re-renders (different
seeds) untouched — and returns the moved paths. Frontend "Dedupe" button beside
Select; the whole sweep pushes ONE undo entry so ⌘Z restores it all. no-store on
the response. Verified in-browser: dedup hides dups, Deleted count updates, ⌘Z
restores. v0.17.7

middleware/studio_home.html

@@ -290,6 +290,7 @@ body.immersive #stage{display:block}
     <button data-tab="deleted" onclick="tab('deleted')" title="Deleted content — recoverable">🗑 Deleted</button>
     <span class="sp"></span>
     <button class="btn" id="selbtn" onclick="toggleSel()" style="padding:6px 12px" title="Select many, then delete/recover them together">Select</button>
+    <button class="btn" onclick="dedup()" style="padding:6px 12px" title="Move exact-duplicate images to Deleted (recoverable, ⌘Z to undo)">Dedupe</button>
     <input class="search" id="q" placeholder="Search…" oninput="render()">
   </div>
   <section id="assets">
@@ -906,6 +907,18 @@ async function quickDel(path){
     if(!r.ok)throw 0;}catch(_){a.status=prev;UNDO.pop();render();paintChips();toast('Delete failed — restored');}
 }
 async function rerun(i){const a=DATA.assets[i];toast('Queueing re-run…');const r=await fetch('/v1/library/rerun',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({path:a.path})});const j=await r.json();toast(r.ok?`Re-run queued — see Runs`:'Failed');}
+// Dedupe: server hashes visible assets, soft-deletes exact byte-identical copies
+// (keeps one), returns the moved paths so ⌘Z undoes the whole sweep at once.
+async function dedup(){
+  toast('Scanning for exact duplicates…');
+  let r; try{ r=await (await fetch('/v1/library/dedup',{method:'POST'})).json(); }catch(_){ toast('Dedup failed'); return; }
+  const moved=r.deleted||[];
+  if(!moved.length){ toast('No exact duplicates found'); return; }
+  pushUndo(moved.map(m=>({path:m.path,prev:m.prev})));
+  moved.forEach(m=>{const a=DATA.assets.find(x=>x.path===m.path); if(a)a.status='deleted';});
+  render(); paintChips();
+  toast('Removed '+moved.length+' duplicate'+(moved.length>1?'s':'')+' → Trash · ⌘Z to undo');
+}
 // ── Fix surface: base image + up to 4 references (history picks + drag/drop
 // uploads). No references → the single-image edit (/v1/library/fix); any
 // reference → the multi-input compose (/v1/library/compose) — same proven Qwen

middleware/studio_home.py

@@ -1226,6 +1226,42 @@ def add_studio_home_routes(routes: web.RouteTableDef, server) -> None:
                 return web.json_response({"ok": True, "path": rel, "status": status})
         return web.json_response({"error": "asset not in library"}, status=404)
 
+    @routes.post("/v1/library/dedup")
+    async def dedup(request: web.Request):
+        """Soft-delete exact-content duplicates — keep one per byte-identical cluster,
+        move the rest to Deleted (recoverable, files kept). Content hash, so only TRUE
+        duplicates go; distinct re-renders (different seeds) are untouched. Returns the
+        moved paths so the UI can offer a one-shot undo."""
+        import hashlib
+        org = _org_of(request)
+        root = _library_root(org)
+        if root is None:
+            return web.json_response({"removed": 0, "deleted": []},
+                                     headers={"Cache-Control": "no-store"})
+        lib = _load_library(root)
+        seen: set[str] = set()
+        moved = []
+        for a in sorted(lib.get("assets", []), key=lambda x: x.get("path", "")):
+            if a.get("status") == "deleted":
+                continue
+            p = _safe_asset(root, a.get("path", ""))
+            if p is None:
+                continue
+            try:
+                h = hashlib.md5(p.read_bytes()).hexdigest()
+            except OSError:
+                continue
+            if h in seen:
+                moved.append({"path": a["path"], "prev": a.get("status", "draft")})
+                a["status"] = "deleted"
+                a["updatedAt"] = int(time.time())
+            else:
+                seen.add(h)
+        if moved:
+            _save_library(root, lib)
+        return web.json_response({"removed": len(moved), "deleted": moved},
+                                 headers={"Cache-Control": "no-store"})
+
     @routes.post("/v1/library/tags")
     async def add_tags(request: web.Request):
         """Merge tags onto a set of library assets (the bulk 'Add tags' action). Tags

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "hanzo-studio"
-version = "0.17.6"
+version = "0.17.7"
 readme = "README.md"
 license = { file = "LICENSE" }
 requires-python = ">=3.10"