qdrant: enforce int/UUID point ids; preserve original id in payload; add error handling

qdrant/main.py

@@ -148,7 +148,27 @@ def _build_filter(metadata: Dict[str, Any]) -> Optional[Filter]:
 
 
 def _point_id(raw: Optional[str]) -> str:
-    return raw or uuid.uuid4().hex
+    """Return a Qdrant-compatible point id.
+
+    Qdrant accepts either an unsigned integer or a UUID string (with hyphens).
+    If the provided `raw` value is an int or valid UUID we return it (as int or str).
+    Otherwise we generate a new UUID string and the caller should store the
+    original `raw` value in the point payload under `_original_id`.
+    """
+    if not raw:
+        return str(uuid.uuid4())
+    # allow integer ids
+    try:
+        return int(raw)
+    except Exception:
+        pass
+    # allow UUID strings
+    try:
+        u = uuid.UUID(raw)
+        return str(u)
+    except Exception:
+        # fallback: generate a UUID
+        return str(uuid.uuid4())
 
 
 # ---------------------------------------------------------------------------
@@ -223,12 +243,19 @@ async def upsert_url(req: UpsertUrlRequest):
     vector = await _embed_url(req.url)
     pid = _point_id(req.id)
     payload = {**req.metadata, "source_url": req.url}
+    # preserve original user-provided id if it wasn't usable as a point id
+    if req.id is not None and str(pid) != str(req.id):
+        payload["_original_id"] = req.id
     col = _col(req.collection)
 
-    client.upsert(
-        collection_name=col,
-        points=[PointStruct(id=pid, vector=vector, payload=payload)],
-    )
+    try:
+        client.upsert(
+            collection_name=col,
+            points=[PointStruct(id=pid, vector=vector, payload=payload)],
+        )
+    except Exception as e:
+        raise HTTPException(500, str(e))
+
     return {"id": pid, "collection": col, "dim": len(vector)}
 
 
@@ -252,12 +279,19 @@ async def upsert_file(
             payload = json.loads(metadata_json)
         except json.JSONDecodeError:
             raise HTTPException(400, "metadata_json must be valid JSON")
+    # preserve original user-provided id if it wasn't usable as a point id
+    if id is not None and str(pid) != str(id):
+        payload["_original_id"] = id
 
     col = _col(collection)
-    client.upsert(
-        collection_name=col,
-        points=[PointStruct(id=pid, vector=vector, payload=payload)],
-    )
+    try:
+        client.upsert(
+            collection_name=col,
+            points=[PointStruct(id=pid, vector=vector, payload=payload)],
+        )
+    except Exception as e:
+        raise HTTPException(500, str(e))
+
     return {"id": pid, "collection": col, "dim": len(vector)}
 
 
@@ -266,10 +300,16 @@ def upsert_vector(req: UpsertVectorRequest):
     """Store a pre-computed vector directly (skip CLIP embedding)."""
     pid = _point_id(req.id)
     col = _col(req.collection)
-    client.upsert(
-        collection_name=col,
-        points=[PointStruct(id=pid, vector=req.vector, payload=req.metadata)],
-    )
+    payload = dict(req.metadata or {})
+    if req.id is not None and str(pid) != str(req.id):
+        payload["_original_id"] = req.id
+    try:
+        client.upsert(
+            collection_name=col,
+            points=[PointStruct(id=pid, vector=req.vector, payload=payload)],
+        )
+    except Exception as e:
+        raise HTTPException(500, str(e))
     return {"id": pid, "collection": col, "dim": len(req.vector)}