Update main.py

main.py

@@ -527,8 +527,46 @@ def run_index_job(job_id: str, req: IndexRequest):
 
         # Warmup -> dimension
         warm = next(_chunk_with_spans(req.files[0].text if req.files else "", req.chunk_size, req.overlap))[2] if req.files else "warmup"
-        embs, _ = _post_embeddings([warm], job_id=job_id)
-        dim = embs.shape[1]
+        # warmup embeddings already computed: embs
+        try:
+            LOG.info(f"[{job_id}] warmup embeddings shape = {embs.shape} dtype={embs.dtype}")
+            _append_log(job_id, f"warmup embeddings shape = {embs.shape}")
+        except Exception:
+            pass
+        
+        # If using QdrantStore: check existing collection vector size and warn if mismatch
+        if isinstance(STORE, QdrantStore):
+            try:
+                # client.get_collection throws if not exists
+                info = STORE.client.get_collection(collection_name=col)
+                existing_size = None
+                # depending on qdrant client version, structure might be different:
+                if hasattr(info, "result") and isinstance(info.result, dict):
+                    cfg = info.result.get("params") or {}
+                    vectors = cfg.get("vectors") or {}
+                    existing_size = int(vectors.get("size")) if vectors.get("size") else None
+                elif isinstance(info, dict):
+                    cfg = info.get("result", info)
+                    vectors = cfg.get("params", {}).get("vectors", {})
+                    existing_size = int(vectors.get("size")) if vectors else None
+        
+                if existing_size and existing_size != dim:
+                    msg = (f"Qdrant collection {col} already exists with dim={existing_size} but embeddings dim={dim}. "
+                           "This will likely cause vectors to be rejected. Consider wiping or recreating collection.")
+                    LOG.error("[%s] %s", job_id, msg)
+                    _append_log(job_id, msg)
+                    # Optional: if WIPE_BEFORE_INDEX True, recreate:
+                    if WIPE_BEFORE_INDEX:
+                        try:
+                            STORE.wipe(col)
+                            STORE.ensure_collection(col, dim)
+                            _append_log(job_id, f"Recreated collection {col} with dim={dim} (WIPE_BEFORE_INDEX).")
+                        except Exception as e:
+                            _append_log(job_id, f"Failed recreate collection: {e}")
+            except Exception as e:
+                # collection not present or unable to introspect -> ok, ensure_collection will create
+                LOG.debug("[%s] Could not introspect collection: %s", job_id, e)
+
         STORE.ensure_collection(col, dim)
         _append_log(job_id, f"Collection ready: {col} (dim={dim})")