fix

ragg/ingest_trigger.py

@@ -41,6 +41,67 @@ def ingest_ping():
     print(info, flush=True)
     return jsonify(info), 200
 
+# ---------- LIGHTWEIGHT DEBUG (no ingestion run) ----------
+@ingest_trigger_bp.route("/ingest/debug", methods=["GET"])
+def ingest_debug():
+    import importlib
+    check_paths = ["ragg/pdfs/low", "ragg/pdfs/mid", "ragg/pdfs/high",
+                   "pdfs/low", "pdfs/mid", "pdfs/high"]
+    paths_info = []
+    for p in check_paths:
+        sample = _list_dir_sample(p)
+        paths_info.append({"path": p, **sample})
+
+    mod_name = os.getenv("INGEST_MODULE", "ragg.ingest_all")
+    import_ok, callable_ok, import_error = False, False, None
+    try:
+        mod = importlib.import_module(mod_name)
+        import_ok = True
+        callable_ok = hasattr(mod, "ingest_all_levels")
+    except Exception as e:
+        import_error = f"{type(e).__name__}: {e}"
+
+    resp = {
+        "cwd": os.getcwd(),
+        "env": {
+            "ENV": os.getenv("ENV"),
+            "INGEST_MODULE": mod_name,
+            "CHROMA_DIR": os.getenv("CHROMA_DIR"),
+            "CHROMA_ROOT": os.getenv("CHROMA_ROOT"),
+        },
+        "paths": paths_info,
+        "import_ok": import_ok,
+        "callable_ok": callable_ok,
+        "import_error": import_error
+    }
+    print("\n=== [INGEST DEBUG] ===", flush=True)
+    print(resp, flush=True)
+    return jsonify(resp), 200
+
+# ---------- INLINE RUN (use only for debugging) ----------
+@ingest_trigger_bp.route("/ingest/run-inline", methods=["POST"])
+def ingest_run_inline():
+    import importlib
+    try:
+        print("\n=== [INGEST INLINE] ===", flush=True)
+        mod_name = os.getenv("INGEST_MODULE", "ragg.ingest_all")
+        print("Importing module:", mod_name, flush=True)
+        mod = importlib.import_module(mod_name)
+        if not hasattr(mod, "ingest_all_levels"):
+            return jsonify({"status": "error", "message": "ingest_all_levels() not found"}), 500
+
+        # Optional: quick preflight
+        for p in ["ragg/pdfs/low", "ragg/pdfs/mid", "ragg/pdfs/high"]:
+            print(f"[INLINE] {p} -> {_list_dir_sample(p)}", flush=True)
+
+        mod.ingest_all_levels()
+        return jsonify({"status": "success", "message": "Ingest completed inline"}), 200
+    except Exception as e:
+        print("[ERROR][INLINE]", e, flush=True)
+        print(traceback.format_exc(), flush=True)
+        return jsonify({"status": "error", "error": f"{type(e).__name__}: {e}",
+                        "traceback": traceback.format_exc()}), 500
+
 @ingest_trigger_bp.route("/ingest", methods=["POST"])
 def run_ingest():
     if not _ingest_lock.acquire(blocking=False):
@@ -92,7 +153,7 @@ def run_ingest():
 
         print("\n[DEBUG] Subprocess completed.", flush=True)
         print("Return code:", result.returncode, flush=True)
-        # Print first and last few lines to avoid massive logs
+
         def _preview(label, text, head=30, tail=30):
             lines = (text or "").splitlines()
             print(f"\n----- {label} (total lines: {len(lines)}) -----", flush=True)
@@ -110,7 +171,6 @@ def run_ingest():
         payload = {
             "status": "success" if result.returncode == 0 else "error",
             "returncode": result.returncode,
-            # Return last ~4000 chars so the client can also see logs in Postman
             "stdout": (result.stdout or "")[-4000:],
             "stderr": (result.stderr or "")[-4000:],
         }
@@ -134,5 +194,4 @@ def run_ingest():
         try:
             _ingest_lock.release()
         except Exception:
-            # Avoid rare "release unlocked lock" noise
             pass