Update utilities/extract_text.py

utilities/extract_text.py

@@ -85,14 +85,15 @@ def extract_text_remote(state):
         print(f"File size in bytes: {file_size}")
         
         # Call API and wait for response
-        # NOTE: Don't set Content-Type header - requests will set it automatically with boundary
+        # NOTE: Increased timeout to 300 seconds (5 minutes) for large files
         try:
-            resp = requests.post(EXTRACT_TEXT_API, files=files, data=data, headers=headers, timeout=120)
+            resp = requests.post(EXTRACT_TEXT_API, files=files, data=data, headers=headers, timeout=300)
         except requests.exceptions.Timeout:
-            print(f"❌ Request timed out after 120 seconds")
-            raise RuntimeError("API request timed out")
+            print(f"❌ Request timed out after 300 seconds")
+            raise RuntimeError("API request timed out after 5 minutes")
         except requests.exceptions.RequestException as e:
             print(f"❌ Request exception: {str(e)}")
+            raise RuntimeError(f"API request failed: {str(e)}")
 
     print(f"\n📥 API Response:")
     print(f"Status Code: {resp.status_code}")
@@ -100,17 +101,21 @@ def extract_text_remote(state):
     
     if resp.status_code != 200:
         print(f"❌ Error Response: {resp.text[:500]}")
-        raise RuntimeError(f"Extract text API failed: {resp.text}")
+        raise RuntimeError(f"Extract text API failed with status {resp.status_code}: {resp.text}")
 
     try:
         response_json = resp.json()
         print(f"Response JSON keys: {list(response_json.keys())}")
         print(f"Response JSON: {str(response_json)[:500]}")
         
-        extracted_text = response_json.get("text", "")
+        # IMPORTANT: The API returns {"status": "completed", "result": "text here", "process_id": "..."}
+        # NOT {"text": "..."}
+        extracted_text = response_json.get("result", "") or response_json.get("text", "")
         text_length = len(extracted_text) if extracted_text else 0
         
         print(f"\n📊 Extraction Result:")
+        print(f"API Status: {response_json.get('status', 'unknown')}")
+        print(f"Process ID: {response_json.get('process_id', 'none')}")
         print(f"Text Length: {text_length} characters")
         if text_length > 0:
             print(f"First 200 chars: {extracted_text[:200]}")