Update utilities/extract_text.py

utilities/extract_text.py

@@ -75,14 +75,24 @@ def extract_text_remote(state):
             "end_page": str(state.get("end_page", 1))  # String, not int
         }
         headers = {"Authorization": f"Bearer {os.getenv('HUGGINGFACE_API_TOKEN')}"}
+        # DON'T set Content-Type - let requests handle it for multipart/form-data
         
         print(f"\n🚀 Sending request to API...")
         print(f"File tuple: ('file', ('{file_basename}', <binary>, 'application/pdf'))")
         print(f"Data params: {data}")
         print(f"Data types: start_page={type(data['start_page'])}, end_page={type(data['end_page'])}")
+        print(f"Headers: {headers}")
+        print(f"File size in bytes: {file_size}")
         
         # Call API and wait for response
-        resp = requests.post(EXTRACT_TEXT_API, files=files, data=data, headers=headers, timeout=120)
+        # NOTE: Don't set Content-Type header - requests will set it automatically with boundary
+        try:
+            resp = requests.post(EXTRACT_TEXT_API, files=files, data=data, headers=headers, timeout=120)
+        except requests.exceptions.Timeout:
+            print(f"❌ Request timed out after 120 seconds")
+            raise RuntimeError("API request timed out")
+        except requests.exceptions.RequestException as e:
+            print(f"❌ Request exception: {str(e)}")
 
     print(f"\n📥 API Response:")
     print(f"Status Code: {resp.status_code}")