Update app.py

app.py

@@ -32,7 +32,7 @@ HF_TOKEN = os.getenv("HF_TOKEN") or os.getenv("HUGGINGFACE_HUB_TOKEN")
 
 def get_space_url():
     """Get the Hugging Face Space URL."""
-    space_id = os.getenv("SPACE_ID", f"{HF_USERNAME}/{HF_SPACE_NAME}")
+    space_id = os.getenv("SPACE_ID", HF_USERNAME)
     return f"https://huggingface.co/spaces/{space_id}/tree/main"
 
 def fetch_questions():
@@ -55,6 +55,49 @@ def fetch_random_question():
     except Exception as e:
         return "", f"Error fetching random question: {str(e)}"
 
+def clean_agent_answer(answer: str) -> str:
+    """
+    Clean agent answer to extract only the final answer.
+    Removes prefixes like "FINAL ANSWER:", explanations, etc.
+    """
+    if not answer:
+        return ""
+    
+    answer = str(answer).strip()
+    
+    # Remove "FINAL ANSWER:" prefix if present
+    prefixes = ["FINAL ANSWER:", "Final Answer:", "final answer:", "ANSWER:", "Answer:"]
+    for prefix in prefixes:
+        if answer.startswith(prefix):
+            answer = answer[len(prefix):].strip()
+    
+    # Try to extract just the answer if there's a lot of explanation
+    # Look for common patterns
+    lines = answer.split('\n')
+    
+    # If answer is very long, try to find the actual answer
+    if len(answer) > 500:
+        # Look for lines that might be the answer (short lines, numbers, etc.)
+        for line in reversed(lines):
+            line = line.strip()
+            if line and len(line) < 200 and not line.startswith(('The', 'This', 'I', 'We')):
+                # Might be the answer
+                if any(char.isdigit() for char in line) or len(line.split()) < 20:
+                    answer = line
+                    break
+    
+    # Remove markdown formatting if present
+    answer = answer.replace('**', '').replace('*', '').replace('`', '')
+    
+    # Take only first line if it seems like the answer
+    if '\n' in answer:
+        first_line = lines[0].strip()
+        # If first line is short and looks like an answer, use it
+        if len(first_line) < 200 and first_line:
+            answer = first_line
+    
+    return answer.strip()
+
 def get_answer_from_metadata(question: str):
     """Get the correct answer from metadata.jsonl if available."""
     if not os.path.exists(METADATA_FILE):
@@ -147,9 +190,15 @@ def process_all_questions(username: str, space_code: str, use_agent: bool = True
                 answer_source = "Error"
             else:
                 try:
-                    answer = run_agent(question)
-                    if not answer or answer.strip() == "":
+                    raw_answer = run_agent(question)
+                    if not raw_answer or raw_answer.strip() == "":
                         answer = "Agent returned empty answer"
+                    else:
+                        # Clean the answer to extract only the final answer
+                        answer = clean_agent_answer(raw_answer)
+                        if not answer or answer.strip() == "":
+                            # If cleaning removed everything, use original
+                            answer = raw_answer.strip()[:500]  # Limit length
                     answer_source = "Agent"
                 except Exception as e:
                     error_msg = str(e)
@@ -199,11 +248,22 @@ def process_all_questions(username: str, space_code: str, use_agent: bool = True
     }
     
     try:
+        # Log submission data for debugging
+        print(f"Submitting {len(answers_payload)} answers for user: {username}")
+        print(f"Space code: {space_code}")
+        
         response = requests.post(
             f"{DEFAULT_API_URL}/submit",
             json=submission_data,
-            timeout=60
+            timeout=300  # Increased timeout for large submissions
         )
+        
+        # Check response status
+        if response.status_code != 200:
+            error_text = response.text
+            print(f"Submission failed with status {response.status_code}: {error_text}")
+            return f"❌ Submission failed with status {response.status_code}: {error_text}", results
+        
         response.raise_for_status()
         result_data = response.json()
         
@@ -216,8 +276,23 @@ def process_all_questions(username: str, space_code: str, use_agent: bool = True
         )
         
         return status, results
+    except requests.exceptions.Timeout:
+        return f"❌ Submission timed out. This may take a while. Please try again or check your agent's response time.", results
+    except requests.exceptions.RequestException as e:
+        error_msg = f"Request error: {str(e)}"
+        print(error_msg)
+        if hasattr(e, 'response') and e.response is not None:
+            try:
+                error_detail = e.response.json()
+                error_msg += f"\nDetails: {error_detail}"
+            except:
+                error_msg += f"\nResponse: {e.response.text[:500]}"
+        return f"❌ Submission failed: {error_msg}", results
     except Exception as e:
-        return f"❌ Submission failed: {str(e)}", results
+        error_msg = f"Unexpected error: {str(e)}"
+        print(error_msg)
+        traceback.print_exc()
+        return f"❌ Submission failed: {error_msg}", results
 
 # Gradio Interface
 with gr.Blocks(title="GAIA Agent") as app:
@@ -279,8 +354,8 @@ with gr.Blocks(title="GAIA Agent") as app:
             
             space_code_input = gr.Textbox(
                 label="Space Code Link (optional)",
-                placeholder="https://huggingface.co/spaces/your-username/your-space/tree/main",
-                value="https://huggingface.co/spaces/ArdaKaratas/agent_hugging/tree/main"
+                placeholder="https://huggingface.co/spaces/your-username/tree/main",
+                value="https://huggingface.co/spaces/ArdaKaratas/tree/main"
             )
             
             use_agent_checkbox = gr.Checkbox(