Update app.py

app.py

@@ -4,7 +4,7 @@ import requests
 import pandas as pd
 from dotenv import load_dotenv
 import traceback
-from typing import Optional # <<< MAKE SURE THIS IMPORT IS PRESENT <<<
+from typing import Optional # Keep this import, good practice
 
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
@@ -12,13 +12,15 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 # --- Agent Integration ---
 AGENT_AVAILABLE = False
 AGENT_LOAD_ERROR = ""
+AGENT_FUNCTION_NAME = "run_gaia_task" # Define the target function name
+
 try:
-    # Ensure this matches the filename and function name exactly
-    from final_agent import answer_gaia_task
-    print("Successfully imported answer_gaia_task from final_agent.py")
+    # --- MODIFIED: Import the correct function ---
+    from final_agent import run_gaia_task
+    print(f"Successfully imported {AGENT_FUNCTION_NAME} from final_agent.py")
     AGENT_AVAILABLE = True
 except ImportError as e:
-    error_msg = f"ERROR: Could not import answer_gaia_task from final_agent.py: {e}"
+    error_msg = f"ERROR: Could not import {AGENT_FUNCTION_NAME} from final_agent.py: {e}"
     print(error_msg)
     AGENT_LOAD_ERROR = error_msg
 except Exception as e:
@@ -28,11 +30,13 @@ except Exception as e:
     traceback.print_exc()
     AGENT_LOAD_ERROR = error_msg
 
-# Define a dummy function if import fails, so Gradio app can still load
+# --- MODIFIED: Define a dummy function matching the new signature ---
 if not AGENT_AVAILABLE:
     # This dummy function will be used if the import fails
-    def answer_gaia_task(question: str, file_path: Optional[str] = None) -> str:
-        return f"ERROR: Agent function could not be loaded. Details: {AGENT_LOAD_ERROR}"
+    def run_gaia_task(task_description: str) -> str:
+        """Dummy function used when the real agent fails to load."""
+        print(f"Executing dummy {AGENT_FUNCTION_NAME} because agent failed to load.")
+        return f"ERROR: Agent function '{AGENT_FUNCTION_NAME}' could not be loaded. Details: {AGENT_LOAD_ERROR}"
 
 # --- Agent Runner Class ---
 class AgentRunner:
@@ -49,20 +53,23 @@ class AgentRunner:
         print(f"\n--- AgentRunner received question: {question[:100]}... ---")
         # Always call the potentially dummy function; it returns error if needed
         try:
-            # Call the imported (or dummy) function
-            # Assuming file_path is handled by the agent based on question text for now
-            final_answer = answer_gaia_task(question=question, file_path=None)
+            # --- MODIFIED: Call the new agent function ---
+            # The 'question' variable holds the task description.
+            # The new agent handles file paths internally based on the description.
+            final_answer = run_gaia_task(task_description=question)
+
             # Ensure result is always a string for submission
             final_answer_str = str(final_answer)
             print(f"--- AgentRunner returning answer: {final_answer_str} ---")
             return final_answer_str
         except Exception as e:
              # Catch unexpected errors during the function call itself
-             print(f"!!! ERROR calling answer_gaia_task function: {e} !!!")
+             print(f"!!! ERROR calling {AGENT_FUNCTION_NAME} function: {e} !!!")
              traceback.print_exc() # Log the full error to Space logs
-             return f"ERROR: Agent function failed during execution - {e}"
+             # --- MODIFIED: Update error message ---
+             return f"ERROR: Agent function '{AGENT_FUNCTION_NAME}' failed during execution - {e}"
 
-# --- Submission Logic ---
+# --- Submission Logic (Largely Unchanged) ---
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """Fetches questions, runs agent, submits answers."""
     space_id = os.getenv("SPACE_ID")
@@ -77,7 +84,8 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         agent = AgentRunner()
         # Check if agent loaded correctly before proceeding
         if not AGENT_AVAILABLE:
-             return f"Agent failed to load. Check logs. Error: {AGENT_LOAD_ERROR}", None
+             # --- MODIFIED: Update error message ---
+             return f"Agent function '{AGENT_FUNCTION_NAME}' failed to load. Check logs. Error: {AGENT_LOAD_ERROR}", None
     except Exception as e: print(f"Error instantiating AgentRunner: {e}"); return f"Init error: {e}", None
 
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id else "Code URL N/A"
@@ -101,7 +109,8 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         print(f"\n--- Processing Question {i+1}/{question_count} (ID: {task_id}) ---") # Add progress logging
         if not task_id or question_text is None: print(f"Skipping item: {item}"); continue
         try:
-            submitted_answer = agent(question_text) # Calls AgentRunner.__call__
+            # Calls AgentRunner.__call__, which now calls run_gaia_task
+            submitted_answer = agent(question_text)
             answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
             results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
         except Exception as e:
@@ -136,28 +145,32 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     print(status_message); results_df = pd.DataFrame(results_log); return status_message, results_df
 
 
-# --- Build Gradio Interface ---
+# --- Build Gradio Interface (Unchanged) ---
 with gr.Blocks() as demo:
     gr.Markdown("# GAIA Agent Evaluation Runner")
     gr.Markdown(
         """
         **Instructions:**
-        1. Ensure your agent logic is in `final_agent.py` and dependencies in `requirements.txt`. Set secrets in Space settings.
+        1. Ensure your agent logic is in `final_agent.py` (exposing the `run_gaia_task` function) and dependencies in `requirements.txt`. Set secrets in Space settings (GROQ_API_KEY, TAVILY_API_KEY, OPENAI_API_KEY).
         2. Log in to Hugging Face using the button below.
         3. Click 'Run Evaluation & Submit All Answers' to run your agent. Check Logs for detailed progress.
         ---
-        **Disclaimers:** Execution can take significant time.
+        **Disclaimers:** Execution can take significant time depending on the number of questions and agent complexity.
         """
     )
     gr.LoginButton()
     run_button = gr.Button("Run Evaluation & Submit All Answers")
     status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
     results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
-    run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table])
+    run_button.click(fn=run_and_submit_all, inputs=gr.State(None), outputs=[status_output, results_table]) # Pass None for profile initially
 
-# --- Main execution block ---
+# --- Main execution block (Unchanged) ---
 if __name__ == "__main__":
     print("\n" + "-"*30 + " App Starting " + "-"*30)
-    # ... Keep startup checks ...
+    # Perform startup checks if needed (e.g., check essential env vars)
+    if not AGENT_AVAILABLE:
+        print(f"CRITICAL WARNING: Agent function '{AGENT_FUNCTION_NAME}' could not be loaded. The app will run but agent calls will fail.")
+        print(f"Load Error Details: {AGENT_LOAD_ERROR}")
     print("Launching Gradio Interface...")
-    demo.launch(debug=True, share=False) # debug=True helps during development
+    # Consider removing debug=True for "production" submission space
+    demo.launch(debug=False, share=False)