Deploy GAIA agent

app.py

@@ -3,126 +3,113 @@
 import os
 import gradio as gr
 import requests
-import inspect
 import pandas as pd
+from bs4 import BeautifulSoup
 
-# SmolAgents imports
-from smolagents import CodeAgent, DuckDuckGoSearchTool, InferenceClientModel
+from smolagents import CodeAgent, tool, InferenceClientModel
 
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# --- Enhanced Agent Definition ---
+# --- Custom Web Search Tool (avoiding proxies) ---
+@tool
+def simple_search(query: str) -> str:
+    """
+    Search DuckDuckGo via HTML parsing.
+
+    Args:
+        query (str): search query string.
+
+    Returns:
+        str: top 3 titles and URLs.
+    """
+    try:
+        resp = requests.get(
+            "https://html.duckduckgo.com/html/",
+            params={"q": query},
+            timeout=10
+        )
+        resp.raise_for_status()
+        soup = BeautifulSoup(resp.text, "html.parser")
+        results = []
+        for a in soup.select("a.result__a")[:3]:
+            title = a.get_text()
+            url = a["href"]
+            results.append(f"- {title}\n  {url}")
+        return "\n".join(results) if results else "No results."
+    except Exception as e:
+        return f"Search error: {e}"
+
+# --- Enhanced Agent ---
 class BasicAgent:
     def __init__(self):
-        print("BasicAgent initialized with real agentic capabilities.")
-
-        # Initialize tools and model
-        self.search_tool = DuckDuckGoSearchTool()
+        print("BasicAgent initialized.")
         self.model = InferenceClientModel()
         self.agent = CodeAgent(
             model=self.model,
-            tools=[self.search_tool]
+            tools=[simple_search]
         )
 
     def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
+        print("Received:", question[:50], "...")
         try:
-            response = self.agent.run(question)
-            print(f"Agent response (first 50 chars): {response[:50]}...")
-            return response
+            return self.agent.run(question)
         except Exception as e:
-            print(f"Agent error during run: {e}")
-            return f"Error in agent: {e}"
+            return f"Agent run error: {e}"
 
 def run_and_submit_all(profile: gr.OAuthProfile | None):
-    """
-    Fetches all questions, runs the BasicAgent on them, submits all answers,
-    and displays the results.
-    """
     space_id = os.getenv("SPACE_ID")
-
-    if profile:
-        username = profile.username
-        print(f"User logged in: {username}")
-    else:
-        print("User not logged in.")
-        return "Please login to Hugging Face to submit answers.", None
-
+    if not profile:
+        return "Please login to Hugging Face", None
+    username = profile.username
     questions_url = f"{DEFAULT_API_URL}/questions"
     submit_url = f"{DEFAULT_API_URL}/submit"
 
-    try:
-        agent = BasicAgent()
+    try: agent = BasicAgent()
     except Exception as e:
-        print(f"Error instantiating agent: {e}")
-        return f"Error initializing agent: {e}", None
+        return f"Agent init error: {e}", None
 
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
-    print(agent_code)
-
-    # Fetch questions
     try:
-        resp = requests.get(questions_url, timeout=15)
-        resp.raise_for_status()
-        questions_data = resp.json()
-        if not questions_data:
-            return "Empty or invalid question list.", None
-        print(f"Fetched {len(questions_data)} questions.")
+        r = requests.get(questions_url, timeout=15)
+        r.raise_for_status()
+        qs = r.json()
     except Exception as e:
-        print(f"Error fetching questions: {e}")
-        return f"Error fetching questions: {e}", None
-
-    # Run agent on questions
-    results_log = []
-    answers_payload = []
-    for item in questions_data:
-        task_id = item.get("task_id")
-        question_text = item.get("question")
-        if not task_id or question_text is None:
-            continue
-        try:
-            submitted = agent(question_text)
-            answers_payload.append({"task_id": task_id, "submitted_answer": submitted})
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted})
-        except Exception as e:
-            print(f"Error on task {task_id}: {e}")
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"ERROR: {e}"})
+        return f"Fetch questions error: {e}", None
+
+    logs, payload = [], []
+    for item in qs:
+        tid, q = item.get("task_id"), item.get("question")
+        if not tid or q is None: continue
+        ans = agent(q)
+        payload.append({"task_id": tid, "submitted_answer": ans})
+        logs.append({"Task ID": tid, "Question": q, "Submitted Answer": ans})
 
-    if not answers_payload:
-        return "Agent did not produce any answers.", pd.DataFrame(results_log)
+    if not payload:
+        return "No answers generated.", pd.DataFrame(logs)
 
-    # Prepare & submit
-    payload = {"username": username, "agent_code": agent_code, "answers": answers_payload}
+    submit = {"username": username, "agent_code": agent_code, "answers": payload}
     try:
-        submit_resp = requests.post(submit_url, json=payload, timeout=60)
-        submit_resp.raise_for_status()
-        result_json = submit_resp.json()
-        final_status = (
-            f"Submission Successful!\n"
-            f"User: {result_json.get('username')}\n"
-            f"Score: {result_json.get('score', 'N/A')}% "
-            f"({result_json.get('correct_count', '?')}/{result_json.get('total_attempted', '?')} correct)\n"
-            f"Message: {result_json.get('message', '')}"
+        resp = requests.post(submit_url, json=submit, timeout=60)
+        resp.raise_for_status()
+        data = resp.json()
+        status = (
+            f"✅ Score: {data.get('score','N/A')}% "
+            f"({data.get('correct_count','?')}/{data.get('total_attempted','?')} correct)\n"
+            f"{data.get('message','No message')}"
         )
-        return final_status, pd.DataFrame(results_log)
+        return status, pd.DataFrame(logs)
     except Exception as e:
-        return f"Submission failed: {e}", pd.DataFrame(results_log)
+        return f"Submit error: {e}", pd.DataFrame(logs)
 
 # --- Gradio UI ---
 with gr.Blocks() as demo:
-    gr.Markdown("# Basic Agent Evaluation Runner")
-    gr.Markdown("""
-    Modify `BasicAgent` to add more tools or logic.
-    Log in, click **Run Evaluation & Submit All Answers**, and watch it process automatically.
-    """)
+    gr.Markdown("# GAIA Evaluation Runner")
     gr.LoginButton()
-    run_btn = gr.Button("Run Evaluation & Submit All Answers")
-    status = gr.Textbox(label="Status / Submission Result", lines=5, interactive=False)
-    results = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
-
-    run_btn.click(fn=run_and_submit_all, outputs=[status, results])
+    run_btn = gr.Button("Run Evaluation & Submit")
+    status = gr.Textbox(lines=5, interactive=False)
+    results = gr.DataFrame(wrap=True)
+    run_btn.click(run_and_submit_all, outputs=[status, results])
 
 if __name__ == "__main__":
-    print("Launching app...")
     demo.launch(debug=True, share=False)