Update app.py

app.py

@@ -1,18 +1,31 @@
+import os
+import gradio as gr
+import requests
+import pandas as pd
+
 from smolagents import CodeAgent, HfApiModel, DuckDuckGoSearchTool, tool
 
-# --- TOOL DEFINITIONS ---
+# --- Constants ---
+DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+
+# --- Tool Definitions ---
 
 @tool
 def summarize_query(query: str) -> str:
     """
-    Provides a structured summary to reframe a query if the search results are unclear or poor.
+    Provides a structured summary to reframe a query if search results are unclear or poor.
+
+    Args:
+        query (str): The search query that needs summarization.
+
+    Returns:
+        str: A concise summary of key facts about the given query.
     """
     return f"Summarize and reframe: {query}"
 
-# Live DuckDuckGo search tool
 search_tool = DuckDuckGoSearchTool()
 
-# --- SYSTEM PROMPT WITH RETRY STRATEGY ---
+# --- System Prompt for ReACT + Scratchpad + Auto-Retry ---
 
 system_prompt = """
 You are a ReACT agent with scratchpad memory and a retry mechanism.
@@ -22,40 +35,167 @@ For every question:
 2. Action: (Optional) Use a tool with a clear query.
 3. Observation: Record what tool returned.
 
-**If the first Observation seems empty, irrelevant, or very weak:**
-4. Thought: "The result was unclear. I should reframe and try again."
-5. Action: Use summarize_query to improve the search query.
-6. Action: Retry DuckDuckGoSearchTool with the improved query.
-7. Observation: Record result.
-8. Thought: Reflect carefully across both Observations.
-9. FINAL ANSWER: Provide the final answer.
+If the first Observation is empty or irrelevant:
+4. Thought: The result was unclear. I should reframe and retry.
+5. Action: summarize_query with the original query.
+6. Action: DuckDuckGoSearchTool with the reframed query.
+7. Observation: Record new result.
 
-**Final Output Rules:**
-- Always start with FINAL ANSWER: [your answer].
-- Numbers: plain (no commas unless list).
-- Strings: no articles unless inside proper names.
-- Lists: comma-separated without extra punctuation.
+Then:
+8. Thought: Reflect on all observations.
+9. FINAL ANSWER: Provide the answer.
 
-**Scratchpad Example:**
+Formatting Rules:
+- Begin with FINAL ANSWER: [your answer]
+- Numbers: plain (no commas unless list)
+- Strings: no articles unless inside proper names
+- Lists: comma-separated without extra punctuation
 
-Thought: Find fruits in Embroidery painting.
+Example scratchpad flow:
+Thought: Need fruits from painting.
 Action: DuckDuckGoSearchTool('fruits in Embroidery from Uzbekistan painting')
-Observation: (empty result)
-Thought: Result was unclear. I should reframe and retry.
-Action: summarize_query('fruits in painting Embroidery from Uzbekistan')
-Observation: fruits: pomegranate, apple, grape
-Thought: Now get the breakfast fruits.
+Observation: (empty)
+Thought: Unclear result, retry.
+Action: summarize_query('fruits in Embroidery painting Uzbekistan')
+Observation: pomegranate, apple, grape
+Thought: Find breakfast fruits.
 Action: DuckDuckGoSearchTool('breakfast menu October 1949 SS Ile de France')
 Observation: grapes, apples, oranges
-Thought: Matching fruits are grapes and apples.
+Thought: Overlap is grapes and apples.
 FINAL ANSWER: grapes, apples
-
-Be methodical, careful, and retry only once if needed.
 """
 
-# --- BUILD THE SMART AGENT ---
+# --- Build the Smart Agent ---
 
 smart_agent = CodeAgent(
     tools=[search_tool, summarize_query],
     model=HfApiModel(system_prompt=system_prompt)
 )
+
+# --- Integrate into Gradio App ---
+
+class BasicAgent:
+    def __init__(self):
+        print("SmolAgent with ReACT, Scratchpad & Retry initialized.")
+
+    def __call__(self, question: str) -> str:
+        print(f"Agent received question (first 50 chars): {question[:50]}...")
+        answer = smart_agent.run(question)
+        print(f"Agent returning answer: {answer}")
+        return answer
+
+def run_and_submit_all(profile: gr.OAuthProfile | None):
+    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 log in to Hugging Face using the button above.", None
+
+    api_url = DEFAULT_API_URL
+    questions_url = f"{api_url}/questions"
+    submit_url = f"{api_url}/submit"
+
+    try:
+        agent = BasicAgent()
+    except Exception as e:
+        return f"Error initializing agent: {e}", None
+
+    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
+    print(f"Agent code URL: {agent_code}")
+
+    # Fetch questions
+    try:
+        response = requests.get(questions_url, timeout=15)
+        response.raise_for_status()
+        questions_data = response.json()
+        if not questions_data:
+            return "Fetched questions list is empty or invalid.", None
+    except Exception as e:
+        return f"Error fetching questions: {e}", None
+
+    # Run agent on each question
+    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_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:
+            results_log.append({
+                "Task ID": task_id,
+                "Question": question_text,
+                "Submitted Answer": f"AGENT ERROR: {e}"
+            })
+
+    if not answers_payload:
+        return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+
+    # Submit answers
+    submission_data = {
+        "username": username,
+        "agent_code": agent_code,
+        "answers": answers_payload
+    }
+    try:
+        response = requests.post(submit_url, json=submission_data, timeout=60)
+        response.raise_for_status()
+        result_data = response.json()
+        final_status = (
+            f"Submission Successful!\n"
+            f"User: {result_data.get('username')}\n"
+            f"Overall Score: {result_data.get('score', 'N/A')}% "
+            f"({result_data.get('correct_count', '?')}/"
+            f"{result_data.get('total_attempted', '?')} correct)\n"
+            f"Message: {result_data.get('message', '')}"
+        )
+        results_df = pd.DataFrame(results_log)
+        return final_status, results_df
+    except Exception as e:
+        results_df = pd.DataFrame(results_log)
+        return f"Submission Failed: {e}", results_df
+
+# --- Gradio Interface ---
+
+with gr.Blocks() as demo:
+    gr.Markdown("# SmolAgent GAIA Evaluation Runner 🚀")
+    gr.Markdown(
+        """
+        **Instructions:**
+        1. Clone this space and modify if needed.
+        2. Log in to Hugging Face.
+        3. Click 'Run Evaluation & Submit All Answers'.
+        **Note:** Evaluation can take a few minutes.
+        """
+    )
+    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]
+    )
+
+if __name__ == "__main__":
+    print("\n" + "-"*30 + " App Starting " + "-"*30)
+    space_host = os.getenv("SPACE_HOST")
+    space_id = os.getenv("SPACE_ID")
+    if space_host:
+        print(f"SPACE_HOST: {space_host}")
+    if space_id:
+        print(f"SPACE_ID: {space_id}")
+    print("Launching Gradio Interface...")
+    demo.launch(debug=True, share=False)