Update app.py

app.py

@@ -1,244 +1,281 @@
 import os
-import gradio as gr
+import re
 import requests
-import inspect
+import traceback
+import gradio as gr
 import pandas as pd
+import google.generativeai as genai
+from google.generativeai.types import Tool, GoogleSearch
 
-# (Keep Constants as is)
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-import os
-import google.generativeai as genai
 
-class BasicAgent:
-    def __init__(self):
-        print("Gemini Agent initialized")
+# ─────────────────────────────────────────────
+#  Helper: strip markdown / fences from output
+# ─────────────────────────────────────────────
+def clean_answer(text: str) -> str:
+    """
+    Normalise the model's raw output into a clean, exact-match-ready string.
+    """
+    text = text.strip()
 
-        genai.configure(api_key=os.getenv("GEMINI_API_KEY"))
+    # Remove code fences if the whole reply is wrapped in one
+    text = re.sub(r"^```[a-zA-Z]*\n?", "", text)
+    text = re.sub(r"```$", "", text)
 
-        # ✅ latest + stable + cheap
-        self.model = genai.GenerativeModel("gemini-2.5-flash")
+    # Remove common label prefixes the model likes to add
+    for prefix in [
+        "Final Answer:", "Answer:", "FINAL ANSWER:", "ANSWER:",
+        "The answer is:", "The final answer is:",
+    ]:
+        if text.lower().startswith(prefix.lower()):
+            text = text[len(prefix):].strip()
 
-    def __call__(self, question: str) -> str:
-        try:
-            base_prompt = f"""
-You are solving benchmark reasoning questions.
+    # Collapse internal newlines to spaces, then trim
+    text = " ".join(text.split())
 
-RULES:
-- Think carefully step by step
-- Extract exact final answer
-- DO NOT explain
-- RETURN ONLY final answer
+    # Do NOT split on "." — it breaks decimals, abbreviations, etc.
+    return text[:200]
+
+
+# ─────────────────────────────────────────────
+#  Agent
+# ─────────────────────────────────────────────
+class BasicAgent:
+    """
+    A Gemini-powered agent that uses:
+      • Google Search grounding  — for real-time factual questions
+      • Code execution           — for maths / data problems
+      • Self-verification pass   — to catch obvious hallucinations
+
+    Falls back gracefully when tools aren't available.
+    """
+
+    SYSTEM_PROMPT = """You are an expert AI assistant solving GAIA benchmark questions.
 
-Question:
-{question}
+RULES:
+1. Think step-by-step before answering.
+2. Use Google Search when you need current facts, dates, or specific data.
+3. Use code execution for any arithmetic, unit conversions, or data processing.
+4. Return ONLY the final answer — no explanation, no preamble, no punctuation tail.
+5. For numbers: no commas, correct decimal places (e.g. 42 not 42.0 if whole number).
+6. For lists: comma-separated values in alphabetical order unless asked otherwise.
+7. For yes/no questions: answer exactly "yes" or "no" (lowercase).
+8. Keep the answer as short as possible while being complete and correct.
 """
 
-        # 🔥 Step 1: First attempt
-            response1 = self.model.generate_content(base_prompt)
-            ans1 = response1.text.strip()
+    VERIFY_PROMPT = """Question: {question}
 
-        # 🔥 Step 2: Self-check (THIS IS THE MAGIC)
-            verify_prompt = f"""
-Question: {question}
+Proposed answer: {answer}
 
-Proposed Answer: {ans1}
+Review the answer carefully:
+- Is it factually correct?
+- Is it in the exact format requested?
+- Is it as concise as possible?
 
-Is this answer correct?
-If wrong, give corrected FINAL answer only.
-If correct, repeat the answer.
+If correct, repeat it unchanged.
+If wrong or badly formatted, return ONLY the corrected answer.
 
-Return ONLY answer.
-"""
+Return ONLY the final answer — nothing else."""
 
-            response2 = self.model.generate_content(verify_prompt)
-            final = response2.text.strip()
+    def __init__(self):
+        api_key = os.getenv("GEMINI_API_KEY")
+        if not api_key:
+            raise EnvironmentError("GEMINI_API_KEY environment variable is not set.")
 
-        # 🔥 Clean output (VERY IMPORTANT for exact match)
-            final = final.replace("Final Answer:", "").strip()
-            final = final.split("\n")[0].strip()
-            final = final.split(".")[0].strip()
+        genai.configure(api_key=api_key)
 
-            return final[:100]
+        # Tools available in gemini-2.5-flash
+        self._search_tool = Tool(google_search=GoogleSearch())
 
-        except Exception as e:
+        # Primary model — with search grounding + code execution
+        self.model = genai.GenerativeModel(
+            model_name="gemini-2.5-flash",
+            system_instruction=self.SYSTEM_PROMPT,
+            tools=[self._search_tool, "code_execution"],
+        )
+
+        # Verification model — plain text, no tools (avoids circular loops)
+        self.verify_model = genai.GenerativeModel(
+            model_name="gemini-2.5-flash",
+            system_instruction="You are a precise answer validator. Return ONLY the final answer.",
+        )
+
+        print("✅ BasicAgent (Gemini 2.5 Flash + Search + Code) initialised.")
+
+    # ------------------------------------------------------------------
+    def __call__(self, question: str) -> str:
+        try:
+            return self._run(question)
+        except Exception as exc:
+            print(f"[AGENT ERROR] {exc}\n{traceback.format_exc()}")
             return "N/A"
 
-    
+    # ------------------------------------------------------------------
+    def _run(self, question: str) -> str:
+        # ── Pass 1: full reasoning with tools ──────────────────────────
+        response1 = self.model.generate_content(question)
+        ans1 = self._extract_text(response1)
 
-def run_and_submit_all( profile: gr.OAuthProfile | None):
-    """
-    Fetches all questions, runs the BasicAgent on them, submits all answers,
-    and displays the results.
-    """
-    # --- Determine HF Space Runtime URL and Repo URL ---
-    space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
+        if not ans1 or ans1 == "N/A":
+            return "N/A"
+
+        ans1 = clean_answer(ans1)
+        print(f"  [Pass 1] {ans1!r}")
+
+        # ── Pass 2: self-verification (plain model, no tools) ──────────
+        verify_prompt = self.VERIFY_PROMPT.format(question=question, answer=ans1)
+        response2 = self.verify_model.generate_content(verify_prompt)
+        ans2 = clean_answer(self._extract_text(response2))
+
+        print(f"  [Pass 2] {ans2!r}")
+
+        return ans2 if ans2 else ans1
+
+    # ------------------------------------------------------------------
+    @staticmethod
+    def _extract_text(response) -> str:
+        """
+        Pull plain text out of a GenerateContentResponse regardless of
+        whether it contains tool-use / code-execution blocks.
+        """
+        try:
+            # Fast path — the .text property works when there's no ambiguity
+            return response.text.strip()
+        except Exception:
+            pass
+
+        # Slow path — iterate parts
+        texts = []
+        for candidate in response.candidates:
+            for part in candidate.content.parts:
+                if hasattr(part, "text") and part.text:
+                    texts.append(part.text.strip())
+        return " ".join(texts).strip() if texts else "N/A"
+
+
+# ─────────────────────────────────────────────
+#  Gradio runner
+# ─────────────────────────────────────────────
+def run_and_submit_all(profile: gr.OAuthProfile | None):
+    space_id = os.getenv("SPACE_ID")
+
+    if not profile:
+        return "Please log in to Hugging Face first.", None
 
-    if profile:
-        username= f"{profile.username}"
-        print(f"User logged in: {username}")
-    else:
-        print("User not logged in.")
-        return "Please Login to Hugging Face with the button.", None
+    username = profile.username
+    print(f"Logged in as: {username}")
 
     api_url = DEFAULT_API_URL
     questions_url = f"{api_url}/questions"
-    submit_url = f"{api_url}/submit"
+    submit_url    = f"{api_url}/submit"
 
-    # 1. Instantiate Agent ( modify this part to create your agent)
+    # ── Instantiate agent ───────────────────────────────────────────────
     try:
         agent = BasicAgent()
     except Exception as e:
-        print(f"Error instantiating agent: {e}")
-        return f"Error initializing agent: {e}", None
-    # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
-    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
-    print(agent_code)
-
-    # 2. Fetch Questions
-    print(f"Fetching questions from: {questions_url}")
+        return f"Error initialising agent: {e}", None
+
+    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id else "unknown"
+    print(f"Agent code URL: {agent_code}")
+
+    # ── Fetch questions ─────────────────────────────────────────────────
+    print(f"Fetching questions from {questions_url} …")
     try:
-        response = requests.get(questions_url, timeout=15)
-        response.raise_for_status()
-        questions_data = response.json()
+        resp = requests.get(questions_url, timeout=15)
+        resp.raise_for_status()
+        questions_data = resp.json()
         if not questions_data:
-             print("Fetched questions list is empty.")
-             return "Fetched questions list is empty or invalid format.", None
+            return "Question list is empty.", None
         print(f"Fetched {len(questions_data)} questions.")
-    except requests.exceptions.RequestException as e:
-        print(f"Error fetching questions: {e}")
-        return f"Error fetching questions: {e}", None
-    except requests.exceptions.JSONDecodeError as e:
-         print(f"Error decoding JSON response from questions endpoint: {e}")
-         print(f"Response text: {response.text[:500]}")
-         return f"Error decoding server response for questions: {e}", None
     except Exception as e:
-        print(f"An unexpected error occurred fetching questions: {e}")
-        return f"An unexpected error occurred fetching questions: {e}", None
+        return f"Error fetching questions: {e}", None
+
+    # ── Run agent ───────────────────────────────────────────────────────
+    results_log      = []
+    answers_payload  = []
 
-    # 3. Run your Agent
-    results_log = []
-    answers_payload = []
-    print(f"Running agent on {len(questions_data)} questions...")
     for item in questions_data:
-        task_id = item.get("task_id")
+        task_id       = item.get("task_id")
         question_text = item.get("question")
         if not task_id or question_text is None:
-            print(f"Skipping item with missing task_id or question: {item}")
+            print(f"Skipping malformed item: {item}")
             continue
+
+        print(f"\n[{task_id}] Q: {question_text[:120]}")
         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})
+            answer = agent(question_text)
         except Exception as e:
-             print(f"Error running agent on task {task_id}: {e}")
-             results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
-
-    if not answers_payload:
-        print("Agent did not produce any answers to submit.")
-        return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+            answer = f"AGENT ERROR: {e}"
+            print(f"  ERROR: {e}")
 
-    # 4. Prepare Submission 
-    submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
-    status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
-    print(status_update)
+        print(f"  → {answer!r}")
+        answers_payload.append({"task_id": task_id, "submitted_answer": answer})
+        results_log.append({
+            "Task ID": task_id,
+            "Question": question_text,
+            "Submitted Answer": answer,
+        })
 
-    # 5. Submit
-    print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
+    if not answers_payload:
+        return "Agent produced no answers.", pd.DataFrame(results_log)
+
+    # ── Submit ──────────────────────────────────────────────────────────
+    submission_data = {
+        "username":   username.strip(),
+        "agent_code": agent_code,
+        "answers":    answers_payload,
+    }
+    print(f"\nSubmitting {len(answers_payload)} answers …")
     try:
-        response = requests.post(submit_url, json=submission_data, timeout=60)
-        response.raise_for_status()
-        result_data = response.json()
+        resp = requests.post(submit_url, json=submission_data, timeout=120)
+        resp.raise_for_status()
+        result = resp.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', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
-            f"Message: {result_data.get('message', 'No message received.')}"
+            f"✅ Submission successful!\n"
+            f"User:  {result.get('username')}\n"
+            f"Score: {result.get('score', 'N/A')}% "
+            f"({result.get('correct_count', '?')}/{result.get('total_attempted', '?')} correct)\n"
+            f"Msg:   {result.get('message', '')}"
         )
-        print("Submission successful.")
-        results_df = pd.DataFrame(results_log)
-        return final_status, results_df
     except requests.exceptions.HTTPError as e:
-        error_detail = f"Server responded with status {e.response.status_code}."
-        try:
-            error_json = e.response.json()
-            error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
-        except requests.exceptions.JSONDecodeError:
-            error_detail += f" Response: {e.response.text[:500]}"
-        status_message = f"Submission Failed: {error_detail}"
-        print(status_message)
-        results_df = pd.DataFrame(results_log)
-        return status_message, results_df
-    except requests.exceptions.Timeout:
-        status_message = "Submission Failed: The request timed out."
-        print(status_message)
-        results_df = pd.DataFrame(results_log)
-        return status_message, results_df
-    except requests.exceptions.RequestException as e:
-        status_message = f"Submission Failed: Network error - {e}"
-        print(status_message)
-        results_df = pd.DataFrame(results_log)
-        return status_message, results_df
+        detail = e.response.text[:500]
+        final_status = f"❌ Submission failed (HTTP {e.response.status_code}): {detail}"
     except Exception as e:
-        status_message = f"An unexpected error occurred during submission: {e}"
-        print(status_message)
-        results_df = pd.DataFrame(results_log)
-        return status_message, results_df
+        final_status = f"❌ Submission error: {e}"
 
+    print(final_status)
+    return final_status, pd.DataFrame(results_log)
 
-# --- Build Gradio Interface using Blocks ---
+
+# ─────────────────────────────────────────────
+#  Gradio UI
+# ─────────────────────────────────────────────
 with gr.Blocks() as demo:
-    gr.Markdown("# Basic Agent Evaluation Runner")
+    gr.Markdown("# GAIA Agent Evaluation Runner")
     gr.Markdown(
         """
-        **Instructions:**
-
-        1.  Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
-        2.  Log in to your Hugging Face account using the button below. This uses your HF username for submission.
-        3.  Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
+        **How to use:**
+        1. Clone this Space and set your `GEMINI_API_KEY` secret.
+        2. Log in with the button below.
+        3. Click **Run Evaluation** — the agent will answer all questions and submit automatically.
 
-        ---
-        **Disclaimers:**
-        Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
-        This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
+        *Note: runs can take several minutes while the agent processes all questions.*
         """
     )
-
     gr.LoginButton()
+    run_btn = gr.Button("▶ Run Evaluation & Submit All Answers", variant="primary")
+    status_out  = gr.Textbox(label="Status / Result", lines=6, interactive=False)
+    results_tbl = gr.DataFrame(label="Questions & Agent Answers", wrap=True)
 
-    run_button = gr.Button("Run Evaluation & Submit All Answers")
-
-    status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
-    # Removed max_rows=10 from DataFrame constructor
-    results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
+    run_btn.click(fn=run_and_submit_all, outputs=[status_out, results_tbl])
 
-    run_button.click(
-        fn=run_and_submit_all,
-        outputs=[status_output, results_table]
-    )
 
 if __name__ == "__main__":
-    print("\n" + "-"*30 + " App Starting " + "-"*30)
-    # Check for SPACE_HOST and SPACE_ID at startup for information
-    space_host_startup = os.getenv("SPACE_HOST")
-    space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
-
-    if space_host_startup:
-        print(f"✅ SPACE_HOST found: {space_host_startup}")
-        print(f"   Runtime URL should be: https://{space_host_startup}.hf.space")
-    else:
-        print("ℹ️  SPACE_HOST environment variable not found (running locally?).")
-
-    if space_id_startup: # Print repo URLs if SPACE_ID is found
-        print(f"✅ SPACE_ID found: {space_id_startup}")
-        print(f"   Repo URL: https://huggingface.co/spaces/{space_id_startup}")
-        print(f"   Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
-    else:
-        print("ℹ️  SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")
-
-    print("-"*(60 + len(" App Starting ")) + "\n")
-
-    print("Launching Gradio Interface for Basic Agent Evaluation...")
+    print("\n" + "─" * 50)
+    for var in ("SPACE_HOST", "SPACE_ID", "GEMINI_API_KEY"):
+        val = os.getenv(var)
+        print(f"{'✅' if val else '⚠️ '} {var}: {'set' if val else 'NOT SET'}")
+    print("─" * 50 + "\n")
     demo.launch(debug=True, share=False)