Update app.py

app.py

@@ -1,196 +1,476 @@
 import os
-import gradio as gr
+import sys
+import json
+import base64
+import tempfile
 import requests
-import inspect
 import pandas as pd
+import gradio as gr
+import anthropic
+from io import StringIO
+from pathlib import Path
 
-# (Keep Constants as is)
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# --- Basic Agent Definition ---
-# ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
-class BasicAgent:
+# ============================================================
+# Tool Implementations
+# ============================================================
+
+def web_search(query: str) -> str:
+    """Search the web using DuckDuckGo (no API key needed)."""
+    try:
+        from duckduckgo_search import DDGS
+        with DDGS() as ddgs:
+            results = list(ddgs.text(query, max_results=6))
+        if not results:
+            return "No results found."
+        return "\n\n".join(
+            f"Title: {r['title']}\nURL: {r['href']}\nSnippet: {r['body']}"
+            for r in results
+        )
+    except Exception as e:
+        return f"Search error: {e}"
+
+
+def visit_webpage(url: str) -> str:
+    """Fetch and return the text content of a webpage."""
+    try:
+        headers = {"User-Agent": "Mozilla/5.0 (compatible; GAIABot/1.0)"}
+        resp = requests.get(url, headers=headers, timeout=15)
+        resp.raise_for_status()
+        try:
+            from bs4 import BeautifulSoup
+            soup = BeautifulSoup(resp.text, "html.parser")
+            for tag in soup(["script", "style", "nav", "footer", "header"]):
+                tag.decompose()
+            text = soup.get_text(separator=" ", strip=True)
+        except ImportError:
+            from html.parser import HTMLParser
+            class _Strip(HTMLParser):
+                def __init__(self):
+                    super().__init__()
+                    self._parts, self._skip = [], False
+                def handle_starttag(self, t, _):
+                    if t in ("script", "style"):
+                        self._skip = True
+                def handle_endtag(self, t):
+                    if t in ("script", "style"):
+                        self._skip = False
+                def handle_data(self, d):
+                    if not self._skip:
+                        self._parts.append(d)
+            p = _Strip()
+            p.feed(resp.text)
+            text = " ".join(p._parts)
+        import re
+        text = re.sub(r"\s+", " ", text).strip()
+        return text[:8000]
+    except Exception as e:
+        return f"Failed to fetch {url}: {e}"
+
+
+def run_python(code: str) -> str:
+    """Execute Python code in a sandboxed namespace and return stdout."""
+    buf_out, buf_err = StringIO(), StringIO()
+    old_out, old_err = sys.stdout, sys.stderr
+    sys.stdout, sys.stderr = buf_out, buf_err
+    try:
+        namespace = {"pd": pd, "__builtins__": __builtins__}
+        exec(code, namespace)
+        out = buf_out.getvalue()
+        err = buf_err.getvalue()
+        if err:
+            out += f"\n[stderr]: {err}"
+        return out.strip() or "(executed — no output)"
+    except Exception as exc:
+        return f"{type(exc).__name__}: {exc}"
+    finally:
+        sys.stdout, sys.stderr = old_out, old_err
+
+
+def read_file_as_text(file_bytes: bytes, file_name: str) -> str:
+    """Convert various file types to a text representation."""
+    ext = Path(file_name).suffix.lower()
+    try:
+        if ext in (".txt", ".py", ".md", ".json", ".xml", ".html", ".css", ".js"):
+            return file_bytes.decode("utf-8", errors="replace")[:6000]
+        elif ext == ".csv":
+            df = pd.read_csv(StringIO(file_bytes.decode("utf-8", errors="replace")))
+            return df.to_string(max_rows=50)
+        elif ext in (".xlsx", ".xls"):
+            import io
+            df = pd.read_excel(io.BytesIO(file_bytes), sheet_name=None)
+            parts = []
+            for sheet, frame in df.items():
+                parts.append(f"=== Sheet: {sheet} ===\n{frame.to_string(max_rows=50)}")
+            return "\n\n".join(parts)[:6000]
+        elif ext == ".pdf":
+            import io
+            try:
+                import pypdf
+                reader = pypdf.PdfReader(io.BytesIO(file_bytes))
+                return "\n".join(p.extract_text() for p in reader.pages)[:6000]
+            except ImportError:
+                return "[PDF reading requires pypdf — install with: pip install pypdf]"
+        elif ext in (".mp3", ".wav", ".m4a", ".flac"):
+            return f"[Audio file: {file_name}, {len(file_bytes):,} bytes — transcription not available without Whisper API]"
+        else:
+            # Try decoding as UTF-8 as a last resort
+            try:
+                return file_bytes.decode("utf-8", errors="replace")[:4000]
+            except Exception:
+                return f"[Binary file: {file_name}, {len(file_bytes):,} bytes]"
+    except Exception as e:
+        return f"Error reading file {file_name}: {e}"
+
+
+# ============================================================
+# Tool Schema (for Anthropic tool_use)
+# ============================================================
+
+TOOLS = [
+    {
+        "name": "web_search",
+        "description": (
+            "Search the web for current information, facts, Wikipedia content, "
+            "news, etc. Returns titles, URLs, and snippets."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "query": {"type": "string", "description": "The search query"}
+            },
+            "required": ["query"],
+        },
+    },
+    {
+        "name": "visit_webpage",
+        "description": (
+            "Fetch the full text of a specific webpage. Use when you need more "
+            "detail than a search snippet, e.g. to read a Wikipedia article."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "url": {"type": "string", "description": "Full URL to fetch"}
+            },
+            "required": ["url"],
+        },
+    },
+    {
+        "name": "run_python",
+        "description": (
+            "Execute Python code. Great for arithmetic, counting, sorting, "
+            "string manipulation, or processing data. Use print() for output. "
+            "pandas (as pd) is pre-imported."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "code": {
+                    "type": "string",
+                    "description": "Python code to run. Always use print() to show results.",
+                }
+            },
+            "required": ["code"],
+        },
+    },
+]
+
+SYSTEM_PROMPT = """You are an expert research assistant solving GAIA benchmark questions.
+These are real-world questions requiring careful research and precise answers.
+
+Strategy:
+- Use web_search to find facts; follow up with visit_webpage for detail
+- Use run_python for any calculation, counting, sorting, or data manipulation
+- For files provided in the question, analyse them carefully
+- Cross-check facts when accuracy is critical
+
+Answer format (VERY IMPORTANT):
+- Provide ONLY the final answer — no preamble, no explanation
+- Give exactly what is asked: a number, a name, a date, a word, a short phrase
+- Numbers: digits only, unless units are part of the question's expected format
+- Lists: comma-separated values unless another format is specified
+- Yes/No questions: just "Yes" or "No"
+
+Think step by step, then output your final concise answer."""
+
+
+# ============================================================
+# Agent
+# ============================================================
+
+class GAIAAgent:
+    """Agentic loop backed by Claude with tool use."""
+
+    MAX_ITERATIONS = 15
+
     def __init__(self):
-        print("BasicAgent initialized.")
+        api_key = os.getenv("ANTHROPIC_API_KEY")
+        if not api_key:
+            raise EnvironmentError("ANTHROPIC_API_KEY environment variable not set.")
+        self.client = anthropic.Anthropic(api_key=api_key)
+        self.model = "claude-sonnet-4-20250514"
+        print(f"GAIAAgent initialised (model: {self.model})")
+
+    # ---- internal helpers ----
+
+    def _dispatch_tool(self, name: str, inputs: dict) -> str:
+        if name == "web_search":
+            return web_search(inputs["query"])
+        if name == "visit_webpage":
+            return visit_webpage(inputs["url"])
+        if name == "run_python":
+            return run_python(inputs["code"])
+        return f"[unknown tool: {name}]"
+
+    def _build_initial_content(
+        self, question: str, file_bytes: bytes | None, file_name: str | None
+    ) -> list:
+        """Return the content list for the first user message."""
+        content = []
+
+        if file_bytes and file_name:
+            ext = Path(file_name).suffix.lower()
+            image_exts = {".jpg", ".jpeg", ".png", ".gif", ".webp"}
+            if ext in image_exts:
+                media_map = {
+                    ".jpg": "image/jpeg", ".jpeg": "image/jpeg",
+                    ".png": "image/png", ".gif": "image/gif",
+                    ".webp": "image/webp",
+                }
+                content.append({
+                    "type": "image",
+                    "source": {
+                        "type": "base64",
+                        "media_type": media_map[ext],
+                        "data": base64.b64encode(file_bytes).decode(),
+                    },
+                })
+                content.append({
+                    "type": "text",
+                    "text": f"The image above is the attached file '{file_name}'.\n\n{question}",
+                })
+            else:
+                file_text = read_file_as_text(file_bytes, file_name)
+                content.append({
+                    "type": "text",
+                    "text": (
+                        f"A file named '{file_name}' is attached. Its contents:\n\n"
+                        f"{file_text}\n\n---\n\nQuestion: {question}"
+                    ),
+                })
+        else:
+            content.append({"type": "text", "text": question})
+
+        return content
+
+    # ---- public interface ----
+
+    def solve(
+        self,
+        question: str,
+        file_bytes: bytes | None = None,
+        file_name: str | None = None,
+    ) -> str:
+        print(f"\n[Agent] Question: {question[:120]}{'...' if len(question)>120 else ''}")
+        messages = [
+            {"role": "user", "content": self._build_initial_content(question, file_bytes, file_name)}
+        ]
+
+        for iteration in range(self.MAX_ITERATIONS):
+            response = self.client.messages.create(
+                model=self.model,
+                max_tokens=4096,
+                system=SYSTEM_PROMPT,
+                tools=TOOLS,
+                messages=messages,
+            )
+
+            if response.stop_reason == "end_turn":
+                for block in response.content:
+                    if hasattr(block, "text"):
+                        answer = block.text.strip()
+                        print(f"[Agent] Answer: {answer[:100]}")
+                        return answer
+                return "No answer generated."
+
+            if response.stop_reason == "tool_use":
+                tool_results = []
+                for block in response.content:
+                    if block.type == "tool_use":
+                        print(f"  [Tool] {block.name}({json.dumps(block.input)[:80]})")
+                        result = self._dispatch_tool(block.name, block.input)
+                        print(f"  [Tool] → {result[:120]}")
+                        tool_results.append({
+                            "type": "tool_result",
+                            "tool_use_id": block.id,
+                            "content": result,
+                        })
+                messages.append({"role": "assistant", "content": response.content})
+                messages.append({"role": "user", "content": tool_results})
+            else:
+                # Unexpected stop reason
+                print(f"[Agent] Unexpected stop_reason: {response.stop_reason}")
+                break
+
+        return "Could not determine answer within iteration limit."
+
     def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
-        fixed_answer = "This is a default answer."
-        print(f"Agent returning fixed answer: {fixed_answer}")
-        return fixed_answer
-
-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
+        """Compatibility shim for the template's agent(question) calls."""
+        return self.solve(question)
+
+
+# ============================================================
+# Evaluation runner
+# ============================================================
+
+def run_and_submit_all(profile: gr.OAuthProfile | None):
+    """Fetch questions, run the agent, submit answers, display results."""
+
+    space_id = os.getenv("SPACE_ID")
 
     if profile:
-        username= f"{profile.username}"
-        print(f"User logged in: {username}")
+        username = profile.username
+        print(f"Logged in as: {username}")
     else:
-        print("User not logged in.")
-        return "Please Login to Hugging Face with the button.", None
+        return "Please log in to Hugging Face first.", None
 
     api_url = DEFAULT_API_URL
     questions_url = f"{api_url}/questions"
     submit_url = f"{api_url}/submit"
 
-    # 1. Instantiate Agent ( modify this part to create your agent)
+    # 1. Build agent
     try:
-        agent = BasicAgent()
+        agent = GAIAAgent()
     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"
+
+    # 2. 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 "Questions 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
 
-    # 3. Run your Agent
+    # 3. Run agent on each question
     results_log = []
     answers_payload = []
-    print(f"Running agent on {len(questions_data)} questions...")
+
     for item in questions_data:
         task_id = item.get("task_id")
         question_text = item.get("question")
+        file_name = item.get("file_name", "")
+
         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
+
+        # Download attached file if present
+        file_bytes = None
+        if file_name:
+            try:
+                file_url = f"{api_url}/files/{task_id}"
+                file_resp = requests.get(file_url, timeout=30)
+                file_resp.raise_for_status()
+                file_bytes = file_resp.content
+                print(f"  Downloaded '{file_name}' ({len(file_bytes):,} bytes)")
+            except Exception as e:
+                print(f"  Could not download file for task {task_id}: {e}")
+
         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})
+            submitted_answer = agent.solve(question_text, file_bytes, file_name)
         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}"})
+            submitted_answer = f"AGENT ERROR: {e}"
+            print(f"  Agent error on {task_id}: {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)
+        answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
+        results_log.append({
+            "Task ID": task_id,
+            "Question": question_text[:120],
+            "File": file_name or "—",
+            "Submitted Answer": submitted_answer,
+        })
 
-    # 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)
+    if not answers_payload:
+        return "Agent produced no answers.", pd.DataFrame(results_log)
 
-    # 5. Submit
-    print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
+    # 4. Submit
+    submission = {
+        "username": username.strip(),
+        "agent_code": agent_code,
+        "answers": answers_payload,
+    }
+    print(f"Submitting {len(answers_payload)} answers …")
     try:
-        response = requests.post(submit_url, json=submission_data, timeout=60)
-        response.raise_for_status()
-        result_data = response.json()
-        final_status = (
+        resp = requests.post(submit_url, json=submission, timeout=120)
+        resp.raise_for_status()
+        result = resp.json()
+        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"User: {result.get('username')}\n"
+            f"Score: {result.get('score', 'N/A')}% "
+            f"({result.get('correct_count', '?')}/{result.get('total_attempted', '?')} correct)\n"
+            f"Message: {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}."
+        detail = ""
         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.json().get("detail", e.response.text)
+        except Exception:
+            detail = e.response.text[:500]
+        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
+        status = f"Submission error: {e}"
+
+    print(status)
+    return 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.
+        **Setup:**
+        1. Set `ANTHROPIC_API_KEY` as a Space secret.
+        2. Log in with your Hugging Face account below.
+        3. Click **Run Evaluation** to fetch questions, run the agent, and submit.
 
-        ---
-        **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.
+        The agent uses Claude with web search, code execution, and file analysis.
         """
     )
 
     gr.LoginButton()
 
-    run_button = gr.Button("Run Evaluation & Submit All Answers")
+    run_btn = gr.Button("Run Evaluation & Submit All Answers", variant="primary")
+    status_box = gr.Textbox(label="Status / Result", lines=6, interactive=False)
+    results_table = gr.DataFrame(label="Questions & Answers", wrap=True)
 
-    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_button.click(
-        fn=run_and_submit_all,
-        outputs=[status_output, results_table]
-    )
+    run_btn.click(fn=run_and_submit_all, outputs=[status_box, 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")
+    print("\n" + "=" * 60)
+    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}")
+    if not os.getenv("ANTHROPIC_API_KEY"):
+        print("⚠️  ANTHROPIC_API_KEY is NOT set — agent will fail.")
     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("✅ ANTHROPIC_API_KEY found.")
+    print("=" * 60 + "\n")
     demo.launch(debug=True, share=False)