Update app.py

app.py

@@ -7,12 +7,17 @@ import re
 import math
 import json
 import unicodedata
+from typing import TypedDict, Annotated
+ 
+from langchain_core.messages import AnyMessage, HumanMessage, SystemMessage
 from langchain_openai import ChatOpenAI
-from langgraph.prebuilt import create_react_agent
-from langchain_core.messages import SystemMessage
 from langchain_core.tools import tool
 from langchain_community.tools import DuckDuckGoSearchRun
 from langchain_community.utilities import WikipediaAPIWrapper
+ 
+from langgraph.graph import START, StateGraph
+from langgraph.graph.message import add_messages
+from langgraph.prebuilt import ToolNode, tools_condition
 
 # (Keep Constants as is)
 # --- Constants ---
@@ -22,8 +27,7 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 def web_search(query: str) -> str:
     """Search the web using DuckDuckGo. Use for current events, facts, and general knowledge."""
     try:
-        search = DuckDuckGoSearchRun()
-        return search.run(query)
+        return DuckDuckGoSearchRun().run(query)
     except Exception as e:
         return f"Search error: {e}"
  
@@ -42,17 +46,15 @@ def wikipedia_search(query: str) -> str:
 def python_repl(code: str) -> str:
     """
     Execute Python code for math calculations, data processing, logic.
-    Always print() the final result.
+    Always use print() to show the result.
     Example: print(2 + 2)
     """
     import io, sys
     old_stdout = sys.stdout
     sys.stdout = io.StringIO()
     try:
-        exec(code, {
-            "math": math, "json": json, "re": re,
-            "unicodedata": unicodedata, "__builtins__": __builtins__
-        })
+        exec(code, {"math": math, "json": json, "re": re,
+                    "unicodedata": unicodedata, "__builtins__": __builtins__})
         output = sys.stdout.getvalue()
         return output.strip() if output.strip() else "Code executed with no output. Use print()."
     except Exception as e:
@@ -68,8 +70,7 @@ def calculator(expression: str) -> str:
     Examples: '2 + 2', '100 * 1.07 ** 5', 'math.sqrt(144)'
     """
     try:
-        result = eval(expression, {"math": math, "__builtins__": {}})
-        return str(result)
+        return str(eval(expression, {"math": math, "__builtins__": {}}))
     except Exception as e:
         return f"Calculation error: {e}"
  
@@ -77,27 +78,26 @@ def calculator(expression: str) -> str:
 @tool
 def get_task_file(task_id: str) -> str:
     """
-    Fetch the file associated with a GAIA task by its task_id.
+    Fetch the file attached to a GAIA task by its task_id.
     Use this when the question mentions an attached file or document.
     """
     try:
+        import requests as req
         url = f"https://agents-course-unit4-scoring.hf.space/files/{task_id}"
-        response = requests.get(url, timeout=15)
+        response = req.get(url, timeout=15)
         if response.status_code == 200:
-            content_type = response.headers.get("Content-Type", "")
-            if "text" in content_type or "json" in content_type:
+            ct = response.headers.get("Content-Type", "")
+            if "text" in ct or "json" in ct:
                 return response.text[:5000]
-            elif "image" in content_type:
-                return f"[Image file - content-type: {content_type}]"
-            elif "audio" in content_type:
-                return f"[Audio file - content-type: {content_type}]"
-            else:
-                return f"[File attached: {content_type}]"
+            return f"[Binary file attached - content-type: {ct}]"
         return f"No file found for task {task_id}"
     except Exception as e:
         return f"Error fetching task file: {e}"
 
 
+class AgentState(TypedDict):
+    messages: Annotated[list[AnyMessage], add_messages]
+
 SYSTEM_PROMPT = """You are a precise expert AI solving GAIA benchmark questions.
  
 ## Answer Format (CRITICAL)
@@ -122,16 +122,8 @@ FINAL ANSWER: <your answer here>
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 class BasicAgent:
     def __init__(self):
-        print("Initializing LangGraph ReAct Agent with Llama 3.3 70B...")
+        print("Initializing LangGraph Agent (HF course pattern)...")
  
-        llm = ChatOpenAI(
-            model="Qwen/Qwen2.5-72B-Instruct",   # supporta tool calling
-            base_url="https://router.huggingface.co/v1",
-            api_key=os.getenv("HF_TOKEN"),
-            temperature=0.1,
-            max_tokens=1024,
-        )
-        
         tools = [
             web_search,
             wikipedia_search,
@@ -140,20 +132,44 @@ class BasicAgent:
             get_task_file,
         ]
  
-        self.agent = create_react_agent(
-            model=llm,
-            tools=tools,
-            prompt=SystemMessage(content=SYSTEM_PROMPT),
+        # Pattern esatto dal notebook del corso:
+        # llm.bind_tools() + StateGraph costruito manualmente
+        llm = ChatOpenAI(
+            model="Qwen/Qwen2.5-72B-Instruct",
+            base_url="https://router.huggingface.co/v1",
+            api_key=os.getenv("HF_TOKEN"),
+            temperature=0.1,
+            max_tokens=1024,
         )
+        self.llm_with_tools = llm.bind_tools(tools, parallel_tool_calls=False)
+ 
+        # Grafo ReAct: assistant → tools → assistant (loop)
+        builder = StateGraph(AgentState)
+        builder.add_node("assistant", self._assistant_node)
+        builder.add_node("tools", ToolNode(tools))
+        builder.add_edge(START, "assistant")
+        builder.add_conditional_edges("assistant", tools_condition)
+        builder.add_edge("tools", "assistant")
+        self.graph = builder.compile()
+ 
         print("Agent ready.")
  
+    def _assistant_node(self, state: AgentState):
+        """Chiama il LLM con system prompt + history dei messaggi."""
+        sys_msg = SystemMessage(content=SYSTEM_PROMPT)
+        response = self.llm_with_tools.invoke([sys_msg] + state["messages"])
+        return {"messages": [response]}
+ 
     def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question[:50]}...")
         try:
-            result = self.agent.invoke({"messages": [("user", question)]})
+            result = self.graph.invoke({
+                "messages": [HumanMessage(content=question)]
+            })
             last_message = result["messages"][-1].content
+            print(f"Agent raw output: {last_message[:200]}...")
  
-            # Estrai FINAL ANSWER se presente
+            # Estrai FINAL ANSWER se presente, altrimenti ultima riga
             match = re.search(r"FINAL ANSWER:\s*(.+?)(?:\n|$)", last_message, re.IGNORECASE)
             answer = match.group(1).strip() if match else last_message.strip().split("\n")[-1]