Fix issues based on testing

agent.py

@@ -8,11 +8,14 @@ This module defines a LangGraph agent that can:
 """
 
 import os
+import re
 import bm25s
 import requests
 from pathlib import Path
 from dotenv import load_dotenv
 
+from pydantic import BaseModel, Field
+
 from langgraph.graph import START, END, StateGraph
 from langgraph.prebuilt import tools_condition, ToolNode
 
@@ -76,6 +79,21 @@ _system_prompt = load_prompt("prompts/prompt.yaml")
 _thinking_enabled = config["models"]["llm"]["parameters"].get("thinking_enabled", True)
 
 
+class FinalAnswer(BaseModel):
+    """Strict GAIA-format answer extracted from the solver's reasoning."""
+    answer: str = Field(
+        description=(
+            "The raw answer value only. "
+            "Numbers: plain digits, no commas, no units, no symbols (write '1000000', not '1,000,000' or '$50'). "
+            "Strings: no articles ('a', 'an', 'the'), no markdown, no surrounding quotes, no trailing punctuation. "
+            "Lists: comma-separated with no extra spaces, in the order requested by the question."
+        )
+    )
+
+
+formatter_llm = agent_llm.with_structured_output(FinalAnswer)
+
+
 # ============================================
 # Graph Nodes
 # ============================================
@@ -265,10 +283,47 @@ def processor_node(state: AgentState) -> AgentState:
     full_messages.extend(messages)
     
     response = agent_with_tools.invoke(full_messages)
-    
+
     return {"messages": [response]}
 
 
+def formatter_node(state: AgentState) -> AgentState:
+    """Extract and reformat the solver's answer into a strict GAIA-compliant value."""
+    print("--- FORMATTER NODE ---")
+    messages = state.get("messages", [])
+    if not messages:
+        return {"final_answer": ""}
+
+    question = ""
+    for m in messages:
+        if isinstance(m, HumanMessage):
+            question = m.content
+            break
+
+    solver_output = messages[-1].content or ""
+
+    prompt = [
+        SystemMessage(content=(
+            "You extract the final answer from an agent's reasoning. "
+            "Apply the GAIA formatting rules exactly. "
+            "If the agent never produced an answer, return an empty string."
+        )),
+        HumanMessage(content=(
+            f"Question:\n{question}\n\n"
+            f"Agent reasoning and conclusion:\n{solver_output}\n\n"
+            "Extract the final answer."
+        )),
+    ]
+
+    try:
+        result = formatter_llm.invoke(prompt)
+        return {"final_answer": result.answer.strip()}
+    except Exception as e:
+        print(f"Formatter error: {e}")
+        match = re.search(r'FINAL ANSWER:\s*(.*)', solver_output, re.DOTALL | re.IGNORECASE)
+        return {"final_answer": (match.group(1).strip() if match else solver_output.strip())}
+
+
 # ============================================
 # Graph Construction
 # ============================================
@@ -285,14 +340,20 @@ def agent_graph():
     workflow.add_node("reranker_node", reranker_node)
     workflow.add_node("processor_node", processor_node)
     workflow.add_node("tools", ToolNode(tools_list))
+    workflow.add_node("formatter_node", formatter_node)
 
     # Add edges
     workflow.add_edge(START, "file_downloader_node")
     workflow.add_edge("file_downloader_node", "retriever_node")
     workflow.add_edge("retriever_node", "reranker_node")
     workflow.add_edge("reranker_node", "processor_node")
-    workflow.add_edge("tools", "processor_node")    
-    workflow.add_conditional_edges("processor_node", tools_condition, {"tools": "tools", END: END})
+    workflow.add_edge("tools", "processor_node")
+    workflow.add_conditional_edges(
+        "processor_node",
+        tools_condition,
+        {"tools": "tools", END: "formatter_node"},
+    )
+    workflow.add_edge("formatter_node", END)
     
     
     compiled = workflow.compile()

app.py

@@ -24,9 +24,11 @@ class BasicAgent:
             "task_id": task_id,
             "file_name": file_name
         })
-        content = response['messages'][-1].content
-        match = re.search(r'FINAL ANSWER:\s*(.*)', content, re.DOTALL)
-        fixed_answer = match.group(1).strip() if match else content.strip()
+        fixed_answer = (response.get("final_answer") or "").strip()
+        if not fixed_answer:
+            content = response['messages'][-1].content
+            match = re.search(r'FINAL ANSWER:\s*(.*)', content, re.DOTALL | re.IGNORECASE)
+            fixed_answer = match.group(1).strip() if match else content.strip()
         print(f"Agent returning fixed answer: {fixed_answer}")
         return fixed_answer
 

prompts/prompt.yaml

@@ -24,30 +24,21 @@ prompt: |
   4. **Observe**: Analyze the tool output. Does it answer the question?
   5. **Refine**: If the output is insufficient, adjust the plan and try a different angle.
 
-  # CRITICAL OUTPUT RULES
-  The automated scoring system is EXTREMELY STRICT. You must follow these formatting rules exactly:
-  - **Numeric Answers**:
-    - Output ONLY the number.
-    - NO commas (e.g., write `1000000`, NOT `1,000,000`).
-    - NO units or symbols (e.g., write `50`, NOT `$50` or `50%`, unless explicitly asked for the unit string).
-  - **String Answers**:
-    - Be concise.
-    - NO articles (a, an, the).
-    - NO abbreviations usually, unless standard (e.g., 'USA' might be okay, but 'Sept' for September is risky).
-  - **Final Format**:
-    - Your final line MUST be exactly: `FINAL ANSWER: <your_answer>`
-    - Do not put proper sentences in the final answer, just the raw value.
+  # FINAL ANSWER
+  When you have reached the answer, end your message with a single concluding line of the form:
+
+      FINAL ANSWER: <value>
+
+  A downstream formatter will normalise the value into the strict GAIA scoring format (no commas in numbers, no articles in strings, comma-separated lists, etc.) — so prioritise correctness of the value itself over format pedantry. Do not wrap the line in markdown, do not add commentary after it, and do not restate it.
 
   # EXAMPLE SCENARIOS
   - Question: "What is the sum of the 'Total' column in the attached file.xlsx?"
-    Thought: I need to read the excel file first using `read_excel`. Then I will sum the values using `calculator`.
+    Thought: I need to read the excel file first using `read_excel`, then check the column stats it returns.
     Action: `read_excel(file_path="...")`
-    Observation: (Dataframe output...)
-    Thought: I have the numbers. Sum is 500 + 200...
-    Action: `calculator(a=500, b=200, type="addition")`
+    Observation: (Dataframe + describe() stats…)
     FINAL ANSWER: 700
 
   - Question: "Which city is the capital of France?"
     FINAL ANSWER: Paris
 
-  Failure to follow these rules will result in a score of 0. Take a deep breath and think step by step.
+  Take a deep breath and think step by step.

states.py

@@ -17,9 +17,11 @@ class AgentState(TypedDict):
         file_name: Name of the attached file (empty string if no file)
         file_path: Local filesystem path to the downloaded file (empty if no file or download failed)
         retrieved_docs: List of candidate documents from the retriever node
+        final_answer: GAIA-formatted answer produced by the formatter node
     """
     messages: Annotated[list[BaseMessage], add_messages]
     task_id: str
     file_name: str
     file_path: str
-    retrieved_docs: List[Document]
+    retrieved_docs: List[Document]
+    final_answer: str

tools.py

@@ -72,7 +72,7 @@ def duck_web_search(query: str) -> str:
     Args:
         query: The search query.
     """
-    search = _get_ddg().invoke(query=query)
+    search = _get_ddg().invoke(input=query)
     
     return {"duckduckgo_web_search": search}