added tools for downloading, files, wikipedia search

app.py

@@ -1,4 +1,5 @@
 import os
+import time
 import gradio as gr
 import requests
 import inspect
@@ -21,7 +22,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     # --- Determine HF Space Runtime URL and Repo URL ---
     space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
-
+    space_id = space_id or "kylea/GAIA_Agent_Space"
     if profile:
         username= f"{profile.username}"
         print(f"User logged in: {username}")
@@ -71,20 +72,40 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     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}")
             continue
         try:
             print(f"Running agent on task {task_id}: {question_text}")
-            submitted_answer = agent.graph.invoke({"messages": [HumanMessage(content=question_text)]})
-            print(submitted_answer)
+            # Retry 3 times with 45 seconds delay between attempts
+            for attempt in range(3):
+                try:
+                    # Call the agent's graph with the question text
+                    submitted_answer = agent.graph.invoke(
+                        {
+                            "messages": [HumanMessage(content=question_text)],
+                            "task_id": task_id,
+                            "file_name": file_name,
+                        }
+                        )
+                    break  # Break if successful
+                except Exception as e:
+                    print(f"Attempt {attempt + 1} failed: {e}")
+                    if attempt < 2:
+                        print("Retrying...")
+                        time.sleep(45)  # Exponential backoff
+            submitted_answer = submitted_answer['messages'][-1].content
+            if "FINAL ANSWER:" in submitted_answer:
+                submitted_answer = submitted_answer.split("FINAL ANSWER:")[-1].strip()
+            print(f"Agent submitted answer for task {task_id}: {submitted_answer}")
             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:
              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}"})
         
-        break
+        # break
 
     if not answers_payload:
         print("Agent did not produce any answers to submit.")
@@ -94,52 +115,52 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     print(f"Results log: {results_log}")
 
     # 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)
-
-    # # 5. Submit
-    # print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
-    # 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', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
-    #         f"Message: {result_data.get('message', 'No message received.')}"
-    #     )
-    #     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
-    # 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
+    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)
+
+    # 5. Submit
+    print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
+    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', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
+            f"Message: {result_data.get('message', 'No message received.')}"
+        )
+        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
+    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
 
 
 # --- Build Gradio Interface using Blocks ---

requirements.txt

@@ -1,7 +1,9 @@
 gradio
 requests
 python-dotenv
+pandas
 langchain
 langchain-google-genai
 langchain-tavily
-langgraph
+langgraph
+openai-whisper

src/custom_tools/downloads.py

@@ -0,0 +1,41 @@
+import requests
+from typing import Annotated
+
+from langchain_core.tools import tool
+from langgraph.prebuilt import InjectedState
+
+from src.state import State
+
+DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+
+@tool
+def download_file(
+     task_id: str,
+     state: Annotated[State, InjectedState]
+     ) -> str:
+        """Download a file specified by using the task id."""
+        file_name = state.file_name
+        task_id = state.task_id
+
+        if not file_name:
+            return "No file name in input, unable to download."
+        if not task_id:
+            return "No task id in input, unable to download."
+        
+        base_url = DEFAULT_API_URL + "/files"
+        url = f"{base_url}/{task_id}" if task_id else None
+        if not url:
+            return "No URL provided."
+        
+        try:
+            response = requests.get(url, stream=True)
+            response.raise_for_status()  # Raise an error for bad responses
+            
+            local_file_path = f"downloads/{file_name}"
+            with open(local_file_path, "wb") as f:
+                for chunk in response.iter_content(chunk_size=8192):
+                    f.write(chunk)
+            
+            return f"File downloaded successfully: {local_file_path}"
+        except requests.exceptions.RequestException as e:
+            return f"Error downloading file: {e}"

src/custom_tools/files.py

@@ -0,0 +1,71 @@
+import os
+
+from langchain.tools import tool
+
+@tool
+def read_python(file_path):
+    """
+    Reads a Python file and returns its content as a string.
+    
+    Args:
+        file_path (str): The path to the Python file.
+        
+    Returns:
+        str: The content of the Python file.
+    """
+    try:
+        if not os.path.exists(file_path):
+            return f"Error: File not found at {file_path}"
+        with open(file_path, "r", encoding="utf-8") as file:
+            content = file.read()
+        return content
+    except Exception as e:
+        return f"Error reading Python file: {str(e)}"
+
+@tool
+def read_excel(file_path):
+    """
+    Reads an Excel file and returns its content as a string.
+    
+    Args:
+        file_path (str): The path to the Excel file.
+        
+    Returns:
+        str: The content of the Excel file.
+    """
+    if not file_path.endswith(('.xls', '.xlsx')):
+        return "Error: File is not an Excel file."
+    try:
+        if not os.path.exists(file_path):
+            return f"Error: File not found at {file_path}"
+        import pandas as pd
+        df = pd.read_excel(file_path)
+        return df.to_string()
+    except Exception as e:
+        return f"Error reading Excel file: {str(e)}"
+    
+
+@tool
+def transcribe_audio(file_path):
+    """
+    Transcribes an audio file and returns its content as a string.
+    
+    Args:
+        file_path (str): The path to the audio file.
+        
+    Returns:
+        str: The transcribed text from the audio file.
+    """
+    if not file_path.endswith(('.wav', '.mp3', '.m4a')):
+        return "Error: File is not an audio file."
+    try:
+        if not os.path.exists(file_path):
+            return f"Error: File not found at {file_path}"
+        
+        import whisper
+        model = whisper.load_model("base")
+        result = model.transcribe(file_path, language="en")
+        text = result["text"]
+        return text
+    except Exception as e:
+        return f"Error transcribing audio file: {str(e)}"

src/custom_tools/wikipedia.py

@@ -0,0 +1,106 @@
+import requests
+from typing_extensions import Any, Annotated
+
+from langchain_core.tools import tool
+from langchain_core.messages import ToolMessage
+from langgraph.types import Command
+from langgraph.prebuilt import InjectedState
+from langchain_core.tools.base import InjectedToolCallId
+
+from src.state import State
+
+@tool
+def get_wiki_page_sections(
+    page_title: str,
+    tool_call_id: Annotated[str, InjectedToolCallId]
+    ) -> Command:
+    """Get sections of a Wikipedia page.
+    
+    This function retrieves the sections of a Wikipedia page.
+    It requires the page title as an input parameter.
+    """
+    page_title = page_title.replace(" ", "_")
+    payload = {
+        "action": "parse",
+        "page": page_title,
+        "prop": "sections",
+        "format": "json",
+    }
+
+    response = requests.get(
+        "https://en.wikipedia.org/w/api.php",
+        params=payload
+    )
+
+    if not response.status_code == 200:
+        return (f"Error fetching sections for {page_title}: {response.test}")
+    
+    data = response.json()
+    sections = data.get("parse", {}).get("sections", [])
+    sections_map = {}
+    for section in sections:
+        section_title = section.get("anchor").lower()
+        section_number = section.get("index")
+        if section_title and section_number:
+            sections_map[section_title] = section_number
+
+    sections_text = "The sections of the page are:\n"
+    for title in sections_map.keys():
+        sections_text += f"{title}\n"
+
+    return Command(
+        update={
+            # update the state keys
+            "wiki_sections": sections_map,
+            # update the message history
+            "messages": [
+                ToolMessage(
+                   sections_text, tool_call_id=tool_call_id
+                )
+            ],
+        }
+    )
+
+@tool
+def get_wiki_page_by_section(
+        page_title: str, 
+        section: str,
+        state: Annotated[State, InjectedState]
+        ) -> str:
+    """Get sections of a Wikipedia page.
+    
+    This function retrieves the content of a specific section from a Wikipedia page.
+    It requires the page title and the section name as input parameters.
+    """
+    wiki_sections = state.wiki_sections
+    if not wiki_sections:
+        return (f"Error: No sections found for {page_title}. Please run get_page_sections first.")
+    
+    page_title = page_title.replace(" ", "_")
+    section = section.replace(" ", "_").lower()
+
+    if section not in wiki_sections:
+        return (f"Error: Section '{section}' not found in {page_title}. Please run get_page_sections first.")
+    
+    payload = {
+        "action": "parse",
+        "page": page_title,
+        "prop": "wikitext",
+        "section": wiki_sections[section],
+        "format": "json",
+    }
+
+    response = requests.get(
+        "https://en.wikipedia.org/w/api.php",
+        params=payload
+    )
+
+    if not response.status_code == 200:
+        return (f"Error fetching sections for {page_title}: {response.test}")
+    
+    data = response.json()
+
+    return data.get("parse", {}).get("wikitext", "No content found.")
+    
+
+        

src/gaia_agent.py

@@ -31,7 +31,7 @@ class GaiaAgent:
         )
         builder.add_edge("tools", "call_model")
         
-        graph = builder.compile(name="GAIA Agent", debug=True)
+        graph = builder.compile(name="GAIA Agent", debug=False)
         
         return graph
         
@@ -59,15 +59,25 @@ class GaiaAgent:
         # Format the system prompt. Customize this to change the agent's behavior.
         system_message = configuration.system_prompt
 
+        if state.file_name:
+            file_prompt = (
+                f"\n\nThe task id is {state.task_id}.\n"
+                f"Please use this to download the file."
+            )
+
+            system_message += file_prompt
+
         # Get the model's response
         response = cast(
             AIMessage,
             model.llm.invoke(
-                [{"role": "system", "content": system_message}, *state.messages]
+                [
+                    {"role": "system", "content": system_message}, 
+                    *state.messages,
+
+                    ]
             ),
         )
-        
-        print(response.tool_calls)
 
         # Handle the case when it's the last step and the model still wants to use a tool
         if state.is_last_step and response.tool_calls:

src/prompts.py

@@ -1,7 +1,23 @@
 SYSTEM_PROMPT = (
     "You are a helpful AI assistant.\n"
-    "Please answer the question to the best of your ability.\n"
-    "Use the tools provided to you to find the answer.\n"
-    "Do not ask for permission to use the tools.\n"
-    "If you think you should use a tool, do so.\n"
+    "Please answer the question to the best of your ability. "
+    "Use the tools provided to you to find the answer. "
+    "Do not ask for permission to use the tools. "
+    "If you think you should use a tool, do so. "
+    "If the user specifies a file to use, use the "
+    "download tool to download the file and then use it. "
+    "If you get a file not found error, please try to download the file. "
+    "Include adjectives in your answer if the user asks shows this in the example. "
+    "Finish your answer with the following template: "
+    "FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER "
+    "should be a number OR as few words as possible OR a "
+    "comma separated list of numbers and/or strings. If "
+    "you are asked for a number, don't use comma to write "
+    "your number neither use units such as $ or percent sign "
+    "unless specified otherwise. If you are asked for a string, "
+    "don't use articles, neither abbreviations (e.g. for cities), "
+    "and write the digits in plain text unless specified otherwise. "
+    "If you are asked for a comma separated list, apply the above rules "
+    "depending of whether the element to be put in the list is a number "
+    "or a string."
 )

src/state.py

@@ -21,6 +21,8 @@ class InputState:
     messages: Annotated[Sequence[AnyMessage], add_messages] = field(
         default_factory=list
     )
+    task_id: str = field(default="")
+    file_name: str = field(default=None)
     """
     Messages tracking the primary execution state of the agent.
 
@@ -46,6 +48,7 @@ class State(InputState):
     """
 
     is_last_step: IsLastStep = field(default=False)
+    wiki_sections: dict[str, int] = field(default_factory=dict)
     """
     Indicates whether the current step is the last one before the graph raises an error.
 

src/tools.py

@@ -3,6 +3,7 @@ from typing import Any, Callable, List, Optional, cast
 from langchain_tavily import TavilySearch  # type: ignore[import-not-found]
 
 from src.config import Configuration
+from src.custom_tools import wikipedia, files, downloads
 
 def search(query: str) -> Optional[dict[str, Any]]:
     """Search for general web results.
@@ -15,5 +16,12 @@ def search(query: str) -> Optional[dict[str, Any]]:
     wrapped = TavilySearch(max_results=configuration.max_search_results)
     return cast(dict[str, Any], wrapped.invoke({"query": query}))
 
-
-TOOLS: List[Callable[..., Any]] = [search]
+TOOLS: List[Callable[..., Any]] = [
+    search,
+    wikipedia.get_wiki_page_sections,
+    wikipedia.get_wiki_page_by_section,
+    downloads.download_file,
+    files.read_python,
+    files.read_excel,
+    files.transcribe_audio,
+    ]