Update app.py

app.py

@@ -1,7 +1,7 @@
 import os
 import gradio as gr
 import requests
-import openai
+import speech_recognition as sr
 from smolagents import OpenAIServerModel, DuckDuckGoSearchTool, CodeAgent, WikipediaSearchTool
 from pathlib import Path
 import tempfile
@@ -11,6 +11,10 @@ from typing import Union, Optional
 import pandas as pd
 from tabulate import tabulate  # pragma: no cover – fallback path
 import re
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from langchain.agents import initialize_agent
+from langchain_community.tools import DuckDuckGoSearchRun, WikipediaQueryRun
+from langchain_community.llms import HuggingFaceHub
 
 # (Keep Constants as is)
 # --- Constants ---
@@ -71,6 +75,21 @@ class SpeechToTextTool(PipelineTool):
         # For response_format="text", `response` is already the raw transcript
         return response
 
+    def transcribe_audio(audio_file_path):
+        recognizer = sr.Recognizer()
+        with sr.AudioFile(audio_file_path) as source:
+        audio_data = recognizer.record(source)
+        try:
+            text = recognizer.recognize_google(audio_data)
+            return text
+        except sr.UnknownValueError:
+            return "Could not understand audio"
+        except sr.RequestError:
+            return "Could not request results (check internet connection)"
+
+
+            
+
 class ExcelToTextTool(Tool):
     """Render an Excel worksheet as Markdown text."""
 
@@ -172,14 +191,12 @@ def download_file_if_any(base_api_url: str, task_id: str) -> str | None:
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 
-
 class BasicAgent:
     def __init__(self):
-        self.agent = CodeAgent(
-            model=OpenAIServerModel(model_id="gpt-4o"),
-            tools=[DuckDuckGoSearchTool(), WikipediaSearchTool(), SpeechToTextTool(), ExcelToTextTool()],
-            add_base_tools=True,
-            additional_authorized_imports=['pandas','numpy','csv','subprocess']
+        # Initialize LLM (requires HuggingFace API token)
+        llm = HuggingFaceHub(
+            repo_id="meta-llama/Meta-Llama-3-8B-Instruct"  #,
+         #   huggingfacehub_api_token="your_token"
         )
 
         print("BasicAgent initialized.")
@@ -190,6 +207,24 @@ class BasicAgent:
         print(f"Agent returning answer: {fixed_answer}")
         return fixed_answer
 
+        
+        # Initialize tools
+        tools = [
+            DuckDuckGoSearchRun(),
+            WikipediaQueryRun()
+            # Would need custom implementations for other tools
+        ]
+        
+        self.agent = initialize_agent(
+            tools=tools,
+            llm=llm,
+            agent="zero-shot-react-description",
+            verbose=True
+        )
+
+    def run(self, prompt):
+        return self.agent.run(prompt)
+
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     Fetches all questions, runs the BasicAgent on them, submits all answers,