Update agents.py

agents.py

@@ -1,62 +1,79 @@
-# agent.py
-
+import importlib
 import os
-from typing import TypedDict, Annotated, List
-from langchain_community.tools import DuckDuckGoSearchResults
-from langchain_groq import ChatGroq
-from langgraph.graph import StateGraph, END, START
-from langgraph.graph.message import AnyMessage, add_messages
-from langchain_core.messages import HumanMessage, AIMessage
-
-# --- Setup API Key ---
-groq_api_key = "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"  # or hardcode if testing
-
-# --- Setup Search Tool ---
-search_tool = DuckDuckGoSearchResults()
-
-# --- Setup Base LLM Model (for generation) ---
-base_llm = ChatGroq(
-    model_name="llama3-8b-8192",
-    api_key=groq_api_key,
-    temperature=0
+
+import requests
+import yaml
+import pandas as pd
+
+from config import DEFAULT_API_URL
+from smolagents import CodeAgent, DuckDuckGoSearchTool, VisitWebpageTool, WikipediaSearchTool, Tool, OpenAIServerModel, SpeechToTextTool
+
+class GetTaskFileTool(Tool):
+    name = "get_task_file_tool"
+    description = """This tool downloads the file content associated with the given task_id if exists. Returns absolute file path"""
+    inputs = {
+        "task_id": {"type": "string", "description": "Task id"},
+        "file_name": {"type": "string", "description": "File name"},
+    }
+    output_type = "string"
+
+    def forward(self, task_id: str, file_name: str) -> str:
+        response = requests.get(f"{DEFAULT_API_URL}/files/{task_id}", timeout=15)
+        response.raise_for_status()
+        with open(file_name, 'wb') as file:
+            file.write(response.content)
+        return os.path.abspath(file_name)
+
+class LoadXlsxFileTool(Tool):
+    name = "load_xlsx_file_tool"
+    description = """This tool loads xlsx file into pandas and returns it"""
+    inputs = {
+        "file_path": {"type": "string", "description": "File path"}
+    }
+    output_type = "object"
+
+    def forward(self, file_path: str) -> object:
+        return pd.read_excel(file_path)
+
+class LoadTextFileTool(Tool):
+    name = "load_text_file_tool"
+    description = """This tool loads any text file"""
+    inputs = {
+        "file_path": {"type": "string", "description": "File path"}
+    }
+    output_type = "string"
+
+    def forward(self, file_path: str) -> object:
+        with open(file_path, 'r', encoding='utf-8') as file:
+            return file.read()
+
+
+prompts = yaml.safe_load(
+    importlib.resources.files("smolagents.prompts").joinpath("code_agent.yaml").read_text()
 )
+prompts["system_prompt"] = ("You are a general AI assistant. I will ask you a question. Report your thoughts, and 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. "
+                            + prompts["system_prompt"])
 
-# --- Final Content Extractor Tool (LLM based) ---
-class FinalContentExtractor:
-    def __init__(self, model):
-        self.model = model
-
-    def extract(self, question: str, search_result: str) -> str:
-        prompt = (
-            f"You are an intelligent assistant. "
-            f"Given the user's question and the search result, extract the most direct and relevant answer.\n\n"
-            f"Question: {question}\n\n"
-            f"Search Result: {search_result}\n\n"
-            f"Answer:"
-        )
-        response = self.model.invoke([HumanMessage(content=prompt)])
-        return response.content.strip()
-
-# --- Define State Class for the Graph ---
-class AgentState(TypedDict):
-    messages: Annotated[List[AnyMessage], add_messages]
-
-# --- Define Search Node ---
-def search_node(state: AgentState):
-    user_message = state["messages"][-1]
-    query = user_message.content
-    search_result = search_tool.invoke({"query": query})
-
-    search_summary = f"Top Search Results:\n{search_result}"
-    ai_message = AIMessage(content=search_summary)
-    return {"messages": state["messages"] + [ai_message]}
-
-# --- Build the Graph ---
-builder = StateGraph(AgentState)
-builder.add_node("search", search_node)
-builder.add_edge(START, "search")
-builder.add_edge("search", END)
-graph = builder.compile()
-
-# --- Export these for app.py ---
-__all__ = ["graph", "FinalContentExtractor", "base_llm"]
+def init_agent():
+    gemini_model = OpenAIServerModel(
+            model_id="gemini-2.0-flash",
+            api_base="https://generativelanguage.googleapis.com/v1beta/openai/",
+            api_key=os.getenv("API_KEY"),
+            temperature=0.7
+    )
+    agent = CodeAgent(
+        tools=[
+            DuckDuckGoSearchTool(),
+            VisitWebpageTool(),
+            WikipediaSearchTool(),
+            GetTaskFileTool(),
+            SpeechToTextTool(),
+            LoadXlsxFileTool(),
+            LoadTextFileTool()
+        ],
+        model=gemini_model,
+        prompt_templates=prompts,
+        max_steps=15,
+        additional_authorized_imports = ["pandas"]
+    )
+    return agent