llms.py

import datetime

from langchain.agents import create_openai_tools_agent, AgentExecutor
from langchain.chat_models import init_chat_model
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_openai import ChatOpenAI
from langchain_tavily import TavilySearch
from langchain.schema import HumanMessage, SystemMessage
from langchain_core.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
from langchain_anthropic import ChatAnthropic
from pydantic import BaseModel, Field
from typing import List
from enum import Enum
from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.output_parsers import JsonOutputParser
import os
from dotenv import load_dotenv
load_dotenv()

from config import settings


class Agents:
    def __init__(self, temperature: float = 0.3, model: str = "gpt-4.1", model_type: str = "openai",
                 is_structured: bool = True):
        """
        Initializes the UserSummaryGenerator with necessary settings and model configuration.
        """

        self.settings = settings
        self.model_type = model_type

        self.llm_tavily = init_chat_model(model="gpt-4o", model_provider="openai", temperature=0)

        self.tavily_search_tool = TavilySearch(max_results=5, topic="general")

        self.tavily_prompt = ChatPromptTemplate.from_messages([
            ("system", self.settings.COUNTY_RISK_ANALYSIS_PROMPT),
            MessagesPlaceholder(variable_name="messages"),
            MessagesPlaceholder(variable_name="agent_scratchpad"),  # Required for tool calls
        ])

        self.tavily_agent = create_openai_tools_agent(
            llm=self.llm_tavily,
            tools=[self.tavily_search_tool],
            prompt=self.tavily_prompt
        )

        self.agent_executor = AgentExecutor(agent=self.tavily_agent, tools=[self.tavily_search_tool], verbose=False)


    def generate_tavily_search(self, user_input: str) -> str:
        response = self.agent_executor.invoke({"messages": [HumanMessage(content=user_input)]},)
        return response