mh_clarification.py

import os
import time
from langchain_openai import ChatOpenAI
from langchain.prompts import PromptTemplate
from langchain.agents import initialize_agent, AgentType
from langchain.tools import Tool
from dotenv import load_dotenv
from langchain_core.messages import HumanMessage

# Load environment variables
load_dotenv()

# Set environment variables for model API key and model type
os.environ.setdefault("DSP_CACHEBOOL", "false")

# Get API key
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
    raise ValueError("OPENAI_API_KEY environment variable is not set")

# Initialize the chat model
model = ChatOpenAI(
    model=os.getenv("MODEL", "gpt-3.5-turbo"),
    openai_api_key=api_key,
    temperature=0.0,
    max_tokens=4000
)

# Define the clarification prompt
clarification_prompt = PromptTemplate(
    input_variables=["checkpoints", "resume"],
    template="""
    You are an expert recruiter specializing in reading resumes against job descriptions. Your task is to read the checkpoints provided and extract objective and factual information (if available) from the resume to clarify these checkpoints.

    **Guidelines:**
    1. Analyze both explicit and implicit meanings from the resume.
    2. For must-have certifications, consider only those explicitly mentioned. Do not assume.
    3. For industry relevance, assess the organizations listed and determine their industries.
    4. For education and certifications, verify if they match stated requirements.
    5. Provide objective reasoning with factual pointers from the resume.
    6. Do not hallucinate or include information not grounded in the resume.
    7. If the resume lacks enough information, mention this explicitly.

    **Checkpoints:**
    {checkpoints}

    **Resume:**
    {resume}

    **Output Format:**
    Checkpoint 1: [Factual reasoning from resume based on checkpoint]
    Checkpoint 2: [Factual reasoning from resume based on checkpoint]
    """
)

# Define a tool to process checkpoints and resume
class ClarificationProcessor(Tool):
    def __init__(self, name: str, description: str, func):
        super().__init__(name=name, func=func, description=description)

    def _run(self, input: str):
        # Extract checkpoints and resume from the input string
        try:
            # Split by "Resume:" and ensure we have both parts
            parts = input.split("Resume:")
            if len(parts) != 2:
                return "Error: Input must contain both checkpoints and resume sections"
            
            # Extract checkpoints and resume, removing any extra whitespace
            checkpoints = parts[0].replace("Checkpoints:", "").strip()
            resume = parts[1].strip()
            
            if not checkpoints or not resume:
                return "Error: Both checkpoints and resume sections must not be empty"
                
            return self.func(checkpoints, resume)
        except Exception as e:
            return f"Error processing input: {str(e)}"

# Define a function to generate clarifications based on checkpoints and resume
def generate_clarifications(checkpoints: str, resume: str) -> str:
    # Create the prompt
    prompt_text = clarification_prompt.format(checkpoints=checkpoints, resume=resume)
    
    # Get response from the model
    response = model.invoke([HumanMessage(content=prompt_text)])
    return response.content

# Define the agent tools
tools = [
    ClarificationProcessor(
        name="ClarificationProcessor",
        description="Generate clarifications based on the provided checkpoints and resume.",
        func=generate_clarifications,
    ),
]

# Initialize the LangChain agent
agent = initialize_agent(
    tools, model, agent_type=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True
)