nlp_parsed.py

import re 
import os
from dotenv import load_dotenv
from openai import AzureOpenAI
import json
from postgres_db import store_prompt, conn

load_dotenv()

SERP_API_KEY = os.getenv("SERP_API_KEY")
endpoint = os.getenv("AZURE_OPENAI_ENDPOINT")
api_key = os.getenv("AZURE_OPENAI_API_KEY")
api_version = os.getenv("AZURE_OPENAI_API_VERSION")
deployment = os.getenv("AZURE_OPENAI_DEPLOYMENT")


try:
    client = AzureOpenAI(api_key=api_key, api_version=api_version, azure_endpoint=endpoint)
except Exception as e:
   print(f"Failed to initialize Azure OpenAI client: {e}")
    

def parse_recruiter_query(query):
    """Parse recruiter query using AI to extract structured data"""
    if not client:
        return {"error": "Azure OpenAI client not available"}

    try:
        system_prompt = """You are an AI assistant that extracts structured recruitment information from natural language queries.

        Fields to extract:
        - job_title: ONLY the exact position title they're hiring for (e.g., "Python Developer", "Data Scientist"). 
          DO NOT include phrases like "looking for", "need a", "hiring", etc.
        - skills: Array of required technical skills mentioned (e.g., ["Python", "Django", "SQL"])
        - experience: Required experience in years (numeric value or range). For fresher candidates, use "fresher" exactly.
        - location: Array of city names if multiple cities are mentioned, or single city name as Array if only one city is mentioned.
        - work_preference: Work mode preference - one of: "remote", "onsite", "hybrid", null
        - job_type: Employment type - one of: "full-time", "part-time", "contract", "internship", null
        - is_indian: true if the job location(s) are in India, false otherwise. 
                     IMPORTANT: If no location is mentioned, always set is_indian = true.

        CRITICAL INSTRUCTIONS:
        1. For job_title, NEVER include phrases like "looking for", "need", "hiring", etc.
        2. For experience, if the query mentions "fresher", "fresh graduate", "entry level", use exactly "fresher"
        3. For is_indian, check the location(s). If the location(s) are Indian cities or the query context is India-based, return true. 
           If no location is mentioned at all, default to true.
        4. Return ONLY valid JSON without any explanation or additional text.
        5. Use your knowledge to recognize job titles across all industries and domains."""

        user_prompt = f"""Extract recruitment information from this query: "{query}"

        Examples of correct extraction:

        Input: "We are looking for a Python developer with 3 years experience from Mumbai"
        Output: {{"job_title": "Python Developer", "skills": ["Python"], "experience": "3", "location": ["Mumbai"], "work_preference": null, "job_type": null, "is_indian": true}}

        Input: "Need a senior React frontend developer with Redux, TypeScript, 5+ years"
        Output: {{"job_title": "React Frontend Developer", "skills": ["React", "Redux", "TypeScript"], "experience": "5+", "location": null, "work_preference": null, "job_type": null, "is_indian": true}}

        Input: "python developer with 2 year of experience from surat, ahmedabad and mumbai"
        Output: {{"job_title": "Python Developer", "skills": ["Python"], "experience": "2", "location": ["Surat", "Ahmedabad", "Mumbai"], "work_preference": null, "job_type": null, "is_indian": true}}

        Input: "Remote React developer needed, 5 years experience, Redux, TypeScript"
        Output: {{"job_title": "React Developer", "skills": ["React", "Redux", "TypeScript"], "experience": "5", "location": null, "work_preference": "remote", "job_type": null, "is_indian": true}}

        Input: "Looking for fresher Java developer from Delhi"
        Output: {{"job_title": "Java Developer", "skills": ["Java"], "experience": "fresher", "location": ["Delhi"], "work_preference": null, "job_type": null, "is_indian": true}}

        Now extract from the query: "{query}"

        Remember: 
        1. Extract ONLY the job title without any prefixes like "looking for", "need", etc.
        2. Extract ONLY the city/location name without additional text.
        3. For fresher candidates, use exactly "fresher" as experience value.
        4. For is_indian: true if job location(s) are Indian, false otherwise. If no location is provided, always return true.
        5. Return ONLY valid JSON."""

        response = client.chat.completions.create(
            model=deployment,
            messages=[
                {"role": "system", "content": system_prompt},
                {"role": "user", "content": user_prompt}
            ],
            temperature=0.0,
            max_tokens=500
        )

        return json.loads(response.choices[0].message.content)

    except json.JSONDecodeError:
        return {"error": "Invalid JSON returned from AI"}
    except Exception as e:
        return {"error": f"Unexpected error: {str(e)}"}



def prompt_enhancer(prompt: str) -> str:
    """Enhance recruiter prompt to be clearer and more structured"""
    if not client:
        return prompt  # fallback: return original if Azure client not available

    try:
        system_prompt = """You are an AI assistant that enhances recruiter job search prompts.
        Your goal is to:
        1. Clean up grammar and spelling mistakes.
        2. Expand shorthand into full professional wording.
        3. Preserve all important details: job title, skills, experience, location, work mode, job type.
        4. Do NOT invent new requirements — only clarify what’s already in the query.
        5. Do not copy examples literally — adapt based on the actual input.
        6. Return ONLY the enhanced recruiter prompt as plain text (no JSON)."""

        user_prompt = f"""Rewrite and enhance this recruiter query for clarity:

        Input: "{prompt}"

        Example Enhancements:
        - "python dev 2yr exp surat" → "Looking for a Python Developer with 2 years of experience in Surat."
        - "need react js fresher remote" → "Hiring a React.js Developer at fresher level for a remote role."
        - "java 5+ exp ahmedabad onsite" → "Looking for a Java Developer with over 5 years of experience for an onsite role in Ahmedabad."
        - "data analyst 3 years bangalore hybrid" → "Seeking a Data Analyst with 3 years of experience for a hybrid position in Bangalore."
        - "ui ux designer fresher mumbai internship" → "Hiring a UI/UX Designer, fresher level, for an internship role in Mumbai."

        Now enhance this query: "{prompt}"
        """

        response = client.chat.completions.create(
            model=deployment,
            messages=[
                {"role": "system", "content": system_prompt},
                {"role": "user", "content": user_prompt}
            ],
            temperature=0.5,   # slightly more creative
            max_tokens=200
        )

        enhanced_prompt = response.choices[0].message.content.strip()
        return enhanced_prompt

    except Exception as e:
        print(f"Error in prompt_enhancer: {e}")
        return prompt  # fallback to original