app.py

import streamlit as st
import os
import re
from groq import Groq
import json

# Set up Groq client
client = Groq(api_key="")

# Define main function
def main():
    st.title("AI-powered Resume Scanner")

    # File upload
    uploaded_file = st.file_uploader("Upload a resume", type=["pdf", "docx"])

    # Job role input
    job_role = st.text_input("Enter the job role")

    if uploaded_file is not None and job_role:
        # Process resume and get results
        resume_text = parse_resume(uploaded_file)
        if resume_text:
            #st.write("Extracted Resume Text:")
            #st.write(resume_text)

            # Get resume analysis from the Groq model
            name, degree, cgpa, skills, experience_score, ats_score = analyze_resume(resume_text, job_role)

            # Display the results
            st.write(f"**Candidate Name:** {name}")
            st.write(f"**Degree:** {degree}")
            st.write(f"**Latest CGPA/Percentage:** {cgpa}")
            if skills:
                st.write("**Skills:**")
                for skill in skills:
                    st.write(f"- {skill}")
            st.write(f"**Experience Score out of 10:** {experience_score}")
            st.write(f"**ATS Score for {job_role} out of 10:** {ats_score}")

# Function to parse PDF/Word file
def parse_resume(uploaded_file):
    # If PDF
    if uploaded_file.type == "application/pdf":
        from PyPDF2 import PdfReader
        reader = PdfReader(uploaded_file)
        text = ""
        for page in reader.pages:
            text += page.extract_text()
        return text
    # If Word
    elif uploaded_file.type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
        from docx import Document
        doc = Document(uploaded_file)
        text = "\n".join([para.text for para in doc.paragraphs])
        return text
    else:
        st.error("Unsupported file type!")
        return None

# Function to analyze resume using Groq API and LLaMA 3.1
# Function to analyze resume using Groq API and LLaMA 3.1
import json

# Function to analyze resume using Groq API and LLaMA 3.1
def analyze_resume(text, job_role):
    # Construct prompt for Groq API
    prompt = (
        f"Extract the following details from the given resume text: \n"
        f"1. Candidate's Name \n"
        f"2. Latest Education CGPA or Percentage \n"
        f"3. List of Skills \n"
        f"4. Rate experience (projects, internships) on a scale from 0 to 10 \n"
        f"5. Provide an ATS score for the job role: {job_role}\n"
        f"Resume Text: {text}\n"
        f"Format your response in a JSON object with the following structure: \n"
        f"{{\n"
        f'  "name": "Candidate Name",\n'
        f'  "Degree": "Latest Education qualification or grade:",\n'
        f'  "cgpa": "Latest Education CGPA or Percentage",\n'
        f'  "skills": ["List of skills"],\n'
        f'  "experience_score": "Experience Score (0 to 10)",\n'
        f'  "ats_score": "ATS Score for the job role"\n'
        f"}}"
    )

    # Call Groq API for chat completion
    chat_completion = client.chat.completions.create(
        messages=[{"role": "user", "content": prompt}],
        model="llama3-8b-8192",
        temperature=0.7,
        max_tokens=1024,
        top_p=1,
        stream=False
    )

    # Get the raw output from the LLM
    output = chat_completion.choices[0].message.content
    # Clean up the output to avoid any parsing issues
    cleaned_output = re.search(r'{[^}]*}', output).group()
    # Try parsing the cleaned JSON
    try:
        response_data = json.loads(cleaned_output)
    except json.JSONDecodeError:
        st.error("Failed to parse response from model.")
        st.write("Model Output:")
        st.write(cleaned_output)
        return None, None, None, None, None

    # Extract information from the parsed JSON
    name = response_data.get("name", "Name not found")
    degree = response_data.get("Degree", "Degree not found")
    cgpa = response_data.get("cgpa", "CGPA/Percentage not found")
    skills = response_data.get("skills", "Skills not found")
    experience_score = response_data.get("experience_score", "Experience score not found")
    ats_score = response_data.get("ats_score", "ATS score not found")

    return name, degree, cgpa, skills, experience_score, ats_score



# Function to extract the candidate's name
def extract_name(text):
    name_pattern = re.compile(r"(Name:?\s*)([A-Z][a-z]+(?:\s[A-Z][a-z]+)*)")
    match = name_pattern.search(text)
    if match:
        return match.group(2)
    return "Name not found"

# Function to extract CGPA or Percentage
def extract_cgpa(text):
    cgpa_pattern = re.compile(r"(\bCGPA\b|\bGPA\b|\bPercentage\b):?\s*(\d+\.?\d*)")
    match = cgpa_pattern.search(text)
    if match:
        return match.group(2)
    return "CGPA/Percentage not found"

# Function to extract skills
def extract_skills(text):
    skills_pattern = re.compile(r"Skills:?\s*(.*?)(?:Experience|Education|$)", re.DOTALL)
    match = skills_pattern.search(text)
    if match:
        skills = match.group(1)
        return [skill.strip() for skill in skills.split(",")]
    return "Skills not found"

# Function to extract experience score
def extract_experience_score(text):
    experience_pattern = re.compile(r"Experience Score:?\s*(\d{1,2})")
    match = experience_pattern.search(text)
    if match:
        return int(match.group(1))
    # Heuristic: If no explicit experience score is given, infer it based on keywords
    experience_keywords = ["internship", "project", "work experience", "employment"]
    experience_count = sum(text.lower().count(keyword) for keyword in experience_keywords)
    return min(10, experience_count)  # Cap at 10

# Function to extract ATS score
def extract_ats_score(text):
    ats_pattern = re.compile(r"ATS Score:?\s*(\d+\.?\d*)")
    match = ats_pattern.search(text)
    if match:
        return float(match.group(1))
    # Heuristic to generate a score based on skill-job match
    return generate_ats_score(text)

# Heuristic function to generate ATS score
def generate_ats_score(text):
    # Just a dummy heuristic for now
    skills = extract_skills(text)
    if not skills:
        return 0

    required_skills = ["Python", "Machine Learning", "Data Analysis"]  # Add job role specific required skills
    match_count = sum(1 for skill in required_skills if skill.lower() in [s.lower() for s in skills])
    return round((match_count / len(required_skills)) * 10, 2)

if __name__ == "__main__":
    main()