app.py

from sentence_transformers import SentenceTransformer, util
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
from fuzzywuzzy import fuzz
import gradio as gr
import fitz  # PyMuPDF for PDF extraction
import pymupdf

# Load the SentenceTransformer model for embeddings
model = SentenceTransformer('fine_tuned_job_resume_similarity_model')

# Load Hugging Face NER model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-large-cased-finetuned-conll03-english")
ner_model = AutoModelForTokenClassification.from_pretrained("dbmdz/bert-large-cased-finetuned-conll03-english")
ner_pipeline = pipeline("ner", model=ner_model, tokenizer=tokenizer, aggregation_strategy="simple")

# Generalized keywords across multiple industries
TARGET_KEYWORDS = [
    "skill", "experience", "education", "certification", "project", "management",
    "sales", "marketing", "customer service", "financial", "analysis", "quality",
    "engineering", "healthcare", "law", "administration", "communication",
    "problem-solving", "teamwork", "leadership", "technical", "planning", "operations"
]

# Function to extract text from PDF files
def extract_text_from_pdf(pdf_file):
    text = ""
    with pymupdf.open(pdf_file) as doc:
        for page in doc:
            text += page.get_text("text")
    return text

# Define function to dynamically extract entities into generalized categories
def extract_entities(text):
    entities = {"qualifications": [], "responsibilities": [], "other": []}
    ner_results = ner_pipeline(text)
    for entity in ner_results:
        word = entity['word'].strip()
        if any(keyword in word.lower() for keyword in ["skill", "degree", "education", "certification", "qualification"]):
            entities["qualifications"].append(word)
        elif any(keyword in word.lower() for keyword in ["experience", "responsibility", "role", "project"]):
            entities["responsibilities"].append(word)
        else:
            entities["other"].append(word)
    return entities

# Function to compute fuzzy matching score for keywords
def fuzzy_match_keywords(cv_text, job_text, keywords):
    match_score = 0
    for keyword in keywords:
        score = fuzz.partial_ratio(cv_text.lower(), keyword.lower())
        match_score += score if score > 60 else 0  # Consider only high-confidence matches
    return match_score / len(keywords) if keywords else 0

def match_cv_to_job(pdf_file, job_description):
    try:
        # Extract text from PDF file
        cv_text = extract_text_from_pdf(pdf_file)
        debug_info = "Debug Info:\n"
        
        # Extract entities from CV and job description
        cv_entities = extract_entities(cv_text)
        job_entities = extract_entities(job_description)
        
        # Calculate NER-based entity match score
        match_score = 0
        for key in cv_entities:
            if key in job_entities:
                matched = set(cv_entities[key]) & set(job_entities[key])
                match_score += len(matched) / len(set(job_entities[key])) if job_entities[key] else 0

        # Normalize NER match score by number of categories
        ner_match_score = (match_score / len(cv_entities)) * 100  
        debug_info += f"NER Match Score: {ner_match_score:.2f}%\n"
        
        # Compute fuzzy matching score for generalized keywords
        fuzzy_keyword_score = fuzzy_match_keywords(cv_text, job_description, TARGET_KEYWORDS)
        debug_info += f"Fuzzy Keyword Score: {fuzzy_keyword_score:.2f}\n"
        
        # Calculate overall similarity score using embeddings
        cv_embedding = model.encode(cv_text, convert_to_tensor=True)
        job_embedding = model.encode(job_description, convert_to_tensor=True)
        similarity_score = util.pytorch_cos_sim(cv_embedding, job_embedding).item()
        debug_info += f"Embedding Similarity Score: {similarity_score:.2f}\n"

        # Adjust weights to balance entity and keyword matching for various industries
        combined_score = (
            similarity_score * 0.5 +      # Embedding similarity
            (ner_match_score / 100) * 0.3 +  # NER-based entity match
            (fuzzy_keyword_score / 100) * 0.2  # Fuzzy matching for keywords
        )
        match_percentage = combined_score * 100
        debug_info += f"Overall Match Percentage: {match_percentage:.2f}%\n"
        
        return {"Match Percentage": f"{match_percentage:.2f}%"}, debug_info
    
    except Exception as e:
        # Capture and display the exception in debug output
        debug_info = f"An error occurred: {str(e)}"
        return {"Match Percentage": "Error"}, debug_info

# Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("# CV and Job Description Matcher for All Industries with NER and Fuzzy Matching")
    
    pdf_input = gr.File(label="Upload CV (PDF format)")
    job_description = gr.Textbox(label="Job Description", placeholder="Enter the job description text here", lines=10)
    
    match_button = gr.Button("Calculate Match Percentage")
    output = gr.JSON(label="Match Result")
    debug_output = gr.Textbox(label="Debug Info", lines=10)
    
    match_button.click(fn=match_cv_to_job, inputs=[pdf_input, job_description], outputs=[output, debug_output])

demo.launch()