app.py

import streamlit as st
import PyPDF2

import pandas as pd
import numpy as np
import re
import os
import  spacy
import matplotlib.pyplot as plt
import seaborn as sns
from sentence_transformers import SentenceTransformer, util

# Load Spacy model
spacy_model = "en_core_web_sm"
try:
    nlp = spacy.load(spacy_model)
except OSError:
    subprocess.run(["python", "-m", "spacy", "download", spacy_model], check=True)
    nlp = spacy.load(spacy_model)

# Load Sentence Transformer Model
model = SentenceTransformer("paraphrase-MiniLM-L6-v2")

# Function to Extract Text from PDF
def extract_text_from_pdf(uploaded_file):
    text = ""
    try:
        pdf_reader = PyPDF2.PdfReader(uploaded_file)
        for page in pdf_reader.pages:
            text += page.extract_text() + " "
    except Exception as e:
        st.error(f"Error reading PDF: {e}")
    return text

# Function to Preprocess Text
def preprocess_text(text):
    text = text.lower()
    text = re.sub(r'[^a-zA-Z0-9\s]', '', text)
    doc = nlp(text)
    tokens = [token.lemma_ for token in doc if not token.is_stop]
    return ' '.join(tokens)

# Function to Extract Keywords
def extract_keywords(text, top_n=10):
    doc = nlp(text)
    words = [token.text for token in doc if token.is_alpha and not token.is_stop]
    word_freq = pd.Series(words).value_counts()
    return word_freq.head(top_n).to_dict()

# Function to Calculate Similarity
def calculate_similarity(resumes, job_desc):
    all_texts = resumes + [job_desc]
    embeddings = model.encode(all_texts, convert_to_tensor=True)
    similarity_scores = util.pytorch_cos_sim(embeddings[:-1], embeddings[-1])
    return similarity_scores.squeeze().tolist()

# Streamlit UI
st.title("📄 AI Resume Screening & Ranking System")
uploaded_files = st.file_uploader("Upload Resumes (Multiple PDFs allowed)", type=["pdf"], accept_multiple_files=True)
job_description = st.text_area("Enter Job Description")

if st.button("Rank Resumes"):
    if uploaded_files and job_description:
        resumes_text = [extract_text_from_pdf(file) for file in uploaded_files]
        resumes_cleaned = [preprocess_text(resume) for resume in resumes_text]
        job_desc_cleaned = preprocess_text(job_description)
        similarity_scores = calculate_similarity(resumes_cleaned, job_desc_cleaned)
        ranked_indices = np.argsort(similarity_scores)[::-1].tolist()
        results = pd.DataFrame({
            'Resume': [uploaded_files[i].name for i in ranked_indices],
            'Similarity Score': [similarity_scores[i] for i in ranked_indices]
        })
        st.write("### 📊 Ranked Resumes:")
        st.dataframe(results)
        
        # Recruiter Dashboard
        st.write("## 📈 Recruiter Dashboard")
        st.write(f"- **Total Resumes Processed:** {len(uploaded_files)}")
        st.write(f"- **Average Similarity Score:** {np.mean(similarity_scores):.2f}")
        
        fig, ax = plt.subplots()
        sns.histplot(similarity_scores, bins=10, kde=True, ax=ax)
        ax.set_title("Similarity Score Distribution")
        ax.set_xlabel("Similarity Score")
        ax.set_ylabel("Count")
        st.pyplot(fig)
        
        # AI Explainability View
        st.write("## 🤖 AI Explainability View")
        job_keywords = extract_keywords(job_desc_cleaned)
        st.write("### 🔑 Top Keywords from Job Description:")
        st.json(job_keywords)
        
        for i, idx in enumerate(ranked_indices):
            resume_keywords = extract_keywords(resumes_cleaned[idx])
            st.write(f"### 📄 {uploaded_files[idx].name}")
            st.write("**Top Keywords in Resume:**")
            st.json(resume_keywords)
            matching_keywords = set(job_keywords.keys()) & set(resume_keywords.keys())
            st.write(f"**Matching Keywords with Job Description:** {list(matching_keywords)}")
    else:
        st.warning("⚠ Please upload resumes and enter a job description.")