app.py

import nltk 
import numpy as np 
import pandas as pd 
import seaborn as sns 
import matplotlib.pyplot as plt
import re 
import streamlit as st 
import pickle as pi

nltk.download('punkt')
nltk.download('stopwords')

# Loading models
clf = pi.load(open('clf.pkl', 'rb'))
tfidf = pi.load(open('tfidf.pkl', 'rb'))


def cleanresume(txt):
    cleantxt = re.sub('http\S+\s'," ", txt)
    cleantxt = re.sub('RT|cc+ '," ", cleantxt)
    cleantxt = re.sub('@\S+'," ", cleantxt)
    cleantxt = re.sub('#\S+\s'," ", cleantxt)
    cleantxt = re.sub('[%s]'% re.escape("""!"#$%&'()*+,./:;<=>?@[\]^_`{|}~""")," ", cleantxt)
    cleantxt = re.sub(r'[^\x00-\x7f]'," ", cleantxt) 
    cleantxt = re.sub('\s+'," ", cleantxt)
    return cleantxt

# Web app
def main():
    st.title("Resume Screening App")
    uploaded_resume = st.file_uploader("Upload Resume", type=['txt', 'pdf'])
    
    if uploaded_resume is not None:
        try: 
            resume_bytes = uploaded_resume.read()
            resume_text = resume_bytes.decode('utf-8')
        except UnicodeDecodeError:
            # If UTF-8 decoding fails, try decoding with 'latin-1'
            resume_text = resume_bytes.decode('latin-1')

        # Clean the input resume
        cleaned_resume = cleanresume(resume_text)

        # Transform the cleaned resume using the trained TfidfVectorizer
        input_features = tfidf.transform([cleaned_resume])

        # Make the prediction using the loaded classifier 
        prediction_id = clf.predict(input_features)[0]
        # st.write(prediction_id)

        # Map category ID to category name
        category_mapping = {
                    15: "Java Developer",
                    23: "Testing",
                    8: "DevOps Engineer",
                    20: "Python Developer",
                    24: "Web Designing",
                    12: "HR",
                    13: "Hadoop",
                    3: "Blockchain",
                    10: "ETL Developer",
                    18: "Operations Manager",
                    6: "Data Science",
                    22: "Sales",
                    16: "Mechanical Engineer",
                    1: "Arts",
                    7: "Database",
                    11: "Electrical Engineering",
                    14: "Health and fitness",
                    19: "PMO",
                    4: "Business Analyst",
                    9: "DotNet Developer",
                    2: "Automation Testing",
                    17: "Network Security Engineer",
                    21: "SAP Developer",
                    5: "Civil Engineer",
                    0: "Advocate",
                }

        category_name = category_mapping.get(prediction_id, "Unknown")

        st.write("Prediction Category : ", category_name)


# Python main
if __name__ == "__main__":
    main()