import streamlit as st
from PyPDF2 import PdfReader
import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer

from sklearn.metrics.pairwise import cosine_similarity

import streamlit as st
from PyPDF2 import PdfReader
import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from gliner import GLiNER


import streamlit as st
import pandas as pd
from PyPDF2 import PdfReader
from gliner import GLiNER

import streamlit as st
import pandas as pd
from PyPDF2 import PdfReader
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
import tempfile

txt1 = st.text_area("Job description", key = "text 1")
job_description_series1 = pd.Series(txt1, name="Text")
st.dataframe(job_description_series1)

uploaded_files = st.file_uploader(
    "Choose a PDF file(s) for candidate profiles",  type="pdf", key = "candidate 1"
)


all_resumes_text1 = []  # Store the text content of each PDF

if uploaded_files:
    for uploaded_file in uploaded_files:
        try:
            pdf_reader = PdfReader(uploaded_file)
            text_data = ""
            for page in pdf_reader.pages:
                text_data += page.extract_text()
                model = GLiNER.from_pretrained("urchade/gliner_base")
                labels = ["person", "country", "organization", "time", "role"]
                entities = model.predict_entities(text_data, labels)

            entity_dict = {}
            for label in labels:
                entity_dict[label] = [entity["text"] for entity in entities if entity["label"] == label]

            data = {"Text": text_data, **entity_dict}
            


            all_resumes_text1.append(data)
        except Exception as e:
            st.error(f"Error processing file {uploaded_file.name}: {e}")

    if all_resumes_text1:
        all_documents = [job_description_series.iloc[0]] + all_resumes_text

        vectorizer = TfidfVectorizer()
        tfidf_matrix = vectorizer.fit_transform(all_documents)

        tfidf_df = pd.DataFrame(tfidf_matrix.toarray(), columns=vectorizer.get_feature_names_out())
        st.subheader("TF-IDF Values:")
        st.dataframe(tfidf_df)

        cosine_sim_matrix = cosine_similarity(tfidf_matrix)
        cosine_sim_df = pd.DataFrame(cosine_sim_matrix)
        st.subheader("Cosine Similarity Matrix:")
        st.dataframe(cosine_sim_df)

        # Display similarity scores between the job description and each resume
        st.subheader("Cosine Similarity Scores (Job Description vs. Resumes):")
        for i, similarity_score in enumerate(cosine_sim_matrix[0][1:]):
            st.write(f"Similarity with Candidate Profile {i + 1}: {similarity_score:.4f}")


       
st.divider()

txt2 = st.text_area("Job description", key = "text 2")
job_description_series2 = pd.Series(txt2, name="Text")
st.dataframe(job_description_series2)

uploaded_files = st.file_uploader(
    "Choose a PDF file(s) for candidate profiles",  type="pdf", key = "candidate 2"
)


all_resumes_text2 = []  # Store the text content of each PDF

if uploaded_files:
    for uploaded_file in uploaded_files:
        try:
            pdf_reader = PdfReader(uploaded_file)
            text_data = ""
            for page in pdf_reader.pages:
                text_data += page.extract_text()
                model = GLiNER.from_pretrained("urchade/gliner_base")
                labels = ["person", "country", "organization", "time", "role"]
                entities = model.predict_entities(text_data, labels)

            entity_dict = {}
            for label in labels:
                entity_dict[label] = [entity["text"] for entity in entities if entity["label"] == label]

            data = {"Text": text_data, **entity_dict}
            


            all_resumes_text2.append(data)
        except Exception as e:
            st.error(f"Error processing file {uploaded_file.name}: {e}")

    if all_resumes_text2:
        all_documents = [job_description_series.iloc[0]] + all_resumes_text

        vectorizer = TfidfVectorizer()
        tfidf_matrix = vectorizer.fit_transform(all_documents)

        tfidf_df = pd.DataFrame(tfidf_matrix.toarray(), columns=vectorizer.get_feature_names_out())
        st.subheader("TF-IDF Values:")
        st.dataframe(tfidf_df)

        cosine_sim_matrix = cosine_similarity(tfidf_matrix)
        cosine_sim_df = pd.DataFrame(cosine_sim_matrix)
        st.subheader("Cosine Similarity Matrix:")
        st.dataframe(cosine_sim_df)

        # Display similarity scores between the job description and each resume
        st.subheader("Cosine Similarity Scores (Job Description vs. Resumes):")
        for i, similarity_score in enumerate(cosine_sim_matrix[0][1:]):
            st.write(f"Similarity with Candidate Profile {i + 1}: {similarity_score:.4f}")