app.py

import gradio as gr
import openai
import fitz  # PyMuPDF
import torch
from transformers import pipeline, BloomForCausalLM, BloomTokenizerFast
from huggingface_hub import login
import requests
import os
from models import evaluate_with_gpt,evaluate_with_bloom,evaluate_with_llama




def extract_text_from_pdf(pdf_file):
    document = fitz.open(pdf_file)
    text = ""
    for page_num in range(len(document)):
        page = document.load_page(page_num)
        text += page.get_text()
    return text



def evaluate_all_models(pdf_file, job_description):
    gpt_result = evaluate_with_gpt(pdf_file, job_description)
    # gemma_result = evaluate_with_gemma(pdf_file, job_description)
    bloom_result = evaluate_with_bloom(pdf_file, job_description)
    jabir_result = evaluate_with_jabir(pdf_file, job_description)
    llama_result=evaluate_with_llama(pdf_file, job_description)
    return f"GPT-4o Result:\n{gpt_result}\n\nBloom Result:\n{bloom_result}\n\njabir Result:\n{jabir_result}\n\nllama Result:\n{llam_result}"
    # return f"\n\nllama Result:\n{llam_result}"

iface = gr.Interface(
    fn=lambda pdf, jd, model: evaluate_with_gpt(pdf, jd) if model == "GPT-4o" else evaluate_with_bloom(pdf, jd) if model == "Bloom" else evaluate_with_jabir(pdf, jd) if model == "jabir" else evaluate_with_llama(pdf, jd) if model == "llama" else evaluate_all_models(pdf, jd),     
    # fn=lambda pdf, jd, model: evaluate_with_llama(pdf, jd),

    inputs=[
        gr.File(label="Upload Resume PDF"),
        gr.Textbox(lines=10, label="Job Description"),
        gr.Radio(choices=["GPT-4o", "Bloom", "jabir","llama", "All"], label="Choose Model")
        # gr.Radio(choices=["llama"], label="Choose Model")
    ],
    outputs="text",
    title="Resume Evaluator"
)

iface.launch()