app.py

import os
import gradio as gr
from pdfminer.high_level import extract_text
from docx import Document
from groq import Groq

key = os.getenv("GroqCVs")
client = Groq(api_key=key)

# Function to read PDF files
def read_pdf(file_path):
    return extract_text(file_path)

# Function to read DOCX files
def read_docx(file_path):
    doc = Document(file_path)
    return "\n".join([paragraph.text for paragraph in doc.paragraphs])

# Function to optimize CV using Llama 3.1 and Groq API
def optimize_cv_with_llama(cv_text, job_description):
    try:
        # Compose the prompt
        system_message = (
            "You are an experienced Human Resource expert with 30 years of experience working for major international companies. "
            "Your task is to optimize the following CV or Resume based on the provided job description. Tailor the CV content to align "
            "with the job requirements and highlight relevant skills, experiences, and achievements. Present the optimized CV in Markdown format. "
            "Ensure the formatting is professional and avoid adding extra comments."
        )

        # Send the job description and CV text to the Groq API
        completion = client.chat.completions.create(
            model="llama-3.3-70b-versatile",
            messages=[
                {"role": "system", "content": system_message},
                {"role": "user", "content": f"Job Description:\n{job_description}\n\nCV:\n{cv_text}"}
            ],
            temperature=0.7,
            max_tokens=2048,
            top_p=0.9,
            stream=False,
        )

        # Return the optimized Markdown text
        return ''.join([chunk.message.content for chunk in completion.choices])
    except Exception as e:
        return f"Error optimizing CV: {e}"

# Function to save optimized Markdown as DOCX
from markdown2 import markdown
from bs4 import BeautifulSoup

def save_markdown_to_docx(markdown_text, output_docx_path="optimized_cv.docx"):
    # Convert Markdown to HTML
    html_content = markdown(markdown_text)

    # Parse HTML with BeautifulSoup
    soup = BeautifulSoup(html_content, "html.parser")

    # Create a new DOCX document
    doc = Document()

    # Iterate through HTML elements and map them to DOCX formatting
    for element in soup.descendants:
        if element.name == "h1":
            para = doc.add_paragraph(element.text, style="Heading 1")
        elif element.name == "h2":
            para = doc.add_paragraph(element.text, style="Heading 2")
        elif element.name == "h3":
            para = doc.add_paragraph(element.text, style="Heading 3")
        elif element.name == "ul":  # Unordered list
            for li in element.find_all("li"):
                doc.add_paragraph(li.text, style="List Bullet")
        elif element.name == "ol":  # Ordered list
            for li in element.find_all("li"):
                doc.add_paragraph(li.text, style="List Number")
        elif element.name == "p":  # Paragraph
            doc.add_paragraph(element.text)
        elif element.name == "strong":  # Bold text
            para = doc.add_paragraph()
            run = para.add_run(element.text)
            run.bold = True
        elif element.name == "em":  # Italics
            para = doc.add_paragraph()
            run = para.add_run(element.text)
            run.italic = True
        elif element.name == "a":  # Hyperlink
            para = doc.add_paragraph(element.text)  # Add the text of the hyperlink
            # Add the URL as well, optionally
            para.add_run(f" ({element.get('href')})").italic = True

    # Save the document
    doc.save(output_docx_path)
    return output_docx_path

# Function to process file upload and optimization
def process_inputs(cv_file, job_description):
    try:
        # Extract text from uploaded file
        file_path = cv_file.name
        file_extension = os.path.splitext(file_path)[1].lower()
        if file_extension == ".pdf":
            cv_text = read_pdf(file_path)
        elif file_extension == ".docx":
            cv_text = read_docx(file_path)
        else:
            return "Unsupported file format", None

        # Optimize the CV based on the job description
        optimized_markdown = optimize_cv_with_llama(cv_text, job_description)

        # Save the optimized Markdown as DOCX
        output_docx_path = "optimized_cv.docx"
        save_markdown_to_docx(optimized_markdown, output_docx_path)

        return optimized_markdown, output_docx_path
    except Exception as e:
        return f"Error processing inputs: {e}", None

# Gradio Interface
def main():
    with gr.Blocks() as app:
        # Title and Description
        gr.Markdown(
            """
            # **AI-Powered CV Optimizer**
            Optimize your CV or Resume to match job descriptions using the power of AI.
            This application processes your uploaded CV along with a job description and generates a tailored, professional CV ready for submission.
            """
        )

        # Input components
        file_input = gr.File(label="Upload Your CV (PDF or DOCX)")


        job_description = gr.Textbox(
            label="Job Description",
            placeholder="Paste the job description here...",
            lines=5
        )


        # Output components
        markdown_output = gr.Markdown(label="Optimized CV Preview")
        download_button = gr.File(label="Download Optimized CV (DOCX)")

        # Button to process inputs
        run_button = gr.Button("Optimize CV")

        # Define the interaction logic
        def handle_inputs(cv_file, job_description):
            markdown_text, docx_path = process_inputs(cv_file, job_description)
            return markdown_text, docx_path

        run_button.click(
            handle_inputs,
            inputs=[file_input, job_description],
            outputs=[markdown_output, download_button]
        )

    app.launch(debug=True)

if __name__ == "__main__":
    main()