app.py

import os
import openai
import PyPDF2
import gradio as gr
import docx

class QuestionsGenerator:
    def __init__(self):
        openai.api_key = os.getenv("OPENAI_API_KEY")

    def extract_text_from_file(self,file_path):
        # Get the file extension
        file_extension = os.path.splitext(file_path)[1]

        if file_extension == '.pdf':
            with open(file_path, 'rb') as file:
                # Create a PDF file reader object
                reader = PyPDF2.PdfFileReader(file)

                # Create an empty string to hold the extracted text
                extracted_text = ""

                # Loop through each page in the PDF and extract the text
                for page_number in range(reader.getNumPages()):
                    page = reader.getPage(page_number)
                    extracted_text += page.extractText()
            return extracted_text

        elif file_extension == '.txt':
            with open(file_path, 'r') as file:
                # Just read the entire contents of the text file
                return file.read()

        elif file_extension == '.docx':
              doc = docx.Document(file_path)
              text = []
              for paragraph in doc.paragraphs:
                  text.append(paragraph.text)
              return '\n'.join(text)

        else:
            return "Unsupported file type"

    def response(self,job_description_path):
        job_description_path = job_description_path.name
        job_description = self.extract_text_from_file(job_description_path)


        # Define the prompt or input for the model
        prompt = f"""Generate interview questions for screening following job_description delimitted by triple backticks. Generate atmost ten questions.
                     ```{job_description}```
                  """

        # Generate a response from the GPT-3 model
        response = openai.Completion.create(
            engine='text-davinci-003',  # Choose the GPT-3 engine you want to use
            prompt=prompt,
            max_tokens=200,  # Set the maximum number of tokens in the generated response
            temperature=0,  # Controls the randomness of the output. Higher values = more random, lower values = more focused
            n=1,  # Generate a single response
            stop=None,  # Specify an optional stop sequence to limit the length of the response
        )

        # Extract the generated text from the API response
        generated_text = response.choices[0].text.strip()

        return generated_text

    def gradio_interface(self):
      with gr.Blocks(css="style.css",theme=gr.themes.Soft()) as app:
            gr.HTML("""<img class="leftimage" align="left" src="https://templates.images.credential.net/1612472097627370951721412474196.png" alt="Image" width="210" height="210">
                       <img class="rightimage" align="right" src="https://companieslogo.com/img/orig/RAND.AS_BIG-0f1935a4.png?t=1651813778" alt="Image" width="210" height="210">""")

            with gr.Row(elem_id="col-container"):
              with gr.Column():
                gr.HTML("<br>")
                gr.HTML(
                    """<h1 style="text-align:center; color:"white">Randstad Questions For Screening</h1> """
                )
              with gr.Column():
                jobDescription = gr.File(label="Job Description",elem_classes="heightfit")

              with gr.Column():
                analyse = gr.Button("Generate")

              with gr.Column():
                result = gr.Textbox(label="Questions For Screening",lines=8)

            analyse.click(self.response, [jobDescription], result)

      app.launch()

ques = QuestionsGenerator()
ques.gradio_interface()