A Simple Python Interface to ChatGPT for a PDF Interrogator - Hufsa Haq

Welcome to my blog! In this post, I will show you my coded version of a simple Python interface to interact with ChatGPT, allowing you to build a PDF interrogator. With this interface, you'll be able to ask questions about the contents of a PDF file and receive AI-generated responses.

Introduction

The goal of my project is to leverage the power of ChatGPT, a language model developed by OpenAI, to create an interface for extracting information from PDF files. By combining the capabilities of ChatGPT with Python and the PyPDF2 library, we can build an application that enables users to ask questions and obtain answers based on the text extracted from the PDF.

Prerequisites

Before we dive into the implementation details, make sure you have the following prerequisites:

  1. Python installed on your machine (version 3.6 or higher)
  2. Flask web framework (install using pip install flask)
  3. PyPDF2 library (install using pip install PyPDF2)
  4. An OpenAI API key for accessing ChatGPT (sign up at OpenAI and retrieve your API key)

Project Structure

Go to this link to view the code at Github : https://github.com/HufsaHaq/A-Simple-Python-Interface-to-ChatGPT

To keep things organized, let's outline the structure of the project:

  • app.py: This is the main Python file that contains the Flask application and handles the routing.
  • templates: This folder will contain the HTML templates for our web interface.
  • static: This folder will hold any static files such as CSS stylesheets.

Now that we have our project structure in place, let's proceed with the implementation.

app.py :

from flask import Flask, render_template, request, redirect, url_for
import PyPDF2
import openai

openai.api_key = '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'

app = Flask(__name__)

file_uploaded = False
text = ""
conversation_history = []

@app.route('/', methods=['GET', 'POST'])
def upload():
    global file_uploaded
    global text

    if request.method == 'POST':
        file = request.files['file']
        if file:
            # Save the uploaded file to the current directory
            file.save(file.filename)

            # Extract text from the PDF file
            text = extract_text_from_pdf(file.filename)
            file_uploaded = True

    return render_template('upload.html', file_uploaded=file_uploaded, conversation=conversation_history)

def extract_text_from_pdf(filename):
    with open(filename, 'rb') as file:
        pdf_reader = PyPDF2.PdfReader(file)
        text = ''
        for page in pdf_reader.pages:
            text += page.extract_text()
        return text

@app.route('/ask', methods=['POST'])
def ask_question():
    global conversation_history

    if request.method == 'POST':
        question = request.form['question']

        # Get a response from ChatGPT
        response = chat_with_gpt(text, question)

        # Append the user's question and AI response to the conversation history
        conversation_history.append({'user': question, 'ai': response})

    return redirect(url_for('upload'))

@app.route('/back', methods=['POST'])
def go_back():
    global file_uploaded
    global conversation_history
    file_uploaded = False
    conversation_history = []
    return redirect(url_for('upload'))

def chat_with_gpt(text, question):
    global conversation_history

    # Prepare the prompt with conversation history
    prompt = ""
    for item in conversation_history:
        prompt += f"User: {item['user']}\nAI: {item['ai']}\n"

    # Send the prompt and the user's current question to ChatGPT for a response
    prompt += f"User: {question}\nAI:"

    # Send the prompt to ChatGPT for a response
    response = openai.Completion.create(
        engine='text-davinci-003',
        prompt=prompt,
        max_tokens=100,
        n=1,
        stop=None,
        temperature=0.7
    )

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

    # Update the conversation history
    conversation_history.append({'user': question, 'ai': answer})

    return answer

########################################################################
if __name__ == '__main__':
    app.run(debug=True)

templates/upload.html:

<!DOCTYPE html>

<html>

<head>

    <title>PDF Interrogation Using CHATGPT by Hufsa Haq</title>

    <link rel="stylesheet" href="{{ url_for('static', filename='style.css') }}">

</head>

<body>

    <div class="container">

        <h1>PDF Interrogation Using CHATGPT by Hufsa Haq</h1>

        <h1>Upload a File</h1>

    </div>

        {% if file_uploaded %}

            <h2>File Uploaded Successfully</h2>

            <div id="conversation">

                {% for item in conversation %}

                <p>User: {{ item['user'] }}</p>

                <p>AI: {{ item['ai'] }}</p>

                {% endfor %}

            </div>

            <form action="/ask" method="POST">

                <label for="question">Question:</label>

                <input type="text" name="question" id="question" required>

                <br><br>

                <input type="submit" value="Ask">

            </form>

            <form action="/back" method="POST">

                <input type="hidden" name="back" value="true">

                <input type="submit" value="Back">

            </form>

        {% else %}

            <form action="/" method="POST" enctype="multipart/form-data">

                <input type="file" name="file" required>

                <br><br>

                <input type="submit" value="Upload">

            </form>

        {% endif %}

</body>

</html>

 static/style.css:

 /* Set the global font family and font size */
body {
  font-family: Arial, sans-serif;
  font-size: 16px;
}

/* Set the background color */
body {
  background-color: #add8d9;
}


/* Style the heading */
h1 {
  color: #333;
  font-size: 32px;
  margin-bottom: 20px;
}

/* Style the paragraph */
p {
  color: #000000;
  line-height: 1.5;
}

/* Style the navigation menu */
.navbar {
  background-color: #333;
  color: #fff;
}

.navbar a {
  color: #fff;
  text-decoration: none;
  padding: 10px;
}

.navbar a:hover {
  background-color: #666;
}

/* Style buttons */
.button {
  display: inline-block;
  padding: 10px 20px;
  background-color: #333;
  color: #fff;
  text-decoration: none;
  border-radius: 4px;
}

.button:hover {
  background-color: #666;
}

/* Style form inputs */
.form-input {
  display: block;
  width: 100%;
  padding: 10px;
  margin-bottom: 10px;
  border: 1px solid #ccc;
  border-radius: 4px;
}

/* Style form submit button */
.form-submit {
  display: inline-block;
  padding: 10px 20px;
  background-color: #333;
  color: #fff;
  text-decoration: none;
  border-radius: 4px;
}

.form-submit:hover {
  background-color: #666;
}

.container {
    display: flex;
    flex-direction: column;
    align-items: center;
    justify-content: center;
    height: 30vh;
    text-align: center;
}

Results:

 

 

 


Comments

Post a Comment

Popular posts from this blog

Hufsa Haq – Report on AI PDF Interrogator Research

  Literature Survey on PDF Interrogation with ChatGPT In order to extract information from documents and utilise this information for answers, researchers have looked into ChatGPT’s potential as a PDF interrogator. ChatGPT interacts with users by being given a document as background and providing answers to their inquiries or information based on that PDF. The model produces answers by extracting information from the paper. Techniques to improve ChatGPT’s document interrogation capabilities have been investigated by researchers. To enhance the model's comprehension of factual data, outside knowledge sources like structured databases or knowledge graphs can be used 1 . ChatGPT’s document interrogation performance has been enhanced through the use of reinforcement learning. The model learns to produce more accurate and detailed responses depending on the document with the use of reward models and reinforcement signals 2 . One of the difficulties in employing ChatGPT for PDF...
Read more

A Reflection On My PDF Interrogator - Hufsa Haq

Lessons from Building a PDF Interrogator and My Work Experience Undertaking the task of creating a PDF interrogator using ChatGPT has been an informative experience, unveiling the immense capabilities of technology, as well as giving me the opportunity to under a professor in a work environment. From the inception of the project to its completion, I have gained insights into the intricacies of programming and the remarkable potential of AI-powered solutions. One of the key lessons learned throughout this journey is the significance of thorough planning and analysis. From the initial stages, I learnt the importance of understanding the project's goals, the specific requirements of the PDF interrogator, and the potential challenges that may arise. This knowledge enabled me to make informed decisions and develop a robust foundation for the project. Another vital lesson was the power of continuous learning and adaptability. As I encountered obstacles and complexities, I embraced a grow...
Read more