app.py

import os
import zipfile
import openai
import gradio as gr
from gradio import components as grc

# Set up OpenAI API credentials
openai.api_key = "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"

# Function to extract text from PDF using OpenAI API
def extract_text_from_pdf(pdf_path):
    with open(pdf_path, "rb") as f:
        pdf_bytes = f.read()
    response = openai.Completion.create(
        engine="text-davinci-003",
        prompt=pdf_bytes.decode("utf-8"),
        max_tokens=2048,
        temperature=0.7,
        n=1,
        stop=None,
        timeout=120,
    )
    return response.choices[0].text.strip()

# Function to extract text from multiple PDFs in a ZIP archive
def extract_text_from_zip(zip_file):
    corpus = ""
    with zipfile.ZipFile(zip_file, "r") as zip_ref:
        for file_name in zip_ref.namelist():
            if file_name.endswith(".pdf"):
                extracted_text = extract_text_from_pdf(zip_ref.read(file_name))
                corpus += extracted_text + "\n"
    return corpus

# Function to split text into chunks based on maximum token length
def split_text_into_chunks(text, max_tokens=2048):
    chunks = []
    words = text.split()
    current_chunk = ""
    for word in words:
        if len(current_chunk) + len(word) <= max_tokens:
            current_chunk += word + " "
        else:
            chunks.append(current_chunk.strip())
            current_chunk = word + " "
    if current_chunk:
        chunks.append(current_chunk.strip())
    return chunks

# Function to process files and query using OpenAI API
def process_files_and_query(zip_file, query):
    # Save uploaded ZIP file
    zip_path = "uploaded.zip"
    with open(zip_path, "wb") as f:
        f.write(zip_file.read())

    # Extract text from PDFs in the ZIP archive
    corpus = extract_text_from_zip(zip_file)

    # Split the corpus into chunks
    chunks = split_text_into_chunks(corpus)

    # Perform OpenAI API query on each chunk
    responses = []
    for chunk in chunks:
        prompt = chunk + "\nQuery: " + query
        response = openai.Completion.create(
            engine="text-davinci-003",
            prompt=prompt,
            max_tokens=2048,
            temperature=0.7,
            n=1,
            stop=None,
            timeout=120,
        )
        responses.append(response.choices[0].text.strip())

    # Combine the responses into a single answer
    answer = " ".join(responses)

    return answer

# Gradio input and output interfaces
zip_file_input = grc.File(label="Upload ZIP File")
query_input = grc.Textbox(label="Enter your query")
output = grc.Textbox(label="Answer")

# Gradio interface configuration
iface = gr.Interface(fn=process_files_and_query, inputs=[zip_file_input, query_input], outputs=output, title="PDF Search", description="Upload a ZIP file containing PDFs, enter your query, and get the answer.")
iface.launch()