Chat-PDF

Chat-PDF is an application designed to facilitate PDF uploads and answer questions related to them.

Frontend Code

For the frontend React code, please refer here.

Installation (Backend)

Prerequisites

• Python: Install Python from its original site.

Setup Instructions

• Fork the Repository Fork the repository into your own GitHub account.

• Clone your newly forked repository from GitHub onto your local computer.

Setup local environment

1. Run python -m venv .venv to create a virtual environment.
2. Download the dependencies mentioned in the requirements.txt file.

Get OpenAI Key

• Obtain your own OpenAI key from here.

Set Up Environment Variables

1. Create a .env file.
2. Set up your OpenAI key within it.

Run the Application

• Run the command uvicorn main:app --reload to start the application.
• Navigate to http://127.0.0.1:8000/docs to test the APIs.

APIs

Our application offers three APIs:

1. PDF Upload API: This API accepts a PDF file and sets up chains to answer questions related to the PDF content.

2. Question Answering API: With this API, you can submit a question, and utilizing the language-based knowledge chains established earlier which returns a suitable answer extracted from your PDF.

3. PDF Retrieval API: This API allows you to retrieve all the PDFs that have been uploaded. ( Note : this can be customized with each user and their pdfs but authentication is not the scope of this project. )

Basic Architecture

File Handling

• API accepts a file and validates if it's a PDF. If not, it returns a 400 error.

File Processing

• The PDF file is read, and its binary content is converted into bytes using IO.

Database Interaction

• Filename and filesize are stored in a PostgreSQL database for future retrieval.

Text Extraction

• The content of pdf is extracted using FileReader from pypdf.

Chunking Text

• The extracted text is divided into smaller chunks for efficient processing.

Embedding Setup

• Embeddings are created from these chunks, establishing a chain to track the conversation.

Semantic Search

• Semantic search is performed based on the user's question.

Answer Retrieval

• Using OpenAI's language model, an appropriate answer is retrieved based on the semantic search results.