CHAT WITH PDF

Project Overview

This project is a FastAPI application integrated with Ollama, PGVectorDB, and Redis, designed to interact with PDF documents and provide AI-generated responses. It includes several endpoints for uploading PDFs and querying information from them, as well as health check mechanisms and rate limiting to ensure stability and performance.

Setup Instructions

Prerequisites

• Docker and Docker Compose installed on your machine.
• Python 3.10.15 or higher installed on your machine.
• A GPU (optional) for enhanced performance.

Environment Configuration

1. Clone the repository:

   git clone https://github.com/huseyindas/chat-with-pdf.git
   cd chat-with-pdf

2. Copy the example environment file:

   • Rename .env.example to .env:

   cp .env.example .env

   Edit .env as necessary to configure your environment settings.

3. Activate the virtual environment:

   • If you have Pipenv installed, you can activate the virtual environment with:

   pipenv shell

   • If you do not have Pipenv, you can install the required packages using requirements.txt:

   pip install -r requirements.txt

Starting the Application

To start the FastAPI application with Docker Compose, you can use the provided start.sh script:

1. Make the script executable:

   chmod +x start.sh

2. Run the application:
   • To build the containers:

   ./start.sh build

   • To run without rebuilding:

   ./start.sh

   The script will check for GPU support and run the appropriate Docker Compose configuration (docker-compose.yml for CPU or docker-compose.gpu.yml for GPU).

API Endpoints

1. Chat with PDF

* Request Method: POST
* Endpoint: /v1/chat/{pdf_id}
* Description: Interact with a specific PDF.

* Input:
    {
    "message": "What is the main topic of this PDF?"
    }

* Output:
    {
    "response": "The main topic of this PDF is <topic>."
    }

2. Upload PDF

* Request Method: POST
* Endpoint: /v1/pdf
* Description: Upload and register a PDF.
* Input: Multipart form data containing the PDF file.

* Example:
    curl -X POST "http://localhost:8000/v1/pdf" \
    -F "file=@/path/to/your/pdf/file.pdf"

* Output:
    {
    "pdf_id": "unique_pdf_identifier"
    }

3. Health Check

* Request Method: GET
* Endpoint: /v1/health
* Description: Check the health status of the application.

Rate Limiting

Upload PDF and chat endpoints are rate-limited to 3 requests per minute. Health check endpoint is limited to 1 request per second.

Middleware

The application includes several middleware components:

CustomTimeoutHandlingMiddleware: Handles a 120-second timeout. CustomErrorHandlingMiddleware: Manages exceptions within the application. CustomHealthCheckMiddleware: Ensures Ollama, PGVector, and Redis are running before processing requests.

Documentation

Access the API documentation at /docs (Swagger UI).

• AI and LangChain codes can be found in src/ai.
• Endpoint implementations are located in src/pdf and src/chat.
• Core operations, including middleware, rate limiting, logging, Redis, and database files, are in src/core.

Testing Procedures

Running Tests

• To run the test suite, ensure that Ollama, PGVector, and Redis are running in Docker or on your system.

• Check the environment file: Ensure that .env.pytest has the necessary configurations.

• The run_tests.sh script exports the environment variables from .env.pytest and runs the test suite using pytest.

Run the tests:

sh run_tests.sh

Conclusion

This FastAPI application provides a robust solution for interacting with PDF documents using AI technologies. Follow the setup instructions and utilize the API endpoints to leverage its capabilities effectively.