Lanvas Backend

This repo contains the code for the backend of Lanvas.

Tech Stack:

Architecture

The architecture for the chat API is based on the Official LangChain guide for RAG with Chat History with a few modifications.

Since Supabase is the database and vectorstore for Lanvas, there is a Custom Retriever and ChatHistory to make the chain work.

Celery is used to handle analytics tasks that aggregate the following:

The analysis is powered by Instructor and OpenAI's GPT LLM.

Install poetry
Install dependencies using poetry
```
poetry install
```
Install and setup RabbitMQ
Install Supabase CLI
Create a new Supabase project using the dashboard
Link with the newly created Supabase project
```
supabase link --project-ref ""
```
Setup Tables and Functions with migrations in the supabase/migrations/ directory
```
supabase db push
```
Setup .env file
```
cp .env.sample .env
```
Activate virtual environment
```
poetry shell
```
Setup Celery worker
```
celery -A celery_tasks worker -l INFO
```
Start the FastAPI server
```
uvicorn main:app --reload
```
Visit the docs at http://localhost:8000/docs

There are a few unit tests setup which you can verify by running the following command:

poetry run pytest -s