SleekReply API

Backend for building and running SleekReply chatbots from company documentation. It is a Django 5 / DRF service with Celery for async work, PostgreSQL for persistence, Weaviate as the vector store, and OpenAI-powered embeddings and responses.

What’s Inside

• Python 3.11, Django 5, Django REST Framework, SimpleJWT
• Celery + Redis (worker queue) and Flower (monitoring)
• PostgreSQL database
• Weaviate vector database for document embeddings
• Optional AWS S3 storage (production), Whitenoise for local static files
• Google OAuth support and Pusher websocket notifications
• Sentry hooks for production observability

Project Layout

• apps/bots: Bot, flow, step, and vector DB logic plus Celery tasks
• apps/accounts: Profiles, API keys, JWT + Google OAuth endpoints
• apps/services: Weaviate and websocket service helpers
• config: Django settings, URLs, ASGI/WSGI, Celery app
• docker-compose.yml: API, Postgres, Weaviate, Redis, Celery worker, Flower
• requirements/*.txt: Base, dev, test, and production dependencies

Prerequisites

• Docker and docker-compose
• Make sure ports 8000 (API), 5432 (Postgres), 8080/50051 (Weaviate), 5555 (Flower) are free

Quick Start (Docker)

1. Copy environment defaults:

cp env.local .env

2. Build and start the stack:

docker-compose build
docker-compose up

3. Run migrations once the API container is up:

docker-compose exec api python manage.py migrate

4. (Optional) Create an admin user:

docker-compose exec api python manage.py createsuperuser

5. Services:

• API: http://localhost:8000
• Flower dashboard: http://localhost:5555
• Weaviate: http://localhost:8080

Development Workflow

• Run tests:

docker-compose exec api pytest

• Lint/format via pre-commit:

docker-compose exec api pre-commit install
docker-compose exec api pre-commit run --all-files

• Migrations:

docker-compose exec api python manage.py makemigrations
docker-compose exec api python manage.py migrate

• Django shell (with django-extensions):

docker-compose exec api python manage.py shell_plus

Environment Variables

Use env.local as a template (cp env.local .env). Key entries:

• ENVIRONMENT, DJANGO_SETTINGS_MODULE (e.g., config.settings.dev)
• DATABASE_URL, REDIS_URL
• OPENAI_API_KEY
• Weaviate: WEAVIATE_LOCAL, WEAVIATE_LOCAL_URL, WEAVIATE_LOCAL_HTTP_PORT, WEAVIATE_LOCAL_GRPC_PORT
• Auth: ACCESS_TOKEN_LIFETIME, REFRESH_TOKEN_LIFETIME, ROTATE_REFRESH_TOKENS, BLACKLIST_AFTER_ROTATION
• Storage: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_STORAGE_BUCKET_NAME, AWS_DEFAULT_REGION
• Google OAuth: GOOGLE_CLIENT_ID, GOOGLE_PROJECT_ID, GOOGLE_CLIENT_SECRET
• Pusher: PUSHER_APP_ID, PUSHER_APP_KEY, PUSHER_APP_SECRET, PUSHER_APP_CLUSTER

API Notes

• JWT endpoints: /api/auth/login, /api/auth/login/refresh/, /api/auth/logout/
• Core resources are namespaced under /api/v1/ (bots, flows, vector DB, schema/vectorial operations, profiles, API keys).

Deployment

• Fly.io deploy helper: make deploy (builds, scales app and Celery, then runs migrations).

Troubleshooting

• If migrations fail on first boot, rerun docker-compose exec api python manage.py migrate.
• Ensure OPENAI_API_KEY is set before using embedding/chat features.
• Weaviate must be up before ingestion tasks; restart the worker if it cannot reach the vector store.