Skip to content

Latest commit

 

History

History
180 lines (133 loc) · 6.39 KB

configuration.md

File metadata and controls

180 lines (133 loc) · 6.39 KB

Configuration Reference

This document lists all environment variables used by Podcast RAG. You can set these via:

  • A .env file in the project root
  • A secrets manager like Doppler, 1Password, or HashiCorp Vault
  • Shell environment variables
  • Container orchestration (Docker Compose, Kubernetes, Cloud Run)

Doppler Integration

When using Doppler with Cloud Run, you can sync all secrets as a single JSON environment variable named ENV. The application automatically parses this JSON blob and loads the values:

# Doppler syncs secrets as:
ENV='{"GEMINI_API_KEY":"xxx","DATABASE_URL":"postgresql://..."}'

Existing environment variables take precedence over values in the ENV blob, allowing local overrides.

Required Variables

These must be set for the application to function:

Variable Description
GEMINI_API_KEY Google Gemini API key for AI features. Get one at Google AI Studio
PODCAST_DOWNLOAD_DIRECTORY Directory for podcast audio files (e.g., /opt/podcasts)

Gemini Configuration

Variable Default Description
GEMINI_API_KEY — Required. Google Gemini API key
GEMINI_MODEL gemini-2.5-flash Model for AI features (gemini-2.5-flash, gemini-2.5-pro)
GEMINI_FILE_SEARCH_STORE_NAME podcast-transcripts Name of the Gemini File Search store

Whisper Transcription

Variable Default Description
WHISPER_MODEL medium Whisper model size: tiny, base, small, medium, large-v3
WHISPER_DEVICE cuda Device for inference: cuda (GPU) or cpu
WHISPER_COMPUTE_TYPE float16 Compute precision: float16 (GPU), int8 (CPU), float32

Note: The medium model offers the best balance of speed and accuracy. Use large-v3 for maximum accuracy (requires ~10GB VRAM).

Database

Variable Default Description
DATABASE_URL sqlite:///./podcast_rag.db SQLAlchemy database URL
DB_POOL_SIZE 3 Connection pool size (PostgreSQL only)
DB_MAX_OVERFLOW 2 Max overflow connections (PostgreSQL only)
DB_POOL_PRE_PING true Test connections before use
DB_ECHO false Log SQL statements (for debugging)

Examples:

# SQLite (local development)
DATABASE_URL=sqlite:///./podcast_rag.db

# PostgreSQL
DATABASE_URL=postgresql://user:password@localhost:5432/podcast_rag

# Supabase (use transaction pooler port 6543)
DATABASE_URL=postgresql://postgres.[PROJECT-REF]:[PASSWORD]@aws-0-[REGION].pooler.supabase.com:6543/postgres

Supabase (Optional)

For future Supabase-specific features:

Variable Default Description
SUPABASE_URL — Supabase project URL
SUPABASE_ANON_KEY — Supabase anonymous key
SUPABASE_SERVICE_ROLE_KEY — Supabase service role key (admin access)

Email (Resend)

Variable Default Description
RESEND_API_KEY — Resend API key. Leave blank to disable email. Get one at resend.com
RESEND_FROM_EMAIL podcast@podcasts.hutchison.org Sender email address (must be verified in Resend)
RESEND_FROM_NAME Podcast RAG Sender display name
WEB_BASE_URL — Base URL for the app (e.g., https://podcasts.example.com). Used for email links and OAuth redirect.
EMAIL_DIGEST_SEND_HOUR 8 Hour (0-23) to send daily digest emails
EMAIL_DIGEST_TIMEZONE America/Los_Angeles Timezone for digest scheduling (IANA format)

Authentication

Google OAuth

Variable Default Description
GOOGLE_CLIENT_ID — Google OAuth client ID
GOOGLE_CLIENT_SECRET — Google OAuth client secret
GOOGLE_REDIRECT_URI {WEB_BASE_URL}/auth/callback OAuth callback URL (auto-derived from WEB_BASE_URL if not set)

JWT

Variable Default Description
JWT_SECRET_KEY — Secret key for JWT signing (use a strong random value)
JWT_EXPIRATION_DAYS 7 JWT token expiration in days

Cookies

Variable Default Description
COOKIE_SECURE true Require HTTPS for cookies (set false for local dev)
COOKIE_DOMAIN — Cookie domain (leave blank for current domain)

Web Server

Variable Default Description
PORT 8080 Web server port
ALLOWED_ORIGINS * CORS allowed origins
RATE_LIMIT 10/minute API rate limit
MAX_CONVERSATION_TOKENS 200000 Max tokens for conversation history
STREAMING_DELAY 0.05 Delay between SSE chunks (seconds)

Podcast Downloads

Variable Default Description
PODCAST_DOWNLOAD_DIRECTORY /opt/podcasts Directory for podcast audio files
PODCAST_MAX_CONCURRENT_DOWNLOADS 10 Max parallel downloads
PODCAST_DOWNLOAD_RETRY_ATTEMPTS 3 Retry attempts for failed downloads
PODCAST_DOWNLOAD_TIMEOUT 300 Download timeout in seconds
PODCAST_CHUNK_SIZE 8192 Download chunk size in bytes

Docker

When running in Docker, edit the volume mounts in docker-compose.yml to set your paths:

volumes:
  - /your/podcast/path:/data/podcasts:ro

Inside the container, PODCAST_DOWNLOAD_DIRECTORY is set to /data/podcasts to match the mount point.

Quick Start Examples

Local Development (SQLite)

GEMINI_API_KEY=your_key_here
PODCAST_DOWNLOAD_DIRECTORY=/path/to/podcasts

Production (Supabase + Resend)

GEMINI_API_KEY=your_key_here
PODCAST_DOWNLOAD_DIRECTORY=/data/podcasts
DATABASE_URL=postgresql://postgres.[ref]:[pass]@aws-0-us-west-1.pooler.supabase.com:6543/postgres
RESEND_API_KEY=re_xxxxx
WEB_BASE_URL=https://podcasts.example.com
JWT_SECRET_KEY=your_secure_random_key
GOOGLE_CLIENT_ID=xxxxx.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=xxxxx
# GOOGLE_REDIRECT_URI is auto-derived from WEB_BASE_URL

Using Doppler

# Install Doppler CLI
brew install dopplerhq/cli/doppler

# Login and setup
doppler login
doppler setup

# Run with secrets injected
doppler run -- python -m src.cli podcast pipeline