CARP Gardener

CARP Gardener collects wearable activity and health data from third-party services and optionally transforms the payloads before publishing them. The repository contains the reusable framework plus a ready-to-run CARP-oriented service.

Modules

• gardener-core (:core): framework for OAuth-based authorization, data collection, and transformer registration. See docs/core_readme.md for architecture and extension points.
• gardener-caws-implementation (:caws-implementation): Ktor service that wires the core to PostgreSQL, RabbitMQ, and CARP payload transformers. See docs/impl_readme.md for deployment details and docs/ktor-routing.md for HTTP endpoints.

Supported data sources

• Fitbit (OAuth2)
• Garmin (OAuth1)
• Withings (OAuth2)
• Dexcom (OAuth2)

Prerequisites

• JDK 21 (the Gradle wrapper configures toolchains automatically).
• Docker + Docker Compose if you want to run the service in a container.
• The carp-webservices stack running: Gardener shares its PostgreSQL and RabbitMQ (containers carp-postgres-<profile> and carp-rabbitmq-<profile> on network carp-<profile>).
• OAuth credentials for the vendors you plan to enable (fill them into gardener-caws-implementation/.env).

Build and test

Run the Gradle wrapper from the repository root:

./gradlew build                # builds everything and runs tests
./gradlew :core:test           # framework-only tests
./gradlew :caws-implementation:test

Run the CARP implementation

• With Docker Compose

  1. Create gardener-caws-implementation/.env (gitignored) with your secrets. Config keys map to variables by upper-casing and replacing . with __, e.g. fitbit.client.id → FITBIT__CLIENT__ID.
  2. Start the stack:
     docker compose -f gardener-caws-implementation/docker/docker-compose.yml --env-file gardener-caws-implementation/.env up --build
     This starts the service on port 8444. It connects to the shared PostgreSQL and RabbitMQ of the carp-webservices stack, which must already be running; POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD, RABBITMQ_USER and RABBITMQ_PASSWORD in .env select the database and credentials (defaults gardener and admin / admin). Gardener's tables, exchange and queue are created on first start if missing.

• Locally with Gradle
  Ensure the carp-webservices stack is running (its PostgreSQL and RabbitMQ are published on localhost). Keep the .env file in gardener-caws-implementation/ for overrides. Then run:
  cd gardener-caws-implementation && profile=local ../gradlew :caws-implementation:run
  Configuration is loaded from conf/config-<profile>.json with environment variables (or .env) overriding the defaults.

Documentation

• docs/core_readme.md – core framework architecture and extension hooks.
• docs/impl_readme.md – deployment architecture and RabbitMQ consumer helper.
• docs/ktor-routing.md – HTTP endpoints exposed by the CARP implementation.
• gardener-caws-implementation/src/main/resources/db/postgres-schema.sql – database schema, applied automatically on startup.

License

MIT License. See LICENSE for details.