This commit adds comprehensive documentation for the new GEMINI project, detailing its architecture, tech stack, key commands, and development workflow. The documentation emphasizes the Clean Architecture principles and modular design of the application. Additionally, AGENTS.md is updated to include information about the composite parser pattern and core domain types related to aviation telegrams and weather reports, enhancing clarity on the project's structure and design choices. New Files: - GEMINI.md: Detailed project overview and architecture documentation. Updates: - AGENTS.md: Added sections on parsers and core domain types. These changes aim to improve onboarding and understanding of the project's architecture and functionality.
4.4 KiB
go-caatsm
Civil Aviation Authority Telegram Message Processor.
High-performance processing for aviation telegrams using Clean Architecture, NATS JetStream, and PostgreSQL/TimescaleDB.
Architecture
Clean Architecture with clear separation of concerns:
- cmd: application entry points
- internal/domain: core domain types
- internal/port: interfaces/contracts
- internal/app: application orchestration
- internal/adapter: parsing, mapping, DTOs
- internal/infra: NATS, PostgreSQL, config, logging, metrics, telemetry
- pkg/di: dependency injection (Wire)
Features
- JetStream ingestion with retries and DLQ support
- PostgreSQL/TimescaleDB persistence
- Structured logging (Zap)
- OpenTelemetry tracing and Prometheus metrics
- Optional AFTN protocol validation
- Batch processing and health monitoring
- Weather report parsing (METAR, SPECI, TAF)
Prerequisites
- Go 1.24+
- PostgreSQL 12+
- NATS Server with JetStream enabled
- Docker (integration tests)
Quick Start
-
Start dependencies:
docker compose -f docker-compose.dev.yml up -d postgres nats nats-box -
Configure the application:
- Copy
configs/config.dev.tomland edit as needed - Or set environment variables with the
CAATSM_prefix
- Copy
-
Run in dev mode:
make run-dev
Build and Run
- Build:
make build - Run (dev):
make run-dev - Run (prod):
make run-prod - Run (local go run):
make run-local
Configuration
Configuration loads from configs/config.{env}.toml, where {env} is GO_ENV (default: dev).
Required values:
nats.urlpostgres.urlpublisher.topic
Defaults:
nats.streamdefaults toTELEGRAMsubscription.topicdefaults totelegram.>nats.modemust bejetstreamor empty (defaults to JetStream)
Minimal example:
[nats]
url = "nats://localhost:4222"
mode = "jetstream"
stream = "TELEGRAM"
consumer = "telegram-consumer"
[publisher]
topic = "telegram.json"
[subscription]
topic = "telegram.serial"
[postgres]
url = "postgres://user:password@localhost:5432/aviation?sslmode=disable"
[app]
batch_size = 50
batch_timeout = "2s"
monitor_interval = "30s"
[log]
level = "info"
format = "json"
Timeouts and Ack Wait
[timeouts] is optional. To tune redelivery, set timeouts.ack_wait and/or nats.consumer_rules.ack_wait. When neither is specified the application defaults to 30s.
JetStream Notes
- JetStream is required; other modes are not supported.
- In dev/test (
GO_ENV=devorGO_ENV=test), the stream and consumer are auto-created. - In production, ensure the stream and consumer exist before starting the service.
- Configure retention and delivery behavior under
[nats.stream_limits]and[nats.consumer_rules].
AFTN Validation
AFTN validation is optional and disabled by default. Enable it with:
[aftn]
validation_enabled = true
message_gap_threshold = "2m"
enable_sequence_gap_detection = true
When enabled, invalid telegrams are logged, recorded with error details, and can be routed to a DLQ if configured.
Observability
- Metrics:
GET /metrics - Liveness:
GET /livez - Readiness:
GET /readyz
Monitoring server settings are under [monitoring]. Tracing is configured via [telemetry].
CLI
./bin/receiver listen --help
Common flags:
--nats-url--subject--stream--consumer--publisher-topic--postgres-url--log-level--replay-from--ack-wait--telemetry-enabled,--telemetry-endpoint,--telemetry-insecure
Testing
- Unit tests (Ginkgo):
make test - Integration tests (Docker):
make test-int - All tests:
make test-all - Lint:
make lint - Coverage:
make coverage
Single test example:
ginkgo -r -v --focus "Test Description" ./path/to/package
Seed Tool
cmd/seed-telegrams publishes synthetic telegrams for development and testing.
Example:
go run ./cmd/seed-telegrams \
--nats-url nats://localhost:4222 \
--jetstream \
--stream TELEGRAM \
--js-subject telegram.serial \
--count 10 \
--category mixed \
--status random
Documentation
docs/dev-guide.mddocs/prod-guide.mddocs/nats.mddocs/observability.mddocs/performance.mddocs/migrations.mddocs/secret-management.mddocs/reliability.mddocs/weather-parser.md
Contributing
See AGENTS.md for coding standards, testing expectations, and release hygiene.
License
This repository has not declared a public license yet.