example-project-structure

README

example-project-structure

A structural reference for a small Go HTTP/RPC service backed by PostgreSQL. The goal is to show one opinionated layout for a production-style service — not to ship a real bookstore.

The service exposes two API styles from a single binary:

• ConnectRPC (over HTTP/2 with h2c) for health.v1.HealthService
• OpenAPI 3.1.1 / REST for the bookstore domain (/books, /books/{id})

The domain is deliberately minimal (book / author / genre) — just enough to show joins and type conversions without drowning the structure in features.

Documentation

• Repository layout — directory tree and what lives where
• Design decisions — why Service (not Repository), package-by-feature, transport-agnostic wiring, and the layering rules that follow
• Server lifecycle and graceful shutdown — signal handling, the conc pool pattern, ordered teardown with errors.Join, and how the sequence scales as resources are added
• Getting started — full prerequisites, oapi-codegen-exp rationale, test layering, codegen targets
• Adding a new entity — step-by-step walkthrough
• Extending this layout — where observability and other cross-cutting concerns slot in

Quickstart

You'll need Go 1.25+, just, and Docker (or any reachable Postgres). For the full prerequisite list and rationale, see Getting started.

1. Start Postgres

just db-up    # Postgres 17 on localhost:5433
just db-down  # tear it down

Migrations are embedded via //go:embed and applied automatically by pgdb.NewPostgres on startup.

2. Run the server

just run

That executes go run ./cmd/bookstore server --config etc/bookstore.yaml. Migrations apply on first start, so the tables exist after the server is up.

3. Seed sample data

just seed

The seed is idempotent (ON CONFLICT DO NOTHING).

4. Hit the endpoints

ConnectRPC health check via buf curl:

buf curl \
  --schema api/health/v1/health.proto \
  --protocol connect \
  --http2-prior-knowledge \
  http://localhost:8080/health.v1.HealthService/Check

REST endpoints (matched against the seed data):

# Fetch a specific book with hydrated author + genre
curl http://localhost:8080/books/2942867b-3a18-4cae-99e1-2f2816110b07

# Substring search on title
curl 'http://localhost:8080/books?query=go'

# Filter by genre slug
curl 'http://localhost:8080/books?genre=fantasy&limit=10'

5. Run the tests

just test

DB-backed tests boot a throwaway Postgres via testcontainers-go and auto-skip if Docker isn't reachable. See Getting started for what each test file demonstrates.

6. Regenerate code

just generate          # everything
just generate-api      # buf generate (protobuf + Connect)
just generate-openapi  # oapi-codegen
just generate-sql      # sqlc generate