Muscle Group Image Generator

HTTP API that renders an anatomical body silhouette with one or more muscle groups highlighted in arbitrary colors. Drop-in compatible with the original PHP API surface; reimplemented in Go for ~20× faster rendering, a ~10 MB static binary, and zero runtime dependencies.

Quick start

make run            # listens on :8080
curl http://localhost:8080/getMuscleGroups
curl -o out.png 'http://localhost:8080/getImage?muscleGroups=chest,triceps&color=FF0000'

Or with Docker:

make docker
docker run --rm -p 8080:8080 mig:dev

API

All endpoints are GET. The legacy camelCase paths are preserved for backward compatibility with existing clients.

Path	Description
/getMuscleGroups	JSON list of supported muscle group names
/getBaseImage	The blank silhouette PNG
/getImage	Highlight one or more groups in a single color
/getMulticolorImage	Highlight two sets of groups in two different colors
/getIndividualColorImage	Assign a distinct color to each group
/healthz	Liveness probe (returns ok)
/openapi.yaml	OpenAPI 3.1 spec
/docs	Interactive Swagger UI

Query parameters

Parameter	Where	Format	Notes
muscleGroups	/getImage, /getIndividualColorImage	comma-separated names	required
primaryMuscleGroups, secondaryMuscleGroups	/getMulticolorImage	comma-separated names	required
color, primaryColor, secondaryColor	image endpoints	hex (#FF0000 / FF0000 / F00) or RGB (255,0,0)	required where listed
colors	/getIndividualColorImage	comma-separated hex or RGB	count must equal muscleGroups count
transparentBackground	image endpoints	0 or 1 (default 0)	optional

Errors are returned as {"error": "..."} JSON with an appropriate 4xx status.

Examples

/getImage?muscleGroups=chest,triceps&color=FF0000&transparentBackground=1
/getMulticolorImage?primaryMuscleGroups=chest&secondaryMuscleGroups=triceps&primaryColor=FF0000&secondaryColor=0000FF
/getIndividualColorImage?muscleGroups=biceps,triceps,abs&colors=FF0000,00FF00,0000FF

Architecture

cmd/server/             # entry point: config, lifecycle, graceful shutdown
internal/
  assets/               # //go:embed PNG library — single binary, no disk I/O at runtime
  compositor/           # color parsing, asset library, image composition
  httpapi/              # HTTP handlers, router, middleware (logging, CORS, recover)

Design highlights

• Zero external dependencies. Standard library only — net/http with Go 1.22+ pattern-matched routing, log/slog for structured logging, image/png + image/draw for compositing.
• Assets are embedded via //go:embed and decoded once at startup. No disk I/O on the request path.
• Recoloring is O(1) per muscle group. Source PNGs are 8-bit paletted; the highlight palette entry is swapped per-request without touching pixel data, so concurrent requests share the same Pix slice safely.
• Graceful shutdown on SIGINT / SIGTERM with a 15s drain window.
• Distroless final image (gcr.io/distroless/static), runs as non-root, ~10 MB total.

Development

make test           # race detector + count=1
make cover          # coverage report
make lint           # golangci-lint
make build          # produces bin/mig
make docker         # local image

CI (.github/workflows/ci.yml) runs vet, golangci-lint, tests with race, build, and Docker build on every push and pull request.

Configuration

Env var	Default	Description
PORT	8080	TCP port to listen on

Deployment (Railway)

1. Connect this repository to a new Railway service.
2. Railway autodetects the Dockerfile.
3. Under Settings → Networking, generate a public domain and set the target port to 8080 (or whatever PORT you configure).

That's it — no Apache, no MPM, no PHP, no surprises.

Credits

Forked from MertenD/gym-api (original PHP implementation). All PNG assets © their original authors.

License

Licensed under the GNU Affero General Public License v3.0 (AGPL-3.0-or-later). If you run a modified version of this software to provide a network service, you must make the corresponding source code available to its users under the same license.