speccy

Make sure your OpenAPI 3.0 specifications are more than just valid, make sure they're useful!

Taking off from where Mike Ralphson started with linting in swagger2openapi, Speccy aims to become the rubocop or eslint of OpenAPI.

Requirements

NodeJS: v8 - v10

OpenAPI Specification

Currently tracking v3.0.0

If you want to run speccy on OpenAPI (f.k.a Swagger) v2.0 specs, run it through swagger2openapi first and speccy can give advice on the output.

Setup

You can install this node module via NPM or Yarn.

$ npm install speccy -g
# or
$ yarn global add speccy

Alternatively, you can use it with Docker (see "Using Docker" below.)

Usage

Usage: speccy <command>


Options:

-V, --version              output the version number
-c, --config [configFile]  config file (containing JSON/YAML). See README for potential values.
-h, --help                 output usage information


Commands:

lint [options] <file-or-url>     ensure specs are not just valid OpenAPI, but lint against specified rules
resolve [options] <file-or-url>  pull in external $ref files to create one mega-file
serve [options] <file-or-url>    view specifications in beautiful human readable documentation

Lint Command

The goal here is to sniff your files for potentially bad things. "Bad" is subjective, but you'll see validation errors, along with special rules for making your APIs better..

Usage: lint [options] <file-or-url>

ensure specs are not just valid OpenAPI, but lint against specified rules

Options:

  -q, --quiet             reduce verbosity
  -r, --rules [ruleFile]  provide multiple rules files
  -s, --skip [ruleName]   provide multiple rules to skip
  -j, --json-schema       treat $ref like JSON Schema and convert to OpenAPI Schema Objects
  -v, --verbose           increase verbosity
  -h, --help              output usage information

You'll see output such as:

#/info  R: info-contact  D: info object should contain contact object

expected Object {
  version: '5.0',
  title: 'Foo API'
} to have property contact

There are going to be different things people are interested in, so the default rules suggest things we think everyone should do; adding descriptions to parameters and operations, and having some sort of contact info.

There are strict rules which demand more contact details, "real" domains, a license, and requires tags have a description!

Resolve Command

Resolving $ref is the art of taking multiple files and squashing them all down into one big OpenAPI file. By default it will output to stdout, but you can pass -o with a file name to write the file locally.

Usage: resolve [options] <file-or-url>

pull in external $ref files to create one mega-file

Options:

  -o, --output <file>  file to output to
  -q, --quiet          reduce verbosity
  -j, --json-schema    treat $ref like JSON Schema and convert to OpenAPI Schema Objects
  -v, --verbose        increase verbosity
  -h, --help           output usage information

Starting with the fantastic resolver logic form swagger2openapi, speccy has one of the most robust resolvers out there. It avoid cyclical dependencies (when A has a property that $refs A, which in turn destroys your CPU), and all sorts of other things.

Thanks to the --json-schema switch, you can have an OpenAPI file which $refs JSON Schema files (not just OpenAPI-flavoured JSON Schema), then resolve them all into one real OpenAPI file, thanks to wework/json-schema-to-openapi-schema.

Serve Command

Using ReDoc, speccy can offer a preview of your specifications, in human-readable format. In the future we'll have speccy outlining improvements right in here, but one thing at a time.

Usage: serve [options] <file-or-url>

view specifications in beautiful human readable documentation

Options:

  -p, --port [value]  port on which the server will listen (default: 5000)
  -q, --quiet         reduce verbosity
  -j, --json-schema   treat $ref like JSON Schema and convert to OpenAPI Schema Objects
  -v, --verbose       increase verbosity
  -h, --help          output usage information

Config File

To avoid needing to send command line options and switches every time, a config file can be used. Create a speccy.yaml in the root of your project.

Example:

# Convert JSON Schema-proper to OpenAPI-flavoured Schema Objects
jsonSchema: true
# Keep the noise down
quiet: true
# Output a lot of information about what is happening (wont work if you have quiet on)
verbose: true
# Rules specific to the lint command
lint:
  # rules files to load
  rules:
  - strict
  - ./some/local/rules.yaml
  - https://example.org/my-rules.yaml
  # rules to skip
  skip:
  - info-contact
# Rules specific to the resolve command
resolve:
  output: foo.yaml
# Rules specific to the serve command
serve:
  port: 8001

Calling Speccy from Code

Not just a command line tool, speccy can be used to normalize machine-readable specifications.

The loader object will return a promise that resolves to an object containing the specification. For example:

const loader = require('speccy/lib/loader');

const options = {
  resolve: true,   // Resolve external references
  jsonSchema: true // Treat $ref like JSON Schema and convert to OpenAPI Schema Objects
};

loader
  .loadSpec('path/to/my/spec', options)            // Load the spec...
  .then(spec => console.log(JSON.stringify(spec))); // ...and print it out.

If options.resolve is truthy, speccy will resolve external references.

Using Docker

To use Speccy without installing any node/npm specific dependencies, you can run it via docker:

docker run wework/speccy lint https://raw.githubusercontent.com/OAI/OpenAPI-Specification/master/examples/v3.0/petstore.yaml

You can work with local files by mounting your spec and any config files to the /project directory when you run the container:

docker run \
  -v openapi.yaml:/project/openapi.yaml \
  wework/speccy lint openapi.yaml

Also, if you want to use a specific version, you can reference tags, so docker run wework/speccy could be docker run wework/speccy:0, docker run wework/speccy:0.8 or docker run wework/speccy:0.8.7. These versions follow semantic versioning.

Using with lint-staged

To lint your specifications before committing them you can use lint-staged to run speccy before each commit. Just install lint-staged and husky as devDependencies and add the following to your package.json:

"husky": {
  "hooks": {
    "pre-commit": "lint-staged"
  }
},
"lint-staged": {
  "*.{yml, yaml}": ["speccy lint openapi.yml", "git add"]
}

You can of course adjust the file filter and the speccy command to fit your setup.

Tests

To run the test-suite:

npm test

Contributing

Contributions are always welcome, no matter how large or small. Before contributing, please read the code of conduct.

Credits

Mike Ralphson for providing the initial linter/validator from swagger2openapi
Roman Gotsiy for the excellent ReDoc
Kenta Mori for providing the serve logic from redocup
All Contributors

License

MIT except the openapi-3.0.json schema, which is taken from the OpenAPI-Specification and the alternative gnostic-3.0.json schema, which is originally from Google Gnostic. Both of these are licensed under the Apache-2 license.

bcrowe/speccy