Skip to content

Generate GraphQL schemas from Swagger OpenAPI specs

Notifications You must be signed in to change notification settings

manifoldco/graphql-gen

Repository files navigation

version (scoped) codecov

⚛️ graphql-gen

Node client for generating crude GraphQL specs from Swagger OpenAPI.

💅 Prettifies output via Prettier.

Support

GraphQL Features
Enum
ID
Union (oneOf)
Non-nullable
Primitives (string, boolean, number)
Query 🚫
Mutation 🚫

To compare actual generated output, see the example folder.

Usage

CLI

npx @manifoldco/graphql-gen schema.yaml --output schema.graphql

# 🚀 schema.yaml -> schema.graphql [2ms]

This will save a schema.graphql file in the current folder. The CLI can accept YAML or JSON for the input file.

Generating multiple schemas

Say you have multiple schemas you need to parse. I’ve found the simplest way to do that is to use npm scripts. In your package.json, you can do something like the following:

"scripts": {
  "generate:specs": "npm run generate:specs:one && npm run generate:specs:two",
  "generate:specs:one": "npx @manifoldco/graphql-gen one.yaml -o one.graphql",
  "generate:specs:two": "npx @manifoldco/graphql-gen two.yaml -o two.graphql"
}

Rinse and repeat for more specs.

For anything more complicated, or for generating specs dynamically, you can also use the Node API (below).

CLI Options

Option Alias Default Description
--output [location] -o (stdout) Where should the output file be saved?
--swagger [version] -s 2 Which Swagger version to use. Currently only supports 2.

Node

npm i --save-dev @manifoldco/graphql-gen
const { readFileSync } = require('fs');
const graphqlGen = require('@manifoldco/graphql-gen');

const input = JSON.parse(readFileSync('spec.json', 'utf8')); // Input be any JS object (OpenAPI format)
const output = graphqlGen(input); // Outputs GraphQL schema as a string (to be parsed, or written to a file)

(OpenAPI format), and return a GraphQL schema in string format. This lets you pull from any source (a Swagger server, local files, etc.), and similarly lets put The Node API is a bit more flexible: it will only take a JS object as input the output anywhere. It even allows for some post-processing in-between if desired.

If you are working with local files, you’ll have to read/write files yourself. Also, if your specs are in YAML, you’ll have to convert them to JS objects. A library such as js-yaml does make this trivial, though! Lastly, if you’re batching large folders of specs, glob may also come in handy.

Node Options

Name Type Default Description
namespace string OpenAPI2 How should the output be namespaced? (namespacing is enforced as there’s a high chance of collision)
swagger number 2 Which Swagger version to use. Currently only supports 2.

FAQ

Why does this exist apart from yarax/swagger-to-graphql ?

That didn’t work for our Swagger 2.0 specs 🤷. While normally a PR is the best course of action, this repo exists because it was less effort to build something that works for us than rewrite a popular library (it wasn’t a quick fix).

Why aren’t Queries and Mutations generated from the OpenAPI spec?

While it’s possible to generate something from OpenAPI, it’s intentionally omitted from autogeneration here. Queries and Mutations are best left up to humans, so you can determine what developers should access, and how. Ultimately generating these will always fall short of how real humans could—and should—use your GraphQL endpoint.

Notes

GraphQL is a spec, just like OpenAPI. For this reason, automatic generation isn’t ideal long-term. This library should probably be used as a first-pass to migrate an OpenAPI endpoint to GraphQL. This can generate types, but can’t intelligently generate the best queries and mutations for your specific endpoint.

A common example of this: Swagger has a concept of format: datetime. GraphQL cares about this, but doesn’t assume the formatting. Is this UNIX time? ISO? Are there timezones? Types can be so much more descriptive than mere string or int, and GraphQL gives you the tools to declare this yourself.