Apollo GraphQL code generator

This is a tool to generate API code or type annotations based on a GraphQL schema and query documents.

It currently generates Swift code, TypeScript annotations, Flow annotations, and Scala code, we hope to add support for other targets in the future.

See Apollo iOS for details on the mapping from GraphQL results to Swift types, as well as runtime support for executing queries and mutations. For Scala, see React Apollo Scala.js for details on how to use generated Scala code in a Scala.js app with Apollo Client.

Usage

If you want to experiment with the tool, you can install the apollo-codegen command globally:

npm install -g apollo-codegen

introspect-schema

The purpose of this command is to create a JSON introspection dump file for a given graphql schema. The input schema can be fetched from a remote graphql server or from a local file. The resulting JSON introspection dump file is needed as input to the generate command.

To download a GraphQL schema by sending an introspection query to a server:

apollo-codegen introspect-schema http://localhost:8080/graphql --output schema.json

You can use the header option to add additional HTTP headers to the request. For example, to include an authentication token, use --header "Authorization: Bearer <token>".

You can use the insecure option to ignore any SSL errors (for example if the server is running with self-signed certificate).

Note: The command for downloading an introspection query was named download-schema but it was renamed to introspect-schema in order to have a single command for introspecting local or remote schemas. The old name download-schema is still available is an alias for backward compatibility.

To generate a GraphQL schema introspection JSON from a local GraphQL schema:

apollo-codegen introspect-schema schema.graphql --output schema.json

generate

The purpose of this command is to generate types for query and mutation operations made against the schema (it will not generate types for the schema itself).

This tool will generate Swift code by default from a set of query definitions in .graphql files:

apollo-codegen generate **/*.graphql --schema schema.json --output API.swift

You can also generate type annotations for TypeScript, Flow, or Scala using the --target option:

# TypeScript
apollo-codegen generate **/*.graphql --schema schema.json --target typescript --output operation-result-types.ts
# Flow
apollo-codegen generate **/*.graphql --schema schema.json --target flow --output operation-result-types.flow.js
# Scala
apollo-codegen generate **/*.graphql --schema schema.json --target scala --output operation-result-types.scala

For the typescript and flow targets, you can choose to output to a directory and Apollo Codegen will split up the types by the queries they are used in.

gql template support

If the source file for generation is a javascript or typescript file, the codegen will try to extrapolate the queries inside the gql tag templates.

The tag name is configurable using the CLI --tag-name option.

.graphqlconfig support

Instead of using the --schema option to point out you GraphQL schema, you can specify it in a .graphqlconfig file.

In case you specify multiple schemas in your .graphqlconfig file, choose which one to pick by using the --project-name option.

Typescript and Flow

When using apollo-codegen with Typescript or Flow, make sure to add the __typename introspection field to every selection set within your graphql operations.

If you're using a client like apollo-client that does this automatically for your GraphQL operations, pass in the --add-typename option to apollo-codegen to make sure the generated Typescript and Flow types have the __typename field as well. This is required to ensure proper type generation support for GraphQLUnionType and GraphQLInterfaceType fields.

Why is the __typename field required?

Using the type information from the GraphQL schema, we can infer the possible types for fields. However, in the case of a GraphQLUnionType or GraphQLInterfaceType, there are multiple types that are possible for that field. This is best modeled using a disjoint union with the __typename as the discriminant.

For example, given a schema:

...

interface Character {
  name: String!
}

type Human implements Character {
  homePlanet: String
}

type Droid implements Character {
  primaryFunction: String
}

...

Whenever a field of type Character is encountered, it could be either a Human or Droid. Human and Droid objects will have a different set of fields. Within your application code, when interacting with a Character you'll want to make sure to handle both of these cases.

Given this query:

query Characters {
  characters(episode: NEW_HOPE) {
    name

    ... on Human {
      homePlanet
    }

    ... on Droid {
      primaryFunction
    }
  }
}

Apollo Codegen will generate a union type for Character.

export type CharactersQuery = {
  characters: Array<{
    __typename: 'Human',
    name: string,
    homePlanet: ?string
  } | {
    __typename: 'Droid',
    name: string,
    primaryFunction: ?string
  }>
}

This type can then be used as follows to ensure that all possible types are handled:

function CharacterFigures({ characters }: CharactersQuery) {
  return characters.map(character => {
    switch(character.__typename) {
      case "Human":
        return <HumanFigure homePlanet={character.homePlanet} name={character.name} />
      case "Droid":
        return <DroidFigure primaryFunction={character.primaryFunction} name={character.name} />
    }
  });
}

Contributing

This repo is composed of multiple packages managed by Lerna. The apollo-codegen-core package contains all the compiler APIs needed to implement support for new languages. The apollo-codegen-cli package contains the final CLI, which combines together the remaining apollo-codegen-* packages that implement language specific code generation.

Running tests locally:

npm install
npm test

You can also run npm commands within package folders after you have bootstrapped the repository (part of npm install).