Options
All
  • Public
  • Public/Protected
  • All
Menu

Package monaco-graphql

Changelog | API Docs | Discord Channel

GraphQL language plugin for the Monaco Editor. You can use it to build vscode/codespaces-like web or desktop IDEs using whatever frontend javascript libraries or frameworks you want, or none!

  • webpack example using plain javascript, shows how to change schemas
  • vite + react example - minimal example with variables (C)JSON support using react
  • live demo of the monaco webpack example (prompts for GitHub access token!)

NOTE: This is in pre-release state as we build towards GraphiQL 2.0.x. codemirror-graphql has more features (such as JSON variables validation) and is more stable.

Features

It provides the following features while editing GraphQL files:

  • Configurable multi-model, multi-schema language worker with fileMatch expressions
  • Code completion (schema driven) for Operation and SDL types
    • Automatic expansion & completion on leaf type completion
  • Hover (schema driven) with markdown support
  • Validation (schema driven)
  • JSON Variables validation and language features (schema driven)
  • Formatting - using prettier
  • Syntax Highlighting & Basic Languages provided by monaco-editor basic languages support
  • Configurable formatting options
  • Providing external fragments
  • create custom workers for custom language service options - parser, validation rules, schemaBuilder, etc

Usage

For now, we use language id of graphql until we can ensure we can dovetail nicely with the official graphql language ID.

To use with webpack, here is an example to get you started:

yarn add monaco-graphql

Sync Example

import * as monaco from 'monaco-editor/esm/vs/editor/editor.api';

import { initializeMode } from 'monaco-graphql/esm/initializeMode';

// you can also configure these using the webpack or vite plugins for `monaco-editor`
import GraphQLWorker from 'worker-loader!monaco-graphql/esm/graphql.worker';

// instantiates the worker & language features with the schema!
const MonacoGraphQLAPI = initializeMode({
  schemas: [
    {
      schema: myGraphqlSchema as GraphQLSchema,
      // anything that monaco.URI.from() is compatible with
      uri: 'https://myschema.com',
      uri: '/myschema.graphql',
      // match the monaco file uris for this schema.
      // accepts specific uris and anything `picomatch` supports.
      // (everything except bracket regular expressions)
      fileMatch: ['**/*.graphql'],
      // note: not sure if ^ works if the graphql model is using urls for uris?
    },
  ],
});

window.MonacoEnvironment = {
  getWorker(_workerId: string, label: string) {
    if (label === 'graphql') {
      return new GraphQLWorker();
    }
    // if you are using vite or webpack plugin, it will be found here
    return new Worker('editor.worker.js');
  },
};
monaco.editor.create(document.getElementById('someElementId'), {
  value: 'query { }',
  language: 'graphql',
  formatOnPaste: true,
});

Lazy Example

The existing API works as before in terms of instantiating the schema

import * as monaco from 'monaco-editor/esm/vs/editor/editor.api';
// enables our language worker right away, despite no schema
import 'monaco-graphql';

// you can also configure these using the webpack or vite plugins for `monaco-editor`
import GraphQLWorker from 'worker-loader!monaco-graphql/esm/graphql.worker';

// lazily invoke the api config methods whenever we want!
monaco.languages.graphql.setSchemaConfig([
  {
    schema: myGraphqlSchema as GraphQLSchema,
    // anything that monaco.URI.from() is compatible with
    uri: 'https://myschema.com',
    uri: '/myschema.graphql',
    // match the monaco file uris for this schema.
    // accepts specific uris and anything `picomatch` supports.
    // (everything except bracket regular expressions)
    fileMatch: ['**/*.graphql'],
    // note: not sure if ^ works if the graphql model is using urls for uris?
  },
]);

window.MonacoEnvironment = {
  getWorker(_workerId: string, label: string) {
    if (label === 'graphql') {
      return new GraphQLWorker();
    }
    return new Worker('editor.worker.js');
  },
};
monaco.editor.create(document.getElementById('someElementId'), {
  value: 'query { }',
  language: 'graphql',
  formatOnPaste: true,
});

This will cover the basics, making an HTTP POST with the default introspectionQuery() operation. To customize the entire fetcher, see advanced customization below. For more customization options, see the Monaco Editor API Docs

Advanced Usage

Variables JSON Support!

In monaco-graphql@0.5.0 we introduced a method getVariablesJSONSchema that allows you to retrieve a JSONSchema description for the declared variables for any given set of operations

Full Sync Demo with Variables JSON

import * as monaco from 'monaco-editor/esm/vs/editor/editor.api';

import { initializeMode } from 'monaco-graphql/esm/initializeMode';

import GraphQLWorker from 'worker-loader!monaco-graphql/esm/graphql.worker';

window.MonacoEnvironment = {
  getWorker(_workerId: string, label: string) {
    if (label === 'graphql') {
      return new GraphQLWorker();
    }
    return new Worker('editor.worker.js');
  },
};

// the language service will be instantiated once the schema is available
const MonacoGraphQLAPI = initializeMode({
  schemas: [
    {
      // anything that monaco.URI.from() is compatible with
      uri: 'https://myschema.com',
      // match the monaco file uris for this schema.
      // accepts specific filenames and anything `picomatch` supports.
      fileMatch: ['**/*.graphql'],
      schema: myGraphqlSchema as GraphQLSchema,
    },
  ],
});

const operationModel = monaco.editor.createModel(
  'query {}',
  'graphql',
  '/operation.graphql',
);

const operationEditor = monaco.editor.create(
  document.getElementById('someElementId'),
  {
    model: operationModel,
    language: 'graphql',
    formatOnPaste: true,
  },
);

const variablesSchemaUri = monaco.editor.URI.file('/variables-schema.json');

const variablesModel = monaco.editor.createModel(
  '{}',
  'json',
  '/variables.json',
);

const variablesEditor = monaco.editor.create(
  document.getElementById('someElementId'),
  {
    model: variablesModel,
    language: 'graphql',
    formatOnPaste: true,
  },
);

// high-level method for configuring json variables validation
MonacoGraphQLAPI.setDiagnosticSettings({
  validateVariablesJson: {
    // Urls, uris, anything that monaco.URI.from() is compatible with.
    // Match operation model to variables editor,
    // and the language service will automatically listen for changes,
    // and compute the json schema using the GraphQLWorker.
    // This is in the main process is applied to the global monaco json settings
    // for validation, completion and more using monaco-json's built-in JSON Schema support.
    [operationModel.uri.toString()]: [variablesModel.uri.toString()],
  },
  jsonDiagnosticSettings: {
    allowComments: true, // allow json, parse with a jsonc parser to make requests
  },
});
// TODO: document manual alternative approach

You can also experiment with the built-in I think jsonc? (MSFT json5-ish syntax, for tsconfig.json etc.) and the 3rd party monaco-yaml language modes for completion of other types of variable input. you can also experiment with editor methods to parse detected input into different formats, etc (yaml pastes as json, etc.)

You could of course prefer to generate a jsonschema form for variables input using a framework of your choice, instead of an editor. Enjoy!

MonacoGraphQLAPI (typedoc)

If you call any of these API methods to modify the language service configuration at any point at runtime, the webworker will reload relevant language features.

If you import 'monaco-graphql' synchronously, you can access the api via monaco.languages.graphql.api.

import 'monaco-graphql';
// now the api will be available on the `monaco.languages` global
monaco.languages.graphql.api;
import 'monaco-graphql';

// also this
import { languages } from 'monaco-editor';
// now the api will be available on the `monaco.languages` global
languages.graphql.api;

Otherwise, you can, like in the sync demo above:

import { initializeMode } from 'monaco-graphql/esm/initializeMode';
const api = initializeMode(config);

monaco.languages.graphql.api.setSchemaConfig([SchemaConfig])

same as the above, except it overwrites the entire schema config.

you can provide multiple, and use fileMatch to map to various uri "directory" globs or specific files. uri can be an url or file path, anything parsable

// you can load it lazily
import 'monaco-graphql';
monaco.languages.graphql.api.setSchemaConfig([
  {
    schema: GraphQLSchema,
    fileMatch: ['**/*.graphql'],
    uri: 'myschema.graphql',
  },
]);

or you can load the language features only when you have your schema

import { initializeMode } from 'monaco-graphql/esm/initializeMode';

const schemas = [
  {
    schema: GraphQLSchema,
    fileMatch: ['operations/*.graphql'],
    uri: 'myschema.graphql',
  },
];
const api = initializeMode({ schemas });

// add another schema. this will cause language workers and features to reset
api.setSchemaConfig([
  ...schemas,
  {
    introspectionJSON: myIntrospectionJSON,
    fileMatch: ['specific/monaco/uri.graphql'],
    uri: 'another-schema.graphql',
  },
]);

or if you want, replace the entire configuration with a single schema. this will cause the worker to be entirely re-created and language services reset

api.setSchemaConfig([
  {
    introspectionJSON: myIntrospectionJSON,
    fileMatch: ['**/*.graphql'],
    uri: 'myschema.graphql',
  },
]);

monaco.languages.graphql.api.setModeConfiguration()

This is where you can toggle monaco language features. all are enabled by default.

monaco.languages.graphql.api.setModeConfiguration({
  documentFormattingEdits: true,
  completionItems: true,
  hovers: true,
  documentSymbols: true,
  diagnostics: true,
});

monaco.languages.graphql.api.setFormattingOptions()

this accepts an object { prettierConfig: prettier.Options }, which accepts any prettier option. it will not re-load the schema or language features, however the new prettier options will take effect.

this method overwrites the previous configuration, and will only accept static values that can be passed between the main/worker process boundary.

monaco.languages.graphql.api.setFormattingOptions({
  // if you wanna be like that
  prettierOptions: { tabWidth: 2, useTabs: true },
});

monaco.languages.graphql.api.setExternalFragmentDefinitions()

Append external fragments to be used by autocomplete and other language features.

This accepts either a string that contains fragment definitions, or TypeDefinitionNode[]

monaco.languages.graphql.api.getDiagnosticOptions

monaco.languages.graphql.api.setDiagnosticSettings({
  validateVariablesJson: {
    // Urls, uris, anything that monaco.URI.from() is compatible with.
    // Match operation model to variables editor,
    // and the language service will automatically listen for changes,
    // and compute the json schema using the GraphQLWorker.
    // This is in the main process is applied to the global monaco json settings
    // for validation, completion and more using monaco-json's built-in JSON Schema support.
    [operationModel.uri.toString()]: [variablesModel.uri.toString()],
  },
  jsonDiagnosticSettings: {
    allowComments: true, // allow json, parse with a jsonc parser to make requests
  },
});

Bundlers

Webpack

you'll can refer to the webpack configuration in the full monaco webpack example to see how it works with webpack and the official monaco-editor-webpack-plugin. there is probably an easier way to configure webpack worker-loader for this.

Vite

You can configure vite to load monaco-editor json mode and even the language editor worker using the example for our mode

Web Frameworks

the plain javascript webpack example should give you a starting point to see how to implement it with

React

  • use-monaco seems to support the custom language worker configuration we want, and seems to be well built! we hope to help them build their
  • when loading it yourself, either dynamic import the mode and/or instantiate it yourself using useEffect on didMount to prevent breaking SSR.
  • it may work with other libraries by using a similar strategy to this. you can also provide MonacoEnvironment.getWorkerUrl which works better as an async import of your pre-build worker files

Custom Webworker (for passing non-static config to worker)

If you want to pass a custom parser and/or validation rules, it is supported, however the setup is a bit more complicated.

You can add any LanguageServiceConfig (typedoc) configuration options you like here to languageConfig as below.

This is because we can't pass non-static configuration to the existing worker programmatically, so you must import these and build the worker custom with those functions. Part of the (worthwhile) cost of crossing runtimes!

you'll want to create your own my-graphql.worker.ts file, and add your custom config such as schemaLoader to createData:

import type { worker as WorkerNamespace } from 'monaco-editor';
// @ts-ignore
import * as worker from 'monaco-editor/esm/vs/editor/editor.worker';

import { GraphQLWorker } from 'monaco-graphql/esm/GraphQLWorker';

import { myValidationRules } from './custom';

self.onmessage = () => {
  try {
    worker.initialize(
      (
        ctx: WorkerNamespace.IWorkerContext,
        createData: monaco.languages.graphql.ICreateData,
      ) => {
        createData.languageConfig.customValidationRules = myValidationRules;
        return new GraphQLWorker(ctx, createData);
      },
    );
  } catch (err) {
    throw err;
  }
};

then, in your application:

import EditorWorker from 'worker-loader!monaco-editor/esm/vs/editor/editor.worker';

// specify the path to your language worker
import GraphQLWorker from 'worker-loader!./my-graphql.worker';

window.MonacoEnvironment = {
  getWorker(_workerId: string, label: string) {
    if (label === 'graphql') {
      return new GraphQLWorker();
    }
    return new EditorWorker();
  },
};

or, if you have webpack configured for it:

window.MonacoEnvironment = {
  getWorkerUrl(_workerId: string, label: string) {
    if (label === 'graphql') {
      return 'my-graphql.worker.js';
    }
    return 'editor.worker.js';
  },
};

with vite you just need:

import { defineConfig } from 'vite';
import monacoEditorPlugin from 'vite-plugin-monaco-editor';

export default defineConfig({
  plugins: [
    monacoEditorPlugin({
      customWorker: [
        {
          label: 'graphql',
          entry: 'my-graphql.worker.js',
        },
      ],
    }),
  ],
});

Monaco Editor Tips

If you are familiar with Codemirror/Atom-era terminology and features, here's some gotchas:

  • "hinting" => "code completion" in LSP terminology
  • "linting" => "diagnostics" in lsp terminology
  • the default keymap is different, more vscode like
  • command palette and right click context menu are important
  • you can extend the standard completion/linting/etc provided. for example, editor.setModelMarkers()
  • Monaco Editor API Docs
  • Monaco Editor Samples repository is great for tips on implementing with different bundlers, runtimes, etc.

TODO

  • variables JSON validation
  • variables completion
  • Symbols & Definitions
  • file uri-driven schema loading
  • op -> schema & schema -> schema references
  • insertText for field and argument completion

Index

References

BaseSchemaConfig

Re-exports BaseSchemaConfig

DiagnosticSettings

Re-exports DiagnosticSettings

FilePointer

Re-exports FilePointer

FormattingOptions

Re-exports FormattingOptions

GraphQLLanguageConfig

Re-exports GraphQLLanguageConfig

ICreateData

Re-exports ICreateData

IDisposable

Re-exports IDisposable

IEvent

Re-exports IEvent

JSONDiagnosticOptions

Re-exports JSONDiagnosticOptions

ModeConfiguration

Re-exports ModeConfiguration

MonacoGraphQLAPI

Re-exports MonacoGraphQLAPI

MonacoGraphQLAPIOptions

Re-exports MonacoGraphQLAPIOptions

MonacoGraphQLInitializeConfig

Re-exports MonacoGraphQLInitializeConfig

SchemaConfig

Re-exports SchemaConfig

SchemaEntry

Re-exports SchemaEntry

SchemaLoader

Re-exports SchemaLoader

diagnosticSettingDefault

Re-exports diagnosticSettingDefault

formattingDefaults

Re-exports formattingDefaults

modeConfigurationDefault

Re-exports modeConfigurationDefault

Type aliases

BaseSchemaConfig

BaseSchemaConfig: { buildSchemaOptions?: BuildSchemaOptions; documentAST?: DocumentNode; documentString?: string; introspectionJSON?: IntrospectionQuery; introspectionJSONString?: string; schema?: GraphQLSchema }

Type declaration

  • Optional buildSchemaOptions?: BuildSchemaOptions
  • Optional documentAST?: DocumentNode
  • Optional documentString?: string
  • Optional introspectionJSON?: IntrospectionQuery
  • Optional introspectionJSONString?: string
  • Optional schema?: GraphQLSchema

DiagnosticSettings

DiagnosticSettings: { jsonDiagnosticSettings?: DiagnosticsOptions; validateVariablesJSON?: Record<string, string[]> }

Type declaration

  • Optional jsonDiagnosticSettings?: DiagnosticsOptions

    the default JSONDiagnosticOptions from monaco-editor's json mode - to use when applying variablesJSON. some examples of settings to provide here:

    • allowComments: true enables jsonc editing
    • validateSchema: 'warning'
    • trailingComments is error by default, and can be warning or ignore {languages.json.DiagnosticsOptions}
  • Optional validateVariablesJSON?: Record<string, string[]>

    whilst editing operations, alongside graphql validation, generate json schema for variables to validate json schema models

    example
    validateVariablesJSON: {
      "monaco://my-operation.graphql": ["monaco://my-variables.json"]
     }

FilePointer

FilePointer: string | string[]

FormattingOptions

FormattingOptions: { prettierConfig?: PrettierConfig }

Type declaration

  • Optional prettierConfig?: PrettierConfig

GraphQLLanguageConfig

GraphQLLanguageConfig: { customValidationRules?: ValidationRule[]; externalFragmentDefinitions?: FragmentDefinitionNode[] | string; parseOptions?: ParseOptions; parser?: typeof parse; schemaLoader?: SchemaLoader; schemas?: SchemaConfig[] }

For the monaco-graphql language worker, these must be specified in a custom webworker. see the readme.

Type declaration

  • Optional customValidationRules?: ValidationRule[]

    Custom validation rules following graphql ValidationRule signature

  • Optional externalFragmentDefinitions?: FragmentDefinitionNode[] | string

    External fragments to be used with completion and validation

  • Optional parseOptions?: ParseOptions

    Custom options passed to parse, whether graphql parse by default or custom parser

  • Optional parser?: typeof parse

    Provide a parser that matches graphql parse() signature Used for internal document parsing operations for autocompletion and hover, graphql-language-service-parser is used via graphql-language-service-interface

  • Optional schemaLoader?: SchemaLoader

    Take a variety of schema inputs common for the language worker, and transform them to at least a schema if not other easily available implementations

  • Optional schemas?: SchemaConfig[]

    An array of schema configurations from which to match files for language features You can provide many formats, see the config for details!

GraphQLWorkerCompletionItem

GraphQLWorkerCompletionItem: GraphQLCompletionItem & { command?: CompletionItem["command"]; range?: IRange }

JSONDiagnosticOptions

JSONDiagnosticOptions: DiagnosticsOptions

MonacoCompletionItem

MonacoCompletionItem: CompletionItem & { deprecationReason?: string | null; isDeprecated?: boolean }

MonacoCompletionItem

MonacoCompletionItem: CompletionItem & { deprecationReason?: string | null; isDeprecated?: boolean }

MonacoGraphQLAPIOptions

MonacoGraphQLAPIOptions: { diagnosticSettings: DiagnosticSettings; formattingOptions: FormattingOptions; languageId: string; modeConfiguration: ModeConfiguration; schemas?: SchemaConfig[] }

Type declaration

MonacoGraphQLInitializeConfig

MonacoGraphQLInitializeConfig: { diagnosticSettings?: DiagnosticSettings; formattingOptions?: FormattingOptions; modeConfiguration?: ModeConfiguration; schemas?: SchemaConfig[] }

Configuration to initialize the editor with

Type declaration

  • Optional diagnosticSettings?: DiagnosticSettings
  • Optional formattingOptions?: FormattingOptions

    provide prettier formatting options as prettierConfig.<option>

    example
     initializeMode({
      formattingOptions: { prettierConfig: { useTabs: true } }
     })
  • Optional modeConfiguration?: ModeConfiguration
  • Optional schemas?: SchemaConfig[]

    Specify array of SchemaConfig items used to initialize the GraphQLWorker if available. You can also api.setSchemaConfig() after instantiating the mode.

SchemaConfig

SchemaConfig: { buildSchemaOptions?: BuildSchemaOptions; documentAST?: DocumentNode; documentString?: string; fileMatch?: string[]; introspectionJSON?: IntrospectionQuery; introspectionJSONString?: string; schema?: GraphQLSchema; uri: string }

Inspired by the monaco-json schema object in DiagnosticSettings["schemas"], which we use :)

You have many schema format options to provide, choose one!

For large schemas, try different formats to see what is most efficient for you.

Type declaration

  • Optional buildSchemaOptions?: BuildSchemaOptions

    provide custom options when using buildClientSchema, buildASTSchema, etc

  • Optional documentAST?: DocumentNode

    A GraphQL DocumentNode AST

  • Optional documentString?: string

    An SDL document string

  • Optional fileMatch?: string[]

    An array of URIs or globs to associate with this schema in the language worker Uses picomatch which supports many common expressions except brackets Only necessary if you provide more than one schema, otherwise it defaults to the sole schema

  • Optional introspectionJSON?: IntrospectionQuery

    A parsed JSON literal of the introspection results

  • Optional introspectionJSONString?: string

    A stringified introspection JSON result

  • Optional schema?: GraphQLSchema

    A GraphQLSchema instance

  • uri: string

    A unique uri string for this schema. Model data will eventually be set for this URI for definition lookup

SchemaEntry

SchemaEntry: { documentString?: string; introspectionJSONString?: string; schema: GraphQLSchema }

Type declaration

  • Optional documentString?: string
  • Optional introspectionJSONString?: string
  • schema: GraphQLSchema

SchemaLoader

SchemaLoader: (schemaConfig: SchemaConfig, parser: GraphQLLanguageConfig["parser"]) => GraphQLSchema

This schema loader is focused on performance for the monaco worker runtime We favor taking in stringified schema representations as they can be used to communicate Across the main/webworker process boundary

param
param
returns

Type declaration

    • (schemaConfig: SchemaConfig, parser: GraphQLLanguageConfig["parser"]): GraphQLSchema
    • Parameters

      • schemaConfig: SchemaConfig
      • parser: GraphQLLanguageConfig["parser"]

      Returns GraphQLSchema

Variables

Const LANGUAGE_ID

LANGUAGE_ID: "graphql" = "graphql"

Functions

create

  • Parameters

    Returns GraphQLWorker

create

Const defaultSchemaLoader

  • defaultSchemaLoader(schemaConfig: SchemaConfig, parser: parse): GraphQLSchema
  • Parameters

    Returns GraphQLSchema

Const getModelLanguageId

  • getModelLanguageId(model: ITextModel): any
  • Parameters

    • model: ITextModel

    Returns any

Const getStringSchema

  • getStringSchema(schemaConfig: SchemaConfig): { documentString: string } | { introspectionJSONString: string }
  • Send the most minimal string representation to the worker for language service instantiation

    Parameters

    Returns { documentString: string } | { introspectionJSONString: string }

initializeMode

setupMode

  • Parameters

    Returns IDisposable

toCompletion

  • Parameters

    • entry: GraphQLCompletionItem
    • Optional range: GraphQLRange

    Returns GraphQLWorkerCompletionItem

toCompletion

toCompletionItemKind

  • Parameters

    • kind: lsCompletionItemKind

    Returns CompletionItemKind

toGraphQLPosition

  • toGraphQLPosition(position: Position): GraphQLPosition
  • Parameters

    • position: Position

    Returns GraphQLPosition

toMarkerData

  • toMarkerData(diagnostic: Diagnostic): IMarkerData
  • Monaco and Vscode have slightly different ideas of marker severity. for example, vscode has Error = 1, whereas monaco has Error = 8. this takes care of that

    Parameters

    • diagnostic: Diagnostic

    Returns IMarkerData

    the matching marker severity level on monaco's terms

toMonacoRange

  • toMonacoRange(range: GraphQLRange): IRange
  • Parameters

    • range: GraphQLRange

    Returns IRange

Object literals

Const diagnosticSettingDefault

diagnosticSettingDefault: object

jsonDiagnosticSettings

jsonDiagnosticSettings: object

schemaValidation

schemaValidation: "error" = "error"

Const formattingDefaults

formattingDefaults: object

prettierConfig

prettierConfig: object

tabWidth

tabWidth: number = 2

Const modeConfigurationDefault

modeConfigurationDefault: object

colors

colors: false = false

completionItems

completionItems: true = true

diagnostics

diagnostics: true = true

documentFormattingEdits

documentFormattingEdits: true = true

documentRangeFormattingEdits

documentRangeFormattingEdits: false = false

documentSymbols

documentSymbols: false = false

foldingRanges

foldingRanges: false = false

hovers

hovers: true = true

selectionRanges

selectionRanges: false = false

tokens

tokens: false = false

Generated using TypeDoc