vscode-textmate-languageservice

vscode-textmate-languageservice

This package is in LTS mode & the Textmate technology is competing with the tree-sitter symbolic-expression parser technology, as used in vscode-anycode.

Language service providers & APIs driven entirely by your Textmate grammar and one configuration file.

To use the API methods and tokenization / outline services, you only need a Textmate grammar. This can be from your extension or one of VS Code's built-in languages.

In order to properly generate full-blown language providers from this module, the Textmate grammar must also include the following features:

• meta declaration scopes for block level declarations
• variable assignment scopes differentiated between multiple and single
• granular keyword control tokens with begin and end scopes

Installation

npm install vscode-textmate-languageservice

Browser support:

• This package supports Webpack and ESBuild.
• If you use a bundler, you need to set crypto as a external (commonjs crypto one in webpack). This allows the library to avoid polyfilling the node:crypto module.

Advisory:

This package is stable with browser compatibility (1.1.0). But I recommend you watch out for tree-sitter native integration into vscode (issue). Maintainable & with faster retokenization, it is a Holy Grail ...

Whereas this package depends on a well-written Textmate grammar and is a band aid of sorts.

If there is native vscode support for the language, find a Tree-sitter syntax online then suggest it in an Anycode issue. Otherwise, please open an issue on the community-maintained Treesitter syntax highlighter extension and someone might deal with it.

Setup

• Language contribution and grammar contribution defined via contributes in the extension manifest (or textmate-languageservice-contributes).
• Your grammar is bundled in the extension source code and is consumable by vscode-textmate (which can load PList XML, JSON or YAML grammars).
• A configuration file is available in the extension, defaulting to ./textmate-configuration.json. You can also use textmate-languageservices property of package.json to map language ID to relative path.

Example language extension manifest - ./package.json:

{
	"name": "lua",
	"displayName": "Textmate language service for Lua",
	"description": "Lua enhanced support for Visual Studio Code",
	"version": "0.0.1",
	"publisher": "",
	"license": "",
	"engines": {
		"vscode": "^1.51.1"
	},
	"categories": [
		"Programming Languages"
	],
	"contributes": {
		"languages": [{
			"id": "lua",
			"aliases": ["Lua"],
			"extensions": [".lua", ".moon", ".luau"],
			"configuration": "./language-configuration.json"
		}],
		"grammars": [{
			"language": "lua",
			"scopeName": "source.lua",
			"path": "./syntaxes/Lua.tmLanguage.json"
		}]
	}
}

Configuration

Create a JSON file named textmate-configuration.json in the extension directory. The file accepts comments and trailing commas.

If you only want to use the document and/or tokenization services, you can skip creating the file!

Textmate configuration fields:

• assignment - optional (object)
  Collection of Textmate scope selectors for variable assignment scopes when including variable symbols:
  Properties:
  • separator: Token to separate multiple assignments (string)
  • single: Token to match single variable assignment. (string)
  • multiple: Token to match multiple variable assignment. (string)
• declarations - optional (array)
  List of Textmate scope selectors for declaration token scopes.
• dedentation - optional (array)
  List of Textmate tokens for dedented code block declarations (e.g. ELSE, ELSEIF).
  Tokens still need to be listed in indentation with the decrementing value -1.
• exclude (string) VS Code glob pattern for files to exclude from workspace symbol search.
• indentation - optional (object)
  Indentation level offset for Textmate token types (used to implement folding).
• punctuation - optional (object)
  Collection of punctuation tokens with a significant effect on syntax providers. Properties:
  • continuation: Token scope selector for line continuation (to use in region matching). (string)
• markers - optional (object)
  Stringified regular expression patterns for folding region comments.
  • start: Escaped regular expression for start region marker. (string)
  • end: Escaped regular expression for end region marker. (string) Properties:
• symbols - optional (object)
  Map of document symbol tokens to their symbol kind (vscode.SymbolKind value).

Configuration examples

Template for textmate-configuration.json file:

{
  "assignment": {
    "single": "",
    "multiple": "",
    "separator": ""
  },
  "declarations": [],
  "dedentation": [
    "keyword.control.elseif.custom",
    "keyword.control.else.custom"
  ],
  "exclude": "{.modules,.includes}/**",
  "indentation": {
    "punctuation.definition.comment.begin.custom": 1,
    "punctuation.definition.comment.end.custom": -1,
    "keyword.control.begin.custom": 1,
    "keyword.control.end.custom": -1
  },
  "punctuation": {
    "continuation": "punctuation.separator.continuation.line.custom"
  },
  "markers": {
    "start": "^\\s*#?region\\b",
    "end": "^\\s*#?end\\s?region\\b"
  },
  "symbols": {
    "keyword.control.custom": 2,
    "entity.name.function.custom": 11
  }
}

An example configuration file that targets Lua:

{
  "assignment": {
    "single": "meta.assignment.variable.single.lua",
    "multiple": "meta.assignment.variable.group.lua",
    "separator": "punctuation.separator.comma.lua"
  },
  "declarations": [
    "meta.declaration.lua entity.name",
    "meta.assignment.definition.lua entity.name"
  ],
  "dedentation": [
    "keyword.control.elseif.lua",
    "keyword.control.else.lua"
  ],
  "exclude": "{.luarocks,lua_modules}/**",
  "indentation": {
    "punctuation.definition.comment.begin.lua": 1,
    "punctuation.definition.comment.end.lua": -1,
    "keyword.control.begin.lua": 1,
    "keyword.control.end.lua": -1
  },
  "markers": {
    "start": "^\\s*#?region\\b",
    "end": "^\\s*#?end\\s?region\\b"
  },
  "symbols": {
    "keyword.control.lua": 2,
    "entity.name.function.lua": 11
  }
}

Usage

TextmateLanguageService

The package exports a default class named TextmateLanguageService.

• Parameter: languageId - Language ID of grammar contribution in VS Code (string).
• Parameter: context? - Extension context from activate entrypoint export (vscode.ExtensionContext).

The library defaults to core behaviour when figuring out which scope name to use - last matching grammar or language wins. If the context parameter is supplied, the extension will first search contributions from the extension itself.

Language extension

Extension code sample - ./src/extension.ts:

import TextmateLanguageService from 'vscode-textmate-languageservice';

export async function activate(context: vscode.ExtensionContext) {
	const selector: vscode.DocumentSelector = 'lua';
	const textmateService = new TextmateLanguageService(selector, context);

	const foldingRangeProvider = await textmateService.createFoldingRangeProvider();
	const documentSymbolProvider = await textmateService.createDocumentSymbolProvider();
	const workspaceSymbolProvider = await textmateService.createWorkspaceSymbolProvider();
	const definitionProvider = await textmateService.createDefinitionProvider();

	context.subscriptions.push(vscode.languages.registerDocumentSymbolProvider(selector, documentSymbolProvider));
	context.subscriptions.push(vscode.languages.registerFoldingRangeProvider(selector, foldingRangeProvider));
	context.subscriptions.push(vscode.languages.registerWorkspaceSymbolProvider(workspaceSymbolProvider));
	context.subscriptions.push(vscode.languages.registerDefinitionProvider(selector, peekDefinitionProvider));
};

Tokenization

Extension code sample - ./src/extension.ts:

import TextmateLanguageService from 'vscode-textmate-languageservice';

export async function activate(context: vscode.ExtensionContext) {
    const selector: vscode.DocumentSelector = 'custom';
    const textmateService = new TextmateLanguageService('custom', context);
    const textmateTokenService = await textmateService.initTokenService();
    const textDocument = vscode.window.activeTextEditor!.document;
    const tokens = textmateTokenService.fetch(textDocument);
};

NB: If you would like to:

• just wire up tokenization or fast document text services to a Textmate grammar,
• without (re-)contributing grammar and language configuration to VS Code,
• or writing a full TextmateLanguageService provider configuration..

You can use the custom "textmate-languageservice-contributes" property in package.json:

{
	"textmate-languageservice-contributes": {
		"languages": [{
			"id": "typescript",
			"aliases": ["TypeScript"],
			"extensions": [".ts", ".tsx", ".cts", ".mts"]
		}],
		"grammars": [{
			"language": "typescript",
			"scopeName": "source.ts",
			"path": "./syntaxes/TypeScript.tmLanguage.json"
		}]
	}
}

API methods

Usage (example is for getting the token at the current cursor position):

const { getScopeInformationAtPosition } = TextmateLanguageService.api;

const editor = vscode.window.activeTextEditor;
const document = editor.document;
const position = editor.selection.active;

const token = await getScopeInformationAtPosition(document, position);

getScopeInformationAtPosition

getScopeInformationAtPosition(document: vscode.TextDocument, position: vscode.Position): Promise<TextmateToken>

Get token scope information at a specific position (caret line and character number).

• Parameter: document - Document to be tokenized (vscode.TextDocument).
• Parameter: position - Zero-indexed caret position of token in document (vscode.Position).
• Returns: Promise resolving to token data for scope selected by caret position ({Promise<TextmateToken>}).

getScopeRangeAtPosition

getScopeRangeAtPosition(document: vscode.TextDocument, position: vscode.Position): vscode.Range;

Get matching scope range of the Textmate token intersecting a caret position.

• Parameter: document - Document to be tokenized (vscode.TextDocument).
• Parameter: position - Zero-indexed caret position to intersect with (vscode.Position).
• Returns: Promise resolving to character and line number of the range (Promise<vscode.Range>).

getTokenInformationAtPosition

getTokenInformationAtPosition(document: vscode.TextDocument, position: vscode.Position): Promise<vscode.TokenInformation>;

VS Code compatible performant API for token information at a caret position.

• Parameter: document - Document to be tokenized (vscode.TextDocument).
• Parameter: position - Zero-indexed caret position of token in document (vscode.Position).
• Returns: Promise resolving to token data compatible with VS Code (Promise<vscode.TokenInformation>).

getLanguageConfiguration

getLanguageConfiguration(languageId: string): LanguageDefinition;

Get the language definition point of a language mode identifier.

• Parameter: languageId - Language ID as shown in brackets in "Change Language Mode" panel (string).
• Returns: Language contribution as configured in source VS Code extension (LanguageDefinition).

getGrammarContribution

getGrammarConfiguration(languageId: string): GrammarLanguageDefinition;

Get the grammar definition point of a language mode identifier.

• Parameter: languageId - Language identifier, shown in brackets in "Change Language Mode" panel (string).
• Returns: Grammar contribution as configured in source VS Code extension (GrammarLanguageDefinition).

getLanguageContribution

getLanguageConfiguration(languageId: string): LanguageDefinition;

Get the language configuration of a language mode identifier.

• Parameter: languageId - Language ID as shown in brackets in "Change Language Mode" panel (string).
• Returns: Language contribution as configured in source VS Code extension (LanguageDefinition).

getContributorExtension

getContributorExtension(languageId: string): vscode.Extension<unknown> | void;

Get the VS Code Extension API entry of the extension that contributed a language mode identifier.

• Parameter: languageId - Language identifier, shown in brackets in "Change Language Mode" panel (string).
• Returns: Extension API instance that contributed the language - (vscode.Extension).

Use Oniguruma WASM buffer

This is the vscode-oniguruma build of Oniguruma written in C, compiled to WASM format with memory hooks to V8.

This is not streaming 🙁 but vscode libs must bundle WebAssembly deps so as to support web ecosystem.

import TextmateLanguageService from 'vscode-textmate-languageservice';
const onigurumaPromise = TextmateLanguageService.utils.getOniguruma();