Deepmark

Translate markdown files correctly with mdast and DeepL.

Getting Started

1. Install deepmark:

# NPM
npm install -D deepmark

# PNPM
pnpm add -D deepmark

# Yarn
yarn add -D deepmark

2. Create a deepmark.config.mjs on your project root:

// deepmark.config.mjs

/** @type {import("deepmark").UserConfig} */
export default {
	sourceLanguage: 'en',
	outputLanguages: ['zh', 'ja'],
	directories: [
		['i18n/$langcode$', 'i18n/$langcode$'],
		['docs', 'i18n/$langcode$/docs'],
		['blog', 'i18n/$langcode$/blog']
	]
};

3. Set DEEPL_AUTH_KEY environment variable containing DeepL developer auth key:

# If you're on Linux
export DEEPL_AUTH_KEY=your_auth_key

You can also use something like dotenv package. For CI and remote environment such as Github Actions and Gitpod, look into their setting page to set environment variables.

4. Set NPM scripts in your package.json:

{
	"scripts": {
		"translate": "deepmark translate"
	}
}

# NPM
npm run translate

# PNPM
pnpm run translate

# Yarn
yarn translate

Notes:

deepmark also supports translating JSON and YAML files.

Limitations

1. It is not possible to add mdast plugins to the workflow.
2. Only support .md and .mdx. Other extended verisons of markdown such as Svelte extended mdsvex are not supported.

Documentation

Translation modes:

deepmark translate --mode hybrid|offline|online

• hybrid (default): Look for translation from local translation memory first before using DeepL API.
• offline: Translate using the local translation memory only, passthrough if not found. This mode will not update the translation memory.
• online: Translate using DeepL API only. Will overwrite existing translation memory.

Configuration:

interface UserConfig {
	/**
	 * Source's language code. Based on DeepL supported languages.
	 */
	sourceLanguage: SourceLanguageCode;
	/**
	 * Output's languages code. Based on DeepL supported languages.
	 */
	outputLanguages: TargetLanguageCode[];
	/**
	 * Sources and ouputs directories pairs. $langcode$ variable
	 * is provided to dynamically define directory.
	 *
	 * e.g. [ ["docs", "i18n/$langcode$/docs"], ["blog", "i18n/$langcode$/blog"] ]
	 */
	directories: [string, string][];
	/**
	 * Override current working directory, defaults to `process.cwd()`.
	 */
	cwd?: string;
	/**
	 * By default, all .md, .mdx, .json, and .yaml|.yml files inside
	 * source directories will be included.
	 *
	 * Define glob patterns to filter what files to include or exclude.
	 * But, the end result is still restricted by file types (.md, .mdx, .json).
	 */
	files?: {
		include?: string[];
		exclude?: string[];
	};
	/**
	 * Frontmatter fields.
	 */
	frontmatterFields?: {
		include?: string[];
		exclude?: string[];
	};
	/**
	 * Markdown node types to include or exclude based on MDAST. Defaults to exclude `code` and `link`.
	 */
	markdownNodes?: {
		default?: boolean;
		include?: MdNodeType[];
		exclude?: MdNodeType[];
	};
	/**
	 * HTML elements to include and exlcude, down to the level of attributes
	 * and children. Include all HTML elements text content
	 * and some global attributes such as title and placeholder.
	 */
	htmlElements?: {
		default?: boolean;
		include?: Partial<{ [Tag in HtmlTag]: { children: boolean; attributes: string[] } }>;
		exclude?: HtmlTag[];
	};
	/**
	 * JSX components to include and exclude, down to the level of attributes
	 * and children. Include all JSX components text children
	 * and exclude all attributes by default.
	 *
	 * Support array, object, and jsx attribute value. For object and array value,
	 * you can specify the access path starting with the attribute name
	 * e.g. `items.description` to translate `items={[{description: "..."}]}.
	 */
	jsxComponents?: {
		default?: boolean;
		include?: { [Name: string]: { children: boolean; attributes: string[] } };
		exclude?: string[];
	};
	/**
	 * JSON or YAML file properties to include and exclude.
	 * Exclude all properties by default.
	 */
	jsonOrYamlProperties?: {
		include?: string[];
		exclude?: string[];
	};
}