Version: 1.9.x

Translations & Localization

Docusaurus allows for easy translation functionality using Crowdin. Documentation files written in English are uploaded to Crowdin for translation by users within a community. Top-level pages written with English strings can be translated by wrapping any strings you want to translate in a <translate> tag. Other titles and labels will also be found and properly translated.

Docusaurus Translation Configurations#

To generate example files for translations with Docusaurus, run the examples script with the command line argument translations:

npm run examples translations

yarn examples translations

This will create the following files:

pages/en/help-with-translations.js
languages.js
../crowdin.yaml

The pages/en/help-with-translations.js file includes the same starter help page generated by the examples script, but now includes translation tags.

Generally, you will use help-with-translations.js as a guide to enable translations in your other pages, but not actually commit the file to your repo (i.e., you can delete it). However, if you want a Help page, and you currently do not have one, you can rename this file to help.js and use it as a starting point.

The languages.js file tells Docusaurus what languages you want to enable for your site. By default, we expect English to be enabled.
The crowdin.yaml file is used to configure Crowdin integration, and is copied up one level into your Docusaurus project repo. If your Docusaurus project resides in /project/website, then crowdin.yaml will be copied to /project/crowdin.yaml.

Translating Your Existing Docs#

Your documentation files (e.g., the .md files that live in your docs directory) do not need to be changed or moved to support translations. They will be uploaded to Crowdin to be translated directly.

Enabling Translations on Pages#

Pages allow you to customize layout and specific content of pages like a custom index page or help page.

Pages with text that you want translated should be placed in website/pages/en directory.

Wrap strings you want translated in a <translate> tag, and add the following require statement to the top of the file:

...
const translate = require('../../server/translate.js').translate;
...
<h2>
  <translate>This header will be translated</translate>
</h2>
...

You can also include an optional description attribute to give more context to a translator about how to translate the string:

<p>
  <translate desc="flower, not verb">Rose</translate>
<p>

The <translate> tag generally works well on pure strings. If you have a string like "Docusaurus currently provides support to help your website use translations", wrapping the <translation> tag around that entire string will cause issues because of the markdown linking, etc. Your options are to not translate those strings, or spread a bunch of <translate> tags amongst the pure substrings of that string.

Gathering Strings to Translate#

The strings within localized Pages must be extracted and provided to Crowdin.

Add the following script to your website/package.json file, if it does not exist already:

{
  ...
  "scripts": {
    "write-translations": "docusaurus-write-translations"
  },
  ...
}

Running the script will generate a website/i18n/en.json file containing all the strings that will be translated from English into other languages.

The script will include text from the following places:

title and sidebar_label strings in document markdown headers
category names in sidebars.json
tagline in siteConfig.js
header link label strings in siteConfig.js
strings wrapped in the <translate> tag in any .js files inside pages

Custom Translation Strings#

If you want to add additional custom translation strings, or override any of the strings that get produced by the script that creates the website/i18n/en.json file, you can add a website/data/custom-translation-strings.json file. The file should have form of:

{
  "localized-strings": {
    "docs": {
      "id": {
        "title": "string1",
        "sidebar_label": "string2"
      },
      "version-0.0.1-id": {
        "title": "string3",
        "sidebar_label": "string4"
      }
    }
  },
  "pages-strings": {
    "id3": "string3",
    "id4": "string4"
  }
}

where localized-strings represent strings in your documentation content and pages-strings represents metadata in your documentation (e.g., title, links, etc).

Here is an example:

{
  "_comment": "This file is used to provide custom strings for translations, including overriding defaults",
  "localized-strings": {
    "translation": "Translations and Localization"
  },
  "pages-strings": {
    "Help Translate|recruit community translators for your project": "Help Us Translate"
  }
}

See the generated website/i18n/en.json for an example.

How Strings Get Translated#

Docusaurus itself does not do any translation from one language to another. Instead, it integrates Crowdin to upload translations and then downloads the appropriately translated files from Crowdin.

How Docusaurus Uses String Translations#

This section provides context about how translations in Docusaurus works.

Strings#

A Docusaurus site has many strings used throughout it that require localization. However, maintaining a list of strings used through out a site can be laborious. Docusaurus simplifies this by centralizing strings.

The header navigation, for example can have links to 'Home' or your 'Blog'. This and other strings found in the headers and sidebars of pages are extracted and placed into i18n/en.json. When your files are translated, say into Spanish, a i18n/es-ES.json file will be downloaded from Crowdin. Then, when the Spanish pages are generated, Docusaurus will replace the English version of corresponding strings with translated strings from the corresponding localized strings file (e.g. In a Spanish enabled site 'Help' will become 'Ayuda').

Markdown Files#

For documentation files themselves, translated versions of these files are downloaded and then rendered through the proper layout template.

Other Pages#

For other pages, Docusaurus will automatically transform all <translate> tags it finds into function calls that return the translated strings from the corresponding localized file locale.json.

Crowdin#

Crowdin is a company that provides translation services. For Open Source projects, Crowdin provides free string translations.

Create your translation project on Crowdin. You can use Crowdin's guides to learn more about the translations work flow. We suggest that you deselect and do not include "English" as a translatable language to prevent the creation of en-US localization files as this can lead to confusion.

Ensure in your Crowdin settings, in the Translations section, that "Duplicate Strings" are set to "Hide - all duplicates will share the same translation". This setting will ensure that identical strings between versions share a single translation.

Your project will need a crowdin.yaml file generated. If you ran yarn examples translations or npm run examples translations, this file was created for you on the same level as your website directory.

You will need to install the crowdin command line interface. Please follow the installation directions.

The example below can be automatically generated by the Docusaurus cli with the examples script. It should be placed in the top level of your project directory to configure how and what files are uploaded/downloaded.

Below is an example Crowdin configuration for the respective languages: German, Spanish, French, Japanese, Korean, Bahasa Indonesia, Portuguese Brazilian, Chinese Simplified, and Chinese Traditional.

project_identifier_env: CROWDIN_DOCUSAURUS_PROJECT_ID
api_key_env: CROWDIN_DOCUSAURUS_API_KEY
base_path: './'
preserve_hierarchy: true

files:
  - source: '/docs/**/*.md'
    translation: '/website/translated_docs/%locale%/**/%original_file_name%'
    languages_mapping: &anchor
      locale:
        'de': 'de'
        'es-ES': 'es-ES'
        'fr': 'fr'
        'ja': 'ja'
        'ko': 'ko'
        'mr': 'mr-IN'
        'pt-BR': 'pt-BR'
        'zh-CN': 'zh-CN'
        'zh-TW': 'zh-TW'

You can go here to learn more about customizing your crowdin.yaml file.

Setup the Crowdin Scripts#

You will want to manually sync your files to and from Crowdin. The sync process will upload any markdown files in /docs as well as translatable strings in website/i18n/en.json. (These strings can be generated by running yarn write-translations.)

You can add the following to your package.json to manually trigger Crowdin.

"scripts": {
  "crowdin-upload": "crowdin --config ../crowdin.yaml upload sources --auto-update -b master",
  "crowdin-download": "crowdin --config ../crowdin.yaml download -b master"
},

Manual File Sync#

You will always want to upload your markdown files and translatable strings first and the download the translations section. So run the commands in this order:

CROWDIN_DOCUSAURUS_PROJECT_ID=YOUR_CROWDIN_PROJECT_ID CROWDIN_DOCUSAURUS_API_KEY=YOUR_CROWDIN_API_KEY yarn run crowdin-upload
CROWDIN_DOCUSAURUS_PROJECT_ID=YOUR_CROWDIN_PROJECT_ID CROWDIN_DOCUSAURUS_API_KEY=YOUR_CROWDIN_API_KEY yarn run crowdin-download

YOUR_CROWDIN_PROJECT_ID is the name of your Crowdin project. e.g., for https://crowdin.com/project/docusaurus/, that variable would be set to docusaurus. YOUR_CROWDIN_API_KEY is a unique key that is like a password. You can find it in the API tab of your Crowdin project's Settings.

These commands require having an environment variable set with your Crowdin project id and api key (CROWDIN_PROJECT_ID, CROWDIN_API_KEY). You can preface them inline as done above or add them permanently to your .bashrc or .bash_profile.

If you run more than one localized Docusaurus project on your computer, you should change the name of the environment variables to something unique (CROWDIN_PROJECTNAME_PROJECT_ID, CROWDIN_PROJECTNAME_API_KEY).

Since the files are generated, you do not need to have any files in your website/i18n or website/translated_docs directory as part of your repo. So you can can add website/i18n/* and website/translated_docs to your .gitignore file.

Automated File Sync Using CircleCI#

You can automate pulling down and uploading translations for your files using the CircleCI web continuous integration service.

First, update the .circleci/config.yml file in your project directory to include steps to upload English files to be translated and download translated files using the Crowdin CLI. Here is an example .circleci/config.yml file:

# If you only want circle to run on direct commits to master, you can uncomment this out
# and uncomment the filters: *filter-only-master down below too
#
# aliases:
#  - &filter-only-master
#    branches:
#      only:
#        - master

version: 2
jobs:
  deploy-website:
    docker:
      # specify the version you desire here
      - image: circleci/node:8.11.1

    steps:
      - checkout
      - run:
          name: Deploying to GitHub Pages
          command: |
            git config --global user.email "<GITHUB_USERNAME>@users.noreply.github.com"
            git config --global user.name "<YOUR_NAME>"
            echo "machine github.com login <GITHUB_USERNAME> password $GITHUB_TOKEN" > ~/.netrc
            # install Docusaurus and generate file of English strings
            - cd website && yarn install && yarn run write-translations && cd ..
            # crowdin install
            - sudo apt-get install default-jre
            - wget https://artifacts.crowdin.com/repo/deb/crowdin.deb -O crowdin.deb
            - sudo dpkg -i crowdin.deb
            # translations upload/download
            - crowdin --config crowdin.yaml upload sources --auto-update -b master
            - crowdin --config crowdin.yaml download -b master
            # build and publish website
            cd website && GIT_USER=<GIT_USER> yarn run publish-gh-pages

workflows:
  version: 2
  build_and_deploy:
    jobs:
      - deploy-website:
#         filters: *filter-only-master

The crowdin command uses the crowdin.yaml file generated with the examples script. It should be placed in your project directory to configure how and what files are uploaded/downloaded.

Note that in the crowdin.yaml file, CROWDIN_PROJECT_ID and CROWDIN_API_KEY are environment variables set-up in Circle for your Crowdin project. They can be found in your Crowdin project settings.

Now, Circle will help you automatically get translations prior to building your website. The provided crowdin.yaml file will copy translated documents into website/translated_docs/, and translated versions of the i18n/en.json strings file will into i18n/${language}.json.

If you wish to use Crowdin on your machine locally, you can install the Crowdin CLI tool and run the same commands found in the circle.yaml file. The only difference is that you must set project_identifier and api_key values in the crowdin.yaml file since you will not have Circle environment variables set up.

Versioned Translations#

If you wish to have translation and versioning for your documentation, add the following section to the end of your crowdin.yaml file:

  -
    source: '/website/versioned_docs/**/*.md'
    translation: '/website/translated_docs/%locale%/**/%original_file_name%'
    languages_mapping: *anchor

Translated, versioned documents will be copied into website/translated_docs/${language}/${version}/.