1.3.10 • Public • Published


Why can't I run my whole app or site through Google Translate and get a basic translation in another language?

Now, you can!

The name hokeylization is a portmanteau, meaning 'hokey localization'

It's somewhat hokey because it's very simple: it sends strings to Google Translate

And it is simple, yet also very powerful. It has special support for HTML documents, HandlebarsJS templates, and Markdown files.

You can translate:

  • a JavaScript object containing messages
  • any number of files or directories, always traversing directories recursively

Read this in another language

This README.md document has been translated, using the hokeylization tool itself, into every language supported by Google Translate!

I'm certain it's not perfect, but I hope it's better than nothing!

   🇸🇦 Arabic    🇧🇩 Bengali    🇩🇪 German    🇺🇸 English    🇪🇸 Spanish    🇫🇷 French    🇹🇩 Hausa    🇮🇳 Hindi    🇮🇩 Indonesian    🇮🇹 Italian    🇯🇵 Japanese    🇰🇷 Korean    🇮🇳 Marathi    🇵🇱 Polish    🇧🇷 Portuguese    🇷🇺 Russian    🇰🇪 Swahili    🇵🇭 Tagalog    🇹🇷 Turkish    🇵🇰 Urdu    🇻🇳 Vietnamese    🇨🇳 Chinese
   📚 ... All Languages ...

Is there a problem with this translation of the README?

This particular translation of the original README may be flawed -- corrections are very welcome! Please send a pull request on GitHub, or if you're not comfortable doing that, open an issue

When you create a new GitHub issue about a translation, please do:

  • include the page URL (copy/paste from browser address bar)
  • include the exact text that is wrong (copy/paste from browser)
  • please describe what is wrong -- is the translation incorrect? is the formatting broken somehow?
  • kindly offer a suggestion of a better translation, or how the text should be properly formatted
  • Thank you!



Support and Funding

I would sincerely appreciate any contribution via Patreon


To use the command line tool, install using npm or yarn:

npm install -g hokeylization
yarn global add hokeylization

To use as a library, install the lite version, which is much smaller:

npm install -g hokeylization-lite
yarn global add hokeylization-lite

Then look at help for the hokey command:

hokey --help
hokey -h

Want to see output in your langauge or another language?

hokey tries to detect the language automatically from your shell's environment variables

You can force a language by setting the LC_ALL environment variable:

LC_ALL=it hokey --help

Note that if you have installed hokeylization-lite, command help is only available in English


Set the GOOGLE_TRANSLATE_PROJECT_ID environment variable to identify your Google Translate project

Set the GOOGLE_APPLICATION_CREDENTIALS environment variable to the JSON credentials you downloaded after figuring out how authentication works on Google cloud (it can be fun)

If you're running from the source code, you can also put these in a .env file in the source directory they'll be loaded at runtime via dotenv

Translating a JavaScript string resource file

Your string table must be in a JavaScript file in one of these two forms:

ES6 export:

export default {
  string_key: "some value",
  another_key: "another value",
  ... more keys ...

CommonJS export

module.exports = {
  string_key: "some value",
  another_key: "another value",
  ... more keys ...

If this file was named myfile.en.js, you can translate it to Spanish and German with:

hokey -l es,de -o myfile.LANG.js myfile.en.js

The LANG in the above is special -- it is a reserved word in this tool!

The LANG is replaced with the language code for the output files

Thus the above command creates the files:


The -l / --languages option is a comma-separated list of ISO language codes supported by Google Translate

If the output file already exists, it will be examined to determine which keys already exist. Existing keys will not be translated. Translations for missing keys will be generated and appended to the end of the JS object. The entire file is always rewritten.

To force retranslation all keys, use the -f / --force option

Translating a directory of text files

You can also translate a directory of files. hokeylization will recursively visit every file in the directory and run its contents through Google Translate, and save the output to an identically named file in a separate directory tree

When the target of your translation is a directory, this mode is enabled

The -o / --outfile option specifies the output directory

BIG WARNING: When translating directories, DO NOT specify an output directory that is within your input directory! If you do this, you will:

  • induce infinite recursion
  • run up your Google bill
  • fill up your disk
  • have less fun

Here is an example of what not to do:

hokey -l es -o templates/es templates    # <--- DON'T DO THIS!

When this runs, translated files are written to templates/es, and thus become new source files to translate, since they are under templates/ -- this process continues forever, don't do it!

Correct usage

OK, let's say you have some email templates in a directory:


To translate all of these to Spanish and German, run:

hokey -l es,de -o templates/email/LANG templates/email/en

In the above, LANG is a reserved word and will be replaced with the ISO language code

What happens when the above runs:

  • The templates/email/es and templates/email/de directories will be created (if they don't exist)
  • Every file in templates/email/en will be translated to Spanish and German
    • Existing output files will not be regenerated unless you use -f / --force
  • You'll end up with an identical directory structure and files within es and de like you have under en

Other options

Dry run

Pass -n / --dry-run to display what would be done, but do not actually make any API calls or write any files


Pass -f / --force to always regenerate translations, even if they already exist


Pass -m / --match to limit the files processed when running in directory-mode

You may not always want to translate every file in your source directory to your target directory

The value of the -m / --match option is a regex (beware shell quoting rules!) that specifies which files should be translated

When in doubt, you can combine this option with -n / --dry-run to see which files would be translated


Sometimes your -m matches too many files. Use the -e / --excludes option to explicitly exclude files that otherwise would have matched

You can list multiple regexes, separated by spaces

A common usage would be: --excludes node_modules dist \.git build tmp


The strings to translate might contain {{ handlebars }} templates, either with two or three curly-braces

You probably DON'T want the stuff inside those templates to be translated

Pass the -H / --handlebars flag, and anything within {{ ... }} will not be translated


Markdown is neither text nor html, so Google Translate has some difficulties with it

The -M / --markdown flag enables special handling for markdown files

With markdown files, if you don't use the -M flag, you will probably find these problems:

  • Broken links. In the translation, a space character appears after a markdown link description ends (with ]) but before its target link begins (with (). This causes the markdown to render incorrectly, and the link is broken when viewing the document.
  • Code blocks get translated. Google translate doesn't know what markdown considers code and what it doesn't
  • Incorrect spacing for indented code blocks. Spacing is difficult to preserve in translation
  • Things inside of backticks will get translated, when you almost always want them to be literal values

When the -M / --markdown flag is enabled:

  • The pattern ] ( will be condensed to ]( thus fixing the broken markdown links
  • A "no translate" wrapper will be placed around indented code blocks, preserving proper indentation and ensuring they are not translated
  • A "no translate" wrapper will be placed around text within backticks to ensure that they are not translated


Normally everything is processed as plain text

If your content is HTML, it will get mangled unless you pass the -p html / --process-as html option


For the adventurous: when processing files in a directory, you can pass the -F / --filter option to filter the output before it is written to the filesystem

The value of this option must be a path to a JS file that exports a function named filter

The filter function must be async because await will be called upon it

Before files are written to disk, the entire file contents will be passed to the filter function as a string

The return value from the filter function is what will actually be written to storage

Thus, you have total control over what will finally be written

The filter script will be looked for in the following locations (with .js will be appended to the filter name, unless it already ends in .js)

  • The current directory
  • A directory named .hokey-filters within the current directory
  • A directory named ${HOME}/.hokey-filters, where ${HOME} is the current user's home directory
  • The built-in filters directory

Filter Parameters

The filter string can be multiple words. In this case, the first word is the filter name, and the remaining words will be passed as arguments to the filter function


Use -h / --help to show help

JSON batch commands

With the -j / --json option, you can run multiple coordinated hokey commands

By convention this file is called hokey.json, but you can name it whatever you want

If you pass a directory as the -j option, hokey will look for a hokey.json in that directory

The JSON file should contain one object. Within that object, its property names are the same as the command-line options, plus one additional property named hokey

The hokey property is an array of commands to run. The properties declared within these commands will override any duplicate declarations in the outer object.

Within each object in the hokey array, you should specify a name, and the input and output files

Here is an example of a hokey.json

    "inputLanguage": "en",
    "languages": "es,fr,ja",  # can also be an array of strings
    "force": false,
    "match": null,
    "processAs": null,
    "excludes": ["exclude-1", "exclude-2"],
    "handlebars": false,
    "markdown": false,
    "regular": false,
    "dryRun": false,
    "filter": "theFilter.js",
    "hokey": [
        "name": "locale names",
        "infile": "messages/locales_en.js",
        "outfile": "messages/locales_LANG.js",
        "handlebars": true
        "name": "CLI messages",
        "infile": "messages/en_messages.js",
        "outfile": "messages/LANG_messages.js",
        "handlebars": true
        "name": "README",
        "infile": "README.md",
        "outfile": "lang/LANG/",
        "excludes": ["lang/", "node_modules/", "\\.git/", "tmp/"],
        "filter": "relativizeMarkdownLinks lang",
        "markdown": true,
        "index": "lang/README.md"

Multiple input files

Pass an array of file paths as infiles instead of a single path infile, as in this example:

  ... [
      "name": "my docs",
      "infiles": ["README.md", "INSTALL.md", "TUTORIAL.md"],
      "outfile": "docs/LANG/",
      "markdown": true


When translating to many languages, hokey can create an index file that lists all the translations made and provides links to them

When generating indexes, you can have only one input source

Pass the -I / --index option, the value is where the index file will be generated, which can be a file or a directory. If it's a directory, a default filename will be used, based on the template (see below)

Use the -A / --index-template to determine how the index output is formatted. You can specify 'html', 'markdown', 'text', or the file path to your own HandlebarsJS template

If you specify your own template, you must also specify a file (not a directory) for the -I / --index option

Have a fun time translating languages!

Package Sidebar


npm i hokeylization

Weekly Downloads






Unpacked Size

3.63 MB

Total Files


Last publish


  • cobbzilla