Giters

ksoichiro / i18n-patch

Replacing codes for i18n with patterns.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

clii18nnodetranslationyaml

i18n-patch

Replacing codes for i18n with patterns.

Travis master Coveralls master npm npm

i18n-patch is a tool to translate source code into your language(locale) for any software that does not provide i18n mechanisms.

This tool enables you to follow the upgrades of the target software: you write translation points using regular expressions, write translations for each points, and execute this tool. This method is better than modifying target source code directly.

Install

$ npm install -g i18n-patch

Usage

  Usage
    $ i18n-patch <locale> <src> [<dest>]

  Options
    --config      Base path for config files.
                  i18n.yml and <locale>.yml is required.
                  json is also available instead of yaml.
                  'config' by default.
    --statistics  Show statistics.
    --condition   Condition value to limit patterns for specific versions.
    --unmatched   Show unmatched lines to stderr.
                  They are scanned and tried to be translated but
                  no suitable translation is found in the config files.
                  These lines indicates that they might
                  have to be translated or they should be skipped.
                  false by default.

  Examples
    $ i18n-patch --config example/config --statistics --condition "version=1.1.0" --unmatched -- ja example/src example/out 2> unmatched.log

Example

You need 2 configuration files: i18n.yml and <locale>.yml like ja.yml for Japanese.

i18n.yml contains translation points that defines "what should be translated".

The tool reads the target source code and when the patterns in i18n.yml are found, it will convert "keys" in translation points and replace them to translations that are provied by <locale>.yml.

Let's see a simple example below.

The following example/i18n.yml defines target source files and patterns to be translated.

translations:
- src: '**/*.js'
  patterns:
  - pattern: preview.text("Nothing to preview.");
    replace: preview.text("${nothingToPreview}");
  - pattern: preview.text("Loading...");
    replace: preview.text("${loading}");

example/ja.yml provides translations for Japanese.
You can see the keys (nothingToPreview, loading) in this file are used in i18n.yml.

nothingToPreview: プレビューする内容がありません
loading: 読み込み中...

And the target file example/src/js/sample.js is like this:

preview.text("Nothing to preview.");
preview.text("Loading...");
console.log('other codes should be untouched.');

Then, by executing i18n-patch ja src out in example directory,
example/out/js/sample.js will be generated:

preview.text("プレビューする内容がありません");
preview.text("読み込み中...");
console.log('other codes should be untouched.');

This is a very simple example, but this tool can handle more complex expressions.

Please check the Configuration details section for further details.

If you want to try it by yourself, clone this repository and execute:

$ npm run build
$ npm start

Then you can confirm the result in example/out directory.

Why?

The main purpose of this project is to provide an external i18n system for any existent source code.

I'm maintaining gitlab-i18n-patch project to provide unofficial Japanese translation patch to GitLab.

In that project, when a new version of GitLab is released, I'm trying to merge big branch(tag) to translated branch.

This method has many problems:

  • Merge operation causes many many conflict files, which are hard to resolve.
  • When merging, some translations are lost because some part of codes are moved into other files. It's so hard to keep tracing these design changes.
  • It's very difficult to provide patches for other languages.
  • It's very difficult to get someone's contribution.
  • This method depends on patch, so one patch cannot be applied to any other versions.

Therefore, I thought it's better to create a new external translation system for providing i18n patch GitLab project without Git branch management.

Configuration details

Basic

The main configuration file is i18n.yml and looks like this:

translations:
- src: '**/*.coffee'
  patterns:
  - pattern: preview.text "Nothing to preview."
    replace: preview.text "${nothingToPreview}"

translations.src will be expanded using node-glob.
For example, src/a.coffee and src/b/c.coffee will match this pattern, but src/d.js will not match.

translations.patterns is an array that includes elements which define target pattern and replacement for it. Each element of patterns that have pattern and replace will be used to replace source code.

pattern is usually just a string value, but you can use regular expressions by js-yaml feature: !!js/regexp /foo/.

replace defines the replacement for the pattern, and this can contain variable expression like ${nothingToPreview}.

Your locale file (like ja.yml) should map this key to a translation.

nothingToPreview: プレビューする内容がありません

With these configurations, the following code named test.coffee

preview.text "Nothing to preview."

will be converted to this:

preview.text "プレビューする内容がありません"

The following test.js will not be changed because it does not match to translation.src.

// preview.text "Nothing to preview."

Arguments

replace elements can have arguments. If you want to aggregate similar patterns, consider using arguments.

For example, you can configure translations for Edit issue and Edit project like this:

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: 'Edit issue'
    replace: '${editIssue}'
  - pattern: 'Edit project'
    replace: '${editProject}'

# ja.yml
editIssue: 課題を編集
editProject: プロジェクトを編集

But you can also write them using arguments:

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: 'Edit issue'
    replace: '${editSomething}'
    args:
    - 'issue'
  - pattern: 'Edit project'
    replace: '${editSomething}'
    args:
    - 'project'
  - pattern: 'issue'
    replace: '${issue}'
  - pattern: 'project'
    replace: '${project}'

# en.yml
editSomething: {0}を編集
issue: 課題
project: プロジェクト

This will be processed like this:

  1. When the expression Edit issue is found, it's replaced into ${editSomething}.
  2. ${NAME} in the replace value is treated as a variable, and in this case, editSomething is resolved to {0}を編集 using ja.yml.
  3. {N} in the translation key is an argument and the values of args are passed to it.
  4. As a result, {0}を編集 will be converted to issueを編集.
  5. Then the third pattern is applied; issue is translated to 課題 and the result will be 課題を編集.

This is useful to write less translations and standardize the expressions.

Exclusion

If you want to exclude some patterns, even when the line matches to a pattern, you can use exclude to skip it.

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: ' user$'
    replace: ' ${user}'
    exclude: '- if'
  - pattern: !!js/regexp /Edit (.*)/
    replace: '${editSomething}'
    args:
    - '$1'

# ja.yml
user: ユーザ
editSomething: '{0}を編集'

If the above configuration is given, then

- if user
    Edit user

will be translated into:

- if user
    ユーザを編集

Sequence

Patterns are processed sequentially, so if you want to apply multiple translations to one line, please check the orders of the patterns are correct.

For example, if you want to translate Edit issue into 課題を編集 (Japanese), the following configurations won't work as expected.

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: 'issue'
    replace: '${issue}'
  - pattern: 'Edit issue'
    replace: '${editSomething}'
    args:
    - 'issue'

# ja.yml
editSomething: '{0}を編集'
issue: 課題

Because Edit issue is translated to Edit 課題 by the first pattern, it won't match to the second expression.

To fix this problem, you could write like this:

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: 'Edit issue'
    replace: '${editSomething}'
    args:
    - 'issue'
  - pattern: 'issue'
    replace: '${issue}'

# ja.yml
editSomething: '{0}を編集'
issue: 課題

Note: to tell you the truth, you can also solve this problem by using regular expressions (!!js/regexp):

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: 'issue'
    replace: '${issue}'
  - pattern: !!js/regexp /Edit (.*)/
    replace: '${editSomething}'
    args:
    - '$1'

# ja.yml
editSomething: '{0}を編集'
issue: 課題

Code insertion

If you want to insert some code snippet into some files, you can use insert.

For example, with the following config files

# i18n.yml
- src: 'foo.js'
  patterns:
  - insert:
      at: end
      value: bar

# ja.yml
bar: |
     // baz
     // qux

the next source file will be like this:

console.log('Hello, world');
console.log('Hello, world');
// baz
// qux

If you want to insert code at the beginning of the file, you should change the value of insert.at from end to begin.

As you can see, insert.value is treated as a translation key, so you must define the value of insert.value to your <locale>.yml.

Conditional insertion for all matching files

If you want to insert some code snippet into some files only when they match some of the patterns, then you can use conditionals and insert.

# i18n.yml
- src: '**/*.rb'
  conditionals:
  - insert:
      at: begin
      value: foo
  patterns:
  - pattern: bar
    replace: baz

# ja.yml
foo: '// This file is edited by i18n-patch'
baz: qux

In i18n.yml, you define what you want to insert when the files match some of the patterns.

conditionals is an array, and can have children that have insert element.

insert should have at and value children. If you'd like to insert codes at the beginning of the file, set begin to insert.at, and set end if you want to insert them at the end of the file.

Suppose you have files like below,

a.js:

console.log('bar');

b.js:

console.log('hello');

then a.js matches bar, and foo will be inserted at the beginning of the file, and won't be inserted to b.js.

// This file is edited by i18n-patch
console.log('qux');

Named patterns

Even if you use arguments and regular expressions,
there would be still many duplicate configurations.
With named patterns, you can aggregate these configurations.

Let's see a more complex example.

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: "notice: 'Project was successfully created.'"
    replace: "notice: '${projectWasSuccessfullyCreated}'"
  - pattern: "notice: 'Project was successfully updated.'"
    replace: "notice: '${projectWasSuccessfullyUpdated}'"
  - pattern: "notice: 'Project was successfully deleted.'"
    replace: "notice: '${projectWasSuccessfullyDeleted}'"
  - pattern: "notice: 'Group was successfully created.'"
    replace: "notice: '${groupWasSuccessfullyCreated}'"
  - pattern: "notice: 'Group was successfully updated.'"
    replace: "notice: '${groupWasSuccessfullyUpdated}'"
  - pattern: "notice: 'Group was successfully deleted.'"
    replace: "notice: '${groupWasSuccessfullyDeleted}'"

# ja.yml
projectWasSuccessfullyCreated: 'プロジェクトが作成されました'
projectWasSuccessfullyUpdated: 'プロジェクトが更新されました'
projectWasSuccessfullyDeleted: 'プロジェクトが削除されました'
groupWasSuccessfullyCreated: 'グループが作成されました'
groupWasSuccessfullyUpdated: 'グループが更新されました'
groupWasSuccessfullyDeleted: 'グループが削除されました'

The above configurations have similar translations, so you could rewrite them using arguments like this:

# i18n.yml
translations:
- src: '**/*'
  patterns:
  - pattern: "notice: 'Project was successfully created.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${project}'
    - '${create}'
  - pattern: "notice: 'Project was successfully updated.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${project}'
    - '${update}'
  - pattern: "notice: 'Project was successfully deleted.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${project}'
    - '${delete}'
  - pattern: "notice: 'Group was successfully created.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${group}'
    - '${create}'
  - pattern: "notice: 'Group was successfully updated.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${group}'
    - '${update}'
  - pattern: "notice: 'Group was successfully deleted.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${group}'
    - '${delete}'

# ja.yml
somethingWasSuccessfullyDone: '{0}が{1}されました'
project: プロジェクト
group: グループ
create: 作成
update: 更新
delete: 削除

ja.yml is much improved - it doesn't contain any duplicate translations.
But as you can see, i18n.yml is much longer than before.

You can use named-patterns to improve this.

The named-patterns pre-define pattern, replace and args like "function" and you can call it with name with params:

# i18n.yml
translations:
- src: '**/*'
  named-patterns:
  - name: somethingWasSuccessfullyDone
    pattern: "notice: '{obj} was successfully {done}\\.'"
    replace: "notice: '${sthWasSuccessfullyDone}'"
    args:
    - '${{objKey}}'
    - '${{doneKey}}'
    params: ['obj', 'objKey', 'done', 'doneKey']

Note: pattern in named-patterns are treated as regular expressions even if you don't write !!js/regexp.

The above configuration defines a pattern named somethingWasSuccessfullyDone.
{obj}, {done}, {objKey} and {doneKey} are parameters, and you can replace it to create a new concrete pattern.

In patterns section, you can write name to use this pattern instead of writing complex pattern, replace and args.
You must also set {obj}, {done}, {objKey} and {doneKey} with params element.

  patterns:
  - name: somethingWasSuccessfullyDone
    params:
    - {obj: Project, objKey: project, done: created, doneKey: create}
    - {obj: Project, objKey: project, done: updated, doneKey: update}
    - {obj: Project, objKey: project, done: deleted, doneKey: delete}
    - {obj: Project, objKey: project, done: created, doneKey: create}
    - {obj: Project, objKey: project, done: created, doneKey: create}

In the example above, this

  patterns:
  - name: somethingWasSuccessfullyDone
    params:
    - {obj: Project, objKey: project, done: created, doneKey: create}

is equivalent to this:

  patterns:
  - pattern: "notice: 'Project was successfully created.'"
    replace: "notice: '${somethingWasSuccessfullyDone}'"
    args:
    - '${project}'
    - '${create}'

Multiline

If you want to use patterns that match two or more lines and change the order of the lines:

  %span.light History for
  = link_to foobar
  = link_to foobar
  %span.light <translation of history for>

then you can write like this:

translations:
- src: 'test.js'
  patterns:
  - pattern: !!js/regexp /^(.*)History for\n([^\n]*)$/m
    replace: '${historyFor}'
historyFor: "$2\n$1の更新履歴"

When test.js is like this,

  %span.light History for
  = link_to foobar

then the result will be:

  = link_to foobar
  %span.light の更新履歴

You must use !!js/regexp to pattern.
To match a line, you must use ([^\n]*), otherwise the tool cannot calculate the required lines.

Include/exclude locales

You can apply a translation config for specific locales using locale.include or locale.exclude option.

For example, when the following configuration is given,

# i18n.yml
translations:
# translation1
- src: 'test.js'
  locale:
    include: ['ja']
  patterns:
  - pattern: foo
    replace: '${foo}'
# translation2
- src: 'test.js'
  locale:
    exclude: ['ja']
  patterns:
  - pattern: bar
    replace: '${bar}'
# translation3
- src: 'test.js'
  patterns:
  - pattern: baz
    replace: '${baz}'

# ja.yml
foo: hoge
bar: fuga
baz: piyo

the following input file

foo
bar
baz

will be converted into:

hoge
bar
piyo

translation1 and translation3 can be applied to ja locale (foo -> hoge, baz -> piyo), but translation2 (bar -> fuga) is not applied to ja config due to locale.exclude.

Add new files

You can add new files using add option.

# i18n.yml
translations:
- add:
    path: 'a/b/test.js'
    value: testContent

# ja.yml
testContent: |
             // foo
             // bar

This will generate the following file a/b/test.js:

// foo
// bar

Skip patterns

If you feel slow to process all files, consider using skip-patterns option. When the lines match these patterns, the subsequent pattern matching process will be skipped, which makes the entire processing faster in some cases.

For example, when the following configuration is given,

# i18n.yml
translations:
- src: 'test.js'
  skip-patterns:
  - !!js/regexp /^$/
  - !!js/regexp /secret/
  patterns:
  - pattern: 'foo'
    replace: '${foo}'
  
# ja.yml
foo: bar

the following input file

/*

  foo
  foo secret
*/

will be converted into:

/*

  bar
  foo secret
*/

You can see the line foo secret is not translated.
This is because it matches one of the skip-patterns: !!js/regexp /secret/.

Match once

If you specifies many files for a pattern and the pattern is aimed to match just one file, you can use match-once flag.

If match-once: true is set for a pattern, then only the first file among all of the candidate files will be replaced, which will contribute to improve performance.

For example, if the following config is given,

# i18n.yml
translations:
- src: 'src/*.js'
  patterns:
  - pattern: 'foo'
    replace: '${foo}'
    match-once: true

# ja.yml
foo: bar

and if there are two files,

// src/a.js
console.log('hello, foo');
console.log('good morning, foo');
// src/b.js
console.log('hi, foo');

then only the first matching expression in the first file (src/a.js) will be replaced:

// src/a.js
console.log('hello, bar');
console.log('good morning, foo');
// src/b.js
console.log('hi, foo');

This option can be used with named-pattern.

Complete pattern

If one of your patterns is complete pattern and don't require other translation, then you can set complete-pattern: true to skip other translations for the matched lines.

This will be useful when you have many patterns and the translations take long time.

For example, if the following config is given,

# i18n.yml
translations:
- src: 'test.js'
  patterns:
  - pattern: 'foo'
    replace: '${foo}'
    complete-pattern: true
  - pattern: 'bar'
    replace: '${bar}'

# ja.yml
foo: FOO
bar: BAR

and if there is a following file,

/*
  bar
  foo bar
  bar baz
*/

then the result will be like this:

/*
  BAR
  FOO bar
  BAR baz
*/

The line foo bar matches the pattern foo and it is defined with complete-pattern: true, so the line foo bar does not match the pattern bar.

Parallel groups

When you want to process several translations in parallel,
you can use parallel-group option.

# i18n.yml
translations:
- name: 'foo'
  parallel-group: 'group1'
- name: 'bar'
  parallel-group: 'group1'
- name: 'baz'

With the above example, foo and bar will be processed in parallel because the same parallel-group value group1 is given, and then baz will be processed.

Split files with suffix

i18n.yml and ja.yml can be split into multiple files using suffix:

# i18n.yml
translations:
- src: '*.md'
  patterns:
  - pattern: 'baz'
    replace: '${baz}'

# i18n-1.yml
translations:
- src: '*.js'
  patterns:
  - pattern: 'foo'
    replace: '${foo}'

# i18n-2.yml
translations:
- src: '*.html'
  patterns:
  - pattern: 'bar'
    replace: '${bar}'

These files will be merged into a single configuration:

translations:
- src: '*.md'
  patterns:
  - pattern: 'baz'
    replace: '${baz}'
- src: '*.js'
  patterns:
  - pattern: 'foo'
    replace: '${foo}'
- src: '*.html'
  patterns:
  - pattern: 'bar'
    replace: '${bar}'

Evaluate only when conditions are satisfied

If you have some translation for older versions of the target software and the recent versions don't require that translation, then you should use evaluate-when configuration.

evaluate-when configuration is a JavaScript expression string to limit evaluation of the translation set.

For example, if you have patterns for versions older than 1.0.0, and the later versions does not need that patterns, then you can write your i18n.yml like this:

translations:
- name: 'example'
  src: '*.js'
  evaluate-when: "semver.lt(version, '1.0.0')"
  patterns:
  - pattern: 'foo'
    replace: '${foo}'

The value of evaluate-when uses semver to compare two versions.

One of the versions is given by command line option --condition.
If you execute this tool like i18n-patch --condition version=1.1.0, then the expression would be evaluated as false and the translation example would be skipped. And if you execute this tool like i18n-patch --condition version=0.9.2, then the expression would be evaluated as true and the translation example would be evaluated and processed as usual.
--condition option can be used multiple times such as i18n-patch --condition foo=1 --condition bar=2.
The value of --condition should be key=value format, and if you omit key= then it would be interpreted to version=value.

evaluate-when is a JavaScript expression and the main use case is that a part of the target software is rewritten with other language from some version, so semver library can be used in this expression to compare two versions.

This configuration can be also applied to patterns:

translations:
- name: 'example'
  src: '*.js'
  patterns:
  - pattern: 'foo'
    replace: '${foo}'
    evaluate-when: "semver.lt(version, '1.0.0')"

License

MIT

About

Replacing codes for i18n with patterns.

clii18nnodetranslationyaml

License:MIT License

Languages

Language:JavaScript 99.6%Language:Ruby 0.3%Language:HTML 0.1%