[feat] Adds language option

.zuul.yml

@@ -19,6 +19,6 @@ browsers:
 browserify:
   - transform: babelify
 scripts:
-  - "build/browser/transliteration.min.js"
+  - "lib/browser/transliteration.min.js"
 username: node-transliteration
 key: 45c0ecd6-f97f-480f-a050-1f35e24e732b

README.md

@@ -102,24 +102,87 @@ __Options:__ (optional)
   replace: { source1: target1, source2: target2, ... }, // Object form of argument
   replace: [[source1, target1], [source2, target2], ... ], // Array form of argument
   /* Strings in the ignore list will be bypassed from transliteration */
-  ignore: [str1, str2] // default: []
+  ignore: [str1, str2], // default: [],
+  /* Source language (allows using custom rules for specific languages) */
+  lang: 'bg' // default: ''
 }
 ```
 
 __transliterate.config([optionsObj])__
 
-Bind options globally so any following calls will be using `optoinsObj` by default. If `optionsObj` argument is omitted, it will return current default option object.
+Bind options globally so any following calls will be using `optionsObj` by default. If `optionsObj` argument is omitted, it will return current default option object.
 ```javascript
 transliterate.config({ replace: [['你好', 'Hello']] });
 transliterate('你好, world!'); // Result: 'Hello, world!'. This equals transliterate('你好, world!', { replace: [['你好', 'Hello']] });
 ```
 
+__transliterate.setLanguagesConfig([languagesConfig])__
+
+You can set a custom languages configuration object in the form of:
+```js
+{
+  'lang-name': {
+    replace: [['a', 'b']],
+    ignore: ['c']
+  },
+  'lang-name2': {
+    replace: [['c', 'd']]
+  }
+}
+```
+
+The need for having control over the source language is brought by the fact that the same characters can have a different transliteration output in different languages. For example, in Russian the character `щ` has a different pronounciation than the same character in the Bulgarian alphabet, thus the difference in transliteration - `shch` versus `sht`.
+
+Sample configuration for the Bulgarian language:
+```javascript
+transliterate.setLanguagesConfig({
+  bg: {
+    replace: [
+      ['ц', 'ts'],
+      ['Ц', 'Ts'],
+      ['щ', 'sht'],
+      ['Щ', 'Sht'],
+      ['ъ', 'a'],
+      ['Ъ', 'A'],
+      ['ь', 'y'],
+      ['ѝ', 'i']
+    ]
+  }
+});
+
+transliterate('щастие'); // 'shchastie'
+transliterate('щастие', { lang: 'bg' }); // 'shtastie'
+```
+
+Currently the following Cyrillic languages are supported:
+- Russian (used by default if you don't specify a language)
+- Bulgarian `'bg'`
+- Macedonian `'mk'`
+- Ukrainian `'ua'`
+- Serbian `'rs'`
+
+If you want to specify your own language rules but still use those provided in the library, you can use something like [deep-assign](https://www.npmjs.com/package/deep-assign) to extend the built-ins:
+
+```javascript
+const builtInConfig = transliterate.setLanguagesConfig();
+transliterate.setLanguagesConfig(
+  deepAssign(
+    builtInConfig, {
+    'my-lang': {
+      replace: [['a', 'b']]
+    }
+  })
+);
+```
+
 __Example__
 ```javascript
 import { transliterate as tr } from 'transliteration';
 tr('你好，世界'); // Ni Hao , Shi Jie
 tr('Γεια σας, τον κόσμο'); // Geia sas, ton kosmo
 tr('안녕하세요, 세계'); // annyeonghaseyo, segye
+tr('цвете'); // 'cvete'
+tr('цвете', { lang: 'bg' }); // 'tsvete'
 tr('你好，世界', { replace: {你: 'You'}, ignore: ['好'] }) // You 好, Shi Jie
 tr('你好，世界', { replace: [['你', 'You']], ignore: ['好'] }) // You 好, Shi Jie (option in array form)
 // or use configurations
@@ -143,14 +206,16 @@ __Options:__ (optional)
   replace: { source1: target1, source2: target2, ... },
   replace: [[source1, target1], [source2, target2], ... ], // default: []
   /* Strings in the ignore list will be bypassed from transliteration */
-  ignore: [str1, str2] // default: []
+  ignore: [str1, str2], // default: []
+  /* Source language */
+  lang: 'bg' // default: ''
 }
 ```
 If `options` is not provided, it will use the above default values.
 
 __slugify.config([optionsObj])__
 
-Bind options globally so any following calls will be using `optoinsObj` by default. If `optionsObj` argument is omitted, it will return current default option object.
+Bind options globally so any following calls will be using `optionsObj` by default. If `optionsObj` argument is omitted, it will return current default option object.
 ```javascript
 slugify.config({ replace: [['你好', 'Hello']] });
 slugify('你好, world!'); // Result: 'hello-world'. This equals slugify('你好, world!', { replace: [['你好', 'Hello']] });
@@ -163,6 +228,7 @@ slugify('你好，世界'); // ni-hao-shi-jie
 slugify('你好，世界', { lowercase: false, separator: '_' }); // Ni_Hao_Shi_Jie
 slugify('你好，世界', { replace: {你好: 'Hello', 世界: 'world'}, separator: '_' }); // hello_world
 slugify('你好，世界', { replace: [['你好', 'Hello'], ['世界', 'world']], separator: '_' }); // hello_world (option in array form)
+slugify('Цветя и щастие.', { lang: 'bg'}); // 'tsvetya-i-shtastie'
 slugify('你好，世界', { ignore: ['你好'] }); // 你好shi-jie
 // or use configurations
 slugify.config({ lowercase: false, separator: '_' });
@@ -203,6 +269,7 @@ Options:
   -u, --unknown  Placeholder for unknown characters                        [string] [default: "[?]"]
   -r, --replace  Custom string replacement                                     [array] [default: []]
   -i, --ignore   String list to ignore                                         [array] [default: []]
+  -l, --lang     Source language                                              [string] [default: ""]
   -S, --stdin      Use stdin as input                                     [boolean] [default: false]
   -h, --help     Show help                                                                 [boolean]
 
@@ -225,6 +292,7 @@ Options:
   -s, --separator  Separator of the slug                                     [string] [default: "-"]
   -r, --replace    Custom string replacement                                   [array] [default: []]
   -i, --ignore     String list to ignore                                       [array] [default: []]
+  --lang           Source language                                            [string] [default: '']
   -S, --stdin      Use stdin as input                                     [boolean] [default: false]
   -h, --help       Show help                                                               [boolean]
 

data/languages-config.js

@@ -0,0 +1,97 @@
+module.exports = {
+  /**
+   * Bulgarian language
+   * https://en.wikipedia.org/wiki/Romanization_of_Bulgarian#Comparison_table
+   * Standard: BGN/PCGN (2013)
+   */
+  bg: {
+    replace: [
+      ['ц', 'ts'],
+      ['Ц', 'Ts'],
+      ['щ', 'sht'],
+      ['Щ', 'Sht'],
+      ['ъ', 'a'],
+      ['Ъ', 'A'],
+      ['ь', 'y'],
+      ['ѝ', 'i'],
+    ],
+  },
+  /**
+   * Macedonian language
+   * https://en.wikipedia.org/wiki/Romanization_of_Macedonian#ISO_9_system
+   * Standard: Official Documents/Cadastre
+   */
+  mk: {
+    replace: [
+      ['Ѓ', 'Gj'],
+      ['ѓ', 'gj'],
+      ['Ѕ', 'Dz'],
+      ['s', 'dz'],
+      ['Љ', 'Lj'],
+      ['љ', 'lj'],
+      ['Њ', 'Nj'],
+      ['њ', 'nj'],
+      ['є', 'ie'],
+      ['Є', 'Ie'],
+      ['Ќ', 'Kj'],
+      ['ќ', 'kj'],
+      ['Џ', 'Dj'],
+      ['џ', 'dj'],
+      ['ц', 'ts'],
+      ['Ц', 'Ts'],
+      ['щ', 'shch'],
+      ['Щ', 'Shch'],
+      ['И', 'Y'],
+      ['и', 'y'],
+    ],
+  },
+  /**
+   * Ukrainian language
+   * https://en.wikipedia.org/wiki/Romanization_of_Ukrainian
+   * Standard: Passport 2007
+   */
+  ua: {
+    replace: [
+      ['зг', 'zgh'],
+      ['Зг', 'Zgh'],
+      ['ґ', 'g'],
+      ['Ґ', 'G'],
+      ['є', 'ie'],
+      ['Є', 'Ie'],
+      ['ї', 'i'],
+      ['Ї', 'I'],
+      ['и', 'y'],
+      ['И', 'Y'],
+      ['х', 'kh'],
+      ['Х', 'Kh'],
+      ['ц', 'ts'],
+      ['Ц', 'Ts'],
+      ['щ', 'shch'],
+      ['Щ', 'Shch'],
+      ['ю', 'iu'],
+      ['Ю', 'Iu'],
+      ['я', 'ia'],
+      ['Я', 'Ia'],
+    ],
+  },
+  /**
+   * Serbian language
+   * https://en.wikipedia.org/wiki/Romanization_of_Serbian
+   */
+  rs: {
+    replace: [
+      ['Ђ', 'Dj'],
+      ['ђ', 'dj'],
+      ['Љ', 'Lj'],
+      ['љ', 'lj'],
+      ['Њ', 'Nj'],
+      ['њ', 'nj'],
+      ['Ћ', 'Ch'],
+      ['ћ', 'ch'],
+      ['Ч', 'Ch'],
+      ['ч', 'ch'],
+      ['Џ', 'Dj'],
+      ['џ', 'dj'],
+    ],
+  },
+};

gulpfile.babel.js

@@ -1,3 +1,4 @@
+/* eslint-disable function-paren-newline */
 import browserify from 'browserify';
 import gulp from 'gulp';
 import source from 'vinyl-source-stream';

lib/bin/slugify

@@ -17,14 +17,17 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 
 const STDIN_ENCODING = 'utf-8'; // eslint-disable-line import/no-unresolved
 
+/* eslint-disable no-console, function-paren-newline */
+
 const options = {
   lowercase: true,
   separator: '-',
   replace: [],
-  ignore: []
+  ignore: [],
+  lang: ''
 };
 
-const argv = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').option('l', {
+const { argv } = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').option('l', {
   alias: 'lowercase',
   default: options.lowercase,
   describe: 'Use lowercase',
@@ -44,14 +47,18 @@ const argv = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').op
   default: options.ignore,
   describe: 'String list to ignore',
   type: 'array'
+}).option('lang', {
+  default: options.lang,
+  describe: 'Source language',
+  type: 'string'
 }).option('S', {
   alias: 'stdin',
   default: false,
   describe: 'Use stdin as input',
   type: 'boolean'
 }).help('h').option('h', {
   alias: 'help'
-}).example('$0 "你好, world!" -r 好=good -r "world=Shi Jie"', 'Replace `,` into `!` and `world` into `shijie`.\nResult: ni-good-shi-jie').example('$0 "你好，世界!" -i 你好 -i ，', 'Ignore `你好` and `，`.\nResult: 你好，shi-jie').wrap(100).argv;
+}).example('$0 "你好, world!" -r 好=good -r "world=Shi Jie"', 'Replace `,` into `!` and `world` into `shijie`.\nResult: ni-good-shi-jie').example('$0 "你好，世界!" -i 你好 -i ，', 'Ignore `你好` and `，`.\nResult: 你好，shi-jie').wrap(100);
 
 options.lowercase = !!argv.l;
 options.separator = argv.separator;
@@ -66,6 +73,7 @@ if (argv.replace.length) {
   }
 }
 options.ignore = argv.ignore;
+options.lang = argv.lang;
 
 if (argv.stdin) {
   process.stdin.setEncoding(STDIN_ENCODING);

lib/bin/transliterate

@@ -15,13 +15,16 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 
 const STDIN_ENCODING = 'utf-8'; // eslint-disable-line import/no-unresolved
 
+/* eslint-disable no-console, function-paren-newline */
+
 const options = {
   unknown: '[?]',
   replace: [],
-  ignore: []
+  ignore: [],
+  lang: ''
 };
 
-const argv = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').option('u', {
+const { argv } = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').option('u', {
   alias: 'unknown',
   default: options.unknown,
   describe: 'Placeholder for unknown characters',
@@ -36,14 +39,19 @@ const argv = _yargs2.default.version().usage('Usage: $0 <unicode> [options]').op
   default: options.ignore,
   describe: 'String list to ignore',
   type: 'array'
+}).option('l', {
+  alias: 'lang',
+  default: options.lang,
+  describe: 'Source language',
+  type: 'string'
 }).option('S', {
   alias: 'stdin',
   default: false,
   describe: 'Use stdin as input',
   type: 'boolean'
 }).help('h').option('h', {
   alias: 'help'
-}).example('$0 "你好, world!" -r 好=good -r "world=Shi Jie"', 'Replace `,` into `!`, `world` into `shijie`.\nResult: Ni good, Shi Jie!').example('$0 "你好，世界!" -i 你好 -i ，', 'Ignore `你好` and `，`.\nResult: 你好，Shi Jie !').wrap(100).argv;
+}).example('$0 "你好, world!" -r 好=good -r "world=Shi Jie"', 'Replace `,` into `!`, `world` into `shijie`.\nResult: Ni good, Shi Jie!').example('$0 "你好，世界!" -i 你好 -i ，', 'Ignore `你好` and `，`.\nResult: 你好，Shi Jie !').wrap(100);
 
 options.unknown = argv.u;
 if (argv.replace.length) {
@@ -57,6 +65,7 @@ if (argv.replace.length) {
   }
 }
 options.ignore = argv.ignore;
+options.lang = argv.lang;
 
 if (argv.stdin) {
   process.stdin.setEncoding(STDIN_ENCODING);