d87ea27fb3
This implements Language::toBcp47Code(), an interface defined in the wikimedia/bcp-47-code library, in order to facilitate interacting with the parsoid library (which shouldn't know anything about mediawiki-internal language codes). Change-Id: Id09218ae97fd8167866064c0a3f6142696d66a8d
153 lines
4.5 KiB
PHP
153 lines
4.5 KiB
PHP
<?php
|
||
|
||
use MediaWiki\Languages\LanguageConverterFactory;
|
||
use MediaWiki\Languages\LanguageFallback;
|
||
use MediaWiki\Languages\LanguageNameUtils;
|
||
|
||
/**
|
||
* @coversDefaultClass Language
|
||
*/
|
||
class LanguageTest extends MediaWikiUnitTestCase {
|
||
/**
|
||
* @param array $options Valid keys:
|
||
* 'code'
|
||
* 'grammarTransformCache'
|
||
* @return Language
|
||
*/
|
||
private function getObj( array $options = [] ) {
|
||
return new Language(
|
||
$options['code'] ?? 'en',
|
||
$this->createNoOpMock( NamespaceInfo::class ),
|
||
$this->createNoOpMock( LocalisationCache::class ),
|
||
$this->createNoOpMock( LanguageNameUtils::class ),
|
||
$this->createNoOpMock( LanguageFallback::class ),
|
||
$this->createNoOpMock( LanguageConverterFactory::class ),
|
||
$this->createHookContainer(),
|
||
new HashConfig( [] )
|
||
);
|
||
}
|
||
|
||
/**
|
||
* @covers ::getCode
|
||
* @dataProvider provideCodes
|
||
*/
|
||
public function testGetCode( $code, $bcp47code ) {
|
||
$lang = $this->getObj( [ 'code' => $code ] );
|
||
$this->assertSame( $code, $lang->getCode() );
|
||
}
|
||
|
||
/**
|
||
* @covers ::getHtmlCode
|
||
* @dataProvider provideCodes
|
||
*/
|
||
public function testGetHtmlCode( $code, $bcp47code ) {
|
||
$lang = $this->getObj( [ 'code' => $code ] );
|
||
$this->assertSame( $bcp47code, $lang->getHtmlCode() );
|
||
}
|
||
|
||
/**
|
||
* @covers ::toBcp47Code
|
||
* @dataProvider provideCodes
|
||
*/
|
||
public function testToBcp47Code( $code, $bcp47code ) {
|
||
$lang = $this->getObj( [ 'code' => $code ] );
|
||
$this->assertSame( $bcp47code, $lang->toBcp47Code() );
|
||
}
|
||
|
||
public function provideCodes() {
|
||
return LanguageCodeTest::provideLanguageCodes();
|
||
}
|
||
|
||
/**
|
||
* @covers ::getGrammarTransformations
|
||
* @todo Test the exception case
|
||
*/
|
||
public function testGetGrammarTransformations() {
|
||
global $IP;
|
||
|
||
// XXX Inject a value here instead of reading the filesystem (T231159)
|
||
$expected = FormatJson::decode(
|
||
file_get_contents( "$IP/languages/data/grammarTransformations/he.json" ), true );
|
||
|
||
$lang = $this->getObj( [ 'code' => 'he' ] );
|
||
|
||
$this->assertSame( $expected, $lang->getGrammarTransformations() );
|
||
$this->assertSame( $expected, $lang->getGrammarTransformations() );
|
||
}
|
||
|
||
/**
|
||
* @covers ::getGrammarTransformations
|
||
*/
|
||
public function testGetGrammarTransformations_empty() {
|
||
$lang = $this->getObj();
|
||
|
||
$this->assertSame( [], $lang->getGrammarTransformations() );
|
||
$this->assertSame( [], $lang->getGrammarTransformations() );
|
||
}
|
||
|
||
/**
|
||
* @covers ::ucwords
|
||
* @dataProvider provideUcwords
|
||
*/
|
||
public function testUcwords( string $input, string $expected ) {
|
||
$lang = $this->getObj();
|
||
$this->assertSame( $expected, $lang->ucwords( $input ) );
|
||
}
|
||
|
||
public function provideUcwords() {
|
||
return [
|
||
'Empty string' => [ '', '' ],
|
||
'Non-alpha only' => [ '3212-353', '3212-353' ],
|
||
'Non-alpha only, multiple words' => [ '@%#, #@$%!', '@%#, #@$%!' ],
|
||
'Single ASCII word' => [ 'teSt', 'Test' ],
|
||
'ASCII words' => [ 'TeSt 123 test foo-bar', 'Test 123 Test Foo-bar' ],
|
||
'Single multibyte word' => [ 'теСт', 'Тест' ],
|
||
'Multibyte words' => [ 'ТесТ 123, тест 测试 test раз-два', 'Тест 123, Тест 测试 Test Раз-два' ],
|
||
];
|
||
}
|
||
|
||
/**
|
||
* @covers ::ucwordbreaks
|
||
* @dataProvider provideUcwordbreaks
|
||
*/
|
||
public function testUcwordbreaks( string $input, string $expected ) {
|
||
$lang = $this->getObj();
|
||
$this->assertSame( $expected, $lang->ucwordbreaks( $input ) );
|
||
}
|
||
|
||
public function provideUcwordbreaks() {
|
||
return [
|
||
'Empty string' => [ '', '' ],
|
||
'Non-alpha only' => [ '3212-353', '3212-353' ],
|
||
'Non-alpha only, multiple words' => [ '@%#, #@$%!', '@%#, #@$%!' ],
|
||
'Single ASCII word' => [ 'teSt', 'TeSt' ],
|
||
'Single ASCII word, prefixed' => [ '-teSt', '-TeSt' ],
|
||
'ASCII words' => [ 'TeSt 123 test foo-bar', 'TeSt 123 Test Foo-Bar' ],
|
||
'Single multibyte word' => [ 'теСт', 'Тест' ],
|
||
'Single multibyte word, prefixed' => [ '-теСт', '-Тест' ],
|
||
'Multibyte words' => [ 'ТесТ 123, тест 测试 test раз-два', 'Тест 123, Тест 测试 Test Раз-Два' ],
|
||
];
|
||
}
|
||
|
||
/**
|
||
* @covers ::firstChar
|
||
* @dataProvider provideFirstChar
|
||
*/
|
||
public function testFirstChar( string $input, string $expected ) {
|
||
$lang = $this->getObj();
|
||
$this->assertSame( $expected, $lang->firstChar( $input ) );
|
||
}
|
||
|
||
public function provideFirstChar() {
|
||
return [
|
||
'Empty string' => [ '', '' ],
|
||
'Single Latin' => [ 'T', 'T' ],
|
||
'Latin' => [ 'TEST', 'T' ],
|
||
'Digit' => [ '123', '1' ],
|
||
'Russian' => [ 'Ёпт', 'Ё' ],
|
||
'Emoji' => [ '😂💀☢️', '😂' ],
|
||
// Korean is special-cased to remove single letters from syllables
|
||
'Korean' => [ '위키백과', 'ㅇ' ],
|
||
];
|
||
}
|
||
}
|