tests/phpunit/languages/LanguageCodeTest.php

   1 <?php
   2
   3 /**
   4  * @covers LanguageCode
   5  * @group Language
   6  *
   7  * @author Thiemo Kreuz
   8  */
   9 class LanguageCodeTest extends PHPUnit\Framework\TestCase {
  10
  11         use MediaWikiCoversValidator;
  12
  13         public function testConstructor() {
  14                 $instance = new LanguageCode();
  15
  16                 $this->assertInstanceOf( LanguageCode::class, $instance );
  17         }
  18
  19         public function testGetDeprecatedCodeMapping() {
  20                 $map = LanguageCode::getDeprecatedCodeMapping();
  21
  22                 $this->assertInternalType( 'array', $map );
  23                 $this->assertContainsOnly( 'string', array_keys( $map ) );
  24                 $this->assertArrayNotHasKey( '', $map );
  25                 $this->assertContainsOnly( 'string', $map );
  26                 $this->assertNotContains( '', $map );
  27
  28                 // Codes special to MediaWiki should never appear in a map of "deprecated" codes
  29                 $this->assertArrayNotHasKey( 'qqq', $map, 'documentation' );
  30                 $this->assertNotContains( 'qqq', $map, 'documentation' );
  31                 $this->assertArrayNotHasKey( 'qqx', $map, 'debug code' );
  32                 $this->assertNotContains( 'qqx', $map, 'debug code' );
  33
  34                 // Valid language codes that are currently not "deprecated"
  35                 $this->assertArrayNotHasKey( 'bh', $map, 'family of Bihari languages' );
  36                 $this->assertArrayNotHasKey( 'no', $map, 'family of Norwegian languages' );
  37                 $this->assertArrayNotHasKey( 'simple', $map );
  38         }
  39
  40         public function testReplaceDeprecatedCodes() {
  41                 $this->assertEquals( 'gsw', LanguageCode::replaceDeprecatedCodes( 'als' ) );
  42                 $this->assertEquals( 'gsw', LanguageCode::replaceDeprecatedCodes( 'gsw' ) );
  43                 $this->assertEquals( null, LanguageCode::replaceDeprecatedCodes( null ) );
  44         }
  45
  46         /**
  47          * test @see LanguageCode::bcp47().
  48          * Please note the BCP 47 explicitly state that language codes are case
  49          * insensitive, there are some exceptions to the rule :)
  50          * This test is used to verify our formatting against all lower and
  51          * all upper cases language code.
  52          *
  53          * @see https://tools.ietf.org/html/bcp47
  54          * @dataProvider provideLanguageCodes()
  55          */
  56         public function testBcp47( $code, $expected ) {
  57                 $this->assertEquals( $expected, LanguageCode::bcp47( $code ),
  58                         "Applying BCP 47 standard to '$code'"
  59                 );
  60
  61                 $code = strtolower( $code );
  62                 $this->assertEquals( $expected, LanguageCode::bcp47( $code ),
  63                         "Applying BCP 47 standard to lower case '$code'"
  64                 );
  65
  66                 $code = strtoupper( $code );
  67                 $this->assertEquals( $expected, LanguageCode::bcp47( $code ),
  68                         "Applying BCP 47 standard to upper case '$code'"
  69                 );
  70         }
  71
  72         /**
  73          * Array format is ($code, $expected)
  74          */
  75         public static function provideLanguageCodes() {
  76                 return [
  77                         // Extracted from BCP 47 (list not exhaustive)
  78                         # 2.1.1
  79                         [ 'en-ca-x-ca', 'en-CA-x-ca' ],
  80                         [ 'sgn-be-fr', 'sgn-BE-FR' ],
  81                         [ 'az-latn-x-latn', 'az-Latn-x-latn' ],
  82                         # 2.2
  83                         [ 'sr-Latn-RS', 'sr-Latn-RS' ],
  84                         [ 'az-arab-ir', 'az-Arab-IR' ],
  85
  86                         # 2.2.5
  87                         [ 'sl-nedis', 'sl-nedis' ],
  88                         [ 'de-ch-1996', 'de-CH-1996' ],
  89
  90                         # 2.2.6
  91                         [
  92                                 'en-latn-gb-boont-r-extended-sequence-x-private',
  93                                 'en-Latn-GB-boont-r-extended-sequence-x-private'
  94                         ],
  95
  96                         // Examples from BCP 47 Appendix A
  97                         # Simple language subtag:
  98                         [ 'DE', 'de' ],
  99                         [ 'fR', 'fr' ],
 100                         [ 'ja', 'ja' ],
 101
 102                         # Language subtag plus script subtag:
 103                         [ 'zh-hans', 'zh-Hans' ],
 104                         [ 'sr-cyrl', 'sr-Cyrl' ],
 105                         [ 'sr-latn', 'sr-Latn' ],
 106
 107                         # Extended language subtags and their primary language subtag
 108                         # counterparts:
 109                         [ 'zh-cmn-hans-cn', 'zh-cmn-Hans-CN' ],
 110                         [ 'cmn-hans-cn', 'cmn-Hans-CN' ],
 111                         [ 'zh-yue-hk', 'zh-yue-HK' ],
 112                         [ 'yue-hk', 'yue-HK' ],
 113
 114                         # Language-Script-Region:
 115                         [ 'zh-hans-cn', 'zh-Hans-CN' ],
 116                         [ 'sr-latn-RS', 'sr-Latn-RS' ],
 117
 118                         # Language-Variant:
 119                         [ 'sl-rozaj', 'sl-rozaj' ],
 120                         [ 'sl-rozaj-biske', 'sl-rozaj-biske' ],
 121                         [ 'sl-nedis', 'sl-nedis' ],
 122
 123                         # Language-Region-Variant:
 124                         [ 'de-ch-1901', 'de-CH-1901' ],
 125                         [ 'sl-it-nedis', 'sl-IT-nedis' ],
 126
 127                         # Language-Script-Region-Variant:
 128                         [ 'hy-latn-it-arevela', 'hy-Latn-IT-arevela' ],
 129
 130                         # Language-Region:
 131                         [ 'de-de', 'de-DE' ],
 132                         [ 'en-us', 'en-US' ],
 133                         [ 'es-419', 'es-419' ],
 134
 135                         # Private use subtags:
 136                         [ 'de-ch-x-phonebk', 'de-CH-x-phonebk' ],
 137                         [ 'az-arab-x-aze-derbend', 'az-Arab-x-aze-derbend' ],
 138                         /**
 139                          * Previous test does not reflect the BCP 47 which states:
 140                          *  az-Arab-x-AZE-derbend
 141                          * AZE being private, it should be lower case, hence the test above
 142                          * should probably be:
 143                          * [ 'az-arab-x-aze-derbend', 'az-Arab-x-AZE-derbend' ],
 144                          */
 145
 146                         # Private use registry values:
 147                         [ 'x-whatever', 'x-whatever' ],
 148                         [ 'qaa-qaaa-qm-x-southern', 'qaa-Qaaa-QM-x-southern' ],
 149                         [ 'de-qaaa', 'de-Qaaa' ],
 150                         [ 'sr-latn-qm', 'sr-Latn-QM' ],
 151                         [ 'sr-qaaa-rs', 'sr-Qaaa-RS' ],
 152
 153                         # Tags that use extensions
 154                         [ 'en-us-u-islamcal', 'en-US-u-islamcal' ],
 155                         [ 'zh-cn-a-myext-x-private', 'zh-CN-a-myext-x-private' ],
 156                         [ 'en-a-myext-b-another', 'en-a-myext-b-another' ],
 157
 158                         # Invalid:
 159                         // de-419-DE
 160                         // a-DE
 161                         // ar-a-aaa-b-bbb-a-ccc
 162
 163                         # Non-standard and deprecated language codes used by MediaWiki
 164                         [ 'als', 'gsw' ],
 165                         [ 'bat-smg', 'sgs' ],
 166                         [ 'be-x-old', 'be-tarask' ],
 167                         [ 'fiu-vro', 'vro' ],
 168                         [ 'roa-rup', 'rup' ],
 169                         [ 'zh-classical', 'lzh' ],
 170                         [ 'zh-min-nan', 'nan' ],
 171                         [ 'zh-yue', 'yue' ],
 172                         [ 'cbk-zam', 'cbk' ],
 173                         [ 'de-formal', 'de-x-formal' ],
 174                         [ 'eml', 'egl' ],
 175                         [ 'en-rtl', 'en-x-rtl' ],
 176                         [ 'es-formal', 'es-x-formal' ],
 177                         [ 'hu-formal', 'hu-x-formal' ],
 178                         [ 'kk-Arab', 'kk-Arab' ],
 179                         [ 'kk-Cyrl', 'kk-Cyrl' ],
 180                         [ 'kk-Latn', 'kk-Latn' ],
 181                         [ 'map-bms', 'jv-x-bms' ],
 182                         [ 'mo', 'ro-Cyrl-MD' ],
 183                         [ 'nrm', 'nrf' ],
 184                         [ 'nl-informal', 'nl-x-informal' ],
 185                         [ 'roa-tara', 'nap-x-tara' ],
 186                         [ 'simple', 'en-simple' ],
 187                         [ 'sr-ec', 'sr-Cyrl' ],
 188                         [ 'sr-el', 'sr-Latn' ],
 189                         [ 'zh-cn', 'zh-Hans-CN' ],
 190                         [ 'zh-sg', 'zh-Hans-SG' ],
 191                         [ 'zh-my', 'zh-Hans-MY' ],
 192                         [ 'zh-tw', 'zh-Hant-TW' ],
 193                         [ 'zh-hk', 'zh-Hant-HK' ],
 194                         [ 'zh-mo', 'zh-Hant-MO' ],
 195                         [ 'zh-hans', 'zh-Hans' ],
 196                         [ 'zh-hant', 'zh-Hant' ],
 197                 ];
 198         }
 199
 200 }