tests/phpunit/unit/languages/LanguageCodeTest.php

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