tests/phpunit/includes/GlobalFunctions/wfBCP47Test.php

   1 <?php
   2 /**
   3  * @group GlobalFunctions
   4  * @covers ::wfBCP47
   5  */
   6 class WfBCP47Test extends MediaWikiTestCase {
   7         /**
   8          * test @see wfBCP47().
   9          * Please note the BCP 47 explicitly state that language codes are case
  10          * insensitive, there are some exceptions to the rule :)
  11          * This test is used to verify our formatting against all lower and
  12          * all upper cases language code.
  13          *
  14          * @see https://tools.ietf.org/html/bcp47
  15          * @dataProvider provideLanguageCodes()
  16          */
  17         public function testBCP47( $code, $expected ) {
  18                 $code = strtolower( $code );
  19                 $this->assertEquals( $expected, wfBCP47( $code ),
  20                         "Applying BCP47 standard to lower case '$code'"
  21                 );
  22
  23                 $code = strtoupper( $code );
  24                 $this->assertEquals( $expected, wfBCP47( $code ),
  25                         "Applying BCP47 standard to upper case '$code'"
  26                 );
  27         }
  28
  29         /**
  30          * Array format is ($code, $expected)
  31          */
  32         public static function provideLanguageCodes() {
  33                 return [
  34                         // Extracted from BCP 47 (list not exhaustive)
  35                         # 2.1.1
  36                         [ 'en-ca-x-ca', 'en-CA-x-ca' ],
  37                         [ 'sgn-be-fr', 'sgn-BE-FR' ],
  38                         [ 'az-latn-x-latn', 'az-Latn-x-latn' ],
  39                         # 2.2
  40                         [ 'sr-Latn-RS', 'sr-Latn-RS' ],
  41                         [ 'az-arab-ir', 'az-Arab-IR' ],
  42
  43                         # 2.2.5
  44                         [ 'sl-nedis', 'sl-nedis' ],
  45                         [ 'de-ch-1996', 'de-CH-1996' ],
  46
  47                         # 2.2.6
  48                         [
  49                                 'en-latn-gb-boont-r-extended-sequence-x-private',
  50                                 'en-Latn-GB-boont-r-extended-sequence-x-private'
  51                         ],
  52
  53                         // Examples from BCP 47 Appendix A
  54                         # Simple language subtag:
  55                         [ 'DE', 'de' ],
  56                         [ 'fR', 'fr' ],
  57                         [ 'ja', 'ja' ],
  58
  59                         # Language subtag plus script subtag:
  60                         [ 'zh-hans', 'zh-Hans' ],
  61                         [ 'sr-cyrl', 'sr-Cyrl' ],
  62                         [ 'sr-latn', 'sr-Latn' ],
  63
  64                         # Extended language subtags and their primary language subtag
  65                         # counterparts:
  66                         [ 'zh-cmn-hans-cn', 'zh-cmn-Hans-CN' ],
  67                         [ 'cmn-hans-cn', 'cmn-Hans-CN' ],
  68                         [ 'zh-yue-hk', 'zh-yue-HK' ],
  69                         [ 'yue-hk', 'yue-HK' ],
  70
  71                         # Language-Script-Region:
  72                         [ 'zh-hans-cn', 'zh-Hans-CN' ],
  73                         [ 'sr-latn-RS', 'sr-Latn-RS' ],
  74
  75                         # Language-Variant:
  76                         [ 'sl-rozaj', 'sl-rozaj' ],
  77                         [ 'sl-rozaj-biske', 'sl-rozaj-biske' ],
  78                         [ 'sl-nedis', 'sl-nedis' ],
  79
  80                         # Language-Region-Variant:
  81                         [ 'de-ch-1901', 'de-CH-1901' ],
  82                         [ 'sl-it-nedis', 'sl-IT-nedis' ],
  83
  84                         # Language-Script-Region-Variant:
  85                         [ 'hy-latn-it-arevela', 'hy-Latn-IT-arevela' ],
  86
  87                         # Language-Region:
  88                         [ 'de-de', 'de-DE' ],
  89                         [ 'en-us', 'en-US' ],
  90                         [ 'es-419', 'es-419' ],
  91
  92                         # Private use subtags:
  93                         [ 'de-ch-x-phonebk', 'de-CH-x-phonebk' ],
  94                         [ 'az-arab-x-aze-derbend', 'az-Arab-x-aze-derbend' ],
  95                         /**
  96                          * Previous test does not reflect the BCP 47 which states:
  97                          *  az-Arab-x-AZE-derbend
  98                          * AZE being private, it should be lower case, hence the test above
  99                          * should probably be:
 100                          * [ 'az-arab-x-aze-derbend', 'az-Arab-x-AZE-derbend' ],
 101                          */
 102
 103                         # Private use registry values:
 104                         [ 'x-whatever', 'x-whatever' ],
 105                         [ 'qaa-qaaa-qm-x-southern', 'qaa-Qaaa-QM-x-southern' ],
 106                         [ 'de-qaaa', 'de-Qaaa' ],
 107                         [ 'sr-latn-qm', 'sr-Latn-QM' ],
 108                         [ 'sr-qaaa-rs', 'sr-Qaaa-RS' ],
 109
 110                         # Tags that use extensions
 111                         [ 'en-us-u-islamcal', 'en-US-u-islamcal' ],
 112                         [ 'zh-cn-a-myext-x-private', 'zh-CN-a-myext-x-private' ],
 113                         [ 'en-a-myext-b-another', 'en-a-myext-b-another' ],
 114
 115                         # Invalid:
 116                         // de-419-DE
 117                         // a-DE
 118                         // ar-a-aaa-b-bbb-a-ccc
 119                 ];
 120         }
 121 }