tests/phpunit/includes/CollationTest.php

   1 <?php
   2 class CollationTest extends MediaWikiLangTestCase {
   3         protected function setUp() {
   4                 parent::setUp();
   5                 if ( !extension_loaded( 'intl' ) ) {
   6                         $this->markTestSkipped( 'These tests require intl extension' );
   7                 }
   8         }
   9
  10         /**
  11          * Test to make sure, that if you
  12          * have "X" and "XY", the binary
  13          * sortkey also has "X" being a
  14          * prefix of "XY". Our collation
  15          * code makes this assumption.
  16          *
  17          * @param $lang String Language code for collator
  18          * @param $base String Base string
  19          * @param $extended String String containing base as a prefix.
  20          *
  21          * @dataProvider prefixDataProvider
  22          */
  23         function testIsPrefix( $lang, $base, $extended ) {
  24                 $cp = Collator::create( $lang );
  25                 $cp->setStrength( Collator::PRIMARY );
  26                 $baseBin = $cp->getSortKey( $base );
  27                 // Remove sortkey terminator
  28                 $baseBin = rtrim( $baseBin, "\0" );
  29                 $extendedBin = $cp->getSortKey( $extended );
  30                 $this->assertStringStartsWith( $baseBin, $extendedBin, "$base is not a prefix of $extended" );
  31         }
  32
  33         function prefixDataProvider() {
  34                 return array(
  35                         array( 'en', 'A', 'AA' ),
  36                         array( 'en', 'A', 'AAA' ),
  37                         array( 'en', 'Д', 'ДЂ' ),
  38                         array( 'en', 'Д', 'ДA' ),
  39                         // 'Ʒ' should expand to 'Z ' (note space).
  40                         array( 'fi', 'Z', 'Ʒ' ),
  41                         // 'Þ' should expand to 'th'
  42                         array( 'sv', 't', 'Þ' ),
  43                         // Javanese is a limited use alphabet, so should have 3 bytes
  44                         // per character, so do some tests with it.
  45                         array( 'en', 'ꦲ', 'ꦲꦤ' ),
  46                         array( 'en', 'ꦲ', 'ꦲД' ),
  47                         array( 'en', 'A', 'Aꦲ' ),
  48                 );
  49         }
  50         /**
  51          * Opposite of testIsPrefix
  52          *
  53          * @dataProvider notPrefixDataProvider
  54          */
  55         function testNotIsPrefix( $lang, $base, $extended ) {
  56                 $cp = Collator::create( $lang );
  57                 $cp->setStrength( Collator::PRIMARY );
  58                 $baseBin = $cp->getSortKey( $base );
  59                 // Remove sortkey terminator
  60                 $baseBin = rtrim( $baseBin, "\0" );
  61                 $extendedBin = $cp->getSortKey( $extended );
  62                 $this->assertStringStartsNotWith( $baseBin, $extendedBin, "$base is a prefix of $extended" );
  63         }
  64
  65         function notPrefixDataProvider() {
  66                 return array(
  67                         array( 'en', 'A', 'B' ),
  68                         array( 'en', 'AC', 'ABC' ),
  69                         array( 'en', 'Z', 'Ʒ' ),
  70                         array( 'en', 'A', 'ꦲ' ),
  71                 );
  72         }
  73
  74         /**
  75          * Test correct first letter is fetched.
  76          *
  77          * @param $collation String Collation name (aka uca-en)
  78          * @param $string String String to get first letter of
  79          * @param $firstLetter String Expected first letter.
  80          *
  81          * @dataProvider firstLetterProvider
  82          */
  83         function testGetFirstLetter( $collation, $string, $firstLetter ) {
  84                 $col = Collation::factory( $collation );
  85                 $this->assertEquals( $firstLetter, $col->getFirstLetter( $string ) );
  86         }
  87         function firstLetterProvider() {
  88                 return array(
  89                         array( 'uppercase', 'Abc', 'A' ),
  90                         array( 'uppercase', 'abc', 'A' ),
  91                         array( 'identity', 'abc', 'a' ),
  92                         array( 'uca-en', 'abc', 'A' ),
  93                         array( 'uca-en', ' ', ' ' ),
  94                         array( 'uca-en', 'Êveryone', 'E' ),
  95                         array( 'uca-vi', 'Êveryone', 'Ê' ),
  96                         // Make sure thorn is not a first letter.
  97                         array( 'uca-sv', 'The', 'T' ),
  98                         array( 'uca-sv', 'Å', 'Å' ),
  99                         array( 'uca-hu', 'dzsdo', 'Dzs' ),
 100                         array( 'uca-hu', 'dzdso', 'Dz' ),
 101                         array( 'uca-hu', 'CSD', 'Cs' ),
 102                         array( 'uca-root', 'CSD', 'C' ),
 103                         array( 'uca-fi', 'Ǥ', 'G' ),
 104                         array( 'uca-fi', 'Ŧ', 'T' ),
 105                         array( 'uca-fi', 'Ʒ', 'Z' ),
 106                         array( 'uca-fi', 'Ŋ', 'N' ),
 107                 );
 108         }
 109 }