MediaWiki
REL1_24
|
00001 <?php 00002 00010 class CollationTest extends MediaWikiLangTestCase { 00011 protected function setUp() { 00012 parent::setUp(); 00013 $this->checkPHPExtension( 'intl' ); 00014 } 00015 00029 public function testIsPrefix( $lang, $base, $extended ) { 00030 $cp = Collator::create( $lang ); 00031 $cp->setStrength( Collator::PRIMARY ); 00032 $baseBin = $cp->getSortKey( $base ); 00033 // Remove sortkey terminator 00034 $baseBin = rtrim( $baseBin, "\0" ); 00035 $extendedBin = $cp->getSortKey( $extended ); 00036 $this->assertStringStartsWith( $baseBin, $extendedBin, "$base is not a prefix of $extended" ); 00037 } 00038 00039 public static function prefixDataProvider() { 00040 return array( 00041 array( 'en', 'A', 'AA' ), 00042 array( 'en', 'A', 'AAA' ), 00043 array( 'en', 'Д', 'ДЂ' ), 00044 array( 'en', 'Д', 'ДA' ), 00045 // 'Ʒ' should expand to 'Z ' (note space). 00046 array( 'fi', 'Z', 'Ʒ' ), 00047 // 'Þ' should expand to 'th' 00048 array( 'sv', 't', 'Þ' ), 00049 // Javanese is a limited use alphabet, so should have 3 bytes 00050 // per character, so do some tests with it. 00051 array( 'en', 'ꦲ', 'ꦲꦤ' ), 00052 array( 'en', 'ꦲ', 'ꦲД' ), 00053 array( 'en', 'A', 'Aꦲ' ), 00054 ); 00055 } 00056 00062 public function testNotIsPrefix( $lang, $base, $extended ) { 00063 $cp = Collator::create( $lang ); 00064 $cp->setStrength( Collator::PRIMARY ); 00065 $baseBin = $cp->getSortKey( $base ); 00066 // Remove sortkey terminator 00067 $baseBin = rtrim( $baseBin, "\0" ); 00068 $extendedBin = $cp->getSortKey( $extended ); 00069 $this->assertStringStartsNotWith( $baseBin, $extendedBin, "$base is a prefix of $extended" ); 00070 } 00071 00072 public static function notPrefixDataProvider() { 00073 return array( 00074 array( 'en', 'A', 'B' ), 00075 array( 'en', 'AC', 'ABC' ), 00076 array( 'en', 'Z', 'Ʒ' ), 00077 array( 'en', 'A', 'ꦲ' ), 00078 ); 00079 } 00080 00090 public function testGetFirstLetter( $collation, $string, $firstLetter ) { 00091 $col = Collation::factory( $collation ); 00092 $this->assertEquals( $firstLetter, $col->getFirstLetter( $string ) ); 00093 } 00094 00095 function firstLetterProvider() { 00096 return array( 00097 array( 'uppercase', 'Abc', 'A' ), 00098 array( 'uppercase', 'abc', 'A' ), 00099 array( 'identity', 'abc', 'a' ), 00100 array( 'uca-en', 'abc', 'A' ), 00101 array( 'uca-en', ' ', ' ' ), 00102 array( 'uca-en', 'Êveryone', 'E' ), 00103 array( 'uca-vi', 'Êveryone', 'Ê' ), 00104 // Make sure thorn is not a first letter. 00105 array( 'uca-sv', 'The', 'T' ), 00106 array( 'uca-sv', 'Å', 'Å' ), 00107 array( 'uca-hu', 'dzsdo', 'Dzs' ), 00108 array( 'uca-hu', 'dzdso', 'Dz' ), 00109 array( 'uca-hu', 'CSD', 'Cs' ), 00110 array( 'uca-root', 'CSD', 'C' ), 00111 array( 'uca-fi', 'Ǥ', 'G' ), 00112 array( 'uca-fi', 'Ŧ', 'T' ), 00113 array( 'uca-fi', 'Ʒ', 'Z' ), 00114 array( 'uca-fi', 'Ŋ', 'N' ), 00115 ); 00116 } 00117 }