tests/phpunit/includes/CollationTest.php

   1 <?php
   2
   3 /**
   4  * Class CollationTest
   5  * @covers Collation
   6  * @covers IcuCollation
   7  * @covers IdentityCollation
   8  * @covers UppercaseCollation
   9  */
  10 class CollationTest extends MediaWikiLangTestCase {
  11         protected function setUp() {
  12                 parent::setUp();
  13                 if ( !extension_loaded( 'intl' ) ) {
  14                         $this->markTestSkipped( 'These tests require intl extension' );
  15                 }
  16         }
  17
  18         /**
  19          * Test to make sure, that if you
  20          * have "X" and "XY", the binary
  21          * sortkey also has "X" being a
  22          * prefix of "XY". Our collation
  23          * code makes this assumption.
  24          *
  25          * @param $lang String Language code for collator
  26          * @param $base String Base string
  27          * @param $extended String String containing base as a prefix.
  28          *
  29          * @dataProvider prefixDataProvider
  30          */
  31         public function testIsPrefix( $lang, $base, $extended ) {
  32                 $cp = Collator::create( $lang );
  33                 $cp->setStrength( Collator::PRIMARY );
  34                 $baseBin = $cp->getSortKey( $base );
  35                 // Remove sortkey terminator
  36                 $baseBin = rtrim( $baseBin, "\0" );
  37                 $extendedBin = $cp->getSortKey( $extended );
  38                 $this->assertStringStartsWith( $baseBin, $extendedBin, "$base is not a prefix of $extended" );
  39         }
  40
  41         public static function prefixDataProvider() {
  42                 return array(
  43                         array( 'en', 'A', 'AA' ),
  44                         array( 'en', 'A', 'AAA' ),
  45                         array( 'en', 'Д', 'ДЂ' ),
  46                         array( 'en', 'Д', 'ДA' ),
  47                         // 'Ʒ' should expand to 'Z ' (note space).
  48                         array( 'fi', 'Z', 'Ʒ' ),
  49                         // 'Þ' should expand to 'th'
  50                         array( 'sv', 't', 'Þ' ),
  51                         // Javanese is a limited use alphabet, so should have 3 bytes
  52                         // per character, so do some tests with it.
  53                         array( 'en', 'ꦲ', 'ꦲꦤ' ),
  54                         array( 'en', 'ꦲ', 'ꦲД' ),
  55                         array( 'en', 'A', 'Aꦲ' ),
  56                 );
  57         }
  58
  59         /**
  60          * Opposite of testIsPrefix
  61          *
  62          * @dataProvider notPrefixDataProvider
  63          */
  64         public function testNotIsPrefix( $lang, $base, $extended ) {
  65                 $cp = Collator::create( $lang );
  66                 $cp->setStrength( Collator::PRIMARY );
  67                 $baseBin = $cp->getSortKey( $base );
  68                 // Remove sortkey terminator
  69                 $baseBin = rtrim( $baseBin, "\0" );
  70                 $extendedBin = $cp->getSortKey( $extended );
  71                 $this->assertStringStartsNotWith( $baseBin, $extendedBin, "$base is a prefix of $extended" );
  72         }
  73
  74         public static function notPrefixDataProvider() {
  75                 return array(
  76                         array( 'en', 'A', 'B' ),
  77                         array( 'en', 'AC', 'ABC' ),
  78                         array( 'en', 'Z', 'Ʒ' ),
  79                         array( 'en', 'A', 'ꦲ' ),
  80                 );
  81         }
  82
  83         /**
  84          * Test correct first letter is fetched.
  85          *
  86          * @param $collation String Collation name (aka uca-en)
  87          * @param $string String String to get first letter of
  88          * @param $firstLetter String Expected first letter.
  89          *
  90          * @dataProvider firstLetterProvider
  91          */
  92         public function testGetFirstLetter( $collation, $string, $firstLetter ) {
  93                 $col = Collation::factory( $collation );
  94                 $this->assertEquals( $firstLetter, $col->getFirstLetter( $string ) );
  95         }
  96
  97         function firstLetterProvider() {
  98                 return array(
  99                         array( 'uppercase', 'Abc', 'A' ),
 100                         array( 'uppercase', 'abc', 'A' ),
 101                         array( 'identity', 'abc', 'a' ),
 102                         array( 'uca-en', 'abc', 'A' ),
 103                         array( 'uca-en', ' ', ' ' ),
 104                         array( 'uca-en', 'Êveryone', 'E' ),
 105                         array( 'uca-vi', 'Êveryone', 'Ê' ),
 106                         // Make sure thorn is not a first letter.
 107                         array( 'uca-sv', 'The', 'T' ),
 108                         array( 'uca-sv', 'Å', 'Å' ),
 109                         array( 'uca-hu', 'dzsdo', 'Dzs' ),
 110                         array( 'uca-hu', 'dzdso', 'Dz' ),
 111                         array( 'uca-hu', 'CSD', 'Cs' ),
 112                         array( 'uca-root', 'CSD', 'C' ),
 113                         array( 'uca-fi', 'Ǥ', 'G' ),
 114                         array( 'uca-fi', 'Ŧ', 'T' ),
 115                         array( 'uca-fi', 'Ʒ', 'Z' ),
 116                         array( 'uca-fi', 'Ŋ', 'N' ),
 117                 );
 118         }
 119 }