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