MediaWiki REL1_35
CustomUppercaseCollation.php
Go to the documentation of this file.
1<?php
41
43 private $alphabet;
44
46 private $puaSubset;
47
50
57 public function __construct( array $alphabet, Language $lang ) {
58 if ( count( $alphabet ) < 1 || count( $alphabet ) >= 4096 ) {
59 throw new UnexpectedValueException( "Alphabet must be < 4096 items" );
60 }
61 $this->firstLetters = $alphabet;
62 // For digraphs, only the first letter is capitalized in input
63 $this->alphabet = array_map( [ $lang, 'uc' ], $alphabet );
64
65 $this->puaSubset = [];
66 $len = count( $alphabet );
67 for ( $i = 0; $i < $len; $i++ ) {
68 $this->puaSubset[] = "\xF3\xB3" . chr( floor( $i / 64 ) + 128 ) . chr( ( $i % 64 ) + 128 );
69 }
70
71 // Sort these arrays so that any trigraphs, digraphs etc. are first
72 // (and they get replaced first in convertToPua()).
73 $lengths = array_map( 'mb_strlen', $this->alphabet );
74 array_multisort( $lengths, SORT_DESC, $this->firstLetters, $this->alphabet, $this->puaSubset );
75
76 parent::__construct( $lang );
77 }
78
79 private function convertToPua( $string ) {
80 return str_replace( $this->alphabet, $this->puaSubset, $string );
81 }
82
83 public function getSortKey( $string ) {
84 return $this->convertToPua( parent::getSortKey( $string ) );
85 }
86
87 public function getFirstLetter( $string ) {
88 $sortkey = $this->getSortKey( $string );
89
90 // In case a title begins with a character from our alphabet, return the corresponding
91 // first-letter. (This also happens if the title has a corresponding PUA code in it, to avoid
92 // inconsistent behaviour. This class mostly assumes that people will not use PUA codes.)
93 $index = array_search( substr( $sortkey, 0, 4 ), $this->puaSubset );
94 if ( $index !== false ) {
95 return $this->firstLetters[ $index ];
96 }
97
98 // String begins with a character outside of our alphabet, fall back
99 return parent::getFirstLetter( $string );
100 }
101}
Resort normal UTF-8 order by putting a bunch of stuff in PUA.
getSortKey( $string)
Given a string, convert it to a (hopefully short) key that can be used for efficient sorting.
array $puaSubset
List of private use area codes.
__construct(array $alphabet, Language $lang)
getFirstLetter( $string)
Given a string, return the logical "first letter" to be used for grouping on category pages and so on...
array $alphabet
Sorted array of letters.
Internationalisation code See https://www.mediawiki.org/wiki/Special:MyLanguage/Localisation for more...
Definition Language.php:41
Collation that orders text with numbers "naturally", so that 'Foo 1' < 'Foo 2' < 'Foo 12'.