summaryrefslogtreecommitdiff
path: root/includes/Collation.php
blob: c1f0b3881a13ef328239189512bca2542726180d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
<?php
/**
 * Database row sorting.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with this program; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
 * http://www.gnu.org/copyleft/gpl.html
 *
 * @file
 */

abstract class Collation {
	private static $instance;

	/**
	 * @return Collation
	 */
	static function singleton() {
		if ( !self::$instance ) {
			global $wgCategoryCollation;
			self::$instance = self::factory( $wgCategoryCollation );
		}
		return self::$instance;
	}

	/**
	 * @throws MWException
	 * @param string $collationName
	 * @return Collation
	 */
	static function factory( $collationName ) {
		switch ( $collationName ) {
			case 'uppercase':
				return new UppercaseCollation;
			case 'identity':
				return new IdentityCollation;
			case 'uca-default':
				return new IcuCollation( 'root' );
			case 'xx-uca-ckb':
				return new CollationCkb;
			case 'xx-uca-et':
				return new CollationEt;
			default:
				$match = array();
				if ( preg_match( '/^uca-([a-z@=-]+)$/', $collationName, $match ) ) {
					return new IcuCollation( $match[1] );
				}

				# Provide a mechanism for extensions to hook in.
				$collationObject = null;
				Hooks::run( 'Collation::factory', array( $collationName, &$collationObject ) );

				if ( $collationObject instanceof Collation ) {
					return $collationObject;
				}

				// If all else fails...
				throw new MWException( __METHOD__ . ": unknown collation type \"$collationName\"" );
		}
	}

	/**
	 * Given a string, convert it to a (hopefully short) key that can be used
	 * for efficient sorting.  A binary sort according to the sortkeys
	 * corresponds to a logical sort of the corresponding strings.  Current
	 * code expects that a line feed character should sort before all others, but
	 * has no other particular expectations (and that one can be changed if
	 * necessary).
	 *
	 * @param string $string UTF-8 string
	 * @return string Binary sortkey
	 */
	abstract function getSortKey( $string );

	/**
	 * Given a string, return the logical "first letter" to be used for
	 * grouping on category pages and so on.  This has to be coordinated
	 * carefully with convertToSortkey(), or else the sorted list might jump
	 * back and forth between the same "initial letters" or other pathological
	 * behavior.  For instance, if you just return the first character, but "a"
	 * sorts the same as "A" based on getSortKey(), then you might get a
	 * list like
	 *
	 * == A ==
	 * * [[Aardvark]]
	 *
	 * == a ==
	 * * [[antelope]]
	 *
	 * == A ==
	 * * [[Ape]]
	 *
	 * etc., assuming for the sake of argument that $wgCapitalLinks is false.
	 *
	 * @param string $string UTF-8 string
	 * @return string UTF-8 string corresponding to the first letter of input
	 */
	abstract function getFirstLetter( $string );
}

class UppercaseCollation extends Collation {
	private $lang;

	function __construct() {
		// Get a language object so that we can use the generic UTF-8 uppercase
		// function there
		$this->lang = Language::factory( 'en' );
	}

	function getSortKey( $string ) {
		return $this->lang->uc( $string );
	}

	function getFirstLetter( $string ) {
		if ( $string[0] == "\0" ) {
			$string = substr( $string, 1 );
		}
		return $this->lang->ucfirst( $this->lang->firstChar( $string ) );
	}
}

/**
 * Collation class that's essentially a no-op.
 *
 * Does sorting based on binary value of the string.
 * Like how things were pre 1.17.
 */
class IdentityCollation extends Collation {

	function getSortKey( $string ) {
		return $string;
	}

	function getFirstLetter( $string ) {
		global $wgContLang;
		// Copied from UppercaseCollation.
		// I'm kind of unclear on when this could happen...
		if ( $string[0] == "\0" ) {
			$string = substr( $string, 1 );
		}
		return $wgContLang->firstChar( $string );
	}
}

class IcuCollation extends Collation {
	const FIRST_LETTER_VERSION = 2;

	/** @var Collator */
	private $primaryCollator;

	/** @var Collator */
	private $mainCollator;

	/** @var string */
	private $locale;

	/** @var Language */
	protected $digitTransformLanguage;

	/** @var array */
	private $firstLetterData;

	/**
	 * Unified CJK blocks.
	 *
	 * The same definition of a CJK block must be used for both Collation and
	 * generateCollationData.php. These blocks are omitted from the first
	 * letter data, as an optimisation measure and because the default UCA table
	 * is pretty useless for sorting Chinese text anyway. Japanese and Korean
	 * blocks are not included here, because they are smaller and more useful.
	 */
	private static $cjkBlocks = array(
		array( 0x2E80, 0x2EFF ), // CJK Radicals Supplement
		array( 0x2F00, 0x2FDF ), // Kangxi Radicals
		array( 0x2FF0, 0x2FFF ), // Ideographic Description Characters
		array( 0x3000, 0x303F ), // CJK Symbols and Punctuation
		array( 0x31C0, 0x31EF ), // CJK Strokes
		array( 0x3200, 0x32FF ), // Enclosed CJK Letters and Months
		array( 0x3300, 0x33FF ), // CJK Compatibility
		array( 0x3400, 0x4DBF ), // CJK Unified Ideographs Extension A
		array( 0x4E00, 0x9FFF ), // CJK Unified Ideographs
		array( 0xF900, 0xFAFF ), // CJK Compatibility Ideographs
		array( 0xFE30, 0xFE4F ), // CJK Compatibility Forms
		array( 0x20000, 0x2A6DF ), // CJK Unified Ideographs Extension B
		array( 0x2A700, 0x2B73F ), // CJK Unified Ideographs Extension C
		array( 0x2B740, 0x2B81F ), // CJK Unified Ideographs Extension D
		array( 0x2F800, 0x2FA1F ), // CJK Compatibility Ideographs Supplement
	);

	/**
	 * Additional characters (or character groups) to be considered separate
	 * letters for given languages, or to be removed from the list of such
	 * letters (denoted by keys starting with '-').
	 *
	 * These are additions to (or subtractions from) the data stored in the
	 * first-letters-root.ser file (which among others includes full basic latin,
	 * cyrillic and greek alphabets).
	 *
	 * "Separate letter" is a letter that would have a separate heading/section
	 * for it in a dictionary or a phone book in this language. This data isn't
	 * used for sorting (the ICU library handles that), only for deciding which
	 * characters (or character groups) to use as headings.
	 *
	 * Initially generated based on the primary level of Unicode collation
	 * tailorings available at http://developer.mimer.com/charts/tailorings.htm ,
	 * later modified.
	 *
	 * Empty arrays are intended; this signifies that the data for the language is
	 * available and that there are, in fact, no additional letters to consider.
	 */
	private static $tailoringFirstLetters = array(
		// Verified by native speakers
		'be' => array( "Ё" ),
		'be-tarask' => array( "Ё" ),
		'cy' => array( "Ch", "Dd", "Ff", "Ng", "Ll", "Ph", "Rh", "Th" ),
		'en' => array(),
		'fa' => array( "آ", "ء", "ه" ),
		'fi' => array( "Å", "Ä", "Ö" ),
		'fr' => array(),
		'hu' => array( "Cs", "Dz", "Dzs", "Gy", "Ly", "Ny", "Ö", "Sz", "Ty", "Ü", "Zs" ),
		'is' => array( "Á", "Ð", "É", "Í", "Ó", "Ú", "Ý", "Þ", "Æ", "Ö", "Å" ),
		'it' => array(),
		'lv' => array( "Č", "Ģ", "Ķ", "Ļ", "Ņ", "Š", "Ž" ),
		'pl' => array( "Ą", "Ć", "Ę", "Ł", "Ń", "Ó", "Ś", "Ź", "Ż" ),
		'pt' => array(),
		'ru' => array(),
		'sv' => array( "Å", "Ä", "Ö" ),
		'sv@collation=standard' => array( "Å", "Ä", "Ö" ),
		'uk' => array( "Ґ", "Ь" ),
		'vi' => array( "Ă", "Â", "Đ", "Ê", "Ô", "Ơ", "Ư" ),
		// Not verified, but likely correct
		'af' => array(),
		'ast' => array( "Ch", "Ll", "Ñ" ),
		'az' => array( "Ç", "Ə", "Ğ", "İ", "Ö", "Ş", "Ü" ),
		'bg' => array(),
		'br' => array( "Ch", "C'h" ),
		'bs' => array( "Č", "Ć", "", "Đ", "Lj", "Nj", "Š", "Ž" ),
		'ca' => array(),
		'co' => array(),
		'cs' => array( "Č", "Ch", "Ř", "Š", "Ž" ),
		'da' => array( "Æ", "Ø", "Å" ),
		'de' => array(),
		'dsb' => array( "Č", "Ć", "", "Ě", "Ch", "Ł", "Ń", "Ŕ", "Š", "Ś", "Ž", "Ź" ),
		'el' => array(),
		'eo' => array( "Ĉ", "Ĝ", "Ĥ", "Ĵ", "Ŝ", "Ŭ" ),
		'es' => array( "Ñ" ),
		'et' => array( "Š", "Ž", "Õ", "Ä", "Ö", "Ü", "W" ), // added W for CollationEt (xx-uca-et)
		'eu' => array( "Ñ" ),
		'fo' => array( "Á", "Ð", "Í", "Ó", "Ú", "Ý", "Æ", "Ø", "Å" ),
		'fur' => array( "À", "Á", "Â", "È", "Ì", "Ò", "Ù" ),
		'fy' => array(),
		'ga' => array(),
		'gd' => array(),
		'gl' => array( "Ch", "Ll", "Ñ" ),
		'hr' => array( "Č", "Ć", "", "Đ", "Lj", "Nj", "Š", "Ž" ),
		'hsb' => array( "Č", "", "Ě", "Ch", "Ł", "Ń", "Ř", "Š", "Ć", "Ž" ),
		'kk' => array( "Ү", "І" ),
		'kl' => array( "Æ", "Ø", "Å" ),
		'ku' => array( "Ç", "Ê", "Î", "Ş", "Û" ),
		'ky' => array( "Ё" ),
		'la' => array(),
		'lb' => array(),
		'lt' => array( "Č", "Š", "Ž" ),
		'mk' => array(),
		'mo' => array( "Ă", "Â", "Î", "Ş", "Ţ" ),
		'mt' => array( "Ċ", "Ġ", "", "Ħ", "Ż" ),
		'nl' => array(),
		'no' => array( "Æ", "Ø", "Å" ),
		'oc' => array(),
		'rm' => array(),
		'ro' => array( "Ă", "Â", "Î", "Ş", "Ţ" ),
		'rup' => array( "Ă", "Â", "Î", "Ľ", "Ń", "Ş", "Ţ" ),
		'sco' => array(),
		'sk' => array( "Ä", "Č", "Ch", "Ô", "Š", "Ž" ),
		'sl' => array( "Č", "Š", "Ž" ),
		'smn' => array( "Á", "Č", "Đ", "Ŋ", "Š", "Ŧ", "Ž", "Æ", "Ø", "Å", "Ä", "Ö" ),
		'sq' => array( "Ç", "Dh", "Ë", "Gj", "Ll", "Nj", "Rr", "Sh", "Th", "Xh", "Zh" ),
		'sr' => array(),
		'tk' => array( "Ç", "Ä", "Ž", "Ň", "Ö", "Ş", "Ü", "Ý" ),
		'tl' => array( "Ñ", "Ng" ),
		'tr' => array( "Ç", "Ğ", "İ", "Ö", "Ş", "Ü" ),
		'tt' => array( "Ә", "Ө", "Ү", "Җ", "Ң", "Һ" ),
		'uz' => array( "Ch", "G'", "Ng", "O'", "Sh" ),
	);

	const RECORD_LENGTH = 14;

	function __construct( $locale ) {
		if ( !extension_loaded( 'intl' ) ) {
			throw new MWException( 'An ICU collation was requested, ' .
				'but the intl extension is not available.' );
		}

		$this->locale = $locale;
		// Drop everything after the '@' in locale's name
		$localeParts = explode( '@', $locale );
		$this->digitTransformLanguage = Language::factory( $locale === 'root' ? 'en' : $localeParts[0] );

		$this->mainCollator = Collator::create( $locale );
		if ( !$this->mainCollator ) {
			throw new MWException( "Invalid ICU locale specified for collation: $locale" );
		}

		$this->primaryCollator = Collator::create( $locale );
		$this->primaryCollator->setStrength( Collator::PRIMARY );
	}

	function getSortKey( $string ) {
		// intl extension produces non null-terminated
		// strings. Appending '' fixes it so that it doesn't generate
		// a warning on each access in debug php.
		MediaWiki\suppressWarnings();
		$key = $this->mainCollator->getSortKey( $string ) . '';
		MediaWiki\restoreWarnings();
		return $key;
	}

	function getPrimarySortKey( $string ) {
		MediaWiki\suppressWarnings();
		$key = $this->primaryCollator->getSortKey( $string ) . '';
		MediaWiki\restoreWarnings();
		return $key;
	}

	function getFirstLetter( $string ) {
		$string = strval( $string );
		if ( $string === '' ) {
			return '';
		}

		// Check for CJK
		$firstChar = mb_substr( $string, 0, 1, 'UTF-8' );
		if ( ord( $firstChar ) > 0x7f && self::isCjk( UtfNormal\Utils::utf8ToCodepoint( $firstChar ) ) ) {
			return $firstChar;
		}

		$sortKey = $this->getPrimarySortKey( $string );

		// Do a binary search to find the correct letter to sort under
		$min = ArrayUtils::findLowerBound(
			array( $this, 'getSortKeyByLetterIndex' ),
			$this->getFirstLetterCount(),
			'strcmp',
			$sortKey );

		if ( $min === false ) {
			// Before the first letter
			return '';
		}
		return $this->getLetterByIndex( $min );
	}

	function getFirstLetterData() {
		if ( $this->firstLetterData !== null ) {
			return $this->firstLetterData;
		}

		$cache = wfGetCache( CACHE_ANYTHING );
		$cacheKey = wfMemcKey(
			'first-letters',
			$this->locale,
			$this->digitTransformLanguage->getCode(),
			self::getICUVersion()
		);
		$cacheEntry = $cache->get( $cacheKey );

		if ( $cacheEntry && isset( $cacheEntry['version'] )
			&& $cacheEntry['version'] == self::FIRST_LETTER_VERSION
		) {
			$this->firstLetterData = $cacheEntry;
			return $this->firstLetterData;
		}

		// Generate data from serialized data file

		if ( isset( self::$tailoringFirstLetters[$this->locale] ) ) {
			$letters = wfGetPrecompiledData( "first-letters-root.ser" );
			// Append additional characters
			$letters = array_merge( $letters, self::$tailoringFirstLetters[$this->locale] );
			// Remove unnecessary ones, if any
			if ( isset( self::$tailoringFirstLetters['-' . $this->locale] ) ) {
				$letters = array_diff( $letters, self::$tailoringFirstLetters['-' . $this->locale] );
			}
			// Apply digit transforms
			$digits = array( '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' );
			$letters = array_diff( $letters, $digits );
			foreach ( $digits as $digit ) {
				$letters[] = $this->digitTransformLanguage->formatNum( $digit, true );
			}
		} else {
			$letters = wfGetPrecompiledData( "first-letters-{$this->locale}.ser" );
			if ( $letters === false ) {
				throw new MWException( "MediaWiki does not support ICU locale " .
					"\"{$this->locale}\"" );
			}
		}

		// Sort the letters.
		//
		// It's impossible to have the precompiled data file properly sorted,
		// because the sort order changes depending on ICU version. If the
		// array is not properly sorted, the binary search will return random
		// results.
		//
		// We also take this opportunity to remove primary collisions.
		$letterMap = array();
		foreach ( $letters as $letter ) {
			$key = $this->getPrimarySortKey( $letter );
			if ( isset( $letterMap[$key] ) ) {
				// Primary collision
				// Keep whichever one sorts first in the main collator
				if ( $this->mainCollator->compare( $letter, $letterMap[$key] ) < 0 ) {
					$letterMap[$key] = $letter;
				}
			} else {
				$letterMap[$key] = $letter;
			}
		}
		ksort( $letterMap, SORT_STRING );
		// Remove duplicate prefixes. Basically if something has a sortkey
		// which is a prefix of some other sortkey, then it is an
		// expansion and probably should not be considered a section
		// header.
		//
		// For example 'þ' is sometimes sorted as if it is the letters
		// 'th'. Other times it is its own primary element. Another
		// example is '₨'. Sometimes its a currency symbol. Sometimes it
		// is an 'R' followed by an 's'.
		//
		// Additionally an expanded element should always sort directly
		// after its first element due to they way sortkeys work.
		//
		// UCA sortkey elements are of variable length but no collation
		// element should be a prefix of some other element, so I think
		// this is safe. See:
		// * https://ssl.icu-project.org/repos/icu/icuhtml/trunk/design/collation/ICU_collation_design.htm
		// * http://site.icu-project.org/design/collation/uca-weight-allocation
		//
		// Additionally, there is something called primary compression to
		// worry about. Basically, if you have two primary elements that
		// are more than one byte and both start with the same byte then
		// the first byte is dropped on the second primary. Additionally
		// either \x03 or \xFF may be added to mean that the next primary
		// does not start with the first byte of the first primary.
		//
		// This shouldn't matter much, as the first primary is not
		// changed, and that is what we are comparing against.
		//
		// tl;dr: This makes some assumptions about how icu implements
		// collations. It seems incredibly unlikely these assumptions
		// will change, but nonetheless they are assumptions.

		$prev = false;
		$duplicatePrefixes = array();
		foreach ( $letterMap as $key => $value ) {
			// Remove terminator byte. Otherwise the prefix
			// comparison will get hung up on that.
			$trimmedKey = rtrim( $key, "\0" );
			if ( $prev === false || $prev === '' ) {
				$prev = $trimmedKey;
				// We don't yet have a collation element
				// to compare against, so continue.
				continue;
			}

			// Due to the fact the array is sorted, we only have
			// to compare with the element directly previous
			// to the current element (skipping expansions).
			// An element "X" will always sort directly
			// before "XZ" (Unless we have "XY", but we
			// do not update $prev in that case).
			if ( substr( $trimmedKey, 0, strlen( $prev ) ) === $prev ) {
				$duplicatePrefixes[] = $key;
				// If this is an expansion, we don't want to
				// compare the next element to this element,
				// but to what is currently $prev
				continue;
			}
			$prev = $trimmedKey;
		}
		foreach ( $duplicatePrefixes as $badKey ) {
			wfDebug( "Removing '{$letterMap[$badKey]}' from first letters.\n" );
			unset( $letterMap[$badKey] );
			// This code assumes that unsetting does not change sort order.
		}
		$data = array(
			'chars' => array_values( $letterMap ),
			'keys' => array_keys( $letterMap ),
			'version' => self::FIRST_LETTER_VERSION,
		);

		// Reduce memory usage before caching
		unset( $letterMap );

		// Save to cache
		$this->firstLetterData = $data;
		$cache->set( $cacheKey, $data, 86400 * 7 /* 1 week */ );
		return $data;
	}

	function getLetterByIndex( $index ) {
		if ( $this->firstLetterData === null ) {
			$this->getFirstLetterData();
		}
		return $this->firstLetterData['chars'][$index];
	}

	function getSortKeyByLetterIndex( $index ) {
		if ( $this->firstLetterData === null ) {
			$this->getFirstLetterData();
		}
		return $this->firstLetterData['keys'][$index];
	}

	function getFirstLetterCount() {
		if ( $this->firstLetterData === null ) {
			$this->getFirstLetterData();
		}
		return count( $this->firstLetterData['chars'] );
	}

	static function isCjk( $codepoint ) {
		foreach ( self::$cjkBlocks as $block ) {
			if ( $codepoint >= $block[0] && $codepoint <= $block[1] ) {
				return true;
			}
		}
		return false;
	}

	/**
	 * Return the version of ICU library used by PHP's intl extension,
	 * or false when the extension is not installed of the version
	 * can't be determined.
	 *
	 * The constant INTL_ICU_VERSION this function refers to isn't really
	 * documented. It is available since PHP 5.3.7 (see PHP bug 54561).
	 * This function will return false on older PHPs.
	 *
	 * @since 1.21
	 * @return string|bool
	 */
	static function getICUVersion() {
		return defined( 'INTL_ICU_VERSION' ) ? INTL_ICU_VERSION : false;
	}

	/**
	 * Return the version of Unicode appropriate for the version of ICU library
	 * currently in use, or false when it can't be determined.
	 *
	 * @since 1.21
	 * @return string|bool
	 */
	static function getUnicodeVersionForICU() {
		$icuVersion = IcuCollation::getICUVersion();
		if ( !$icuVersion ) {
			return false;
		}

		$versionPrefix = substr( $icuVersion, 0, 3 );
		// Source: http://site.icu-project.org/download
		$map = array(
			'50.' => '6.2',
			'49.' => '6.1',
			'4.8' => '6.0',
			'4.6' => '6.0',
			'4.4' => '5.2',
			'4.2' => '5.1',
			'4.0' => '5.1',
			'3.8' => '5.0',
			'3.6' => '5.0',
			'3.4' => '4.1',
		);

		if ( isset( $map[$versionPrefix] ) ) {
			return $map[$versionPrefix];
		} else {
			return false;
		}
	}
}

/**
 * Workaround for the lack of support of Sorani Kurdish / Central Kurdish language ('ckb') in ICU.
 *
 * Uses the same collation rules as Persian / Farsi ('fa'), but different characters for digits.
 */
class CollationCkb extends IcuCollation {
	function __construct() {
		// This will set $locale and collators, which affect the actual sorting order
		parent::__construct( 'fa' );
		// Override the 'fa' language set by parent constructor, which affects #getFirstLetterData()
		$this->digitTransformLanguage = Language::factory( 'ckb' );
	}
}

/**
 * Workaround for incorrect collation of Estonian language ('et') in ICU (bug 54168).
 *
 * 'W' and 'V' should not be considered the same letter for the purposes of collation in modern
 * Estonian. We work around this by replacing 'W' and 'w' with 'ᴡ' U+1D21 'LATIN LETTER SMALL
 * CAPITAL W' for sortkey generation, which is collated like 'W' and is not tailored to have the
 * same primary weight as 'V' in Estonian.
 */
class CollationEt extends IcuCollation {
	function __construct() {
		parent::__construct( 'et' );
	}

	private static function mangle( $string ) {
		return str_replace(
			array( 'w', 'W' ),
			'ᴡ', // U+1D21 'LATIN LETTER SMALL CAPITAL W'
			$string
		);
	}

	private static function unmangle( $string ) {
		// Casing data is lost…
		return str_replace(
			'ᴡ', // U+1D21 'LATIN LETTER SMALL CAPITAL W'
			'W',
			$string
		);
	}

	function getSortKey( $string ) {
		return parent::getSortKey( self::mangle( $string ) );
	}

	function getFirstLetter( $string ) {
		return self::unmangle( parent::getFirstLetter( self::mangle( $string ) ) );
	}
}