includes/collation/IcuCollation.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  */
  20
  21 /**
  22  * @since 1.16.3
  23  */
  24 class IcuCollation extends Collation {
  25         const FIRST_LETTER_VERSION = 2;
  26
  27         /** @var Collator */
  28         private $primaryCollator;
  29
  30         /** @var Collator */
  31         private $mainCollator;
  32
  33         /** @var string */
  34         private $locale;
  35
  36         /** @var Language */
  37         protected $digitTransformLanguage;
  38
  39         /** @var boolean */
  40         private $useNumericCollation = false;
  41
  42         /** @var array */
  43         private $firstLetterData;
  44
  45         /**
  46          * Unified CJK blocks.
  47          *
  48          * The same definition of a CJK block must be used for both Collation and
  49          * generateCollationData.php. These blocks are omitted from the first
  50          * letter data, as an optimisation measure and because the default UCA table
  51          * is pretty useless for sorting Chinese text anyway. Japanese and Korean
  52          * blocks are not included here, because they are smaller and more useful.
  53          */
  54         private static $cjkBlocks = [
  55                 [ 0x2E80, 0x2EFF ], // CJK Radicals Supplement
  56                 [ 0x2F00, 0x2FDF ], // Kangxi Radicals
  57                 [ 0x2FF0, 0x2FFF ], // Ideographic Description Characters
  58                 [ 0x3000, 0x303F ], // CJK Symbols and Punctuation
  59                 [ 0x31C0, 0x31EF ], // CJK Strokes
  60                 [ 0x3200, 0x32FF ], // Enclosed CJK Letters and Months
  61                 [ 0x3300, 0x33FF ], // CJK Compatibility
  62                 [ 0x3400, 0x4DBF ], // CJK Unified Ideographs Extension A
  63                 [ 0x4E00, 0x9FFF ], // CJK Unified Ideographs
  64                 [ 0xF900, 0xFAFF ], // CJK Compatibility Ideographs
  65                 [ 0xFE30, 0xFE4F ], // CJK Compatibility Forms
  66                 [ 0x20000, 0x2A6DF ], // CJK Unified Ideographs Extension B
  67                 [ 0x2A700, 0x2B73F ], // CJK Unified Ideographs Extension C
  68                 [ 0x2B740, 0x2B81F ], // CJK Unified Ideographs Extension D
  69                 [ 0x2F800, 0x2FA1F ], // CJK Compatibility Ideographs Supplement
  70         ];
  71
  72         /**
  73          * Additional characters (or character groups) to be considered separate
  74          * letters for given languages, or to be removed from the list of such
  75          * letters (denoted by keys starting with '-').
  76          *
  77          * These are additions to (or subtractions from) the data stored in the
  78          * first-letters-root.ser file (which among others includes full basic latin,
  79          * cyrillic and greek alphabets).
  80          *
  81          * "Separate letter" is a letter that would have a separate heading/section
  82          * for it in a dictionary or a phone book in this language. This data isn't
  83          * used for sorting (the ICU library handles that), only for deciding which
  84          * characters (or character groups) to use as headings.
  85          *
  86          * Initially generated based on the primary level of Unicode collation
  87          * tailorings available at http://developer.mimer.com/charts/tailorings.htm ,
  88          * later modified.
  89          *
  90          * Empty arrays are intended; this signifies that the data for the language is
  91          * available and that there are, in fact, no additional letters to consider.
  92          */
  93         private static $tailoringFirstLetters = [
  94                 // Verified by native speakers
  95                 'be' => [ "Ё" ],
  96                 'be-tarask' => [ "Ё" ],
  97                 'bs' => [ "Č", "Ć", "Dž", "Đ", "Lj", "Nj", "Š", "Ž" ],
  98                 'cs' => [ "Č", "Ch", "Ř", "Š", "Ž" ],
  99                 'cy' => [ "Ch", "Dd", "Ff", "Ng", "Ll", "Ph", "Rh", "Th" ],
 100                 'en' => [],
 101                 'fa' => [
 102                         // RTL, let's put each letter on a new line
 103                         "آ",
 104                         "ء",
 105                         "ه",
 106                         "ا",
 107                         "و"
 108                 ],
 109                 'fi' => [ "Å", "Ä", "Ö" ],
 110                 'fr' => [],
 111                 'hr' => [ "Č", "Ć", "Dž", "Đ", "Lj", "Nj", "Š", "Ž" ],
 112                 'hsb' => [ "Č", "Dź", "Ě", "Ch", "Ł", "Ń", "Ř", "Š", "Ć", "Ž" ],
 113                 'hu' => [ "Cs", "Dz", "Dzs", "Gy", "Ly", "Ny", "Ö", "Sz", "Ty", "Ü", "Zs" ],
 114                 'is' => [ "Á", "Ð", "É", "Í", "Ó", "Ú", "Ý", "Þ", "Æ", "Ö", "Å" ],
 115                 'it' => [],
 116                 'lt' => [ "Č", "Š", "Ž" ],
 117                 'lv' => [ "Č", "Ģ", "Ķ", "Ļ", "Ņ", "Š", "Ž" ],
 118                 'mk' => [ "Ѓ", "Ќ" ],
 119                 'nl' => [],
 120                 'pl' => [ "Ą", "Ć", "Ę", "Ł", "Ń", "Ó", "Ś", "Ź", "Ż" ],
 121                 'pt' => [],
 122                 'ru' => [],
 123                 'sk' => [ "Ä", "Č", "Ch", "Ô", "Š", "Ž" ],
 124                 'sr' => [],
 125                 'sv' => [ "Å", "Ä", "Ö" ],
 126                 'sv@collation=standard' => [ "Å", "Ä", "Ö" ],
 127                 'ta' => [
 128                         "\xE0\xAE\x82", "ஃ", "க்ஷ", "க்", "ங்", "ச்", "ஞ்", "ட்", "ண்", "த்", "ந்",
 129                         "ப்", "ம்", "ய்", "ர்", "ல்", "வ்", "ழ்", "ள்", "ற்", "ன்", "ஜ்", "ஶ்", "ஷ்",
 130                         "ஸ்", "ஹ்", "க்ஷ்"
 131                 ],
 132                 'uk' => [ "Ґ", "Ь" ],
 133                 'vi' => [ "Ă", "Â", "Đ", "Ê", "Ô", "Ơ", "Ư" ],
 134                 // Not verified, but likely correct
 135                 'af' => [],
 136                 'ast' => [ "Ch", "Ll", "Ñ" ],
 137                 'az' => [ "Ç", "Ə", "Ğ", "İ", "Ö", "Ş", "Ü" ],
 138                 'bg' => [],
 139                 'br' => [ "Ch", "C'h" ],
 140                 'ca' => [],
 141                 'co' => [],
 142                 'da' => [ "Æ", "Ø", "Å" ],
 143                 'de' => [],
 144                 'dsb' => [ "Č", "Ć", "Dź", "Ě", "Ch", "Ł", "Ń", "Ŕ", "Š", "Ś", "Ž", "Ź" ],
 145                 'el' => [],
 146                 'eo' => [ "Ĉ", "Ĝ", "Ĥ", "Ĵ", "Ŝ", "Ŭ" ],
 147                 'es' => [ "Ñ" ],
 148                 'et' => [ "Š", "Ž", "Õ", "Ä", "Ö", "Ü", "W" ], // added W for CollationEt (xx-uca-et)
 149                 'eu' => [ "Ñ" ],
 150                 'fo' => [ "Á", "Ð", "Í", "Ó", "Ú", "Ý", "Æ", "Ø", "Å" ],
 151                 'fur' => [ "À", "Á", "Â", "È", "Ì", "Ò", "Ù" ],
 152                 'fy' => [],
 153                 'ga' => [],
 154                 'gd' => [],
 155                 'gl' => [ "Ch", "Ll", "Ñ" ],
 156                 'kk' => [ "Ү", "І" ],
 157                 'kl' => [ "Æ", "Ø", "Å" ],
 158                 'ku' => [ "Ç", "Ê", "Î", "Ş", "Û" ],
 159                 'ky' => [ "Ё" ],
 160                 'la' => [],
 161                 'lb' => [],
 162                 'mo' => [ "Ă", "Â", "Î", "Ş", "Ţ" ],
 163                 'mt' => [ "Ċ", "Ġ", "Għ", "Ħ", "Ż" ],
 164                 'no' => [ "Æ", "Ø", "Å" ],
 165                 'oc' => [],
 166                 'rm' => [],
 167                 'ro' => [ "Ă", "Â", "Î", "Ş", "Ţ" ],
 168                 'rup' => [ "Ă", "Â", "Î", "Ľ", "Ń", "Ş", "Ţ" ],
 169                 'sco' => [],
 170                 'sl' => [ "Č", "Š", "Ž" ],
 171                 'smn' => [ "Á", "Č", "Đ", "Ŋ", "Š", "Ŧ", "Ž", "Æ", "Ø", "Å", "Ä", "Ö" ],
 172                 'sq' => [ "Ç", "Dh", "Ë", "Gj", "Ll", "Nj", "Rr", "Sh", "Th", "Xh", "Zh" ],
 173                 'tk' => [ "Ç", "Ä", "Ž", "Ň", "Ö", "Ş", "Ü", "Ý" ],
 174                 'tl' => [ "Ñ", "Ng" ],
 175                 'tr' => [ "Ç", "Ğ", "İ", "Ö", "Ş", "Ü" ],
 176                 'tt' => [ "Ә", "Ө", "Ү", "Җ", "Ң", "Һ" ],
 177                 'uz' => [ "Ch", "G'", "Ng", "O'", "Sh" ],
 178         ];
 179
 180         /**
 181          * @since 1.16.3
 182          */
 183         const RECORD_LENGTH = 14;
 184
 185         public function __construct( $locale ) {
 186                 if ( !extension_loaded( 'intl' ) ) {
 187                         throw new MWException( 'An ICU collation was requested, ' .
 188                                 'but the intl extension is not available.' );
 189                 }
 190
 191                 $this->locale = $locale;
 192                 // Drop everything after the '@' in locale's name
 193                 $localeParts = explode( '@', $locale );
 194                 $this->digitTransformLanguage = Language::factory( $locale === 'root' ? 'en' : $localeParts[0] );
 195
 196                 $this->mainCollator = Collator::create( $locale );
 197                 if ( !$this->mainCollator ) {
 198                         throw new MWException( "Invalid ICU locale specified for collation: $locale" );
 199                 }
 200
 201                 $this->primaryCollator = Collator::create( $locale );
 202                 $this->primaryCollator->setStrength( Collator::PRIMARY );
 203
 204                 // If the special suffix for numeric collation is present, turn on numeric collation.
 205                 if ( substr( $locale, -5, 5 ) === '-u-kn' ) {
 206                         $this->useNumericCollation = true;
 207                         // Strip off the special suffix so it doesn't trip up fetchFirstLetterData().
 208                         $this->locale = substr( $this->locale, 0, -5 );
 209                         $this->mainCollator->setAttribute( Collator::NUMERIC_COLLATION, Collator::ON );
 210                         $this->primaryCollator->setAttribute( Collator::NUMERIC_COLLATION, Collator::ON );
 211                 }
 212         }
 213
 214         public function getSortKey( $string ) {
 215                 return $this->mainCollator->getSortKey( $string );
 216         }
 217
 218         public function getPrimarySortKey( $string ) {
 219                 return $this->primaryCollator->getSortKey( $string );
 220         }
 221
 222         public function getFirstLetter( $string ) {
 223                 $string = strval( $string );
 224                 if ( $string === '' ) {
 225                         return '';
 226                 }
 227
 228                 $firstChar = mb_substr( $string, 0, 1, 'UTF-8' );
 229
 230                 // If the first character is a CJK character, just return that character.
 231                 if ( ord( $firstChar ) > 0x7f && self::isCjk( UtfNormal\Utils::utf8ToCodepoint( $firstChar ) ) ) {
 232                         return $firstChar;
 233                 }
 234
 235                 $sortKey = $this->getPrimarySortKey( $string );
 236
 237                 // Do a binary search to find the correct letter to sort under
 238                 $min = ArrayUtils::findLowerBound(
 239                         [ $this, 'getSortKeyByLetterIndex' ],
 240                         $this->getFirstLetterCount(),
 241                         'strcmp',
 242                         $sortKey );
 243
 244                 if ( $min === false ) {
 245                         // Before the first letter
 246                         return '';
 247                 }
 248
 249                 $sortLetter = $this->getLetterByIndex( $min );
 250
 251                 if ( $this->useNumericCollation ) {
 252                         // If the sort letter is a number, return '0–9' (or localized equivalent).
 253                         // ASCII value of 0 is 48. ASCII value of 9 is 57.
 254                         // Note that this also applies to non-Arabic numerals since they are
 255                         // mapped to Arabic numeral sort letters. For example, ২ sorts as 2.
 256                         if ( ord( $sortLetter ) >= 48 && ord( $sortLetter ) <= 57 ) {
 257                                 $sortLetter = wfMessage( 'category-header-numerals' )->numParams( 0, 9 )->text();
 258                         }
 259                 }
 260                 return $sortLetter;
 261         }
 262
 263         /**
 264          * @since 1.16.3
 265          * @return array
 266          */
 267         public function getFirstLetterData() {
 268                 if ( $this->firstLetterData === null ) {
 269                         $cache = ObjectCache::getLocalServerInstance( CACHE_ANYTHING );
 270                         $cacheKey = $cache->makeKey(
 271                                 'first-letters',
 272                                 $this->locale,
 273                                 $this->digitTransformLanguage->getCode(),
 274                                 self::getICUVersion(),
 275                                 self::FIRST_LETTER_VERSION
 276                         );
 277                         $this->firstLetterData = $cache->getWithSetCallback( $cacheKey, $cache::TTL_WEEK, function () {
 278                                 return $this->fetchFirstLetterData();
 279                         } );
 280                 }
 281                 return $this->firstLetterData;
 282         }
 283
 284         /**
 285          * @return array
 286          * @throws MWException
 287          */
 288         private function fetchFirstLetterData() {
 289                 // Generate data from serialized data file
 290                 if ( isset( self::$tailoringFirstLetters[$this->locale] ) ) {
 291                         $letters = wfGetPrecompiledData( 'first-letters-root.ser' );
 292                         // Append additional characters
 293                         $letters = array_merge( $letters, self::$tailoringFirstLetters[$this->locale] );
 294                         // Remove unnecessary ones, if any
 295                         if ( isset( self::$tailoringFirstLetters['-' . $this->locale] ) ) {
 296                                 $letters = array_diff( $letters, self::$tailoringFirstLetters['-' . $this->locale] );
 297                         }
 298                         // Apply digit transforms
 299                         $digits = [ '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' ];
 300                         $letters = array_diff( $letters, $digits );
 301                         foreach ( $digits as $digit ) {
 302                                 $letters[] = $this->digitTransformLanguage->formatNum( $digit, true );
 303                         }
 304                 } else {
 305                         $letters = wfGetPrecompiledData( "first-letters-{$this->locale}.ser" );
 306                         if ( $letters === false ) {
 307                                 throw new MWException( "MediaWiki does not support ICU locale " .
 308                                         "\"{$this->locale}\"" );
 309                         }
 310                 }
 311
 312                 /* Sort the letters.
 313                  *
 314                  * It's impossible to have the precompiled data file properly sorted,
 315                  * because the sort order changes depending on ICU version. If the
 316                  * array is not properly sorted, the binary search will return random
 317                  * results.
 318                  *
 319                  * We also take this opportunity to remove primary collisions.
 320                  */
 321                 $letterMap = [];
 322                 foreach ( $letters as $letter ) {
 323                         $key = $this->getPrimarySortKey( $letter );
 324                         if ( isset( $letterMap[$key] ) ) {
 325                                 // Primary collision
 326                                 // Keep whichever one sorts first in the main collator
 327                                 if ( $this->mainCollator->compare( $letter, $letterMap[$key] ) < 0 ) {
 328                                         $letterMap[$key] = $letter;
 329                                 }
 330                         } else {
 331                                 $letterMap[$key] = $letter;
 332                         }
 333                 }
 334                 ksort( $letterMap, SORT_STRING );
 335
 336                 /* Remove duplicate prefixes. Basically if something has a sortkey
 337                  * which is a prefix of some other sortkey, then it is an
 338                  * expansion and probably should not be considered a section
 339                  * header.
 340                  *
 341                  * For example 'þ' is sometimes sorted as if it is the letters
 342                  * 'th'. Other times it is its own primary element. Another
 343                  * example is '₨'. Sometimes its a currency symbol. Sometimes it
 344                  * is an 'R' followed by an 's'.
 345                  *
 346                  * Additionally an expanded element should always sort directly
 347                  * after its first element due to they way sortkeys work.
 348                  *
 349                  * UCA sortkey elements are of variable length but no collation
 350                  * element should be a prefix of some other element, so I think
 351                  * this is safe. See:
 352                  * - https://ssl.icu-project.org/repos/icu/icuhtml/trunk/design/collation/ICU_collation_design.htm
 353                  * - http://site.icu-project.org/design/collation/uca-weight-allocation
 354                  *
 355                  * Additionally, there is something called primary compression to
 356                  * worry about. Basically, if you have two primary elements that
 357                  * are more than one byte and both start with the same byte then
 358                  * the first byte is dropped on the second primary. Additionally
 359                  * either \x03 or \xFF may be added to mean that the next primary
 360                  * does not start with the first byte of the first primary.
 361                  *
 362                  * This shouldn't matter much, as the first primary is not
 363                  * changed, and that is what we are comparing against.
 364                  *
 365                  * tl;dr: This makes some assumptions about how icu implements
 366                  * collations. It seems incredibly unlikely these assumptions
 367                  * will change, but nonetheless they are assumptions.
 368                  */
 369
 370                 $prev = false;
 371                 $duplicatePrefixes = [];
 372                 foreach ( $letterMap as $key => $value ) {
 373                         // Remove terminator byte. Otherwise the prefix
 374                         // comparison will get hung up on that.
 375                         $trimmedKey = rtrim( $key, "\0" );
 376                         if ( $prev === false || $prev === '' ) {
 377                                 $prev = $trimmedKey;
 378                                 // We don't yet have a collation element
 379                                 // to compare against, so continue.
 380                                 continue;
 381                         }
 382
 383                         // Due to the fact the array is sorted, we only have
 384                         // to compare with the element directly previous
 385                         // to the current element (skipping expansions).
 386                         // An element "X" will always sort directly
 387                         // before "XZ" (Unless we have "XY", but we
 388                         // do not update $prev in that case).
 389                         if ( substr( $trimmedKey, 0, strlen( $prev ) ) === $prev ) {
 390                                 $duplicatePrefixes[] = $key;
 391                                 // If this is an expansion, we don't want to
 392                                 // compare the next element to this element,
 393                                 // but to what is currently $prev
 394                                 continue;
 395                         }
 396                         $prev = $trimmedKey;
 397                 }
 398                 foreach ( $duplicatePrefixes as $badKey ) {
 399                         wfDebug( "Removing '{$letterMap[$badKey]}' from first letters.\n" );
 400                         unset( $letterMap[$badKey] );
 401                         // This code assumes that unsetting does not change sort order.
 402                 }
 403                 $data = [
 404                         'chars' => array_values( $letterMap ),
 405                         'keys' => array_keys( $letterMap ),
 406                 ];
 407
 408                 // Reduce memory usage before caching
 409                 unset( $letterMap );
 410
 411                 return $data;
 412         }
 413
 414         /**
 415          * @since 1.16.3
 416          */
 417         public function getLetterByIndex( $index ) {
 418                 return $this->getFirstLetterData()['chars'][$index];
 419         }
 420
 421         /**
 422          * @since 1.16.3
 423          */
 424         public function getSortKeyByLetterIndex( $index ) {
 425                 return $this->getFirstLetterData()['keys'][$index];
 426         }
 427
 428         /**
 429          * @since 1.16.3
 430          */
 431         public function getFirstLetterCount() {
 432                 return count( $this->getFirstLetterData()['chars'] );
 433         }
 434
 435         /**
 436          * Test if a code point is a CJK (Chinese, Japanese, Korean) character
 437          * @since 1.16.3
 438          */
 439         public static function isCjk( $codepoint ) {
 440                 foreach ( self::$cjkBlocks as $block ) {
 441                         if ( $codepoint >= $block[0] && $codepoint <= $block[1] ) {
 442                                 return true;
 443                         }
 444                 }
 445                 return false;
 446         }
 447
 448         /**
 449          * Return the version of ICU library used by PHP's intl extension,
 450          * or false when the extension is not installed of the version
 451          * can't be determined.
 452          *
 453          * The constant INTL_ICU_VERSION this function refers to isn't really
 454          * documented. It is available since PHP 5.3.7 (see PHP bug 54561).
 455          * This function will return false on older PHPs.
 456          *
 457          * @since 1.21
 458          * @return string|bool
 459          */
 460         static function getICUVersion() {
 461                 return defined( 'INTL_ICU_VERSION' ) ? INTL_ICU_VERSION : false;
 462         }
 463
 464         /**
 465          * Return the version of Unicode appropriate for the version of ICU library
 466          * currently in use, or false when it can't be determined.
 467          *
 468          * @since 1.21
 469          * @return string|bool
 470          */
 471         static function getUnicodeVersionForICU() {
 472                 $icuVersion = IcuCollation::getICUVersion();
 473                 if ( !$icuVersion ) {
 474                         return false;
 475                 }
 476
 477                 $versionPrefix = substr( $icuVersion, 0, 3 );
 478                 // Source: http://site.icu-project.org/download
 479                 $map = [
 480                         '57.' => '8.0',
 481                         '56.' => '8.0',
 482                         '55.' => '7.0',
 483                         '54.' => '7.0',
 484                         '53.' => '6.3',
 485                         '52.' => '6.3',
 486                         '51.' => '6.2',
 487                         '50.' => '6.2',
 488                         '49.' => '6.1',
 489                         '4.8' => '6.0',
 490                         '4.6' => '6.0',
 491                         '4.4' => '5.2',
 492                         '4.2' => '5.1',
 493                         '4.0' => '5.1',
 494                         '3.8' => '5.0',
 495                         '3.6' => '5.0',
 496                         '3.4' => '4.1',
 497                 ];
 498
 499                 if ( isset( $map[$versionPrefix] ) ) {
 500                         return $map[$versionPrefix];
 501                 } else {
 502                         return false;
 503                 }
 504         }
 505 }