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