Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 155 |
|
0.00% |
0 / 19 |
CRAP | |
0.00% |
0 / 1 |
| SearchMySQL | |
0.00% |
0 / 155 |
|
0.00% |
0 / 19 |
2256 | |
0.00% |
0 / 1 |
| parseQuery | |
0.00% |
0 / 50 |
|
0.00% |
0 / 1 |
182 | |||
| regexTerm | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| legalSearchChars | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| doSearchTextInDB | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchTitleInDB | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| searchInternal | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| supports | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| queryFeatures | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
20 | |||
| queryNamespaces | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| getQueryBuilder | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| getIndexField | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| queryMain | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| getCountQueryBuilder | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
2 | |||
| update | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
2 | |||
| updateTitle | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| delete | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| normalizeText | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
6 | |||
| stripForSearchCallback | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| minSearchLength | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * MySQL search engine |
| 4 | * |
| 5 | * Copyright (C) 2004 Brooke Vibber <bvibber@wikimedia.org> |
| 6 | * https://www.mediawiki.org/ |
| 7 | * |
| 8 | * @license GPL-2.0-or-later |
| 9 | * @file |
| 10 | * @ingroup Search |
| 11 | */ |
| 12 | |
| 13 | use MediaWiki\MediaWikiServices; |
| 14 | use Wikimedia\AtEase\AtEase; |
| 15 | use Wikimedia\Rdbms\IDatabase; |
| 16 | use Wikimedia\Rdbms\IExpression; |
| 17 | use Wikimedia\Rdbms\LikeValue; |
| 18 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 19 | |
| 20 | /** |
| 21 | * Search engine hook for MySQL |
| 22 | * @ingroup Search |
| 23 | */ |
| 24 | class SearchMySQL extends SearchDatabase { |
| 25 | /** @var bool */ |
| 26 | protected $strictMatching = true; |
| 27 | |
| 28 | /** @var int|null */ |
| 29 | private static $mMinSearchLength; |
| 30 | |
| 31 | /** |
| 32 | * Parse the user's query and transform it into two SQL fragments: |
| 33 | * a WHERE condition and an ORDER BY expression |
| 34 | * |
| 35 | * @param string $filteredText |
| 36 | * @param bool $fulltext |
| 37 | * |
| 38 | * @return array |
| 39 | */ |
| 40 | private function parseQuery( $filteredText, $fulltext ) { |
| 41 | $lc = $this->legalSearchChars( self::CHARS_NO_SYNTAX ); // Minus syntax chars (" and *) |
| 42 | $searchon = ''; |
| 43 | $this->searchTerms = []; |
| 44 | |
| 45 | # @todo FIXME: This doesn't handle parenthetical expressions. |
| 46 | $m = []; |
| 47 | if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| 48 | $filteredText, $m, PREG_SET_ORDER ) |
| 49 | ) { |
| 50 | $services = MediaWikiServices::getInstance(); |
| 51 | $contLang = $services->getContentLanguage(); |
| 52 | $langConverter = $services->getLanguageConverterFactory()->getLanguageConverter( $contLang ); |
| 53 | foreach ( $m as $bits ) { |
| 54 | AtEase::suppressWarnings(); |
| 55 | [ /* all */, $modifier, $term, $nonQuoted, $wildcard ] = $bits; |
| 56 | AtEase::restoreWarnings(); |
| 57 | |
| 58 | if ( $nonQuoted != '' ) { |
| 59 | $term = $nonQuoted; |
| 60 | $quote = ''; |
| 61 | } else { |
| 62 | $term = str_replace( '"', '', $term ); |
| 63 | $quote = '"'; |
| 64 | } |
| 65 | |
| 66 | if ( $searchon !== '' ) { |
| 67 | $searchon .= ' '; |
| 68 | } |
| 69 | if ( $this->strictMatching && ( $modifier == '' ) ) { |
| 70 | // If we leave this out, boolean op defaults to OR which is rarely helpful. |
| 71 | $modifier = '+'; |
| 72 | } |
| 73 | |
| 74 | // Some languages such as Serbian store the input form in the search index, |
| 75 | // so we may need to search for matches in multiple writing system variants. |
| 76 | $convertedVariants = $langConverter->autoConvertToAllVariants( $term ); |
| 77 | if ( is_array( $convertedVariants ) ) { |
| 78 | $variants = array_unique( array_values( $convertedVariants ) ); |
| 79 | } else { |
| 80 | $variants = [ $term ]; |
| 81 | } |
| 82 | |
| 83 | // The low-level search index does some processing on input to work |
| 84 | // around problems with minimum lengths and encoding in MySQL's |
| 85 | // fulltext engine. |
| 86 | // For Chinese this also inserts spaces between adjacent Han characters. |
| 87 | $strippedVariants = array_map( $contLang->normalizeForSearch( ... ), $variants ); |
| 88 | |
| 89 | // Some languages such as Chinese force all variants to a canonical |
| 90 | // form when stripping to the low-level search index, so to be sure |
| 91 | // let's check our variants list for unique items after stripping. |
| 92 | $strippedVariants = array_unique( $strippedVariants ); |
| 93 | |
| 94 | $searchon .= $modifier; |
| 95 | if ( count( $strippedVariants ) > 1 ) { |
| 96 | $searchon .= '('; |
| 97 | } |
| 98 | foreach ( $strippedVariants as $stripped ) { |
| 99 | $stripped = $this->normalizeText( $stripped ); |
| 100 | if ( $nonQuoted && str_contains( $stripped, ' ' ) ) { |
| 101 | // Hack for Chinese: we need to toss in quotes for |
| 102 | // multiple-character phrases since normalizeForSearch() |
| 103 | // added spaces between them to make word breaks. |
| 104 | $stripped = '"' . trim( $stripped ) . '"'; |
| 105 | } |
| 106 | $searchon .= "$quote$stripped$quote$wildcard "; |
| 107 | } |
| 108 | if ( count( $strippedVariants ) > 1 ) { |
| 109 | $searchon .= ')'; |
| 110 | } |
| 111 | |
| 112 | // Match individual terms or quoted phrase in result highlighting... |
| 113 | // Note that variants will be introduced in a later stage for highlighting! |
| 114 | $regexp = $this->regexTerm( $term, $wildcard ); |
| 115 | $this->searchTerms[] = $regexp; |
| 116 | } |
| 117 | wfDebug( __METHOD__ . ": Would search with '$searchon'" ); |
| 118 | wfDebug( __METHOD__ . ': Match with /' . implode( '|', $this->searchTerms ) . "/" ); |
| 119 | } else { |
| 120 | wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'" ); |
| 121 | } |
| 122 | |
| 123 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 124 | $searchon = $dbr->addQuotes( $searchon ); |
| 125 | $field = $this->getIndexField( $fulltext ); |
| 126 | return [ |
| 127 | " MATCH($field) AGAINST($searchon IN BOOLEAN MODE) ", |
| 128 | " MATCH($field) AGAINST($searchon IN NATURAL LANGUAGE MODE) DESC " |
| 129 | ]; |
| 130 | } |
| 131 | |
| 132 | private function regexTerm( string $string, ?string $wildcard ): string { |
| 133 | $regex = preg_quote( $string, '/' ); |
| 134 | if ( MediaWikiServices::getInstance()->getContentLanguage()->hasWordBreaks() ) { |
| 135 | if ( $wildcard ) { |
| 136 | // Don't cut off the final bit! |
| 137 | $regex = "\b$regex"; |
| 138 | } else { |
| 139 | $regex = "\b$regex\b"; |
| 140 | } |
| 141 | } else { |
| 142 | // For Chinese, words may legitimately abut other words in the text literal. |
| 143 | // Don't add \b boundary checks... note this could cause false positives |
| 144 | // for Latin chars. |
| 145 | } |
| 146 | return $regex; |
| 147 | } |
| 148 | |
| 149 | /** @inheritDoc */ |
| 150 | public function legalSearchChars( $type = self::CHARS_ALL ) { |
| 151 | $searchChars = parent::legalSearchChars( $type ); |
| 152 | |
| 153 | // In the MediaWiki UI, search strings containing (just) a hyphen are translated into |
| 154 | // MATCH(si_title) AGAINST('+- ' IN BOOLEAN MODE) |
| 155 | // which is not valid. |
| 156 | |
| 157 | // From <https://dev.mysql.com/doc/refman/8.0/en/fulltext-boolean.html>: |
| 158 | // "InnoDB full-text search does not support... a plus and minus sign combination ('+-')" |
| 159 | |
| 160 | // See also https://phabricator.wikimedia.org/T221560 |
| 161 | $searchChars = preg_replace( '/\\\\-/', '', $searchChars ); |
| 162 | |
| 163 | if ( $type === self::CHARS_ALL ) { |
| 164 | // " for phrase, * for wildcard |
| 165 | $searchChars = "\"*" . $searchChars; |
| 166 | } |
| 167 | return $searchChars; |
| 168 | } |
| 169 | |
| 170 | /** |
| 171 | * Perform a full text search query and return a result set. |
| 172 | * |
| 173 | * @param string $term Raw search term |
| 174 | * @return SqlSearchResultSet|null |
| 175 | */ |
| 176 | protected function doSearchTextInDB( $term ) { |
| 177 | return $this->searchInternal( $term, true ); |
| 178 | } |
| 179 | |
| 180 | /** |
| 181 | * Perform a title-only search query and return a result set. |
| 182 | * |
| 183 | * @param string $term Raw search term |
| 184 | * @return SqlSearchResultSet|null |
| 185 | */ |
| 186 | protected function doSearchTitleInDB( $term ) { |
| 187 | return $this->searchInternal( $term, false ); |
| 188 | } |
| 189 | |
| 190 | protected function searchInternal( string $term, bool $fulltext ): ?SqlSearchResultSet { |
| 191 | // This seems out of place, why is this called with empty term? |
| 192 | if ( trim( $term ) === '' ) { |
| 193 | return null; |
| 194 | } |
| 195 | |
| 196 | $filteredTerm = $this->filter( $term ); |
| 197 | $queryBuilder = $this->getQueryBuilder( $filteredTerm, $fulltext ); |
| 198 | $resultSet = $queryBuilder->caller( __METHOD__ )->fetchResultSet(); |
| 199 | |
| 200 | $queryBuilder = $this->getCountQueryBuilder( $filteredTerm, $fulltext ); |
| 201 | $total = (int)$queryBuilder->caller( __METHOD__ )->fetchField(); |
| 202 | |
| 203 | return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total ); |
| 204 | } |
| 205 | |
| 206 | /** @inheritDoc */ |
| 207 | public function supports( $feature ) { |
| 208 | switch ( $feature ) { |
| 209 | case 'title-suffix-filter': |
| 210 | return true; |
| 211 | default: |
| 212 | return parent::supports( $feature ); |
| 213 | } |
| 214 | } |
| 215 | |
| 216 | /** |
| 217 | * Add special conditions |
| 218 | * @param SelectQueryBuilder $queryBuilder |
| 219 | * @since 1.18 |
| 220 | */ |
| 221 | protected function queryFeatures( SelectQueryBuilder $queryBuilder ) { |
| 222 | foreach ( $this->features as $feature => $value ) { |
| 223 | if ( $feature === 'title-suffix-filter' && $value ) { |
| 224 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 225 | $queryBuilder->andWhere( |
| 226 | $dbr->expr( 'page_title', IExpression::LIKE, new LikeValue( $dbr->anyString(), $value ) ) |
| 227 | ); |
| 228 | } |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | /** |
| 233 | * Add namespace conditions |
| 234 | * @param SelectQueryBuilder $queryBuilder |
| 235 | * @since 1.18 (changed) |
| 236 | */ |
| 237 | private function queryNamespaces( $queryBuilder ) { |
| 238 | if ( is_array( $this->namespaces ) ) { |
| 239 | if ( count( $this->namespaces ) === 0 ) { |
| 240 | $this->namespaces[] = NS_MAIN; |
| 241 | } |
| 242 | $queryBuilder->andWhere( [ 'page_namespace' => $this->namespaces ] ); |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | /** |
| 247 | * Construct the SQL query builder to do the search. |
| 248 | * @param string $filteredTerm |
| 249 | * @param bool $fulltext |
| 250 | * @return SelectQueryBuilder |
| 251 | * @since 1.41 |
| 252 | */ |
| 253 | private function getQueryBuilder( $filteredTerm, $fulltext ): SelectQueryBuilder { |
| 254 | $queryBuilder = $this->dbProvider->getReplicaDatabase()->newSelectQueryBuilder(); |
| 255 | |
| 256 | $this->queryMain( $queryBuilder, $filteredTerm, $fulltext ); |
| 257 | $this->queryFeatures( $queryBuilder ); |
| 258 | $this->queryNamespaces( $queryBuilder ); |
| 259 | $queryBuilder->limit( $this->limit ) |
| 260 | ->offset( $this->offset ); |
| 261 | |
| 262 | return $queryBuilder; |
| 263 | } |
| 264 | |
| 265 | /** |
| 266 | * Picks which field to index on, depending on what type of query. |
| 267 | * @param bool $fulltext |
| 268 | * @return string |
| 269 | */ |
| 270 | private function getIndexField( $fulltext ) { |
| 271 | return $fulltext ? 'si_text' : 'si_title'; |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * Get the base part of the search query. |
| 276 | * |
| 277 | * @param SelectQueryBuilder $queryBuilder Search query builder |
| 278 | * @param string $filteredTerm |
| 279 | * @param bool $fulltext |
| 280 | * @since 1.18 (changed) |
| 281 | */ |
| 282 | private function queryMain( SelectQueryBuilder $queryBuilder, $filteredTerm, $fulltext ) { |
| 283 | $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 284 | $queryBuilder->select( [ 'page_id', 'page_namespace', 'page_title' ] ) |
| 285 | ->from( 'page' ) |
| 286 | ->join( 'searchindex', null, 'page_id=si_page' ) |
| 287 | ->where( $match[0] ) |
| 288 | ->orderBy( $match[1] ); |
| 289 | } |
| 290 | |
| 291 | /** |
| 292 | * @since 1.41 (changed) |
| 293 | * @param string $filteredTerm |
| 294 | * @param bool $fulltext |
| 295 | * @return SelectQueryBuilder |
| 296 | */ |
| 297 | private function getCountQueryBuilder( $filteredTerm, $fulltext ): SelectQueryBuilder { |
| 298 | $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 299 | $queryBuilder = $this->dbProvider->getReplicaDatabase()->newSelectQueryBuilder() |
| 300 | ->select( 'COUNT(*)' ) |
| 301 | ->from( 'page' ) |
| 302 | ->join( 'searchindex', null, 'page_id=si_page' ) |
| 303 | ->where( $match[0] ); |
| 304 | |
| 305 | $this->queryFeatures( $queryBuilder ); |
| 306 | $this->queryNamespaces( $queryBuilder ); |
| 307 | |
| 308 | return $queryBuilder; |
| 309 | } |
| 310 | |
| 311 | /** |
| 312 | * Create or update the search index record for the given page. |
| 313 | * Title and text should be pre-processed. |
| 314 | * |
| 315 | * @param int $id |
| 316 | * @param string $title |
| 317 | * @param string $text |
| 318 | */ |
| 319 | public function update( $id, $title, $text ) { |
| 320 | $this->dbProvider->getPrimaryDatabase()->newReplaceQueryBuilder() |
| 321 | ->replaceInto( 'searchindex' ) |
| 322 | ->uniqueIndexFields( [ 'si_page' ] ) |
| 323 | ->row( [ |
| 324 | 'si_page' => $id, |
| 325 | 'si_title' => $this->normalizeText( $title ), |
| 326 | 'si_text' => $this->normalizeText( $text ) |
| 327 | ] ) |
| 328 | ->caller( __METHOD__ )->execute(); |
| 329 | } |
| 330 | |
| 331 | /** |
| 332 | * Update a search index record's title only. |
| 333 | * Title should be pre-processed. |
| 334 | * |
| 335 | * @param int $id |
| 336 | * @param string $title |
| 337 | */ |
| 338 | public function updateTitle( $id, $title ) { |
| 339 | $this->dbProvider->getPrimaryDatabase()->newUpdateQueryBuilder() |
| 340 | ->update( 'searchindex' ) |
| 341 | ->set( [ 'si_title' => $this->normalizeText( $title ) ] ) |
| 342 | ->where( [ 'si_page' => $id ] ) |
| 343 | ->caller( __METHOD__ )->execute(); |
| 344 | } |
| 345 | |
| 346 | /** |
| 347 | * Delete an indexed page |
| 348 | * Title should be pre-processed. |
| 349 | * |
| 350 | * @param int $id Page id that was deleted |
| 351 | * @param string $title Title of page that was deleted |
| 352 | */ |
| 353 | public function delete( $id, $title ) { |
| 354 | $this->dbProvider->getPrimaryDatabase()->newDeleteQueryBuilder() |
| 355 | ->deleteFrom( 'searchindex' ) |
| 356 | ->where( [ 'si_page' => $id ] ) |
| 357 | ->caller( __METHOD__ )->execute(); |
| 358 | } |
| 359 | |
| 360 | /** |
| 361 | * Converts some characters for MySQL's indexing to grok it correctly, |
| 362 | * and pads short words to overcome limitations. |
| 363 | * @param string $string |
| 364 | * @return string |
| 365 | */ |
| 366 | public function normalizeText( $string ) { |
| 367 | $out = parent::normalizeText( $string ); |
| 368 | |
| 369 | // MySQL fulltext index doesn't grok utf-8, so we |
| 370 | // need to fold cases and convert to hex |
| 371 | $out = preg_replace_callback( |
| 372 | "/([\\xc0-\\xff][\\x80-\\xbf]*)/", |
| 373 | $this->stripForSearchCallback( ... ), |
| 374 | MediaWikiServices::getInstance()->getContentLanguage()->lc( $out ) ); |
| 375 | |
| 376 | // And to add insult to injury, the default indexing |
| 377 | // ignores short words... Pad them so we can pass them |
| 378 | // through without reconfiguring the server... |
| 379 | $minLength = $this->minSearchLength(); |
| 380 | if ( $minLength > 1 ) { |
| 381 | $n = $minLength - 1; |
| 382 | $out = preg_replace( |
| 383 | "/\b(\w{1,$n})\b/", |
| 384 | "$1u800", |
| 385 | $out ); |
| 386 | } |
| 387 | |
| 388 | // Periods within things like hostnames and IP addresses |
| 389 | // are also important -- we want a search for "example.com" |
| 390 | // or "192.168.1.1" to work sensibly. |
| 391 | // MySQL's search seems to ignore them, so you'd match on |
| 392 | // "example.wikipedia.com" and "192.168.83.1" as well. |
| 393 | return preg_replace( |
| 394 | "/(\w)\.(\w|\*)/u", |
| 395 | "$1u82e$2", |
| 396 | $out |
| 397 | ); |
| 398 | } |
| 399 | |
| 400 | /** |
| 401 | * Armor a case-folded UTF-8 string to get through MySQL's |
| 402 | * fulltext search without being mucked up by funny charset |
| 403 | * settings or anything else of the sort. |
| 404 | * @param array $matches |
| 405 | * @return string |
| 406 | */ |
| 407 | protected function stripForSearchCallback( $matches ) { |
| 408 | return 'u8' . bin2hex( $matches[1] ); |
| 409 | } |
| 410 | |
| 411 | /** |
| 412 | * Check MySQL server's ft_min_word_len setting so we know |
| 413 | * if we need to pad short words... |
| 414 | * |
| 415 | * @return int |
| 416 | */ |
| 417 | protected function minSearchLength() { |
| 418 | if ( self::$mMinSearchLength === null ) { |
| 419 | $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'"; |
| 420 | |
| 421 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 422 | // The real type is still IDatabase, but IReplicaDatabase is used for safety. |
| 423 | '@phan-var IDatabase $dbr'; |
| 424 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 425 | $result = $dbr->query( $sql, __METHOD__ ); |
| 426 | $row = $result->fetchObject(); |
| 427 | $result->free(); |
| 428 | |
| 429 | if ( $row && $row->Variable_name == 'ft_min_word_len' ) { |
| 430 | self::$mMinSearchLength = intval( $row->Value ); |
| 431 | } else { |
| 432 | self::$mMinSearchLength = 0; |
| 433 | } |
| 434 | } |
| 435 | return self::$mMinSearchLength; |
| 436 | } |
| 437 | } |