includes/search/SearchMySQL.php

   1 <?php
   2 /**
   3  * MySQL search engine
   4  *
   5  * Copyright (C) 2004 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Search
  25  */
  26
  27 use MediaWiki\MediaWikiServices;
  28
  29 /**
  30  * Search engine hook for MySQL 4+
  31  * @ingroup Search
  32  */
  33 class SearchMySQL extends SearchDatabase {
  34         protected $strictMatching = true;
  35
  36         private static $mMinSearchLength;
  37
  38         /**
  39          * Parse the user's query and transform it into two SQL fragments:
  40          * a WHERE condition and an ORDER BY expression
  41          *
  42          * @param string $filteredText
  43          * @param string $fulltext
  44          *
  45          * @return array
  46          */
  47         private function parseQuery( $filteredText, $fulltext ) {
  48                 $lc = $this->legalSearchChars( self::CHARS_NO_SYNTAX ); // Minus syntax chars (" and *)
  49                 $searchon = '';
  50                 $this->searchTerms = [];
  51
  52                 # @todo FIXME: This doesn't handle parenthetical expressions.
  53                 $m = [];
  54                 if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
  55                                 $filteredText, $m, PREG_SET_ORDER ) ) {
  56                         foreach ( $m as $bits ) {
  57                                 Wikimedia\suppressWarnings();
  58                                 list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits;
  59                                 Wikimedia\restoreWarnings();
  60
  61                                 if ( $nonQuoted != '' ) {
  62                                         $term = $nonQuoted;
  63                                         $quote = '';
  64                                 } else {
  65                                         $term = str_replace( '"', '', $term );
  66                                         $quote = '"';
  67                                 }
  68
  69                                 if ( $searchon !== '' ) {
  70                                         $searchon .= ' ';
  71                                 }
  72                                 if ( $this->strictMatching && ( $modifier == '' ) ) {
  73                                         // If we leave this out, boolean op defaults to OR which is rarely helpful.
  74                                         $modifier = '+';
  75                                 }
  76
  77                                 // Some languages such as Serbian store the input form in the search index,
  78                                 // so we may need to search for matches in multiple writing system variants.
  79                                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
  80                                 $convertedVariants = $contLang->autoConvertToAllVariants( $term );
  81                                 if ( is_array( $convertedVariants ) ) {
  82                                         $variants = array_unique( array_values( $convertedVariants ) );
  83                                 } else {
  84                                         $variants = [ $term ];
  85                                 }
  86
  87                                 // The low-level search index does some processing on input to work
  88                                 // around problems with minimum lengths and encoding in MySQL's
  89                                 // fulltext engine.
  90                                 // For Chinese this also inserts spaces between adjacent Han characters.
  91                                 $strippedVariants = array_map( [ $contLang, 'normalizeForSearch' ], $variants );
  92
  93                                 // Some languages such as Chinese force all variants to a canonical
  94                                 // form when stripping to the low-level search index, so to be sure
  95                                 // let's check our variants list for unique items after stripping.
  96                                 $strippedVariants = array_unique( $strippedVariants );
  97
  98                                 $searchon .= $modifier;
  99                                 if ( count( $strippedVariants ) > 1 ) {
 100                                         $searchon .= '(';
 101                                 }
 102                                 foreach ( $strippedVariants as $stripped ) {
 103                                         $stripped = $this->normalizeText( $stripped );
 104                                         if ( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
 105                                                 // Hack for Chinese: we need to toss in quotes for
 106                                                 // multiple-character phrases since normalizeForSearch()
 107                                                 // added spaces between them to make word breaks.
 108                                                 $stripped = '"' . trim( $stripped ) . '"';
 109                                         }
 110                                         $searchon .= "$quote$stripped$quote$wildcard ";
 111                                 }
 112                                 if ( count( $strippedVariants ) > 1 ) {
 113                                         $searchon .= ')';
 114                                 }
 115
 116                                 // Match individual terms or quoted phrase in result highlighting...
 117                                 // Note that variants will be introduced in a later stage for highlighting!
 118                                 $regexp = $this->regexTerm( $term, $wildcard );
 119                                 $this->searchTerms[] = $regexp;
 120                         }
 121                         wfDebug( __METHOD__ . ": Would search with '$searchon'\n" );
 122                         wfDebug( __METHOD__ . ': Match with /' . implode( '|', $this->searchTerms ) . "/\n" );
 123                 } else {
 124                         wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" );
 125                 }
 126
 127                 $dbr = $this->lb->getConnectionRef( DB_REPLICA );
 128                 $searchon = $dbr->addQuotes( $searchon );
 129                 $field = $this->getIndexField( $fulltext );
 130                 return [
 131                         " MATCH($field) AGAINST($searchon IN BOOLEAN MODE) ",
 132                         " MATCH($field) AGAINST($searchon IN NATURAL LANGUAGE MODE) DESC "
 133                 ];
 134         }
 135
 136         private function regexTerm( $string, $wildcard ) {
 137                 $regex = preg_quote( $string, '/' );
 138                 if ( MediaWikiServices::getInstance()->getContentLanguage()->hasWordBreaks() ) {
 139                         if ( $wildcard ) {
 140                                 // Don't cut off the final bit!
 141                                 $regex = "\b$regex";
 142                         } else {
 143                                 $regex = "\b$regex\b";
 144                         }
 145                 } else {
 146                         // For Chinese, words may legitimately abut other words in the text literal.
 147                         // Don't add \b boundary checks... note this could cause false positives
 148                         // for Latin chars.
 149                 }
 150                 return $regex;
 151         }
 152
 153         public function legalSearchChars( $type = self::CHARS_ALL ) {
 154                 $searchChars = parent::legalSearchChars( $type );
 155                 if ( $type === self::CHARS_ALL ) {
 156                         // " for phrase, * for wildcard
 157                         $searchChars = "\"*" . $searchChars;
 158                 }
 159                 return $searchChars;
 160         }
 161
 162         /**
 163          * Perform a full text search query and return a result set.
 164          *
 165          * @param string $term Raw search term
 166          * @return SqlSearchResultSet
 167          */
 168         protected function doSearchTextInDB( $term ) {
 169                 return $this->searchInternal( $term, true );
 170         }
 171
 172         /**
 173          * Perform a title-only search query and return a result set.
 174          *
 175          * @param string $term Raw search term
 176          * @return SqlSearchResultSet
 177          */
 178         protected function doSearchTitleInDB( $term ) {
 179                 return $this->searchInternal( $term, false );
 180         }
 181
 182         protected function searchInternal( $term, $fulltext ) {
 183                 // This seems out of place, why is this called with empty term?
 184                 if ( trim( $term ) === '' ) {
 185                         return null;
 186                 }
 187
 188                 $filteredTerm = $this->filter( $term );
 189                 $query = $this->getQuery( $filteredTerm, $fulltext );
 190                 $dbr = $this->lb->getConnectionRef( DB_REPLICA );
 191                 $resultSet = $dbr->select(
 192                         $query['tables'], $query['fields'], $query['conds'],
 193                         __METHOD__, $query['options'], $query['joins']
 194                 );
 195
 196                 $total = null;
 197                 $query = $this->getCountQuery( $filteredTerm, $fulltext );
 198                 $totalResult = $dbr->select(
 199                         $query['tables'], $query['fields'], $query['conds'],
 200                         __METHOD__, $query['options'], $query['joins']
 201                 );
 202
 203                 $row = $totalResult->fetchObject();
 204                 if ( $row ) {
 205                         $total = intval( $row->c );
 206                 }
 207                 $totalResult->free();
 208
 209                 return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total );
 210         }
 211
 212         public function supports( $feature ) {
 213                 switch ( $feature ) {
 214                         case 'title-suffix-filter':
 215                                 return true;
 216                         default:
 217                                 return parent::supports( $feature );
 218                 }
 219         }
 220
 221         /**
 222          * Add special conditions
 223          * @param array &$query
 224          * @since 1.18
 225          */
 226         protected function queryFeatures( &$query ) {
 227                 foreach ( $this->features as $feature => $value ) {
 228                         if ( $feature === 'title-suffix-filter' && $value ) {
 229                                 $dbr = $this->lb->getConnectionRef( DB_REPLICA );
 230                                 $query['conds'][] = 'page_title' . $dbr->buildLike( $dbr->anyString(), $value );
 231                         }
 232                 }
 233         }
 234
 235         /**
 236          * Add namespace conditions
 237          * @param array &$query
 238          * @since 1.18 (changed)
 239          */
 240         function queryNamespaces( &$query ) {
 241                 if ( is_array( $this->namespaces ) ) {
 242                         if ( count( $this->namespaces ) === 0 ) {
 243                                 $this->namespaces[] = '0';
 244                         }
 245                         $query['conds']['page_namespace'] = $this->namespaces;
 246                 }
 247         }
 248
 249         /**
 250          * Add limit options
 251          * @param array &$query
 252          * @since 1.18
 253          */
 254         protected function limitResult( &$query ) {
 255                 $query['options']['LIMIT'] = $this->limit;
 256                 $query['options']['OFFSET'] = $this->offset;
 257         }
 258
 259         /**
 260          * Construct the SQL query to do the search.
 261          * The guts shoulds be constructed in queryMain()
 262          * @param string $filteredTerm
 263          * @param bool $fulltext
 264          * @return array
 265          * @since 1.18 (changed)
 266          */
 267         private function getQuery( $filteredTerm, $fulltext ) {
 268                 $query = [
 269                         'tables' => [],
 270                         'fields' => [],
 271                         'conds' => [],
 272                         'options' => [],
 273                         'joins' => [],
 274                 ];
 275
 276                 $this->queryMain( $query, $filteredTerm, $fulltext );
 277                 $this->queryFeatures( $query );
 278                 $this->queryNamespaces( $query );
 279                 $this->limitResult( $query );
 280
 281                 return $query;
 282         }
 283
 284         /**
 285          * Picks which field to index on, depending on what type of query.
 286          * @param bool $fulltext
 287          * @return string
 288          */
 289         private function getIndexField( $fulltext ) {
 290                 return $fulltext ? 'si_text' : 'si_title';
 291         }
 292
 293         /**
 294          * Get the base part of the search query.
 295          *
 296          * @param array &$query Search query array
 297          * @param string $filteredTerm
 298          * @param bool $fulltext
 299          * @since 1.18 (changed)
 300          */
 301         private function queryMain( &$query, $filteredTerm, $fulltext ) {
 302                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 303                 $query['tables'][] = 'page';
 304                 $query['tables'][] = 'searchindex';
 305                 $query['fields'][] = 'page_id';
 306                 $query['fields'][] = 'page_namespace';
 307                 $query['fields'][] = 'page_title';
 308                 $query['conds'][] = 'page_id=si_page';
 309                 $query['conds'][] = $match[0];
 310                 $query['options']['ORDER BY'] = $match[1];
 311         }
 312
 313         /**
 314          * @since 1.18 (changed)
 315          * @param string $filteredTerm
 316          * @param bool $fulltext
 317          * @return array
 318          */
 319         private function getCountQuery( $filteredTerm, $fulltext ) {
 320                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 321
 322                 $query = [
 323                         'tables' => [ 'page', 'searchindex' ],
 324                         'fields' => [ 'COUNT(*) as c' ],
 325                         'conds' => [ 'page_id=si_page', $match[0] ],
 326                         'options' => [],
 327                         'joins' => [],
 328                 ];
 329
 330                 $this->queryFeatures( $query );
 331                 $this->queryNamespaces( $query );
 332
 333                 return $query;
 334         }
 335
 336         /**
 337          * Create or update the search index record for the given page.
 338          * Title and text should be pre-processed.
 339          *
 340          * @param int $id
 341          * @param string $title
 342          * @param string $text
 343          */
 344         function update( $id, $title, $text ) {
 345                 $dbw = $this->lb->getConnectionRef( DB_MASTER );
 346                 $dbw->replace( 'searchindex',
 347                         [ 'si_page' ],
 348                         [
 349                                 'si_page' => $id,
 350                                 'si_title' => $this->normalizeText( $title ),
 351                                 'si_text' => $this->normalizeText( $text )
 352                         ], __METHOD__ );
 353         }
 354
 355         /**
 356          * Update a search index record's title only.
 357          * Title should be pre-processed.
 358          *
 359          * @param int $id
 360          * @param string $title
 361          */
 362         function updateTitle( $id, $title ) {
 363                 $dbw = $this->lb->getConnectionRef( DB_MASTER );
 364                 $dbw->update( 'searchindex',
 365                         [ 'si_title' => $this->normalizeText( $title ) ],
 366                         [ 'si_page' => $id ],
 367                         __METHOD__
 368                 );
 369         }
 370
 371         /**
 372          * Delete an indexed page
 373          * Title should be pre-processed.
 374          *
 375          * @param int $id Page id that was deleted
 376          * @param string $title Title of page that was deleted
 377          */
 378         function delete( $id, $title ) {
 379                 $dbw = $this->lb->getConnectionRef( DB_MASTER );
 380                 $dbw->delete( 'searchindex', [ 'si_page' => $id ], __METHOD__ );
 381         }
 382
 383         /**
 384          * Converts some characters for MySQL's indexing to grok it correctly,
 385          * and pads short words to overcome limitations.
 386          * @param string $string
 387          * @return mixed|string
 388          */
 389         function normalizeText( $string ) {
 390                 $out = parent::normalizeText( $string );
 391
 392                 // MySQL fulltext index doesn't grok utf-8, so we
 393                 // need to fold cases and convert to hex
 394                 $out = preg_replace_callback(
 395                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 396                         [ $this, 'stripForSearchCallback' ],
 397                         MediaWikiServices::getInstance()->getContentLanguage()->lc( $out ) );
 398
 399                 // And to add insult to injury, the default indexing
 400                 // ignores short words... Pad them so we can pass them
 401                 // through without reconfiguring the server...
 402                 $minLength = $this->minSearchLength();
 403                 if ( $minLength > 1 ) {
 404                         $n = $minLength - 1;
 405                         $out = preg_replace(
 406                                 "/\b(\w{1,$n})\b/",
 407                                 "$1u800",
 408                                 $out );
 409                 }
 410
 411                 // Periods within things like hostnames and IP addresses
 412                 // are also important -- we want a search for "example.com"
 413                 // or "192.168.1.1" to work sanely.
 414                 // MySQL's search seems to ignore them, so you'd match on
 415                 // "example.wikipedia.com" and "192.168.83.1" as well.
 416                 $out = preg_replace(
 417                         "/(\w)\.(\w|\*)/u",
 418                         "$1u82e$2",
 419                         $out );
 420
 421                 return $out;
 422         }
 423
 424         /**
 425          * Armor a case-folded UTF-8 string to get through MySQL's
 426          * fulltext search without being mucked up by funny charset
 427          * settings or anything else of the sort.
 428          * @param array $matches
 429          * @return string
 430          */
 431         protected function stripForSearchCallback( $matches ) {
 432                 return 'u8' . bin2hex( $matches[1] );
 433         }
 434
 435         /**
 436          * Check MySQL server's ft_min_word_len setting so we know
 437          * if we need to pad short words...
 438          *
 439          * @return int
 440          */
 441         protected function minSearchLength() {
 442                 if ( is_null( self::$mMinSearchLength ) ) {
 443                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 444
 445                         $dbr = $this->lb->getConnectionRef( DB_REPLICA );
 446                         $result = $dbr->query( $sql, __METHOD__ );
 447                         $row = $result->fetchObject();
 448                         $result->free();
 449
 450                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 451                                 self::$mMinSearchLength = intval( $row->Value );
 452                         } else {
 453                                 self::$mMinSearchLength = 0;
 454                         }
 455                 }
 456                 return self::$mMinSearchLength;
 457         }
 458 }