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 /**
  28  * Search engine hook for MySQL 4+
  29  * @ingroup Search
  30  */
  31 class SearchMySQL extends SearchDatabase {
  32         protected $strictMatching = true;
  33
  34         private static $mMinSearchLength;
  35
  36         /**
  37          * Parse the user's query and transform it into two SQL fragments:
  38          * a WHERE condition and an ORDER BY expression
  39          *
  40          * @param string $filteredText
  41          * @param string $fulltext
  42          *
  43          * @return array
  44          */
  45         function parseQuery( $filteredText, $fulltext ) {
  46                 global $wgContLang;
  47
  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                                 $convertedVariants = $wgContLang->autoConvertToAllVariants( $term );
  80                                 if ( is_array( $convertedVariants ) ) {
  81                                         $variants = array_unique( array_values( $convertedVariants ) );
  82                                 } else {
  83                                         $variants = [ $term ];
  84                                 }
  85
  86                                 // The low-level search index does some processing on input to work
  87                                 // around problems with minimum lengths and encoding in MySQL's
  88                                 // fulltext engine.
  89                                 // For Chinese this also inserts spaces between adjacent Han characters.
  90                                 $strippedVariants = array_map(
  91                                         [ $wgContLang, 'normalizeForSearch' ],
  92                                         $variants );
  93
  94                                 // Some languages such as Chinese force all variants to a canonical
  95                                 // form when stripping to the low-level search index, so to be sure
  96                                 // let's check our variants list for unique items after stripping.
  97                                 $strippedVariants = array_unique( $strippedVariants );
  98
  99                                 $searchon .= $modifier;
 100                                 if ( count( $strippedVariants ) > 1 ) {
 101                                         $searchon .= '(';
 102                                 }
 103                                 foreach ( $strippedVariants as $stripped ) {
 104                                         $stripped = $this->normalizeText( $stripped );
 105                                         if ( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
 106                                                 // Hack for Chinese: we need to toss in quotes for
 107                                                 // multiple-character phrases since normalizeForSearch()
 108                                                 // added spaces between them to make word breaks.
 109                                                 $stripped = '"' . trim( $stripped ) . '"';
 110                                         }
 111                                         $searchon .= "$quote$stripped$quote$wildcard ";
 112                                 }
 113                                 if ( count( $strippedVariants ) > 1 ) {
 114                                         $searchon .= ')';
 115                                 }
 116
 117                                 // Match individual terms or quoted phrase in result highlighting...
 118                                 // Note that variants will be introduced in a later stage for highlighting!
 119                                 $regexp = $this->regexTerm( $term, $wildcard );
 120                                 $this->searchTerms[] = $regexp;
 121                         }
 122                         wfDebug( __METHOD__ . ": Would search with '$searchon'\n" );
 123                         wfDebug( __METHOD__ . ': Match with /' . implode( '|', $this->searchTerms ) . "/\n" );
 124                 } else {
 125                         wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" );
 126                 }
 127
 128                 $searchon = $this->db->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         function regexTerm( $string, $wildcard ) {
 137                 global $wgContLang;
 138
 139                 $regex = preg_quote( $string, '/' );
 140                 if ( $wgContLang->hasWordBreaks() ) {
 141                         if ( $wildcard ) {
 142                                 // Don't cut off the final bit!
 143                                 $regex = "\b$regex";
 144                         } else {
 145                                 $regex = "\b$regex\b";
 146                         }
 147                 } else {
 148                         // For Chinese, words may legitimately abut other words in the text literal.
 149                         // Don't add \b boundary checks... note this could cause false positives
 150                         // for latin chars.
 151                 }
 152                 return $regex;
 153         }
 154
 155         public static function legalSearchChars( $type = self::CHARS_ALL ) {
 156                 $searchChars = parent::legalSearchChars( $type );
 157                 if ( $type === self::CHARS_ALL ) {
 158                         // " for phrase, * for wildcard
 159                         $searchChars = "\"*" . $searchChars;
 160                 }
 161                 return $searchChars;
 162         }
 163
 164         /**
 165          * Perform a full text search query and return a result set.
 166          *
 167          * @param string $term Raw search term
 168          * @return SqlSearchResultSet
 169          */
 170         function searchText( $term ) {
 171                 return $this->searchInternal( $term, true );
 172         }
 173
 174         /**
 175          * Perform a title-only search query and return a result set.
 176          *
 177          * @param string $term Raw search term
 178          * @return SqlSearchResultSet
 179          */
 180         function searchTitle( $term ) {
 181                 return $this->searchInternal( $term, false );
 182         }
 183
 184         protected function searchInternal( $term, $fulltext ) {
 185                 // This seems out of place, why is this called with empty term?
 186                 if ( trim( $term ) === '' ) {
 187                         return null;
 188                 }
 189
 190                 $filteredTerm = $this->filter( $term );
 191                 $query = $this->getQuery( $filteredTerm, $fulltext );
 192                 $resultSet = $this->db->select(
 193                         $query['tables'], $query['fields'], $query['conds'],
 194                         __METHOD__, $query['options'], $query['joins']
 195                 );
 196
 197                 $total = null;
 198                 $query = $this->getCountQuery( $filteredTerm, $fulltext );
 199                 $totalResult = $this->db->select(
 200                         $query['tables'], $query['fields'], $query['conds'],
 201                         __METHOD__, $query['options'], $query['joins']
 202                 );
 203
 204                 $row = $totalResult->fetchObject();
 205                 if ( $row ) {
 206                         $total = intval( $row->c );
 207                 }
 208                 $totalResult->free();
 209
 210                 return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total );
 211         }
 212
 213         public function supports( $feature ) {
 214                 switch ( $feature ) {
 215                         case 'title-suffix-filter':
 216                                 return true;
 217                         default:
 218                                 return parent::supports( $feature );
 219                 }
 220         }
 221
 222         /**
 223          * Add special conditions
 224          * @param array &$query
 225          * @since 1.18
 226          */
 227         protected function queryFeatures( &$query ) {
 228                 foreach ( $this->features as $feature => $value ) {
 229                         if ( $feature === 'title-suffix-filter' && $value ) {
 230                                 $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->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         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         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 $fulltext
 299          * @since 1.18 (changed)
 300          */
 301         function queryMain( &$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         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 = wfGetDB( 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 = wfGetDB( DB_MASTER );
 364
 365                 $dbw->update( 'searchindex',
 366                         [ 'si_title' => $this->normalizeText( $title ) ],
 367                         [ 'si_page' => $id ],
 368                         __METHOD__,
 369                         [ $dbw->lowPriorityOption() ] );
 370         }
 371
 372         /**
 373          * Delete an indexed page
 374          * Title should be pre-processed.
 375          *
 376          * @param int $id Page id that was deleted
 377          * @param string $title Title of page that was deleted
 378          */
 379         function delete( $id, $title ) {
 380                 $dbw = wfGetDB( DB_MASTER );
 381
 382                 $dbw->delete( 'searchindex', [ 'si_page' => $id ], __METHOD__ );
 383         }
 384
 385         /**
 386          * Converts some characters for MySQL's indexing to grok it correctly,
 387          * and pads short words to overcome limitations.
 388          * @param string $string
 389          * @return mixed|string
 390          */
 391         function normalizeText( $string ) {
 392                 global $wgContLang;
 393
 394                 $out = parent::normalizeText( $string );
 395
 396                 // MySQL fulltext index doesn't grok utf-8, so we
 397                 // need to fold cases and convert to hex
 398                 $out = preg_replace_callback(
 399                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 400                         [ $this, 'stripForSearchCallback' ],
 401                         $wgContLang->lc( $out ) );
 402
 403                 // And to add insult to injury, the default indexing
 404                 // ignores short words... Pad them so we can pass them
 405                 // through without reconfiguring the server...
 406                 $minLength = $this->minSearchLength();
 407                 if ( $minLength > 1 ) {
 408                         $n = $minLength - 1;
 409                         $out = preg_replace(
 410                                 "/\b(\w{1,$n})\b/",
 411                                 "$1u800",
 412                                 $out );
 413                 }
 414
 415                 // Periods within things like hostnames and IP addresses
 416                 // are also important -- we want a search for "example.com"
 417                 // or "192.168.1.1" to work sanely.
 418                 // MySQL's search seems to ignore them, so you'd match on
 419                 // "example.wikipedia.com" and "192.168.83.1" as well.
 420                 $out = preg_replace(
 421                         "/(\w)\.(\w|\*)/u",
 422                         "$1u82e$2",
 423                         $out );
 424
 425                 return $out;
 426         }
 427
 428         /**
 429          * Armor a case-folded UTF-8 string to get through MySQL's
 430          * fulltext search without being mucked up by funny charset
 431          * settings or anything else of the sort.
 432          * @param array $matches
 433          * @return string
 434          */
 435         protected function stripForSearchCallback( $matches ) {
 436                 return 'u8' . bin2hex( $matches[1] );
 437         }
 438
 439         /**
 440          * Check MySQL server's ft_min_word_len setting so we know
 441          * if we need to pad short words...
 442          *
 443          * @return int
 444          */
 445         protected function minSearchLength() {
 446                 if ( is_null( self::$mMinSearchLength ) ) {
 447                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 448
 449                         $dbr = wfGetDB( DB_REPLICA );
 450                         $result = $dbr->query( $sql, __METHOD__ );
 451                         $row = $result->fetchObject();
 452                         $result->free();
 453
 454                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 455                                 self::$mMinSearchLength = intval( $row->Value );
 456                         } else {
 457                                 self::$mMinSearchLength = 0;
 458                         }
 459                 }
 460                 return self::$mMinSearchLength;
 461         }
 462 }