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                 $searchon = $this->db->addQuotes( $searchon );
 128                 $field = $this->getIndexField( $fulltext );
 129                 return [
 130                         " MATCH($field) AGAINST($searchon IN BOOLEAN MODE) ",
 131                         " MATCH($field) AGAINST($searchon IN NATURAL LANGUAGE MODE) DESC "
 132                 ];
 133         }
 134
 135         private function regexTerm( $string, $wildcard ) {
 136                 $regex = preg_quote( $string, '/' );
 137                 if ( MediaWikiServices::getInstance()->getContentLanguage()->hasWordBreaks() ) {
 138                         if ( $wildcard ) {
 139                                 // Don't cut off the final bit!
 140                                 $regex = "\b$regex";
 141                         } else {
 142                                 $regex = "\b$regex\b";
 143                         }
 144                 } else {
 145                         // For Chinese, words may legitimately abut other words in the text literal.
 146                         // Don't add \b boundary checks... note this could cause false positives
 147                         // for Latin chars.
 148                 }
 149                 return $regex;
 150         }
 151
 152         public function legalSearchChars( $type = self::CHARS_ALL ) {
 153                 $searchChars = parent::legalSearchChars( $type );
 154                 if ( $type === self::CHARS_ALL ) {
 155                         // " for phrase, * for wildcard
 156                         $searchChars = "\"*" . $searchChars;
 157                 }
 158                 return $searchChars;
 159         }
 160
 161         /**
 162          * Perform a full text search query and return a result set.
 163          *
 164          * @param string $term Raw search term
 165          * @return SqlSearchResultSet
 166          */
 167         protected function doSearchTextInDB( $term ) {
 168                 return $this->searchInternal( $term, true );
 169         }
 170
 171         /**
 172          * Perform a title-only search query and return a result set.
 173          *
 174          * @param string $term Raw search term
 175          * @return SqlSearchResultSet
 176          */
 177         protected function doSearchTitleInDB( $term ) {
 178                 return $this->searchInternal( $term, false );
 179         }
 180
 181         protected function searchInternal( $term, $fulltext ) {
 182                 // This seems out of place, why is this called with empty term?
 183                 if ( trim( $term ) === '' ) {
 184                         return null;
 185                 }
 186
 187                 $filteredTerm = $this->filter( $term );
 188                 $query = $this->getQuery( $filteredTerm, $fulltext );
 189                 $resultSet = $this->db->select(
 190                         $query['tables'], $query['fields'], $query['conds'],
 191                         __METHOD__, $query['options'], $query['joins']
 192                 );
 193
 194                 $total = null;
 195                 $query = $this->getCountQuery( $filteredTerm, $fulltext );
 196                 $totalResult = $this->db->select(
 197                         $query['tables'], $query['fields'], $query['conds'],
 198                         __METHOD__, $query['options'], $query['joins']
 199                 );
 200
 201                 $row = $totalResult->fetchObject();
 202                 if ( $row ) {
 203                         $total = intval( $row->c );
 204                 }
 205                 $totalResult->free();
 206
 207                 return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total );
 208         }
 209
 210         public function supports( $feature ) {
 211                 switch ( $feature ) {
 212                         case 'title-suffix-filter':
 213                                 return true;
 214                         default:
 215                                 return parent::supports( $feature );
 216                 }
 217         }
 218
 219         /**
 220          * Add special conditions
 221          * @param array &$query
 222          * @since 1.18
 223          */
 224         protected function queryFeatures( &$query ) {
 225                 foreach ( $this->features as $feature => $value ) {
 226                         if ( $feature === 'title-suffix-filter' && $value ) {
 227                                 $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->anyString(), $value );
 228                         }
 229                 }
 230         }
 231
 232         /**
 233          * Add namespace conditions
 234          * @param array &$query
 235          * @since 1.18 (changed)
 236          */
 237         function queryNamespaces( &$query ) {
 238                 if ( is_array( $this->namespaces ) ) {
 239                         if ( count( $this->namespaces ) === 0 ) {
 240                                 $this->namespaces[] = '0';
 241                         }
 242                         $query['conds']['page_namespace'] = $this->namespaces;
 243                 }
 244         }
 245
 246         /**
 247          * Add limit options
 248          * @param array &$query
 249          * @since 1.18
 250          */
 251         protected function limitResult( &$query ) {
 252                 $query['options']['LIMIT'] = $this->limit;
 253                 $query['options']['OFFSET'] = $this->offset;
 254         }
 255
 256         /**
 257          * Construct the SQL query to do the search.
 258          * The guts shoulds be constructed in queryMain()
 259          * @param string $filteredTerm
 260          * @param bool $fulltext
 261          * @return array
 262          * @since 1.18 (changed)
 263          */
 264         private function getQuery( $filteredTerm, $fulltext ) {
 265                 $query = [
 266                         'tables' => [],
 267                         'fields' => [],
 268                         'conds' => [],
 269                         'options' => [],
 270                         'joins' => [],
 271                 ];
 272
 273                 $this->queryMain( $query, $filteredTerm, $fulltext );
 274                 $this->queryFeatures( $query );
 275                 $this->queryNamespaces( $query );
 276                 $this->limitResult( $query );
 277
 278                 return $query;
 279         }
 280
 281         /**
 282          * Picks which field to index on, depending on what type of query.
 283          * @param bool $fulltext
 284          * @return string
 285          */
 286         private function getIndexField( $fulltext ) {
 287                 return $fulltext ? 'si_text' : 'si_title';
 288         }
 289
 290         /**
 291          * Get the base part of the search query.
 292          *
 293          * @param array &$query Search query array
 294          * @param string $filteredTerm
 295          * @param bool $fulltext
 296          * @since 1.18 (changed)
 297          */
 298         private function queryMain( &$query, $filteredTerm, $fulltext ) {
 299                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 300                 $query['tables'][] = 'page';
 301                 $query['tables'][] = 'searchindex';
 302                 $query['fields'][] = 'page_id';
 303                 $query['fields'][] = 'page_namespace';
 304                 $query['fields'][] = 'page_title';
 305                 $query['conds'][] = 'page_id=si_page';
 306                 $query['conds'][] = $match[0];
 307                 $query['options']['ORDER BY'] = $match[1];
 308         }
 309
 310         /**
 311          * @since 1.18 (changed)
 312          * @param string $filteredTerm
 313          * @param bool $fulltext
 314          * @return array
 315          */
 316         private function getCountQuery( $filteredTerm, $fulltext ) {
 317                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 318
 319                 $query = [
 320                         'tables' => [ 'page', 'searchindex' ],
 321                         'fields' => [ 'COUNT(*) as c' ],
 322                         'conds' => [ 'page_id=si_page', $match[0] ],
 323                         'options' => [],
 324                         'joins' => [],
 325                 ];
 326
 327                 $this->queryFeatures( $query );
 328                 $this->queryNamespaces( $query );
 329
 330                 return $query;
 331         }
 332
 333         /**
 334          * Create or update the search index record for the given page.
 335          * Title and text should be pre-processed.
 336          *
 337          * @param int $id
 338          * @param string $title
 339          * @param string $text
 340          */
 341         function update( $id, $title, $text ) {
 342                 $dbw = wfGetDB( DB_MASTER );
 343                 $dbw->replace( 'searchindex',
 344                         [ 'si_page' ],
 345                         [
 346                                 'si_page' => $id,
 347                                 'si_title' => $this->normalizeText( $title ),
 348                                 'si_text' => $this->normalizeText( $text )
 349                         ], __METHOD__ );
 350         }
 351
 352         /**
 353          * Update a search index record's title only.
 354          * Title should be pre-processed.
 355          *
 356          * @param int $id
 357          * @param string $title
 358          */
 359         function updateTitle( $id, $title ) {
 360                 $dbw = wfGetDB( DB_MASTER );
 361
 362                 $dbw->update( 'searchindex',
 363                         [ 'si_title' => $this->normalizeText( $title ) ],
 364                         [ 'si_page' => $id ],
 365                         __METHOD__,
 366                         [ $dbw->lowPriorityOption() ] );
 367         }
 368
 369         /**
 370          * Delete an indexed page
 371          * Title should be pre-processed.
 372          *
 373          * @param int $id Page id that was deleted
 374          * @param string $title Title of page that was deleted
 375          */
 376         function delete( $id, $title ) {
 377                 $dbw = wfGetDB( DB_MASTER );
 378
 379                 $dbw->delete( 'searchindex', [ 'si_page' => $id ], __METHOD__ );
 380         }
 381
 382         /**
 383          * Converts some characters for MySQL's indexing to grok it correctly,
 384          * and pads short words to overcome limitations.
 385          * @param string $string
 386          * @return mixed|string
 387          */
 388         function normalizeText( $string ) {
 389                 $out = parent::normalizeText( $string );
 390
 391                 // MySQL fulltext index doesn't grok utf-8, so we
 392                 // need to fold cases and convert to hex
 393                 $out = preg_replace_callback(
 394                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 395                         [ $this, 'stripForSearchCallback' ],
 396                         MediaWikiServices::getInstance()->getContentLanguage()->lc( $out ) );
 397
 398                 // And to add insult to injury, the default indexing
 399                 // ignores short words... Pad them so we can pass them
 400                 // through without reconfiguring the server...
 401                 $minLength = $this->minSearchLength();
 402                 if ( $minLength > 1 ) {
 403                         $n = $minLength - 1;
 404                         $out = preg_replace(
 405                                 "/\b(\w{1,$n})\b/",
 406                                 "$1u800",
 407                                 $out );
 408                 }
 409
 410                 // Periods within things like hostnames and IP addresses
 411                 // are also important -- we want a search for "example.com"
 412                 // or "192.168.1.1" to work sanely.
 413                 // MySQL's search seems to ignore them, so you'd match on
 414                 // "example.wikipedia.com" and "192.168.83.1" as well.
 415                 $out = preg_replace(
 416                         "/(\w)\.(\w|\*)/u",
 417                         "$1u82e$2",
 418                         $out );
 419
 420                 return $out;
 421         }
 422
 423         /**
 424          * Armor a case-folded UTF-8 string to get through MySQL's
 425          * fulltext search without being mucked up by funny charset
 426          * settings or anything else of the sort.
 427          * @param array $matches
 428          * @return string
 429          */
 430         protected function stripForSearchCallback( $matches ) {
 431                 return 'u8' . bin2hex( $matches[1] );
 432         }
 433
 434         /**
 435          * Check MySQL server's ft_min_word_len setting so we know
 436          * if we need to pad short words...
 437          *
 438          * @return int
 439          */
 440         protected function minSearchLength() {
 441                 if ( is_null( self::$mMinSearchLength ) ) {
 442                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 443
 444                         $dbr = wfGetDB( DB_REPLICA );
 445                         $result = $dbr->query( $sql, __METHOD__ );
 446                         $row = $result->fetchObject();
 447                         $result->free();
 448
 449                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 450                                 self::$mMinSearchLength = intval( $row->Value );
 451                         } else {
 452                                 self::$mMinSearchLength = 0;
 453                         }
 454                 }
 455                 return self::$mMinSearchLength;
 456         }
 457 }