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 an SQL fragment which will
  38          * become part of a WHERE clause
  39          *
  40          * @param string $filteredText
  41          * @param string $fulltext
  42          *
  43          * @return string
  44          */
  45         function parseQuery( $filteredText, $fulltext ) {
  46                 global $wgContLang;
  47
  48                 $lc = $this->legalSearchChars(); // Minus format chars
  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                                 MediaWiki\suppressWarnings();
  58                                 list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits;
  59                                 MediaWiki\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( $fulltext );
 130                 return " MATCH($field) AGAINST($searchon IN BOOLEAN MODE) ";
 131         }
 132
 133         function regexTerm( $string, $wildcard ) {
 134                 global $wgContLang;
 135
 136                 $regex = preg_quote( $string, '/' );
 137                 if ( $wgContLang->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 static function legalSearchChars() {
 153                 return "\"*" . parent::legalSearchChars();
 154         }
 155
 156         /**
 157          * Perform a full text search query and return a result set.
 158          *
 159          * @param string $term Raw search term
 160          * @return SqlSearchResultSet
 161          */
 162         function searchText( $term ) {
 163                 return $this->searchInternal( $term, true );
 164         }
 165
 166         /**
 167          * Perform a title-only search query and return a result set.
 168          *
 169          * @param string $term Raw search term
 170          * @return SqlSearchResultSet
 171          */
 172         function searchTitle( $term ) {
 173                 return $this->searchInternal( $term, false );
 174         }
 175
 176         protected function searchInternal( $term, $fulltext ) {
 177                 // This seems out of place, why is this called with empty term?
 178                 if ( trim( $term ) === '' ) {
 179                         return null;
 180                 }
 181
 182                 $filteredTerm = $this->filter( $term );
 183                 $query = $this->getQuery( $filteredTerm, $fulltext );
 184                 $resultSet = $this->db->select(
 185                         $query['tables'], $query['fields'], $query['conds'],
 186                         __METHOD__, $query['options'], $query['joins']
 187                 );
 188
 189                 $total = null;
 190                 $query = $this->getCountQuery( $filteredTerm, $fulltext );
 191                 $totalResult = $this->db->select(
 192                         $query['tables'], $query['fields'], $query['conds'],
 193                         __METHOD__, $query['options'], $query['joins']
 194                 );
 195
 196                 $row = $totalResult->fetchObject();
 197                 if ( $row ) {
 198                         $total = intval( $row->c );
 199                 }
 200                 $totalResult->free();
 201
 202                 return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total );
 203         }
 204
 205         public function supports( $feature ) {
 206                 switch ( $feature ) {
 207                 case 'title-suffix-filter':
 208                         return true;
 209                 default:
 210                         return parent::supports( $feature );
 211                 }
 212         }
 213
 214         /**
 215          * Add special conditions
 216          * @param array $query
 217          * @since 1.18
 218          */
 219         protected function queryFeatures( &$query ) {
 220                 foreach ( $this->features as $feature => $value ) {
 221                         if ( $feature === 'title-suffix-filter' && $value ) {
 222                                 $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->anyString(), $value );
 223                         }
 224                 }
 225         }
 226
 227         /**
 228          * Add namespace conditions
 229          * @param array $query
 230          * @since 1.18 (changed)
 231          */
 232         function queryNamespaces( &$query ) {
 233                 if ( is_array( $this->namespaces ) ) {
 234                         if ( count( $this->namespaces ) === 0 ) {
 235                                 $this->namespaces[] = '0';
 236                         }
 237                         $query['conds']['page_namespace'] = $this->namespaces;
 238                 }
 239         }
 240
 241         /**
 242          * Add limit options
 243          * @param array $query
 244          * @since 1.18
 245          */
 246         protected function limitResult( &$query ) {
 247                 $query['options']['LIMIT'] = $this->limit;
 248                 $query['options']['OFFSET'] = $this->offset;
 249         }
 250
 251         /**
 252          * Construct the SQL query to do the search.
 253          * The guts shoulds be constructed in queryMain()
 254          * @param string $filteredTerm
 255          * @param bool $fulltext
 256          * @return array
 257          * @since 1.18 (changed)
 258          */
 259         function getQuery( $filteredTerm, $fulltext ) {
 260                 $query = [
 261                         'tables' => [],
 262                         'fields' => [],
 263                         'conds' => [],
 264                         'options' => [],
 265                         'joins' => [],
 266                 ];
 267
 268                 $this->queryMain( $query, $filteredTerm, $fulltext );
 269                 $this->queryFeatures( $query );
 270                 $this->queryNamespaces( $query );
 271                 $this->limitResult( $query );
 272
 273                 return $query;
 274         }
 275
 276         /**
 277          * Picks which field to index on, depending on what type of query.
 278          * @param bool $fulltext
 279          * @return string
 280          */
 281         function getIndexField( $fulltext ) {
 282                 return $fulltext ? 'si_text' : 'si_title';
 283         }
 284
 285         /**
 286          * Get the base part of the search query.
 287          *
 288          * @param array &$query Search query array
 289          * @param string $filteredTerm
 290          * @param bool $fulltext
 291          * @since 1.18 (changed)
 292          */
 293         function queryMain( &$query, $filteredTerm, $fulltext ) {
 294                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 295                 $query['tables'][] = 'page';
 296                 $query['tables'][] = 'searchindex';
 297                 $query['fields'][] = 'page_id';
 298                 $query['fields'][] = 'page_namespace';
 299                 $query['fields'][] = 'page_title';
 300                 $query['conds'][] = 'page_id=si_page';
 301                 $query['conds'][] = $match;
 302         }
 303
 304         /**
 305          * @since 1.18 (changed)
 306          * @param string $filteredTerm
 307          * @param bool $fulltext
 308          * @return array
 309          */
 310         function getCountQuery( $filteredTerm, $fulltext ) {
 311                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 312
 313                 $query = [
 314                         'tables' => [ 'page', 'searchindex' ],
 315                         'fields' => [ 'COUNT(*) as c' ],
 316                         'conds' => [ 'page_id=si_page', $match ],
 317                         'options' => [],
 318                         'joins' => [],
 319                 ];
 320
 321                 $this->queryFeatures( $query );
 322                 $this->queryNamespaces( $query );
 323
 324                 return $query;
 325         }
 326
 327         /**
 328          * Create or update the search index record for the given page.
 329          * Title and text should be pre-processed.
 330          *
 331          * @param int $id
 332          * @param string $title
 333          * @param string $text
 334          */
 335         function update( $id, $title, $text ) {
 336                 $dbw = wfGetDB( DB_MASTER );
 337                 $dbw->replace( 'searchindex',
 338                         [ 'si_page' ],
 339                         [
 340                                 'si_page' => $id,
 341                                 'si_title' => $this->normalizeText( $title ),
 342                                 'si_text' => $this->normalizeText( $text )
 343                         ], __METHOD__ );
 344         }
 345
 346         /**
 347          * Update a search index record's title only.
 348          * Title should be pre-processed.
 349          *
 350          * @param int $id
 351          * @param string $title
 352          */
 353         function updateTitle( $id, $title ) {
 354                 $dbw = wfGetDB( DB_MASTER );
 355
 356                 $dbw->update( 'searchindex',
 357                         [ 'si_title' => $this->normalizeText( $title ) ],
 358                         [ 'si_page' => $id ],
 359                         __METHOD__,
 360                         [ $dbw->lowPriorityOption() ] );
 361         }
 362
 363         /**
 364          * Delete an indexed page
 365          * Title should be pre-processed.
 366          *
 367          * @param int $id Page id that was deleted
 368          * @param string $title Title of page that was deleted
 369          */
 370         function delete( $id, $title ) {
 371                 $dbw = wfGetDB( DB_MASTER );
 372
 373                 $dbw->delete( 'searchindex', [ 'si_page' => $id ], __METHOD__ );
 374         }
 375
 376         /**
 377          * Converts some characters for MySQL's indexing to grok it correctly,
 378          * and pads short words to overcome limitations.
 379          * @param string $string
 380          * @return mixed|string
 381          */
 382         function normalizeText( $string ) {
 383                 global $wgContLang;
 384
 385                 $out = parent::normalizeText( $string );
 386
 387                 // MySQL fulltext index doesn't grok utf-8, so we
 388                 // need to fold cases and convert to hex
 389                 $out = preg_replace_callback(
 390                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 391                         [ $this, 'stripForSearchCallback' ],
 392                         $wgContLang->lc( $out ) );
 393
 394                 // And to add insult to injury, the default indexing
 395                 // ignores short words... Pad them so we can pass them
 396                 // through without reconfiguring the server...
 397                 $minLength = $this->minSearchLength();
 398                 if ( $minLength > 1 ) {
 399                         $n = $minLength - 1;
 400                         $out = preg_replace(
 401                                 "/\b(\w{1,$n})\b/",
 402                                 "$1u800",
 403                                 $out );
 404                 }
 405
 406                 // Periods within things like hostnames and IP addresses
 407                 // are also important -- we want a search for "example.com"
 408                 // or "192.168.1.1" to work sanely.
 409                 // MySQL's search seems to ignore them, so you'd match on
 410                 // "example.wikipedia.com" and "192.168.83.1" as well.
 411                 $out = preg_replace(
 412                         "/(\w)\.(\w|\*)/u",
 413                         "$1u82e$2",
 414                         $out );
 415
 416                 return $out;
 417         }
 418
 419         /**
 420          * Armor a case-folded UTF-8 string to get through MySQL's
 421          * fulltext search without being mucked up by funny charset
 422          * settings or anything else of the sort.
 423          * @param array $matches
 424          * @return string
 425          */
 426         protected function stripForSearchCallback( $matches ) {
 427                 return 'u8' . bin2hex( $matches[1] );
 428         }
 429
 430         /**
 431          * Check MySQL server's ft_min_word_len setting so we know
 432          * if we need to pad short words...
 433          *
 434          * @return int
 435          */
 436         protected function minSearchLength() {
 437                 if ( is_null( self::$mMinSearchLength ) ) {
 438                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 439
 440                         $dbr = wfGetDB( DB_REPLICA );
 441                         $result = $dbr->query( $sql, __METHOD__ );
 442                         $row = $result->fetchObject();
 443                         $result->free();
 444
 445                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 446                                 self::$mMinSearchLength = intval( $row->Value );
 447                         } else {
 448                                 self::$mMinSearchLength = 0;
 449                         }
 450                 }
 451                 return self::$mMinSearchLength;
 452         }
 453 }