includes/search/SearchMySQL.php

   1 <?php
   2 /**
   3  * MySQL search engine
   4  *
   5  * Copyright (C) 2004 Brion Vibber <brion@pobox.com>
   6  * http://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 SearchEngine {
  32         var $strictMatching = true;
  33         static $mMinSearchLength;
  34
  35         /**
  36          * Creates an instance of this class
  37          * @param $db DatabaseMysql: database object
  38          */
  39         function __construct( $db ) {
  40                 parent::__construct( $db );
  41         }
  42
  43         /**
  44          * Parse the user's query and transform it into an SQL fragment which will
  45          * become part of a WHERE clause
  46          *
  47          * @param $filteredText string
  48          * @param $fulltext string
  49          *
  50          * @return string
  51          */
  52         function parseQuery( $filteredText, $fulltext ) {
  53                 global $wgContLang;
  54                 $lc = SearchEngine::legalSearchChars(); // Minus format chars
  55                 $searchon = '';
  56                 $this->searchTerms = array();
  57
  58                 # @todo FIXME: This doesn't handle parenthetical expressions.
  59                 $m = array();
  60                 if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
  61                                 $filteredText, $m, PREG_SET_ORDER ) ) {
  62                         foreach ( $m as $bits ) {
  63                                 @list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits;
  64
  65                                 if ( $nonQuoted != '' ) {
  66                                         $term = $nonQuoted;
  67                                         $quote = '';
  68                                 } else {
  69                                         $term = str_replace( '"', '', $term );
  70                                         $quote = '"';
  71                                 }
  72
  73                                 if ( $searchon !== '' ) {
  74                                         $searchon .= ' ';
  75                                 }
  76                                 if ( $this->strictMatching && ( $modifier == '' ) ) {
  77                                         // If we leave this out, boolean op defaults to OR which is rarely helpful.
  78                                         $modifier = '+';
  79                                 }
  80
  81                                 // Some languages such as Serbian store the input form in the search index,
  82                                 // so we may need to search for matches in multiple writing system variants.
  83                                 $convertedVariants = $wgContLang->autoConvertToAllVariants( $term );
  84                                 if ( is_array( $convertedVariants ) ) {
  85                                         $variants = array_unique( array_values( $convertedVariants ) );
  86                                 } else {
  87                                         $variants = array( $term );
  88                                 }
  89
  90                                 // The low-level search index does some processing on input to work
  91                                 // around problems with minimum lengths and encoding in MySQL's
  92                                 // fulltext engine.
  93                                 // For Chinese this also inserts spaces between adjacent Han characters.
  94                                 $strippedVariants = array_map(
  95                                         array( $wgContLang, 'normalizeForSearch' ),
  96                                         $variants );
  97
  98                                 // Some languages such as Chinese force all variants to a canonical
  99                                 // form when stripping to the low-level search index, so to be sure
 100                                 // let's check our variants list for unique items after stripping.
 101                                 $strippedVariants = array_unique( $strippedVariants );
 102
 103                                 $searchon .= $modifier;
 104                                 if ( count( $strippedVariants ) > 1 ) {
 105                                         $searchon .= '(';
 106                                 }
 107                                 foreach ( $strippedVariants as $stripped ) {
 108                                         $stripped = $this->normalizeText( $stripped );
 109                                         if ( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
 110                                                 // Hack for Chinese: we need to toss in quotes for
 111                                                 // multiple-character phrases since normalizeForSearch()
 112                                                 // added spaces between them to make word breaks.
 113                                                 $stripped = '"' . trim( $stripped ) . '"';
 114                                         }
 115                                         $searchon .= "$quote$stripped$quote$wildcard ";
 116                                 }
 117                                 if ( count( $strippedVariants ) > 1 ) {
 118                                         $searchon .= ')';
 119                                 }
 120
 121                                 // Match individual terms or quoted phrase in result highlighting...
 122                                 // Note that variants will be introduced in a later stage for highlighting!
 123                                 $regexp = $this->regexTerm( $term, $wildcard );
 124                                 $this->searchTerms[] = $regexp;
 125                         }
 126                         wfDebug( __METHOD__ . ": Would search with '$searchon'\n" );
 127                         wfDebug( __METHOD__ . ': Match with /' . implode( '|', $this->searchTerms ) . "/\n" );
 128                 } else {
 129                         wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" );
 130                 }
 131
 132                 $searchon = $this->db->strencode( $searchon );
 133                 $field = $this->getIndexField( $fulltext );
 134                 return " MATCH($field) AGAINST('$searchon' IN BOOLEAN MODE) ";
 135         }
 136
 137         function regexTerm( $string, $wildcard ) {
 138                 global $wgContLang;
 139
 140                 $regex = preg_quote( $string, '/' );
 141                 if ( $wgContLang->hasWordBreaks() ) {
 142                         if ( $wildcard ) {
 143                                 // Don't cut off the final bit!
 144                                 $regex = "\b$regex";
 145                         } else {
 146                                 $regex = "\b$regex\b";
 147                         }
 148                 } else {
 149                         // For Chinese, words may legitimately abut other words in the text literal.
 150                         // Don't add \b boundary checks... note this could cause false positives
 151                         // for latin chars.
 152                 }
 153                 return $regex;
 154         }
 155
 156         public static function legalSearchChars() {
 157                 return "\"*" . parent::legalSearchChars();
 158         }
 159
 160         /**
 161          * Perform a full text search query and return a result set.
 162          *
 163          * @param string $term raw search term
 164          * @return MySQLSearchResultSet
 165          */
 166         function searchText( $term ) {
 167                 return $this->searchInternal( $term, true );
 168         }
 169
 170         /**
 171          * Perform a title-only search query and return a result set.
 172          *
 173          * @param string $term raw search term
 174          * @return MySQLSearchResultSet
 175          */
 176         function searchTitle( $term ) {
 177                 return $this->searchInternal( $term, false );
 178         }
 179
 180         protected function searchInternal( $term, $fulltext ) {
 181                 global $wgCountTotalSearchHits;
 182
 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                 $resultSet = $this->db->select(
 191                         $query['tables'], $query['fields'], $query['conds'],
 192                         __METHOD__, $query['options'], $query['joins']
 193                 );
 194
 195                 $total = null;
 196                 if ( $wgCountTotalSearchHits ) {
 197                         $query = $this->getCountQuery( $filteredTerm, $fulltext );
 198                         $totalResult = $this->db->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
 210                 return new MySQLSearchResultSet( $resultSet, $this->searchTerms, $total );
 211         }
 212
 213         public function supports( $feature ) {
 214                 switch ( $feature ) {
 215                 case 'list-redirects':
 216                 case 'title-suffix-filter':
 217                         return true;
 218                 default:
 219                         return false;
 220                 }
 221         }
 222
 223         /**
 224          * Add special conditions
 225          * @param $query Array
 226          * @since 1.18
 227          */
 228         protected function queryFeatures( &$query ) {
 229                 foreach ( $this->features as $feature => $value ) {
 230                         if ( $feature === 'list-redirects' && !$value ) {
 231                                 $query['conds']['page_is_redirect'] = 0;
 232                         } elseif ( $feature === 'title-suffix-filter' && $value ) {
 233                                 $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->anyString(), $value );
 234                         }
 235                 }
 236         }
 237
 238         /**
 239          * Add namespace conditions
 240          * @param $query Array
 241          * @since 1.18 (changed)
 242          */
 243         function queryNamespaces( &$query ) {
 244                 if ( is_array( $this->namespaces ) ) {
 245                         if ( count( $this->namespaces ) === 0 ) {
 246                                 $this->namespaces[] = '0';
 247                         }
 248                         $query['conds']['page_namespace'] = $this->namespaces;
 249                 }
 250         }
 251
 252         /**
 253          * Add limit options
 254          * @param $query Array
 255          * @since 1.18
 256          */
 257         protected function limitResult( &$query ) {
 258                 $query['options']['LIMIT'] = $this->limit;
 259                 $query['options']['OFFSET'] = $this->offset;
 260         }
 261
 262         /**
 263          * Construct the SQL query to do the search.
 264          * The guts shoulds be constructed in queryMain()
 265          * @param $filteredTerm String
 266          * @param $fulltext Boolean
 267          * @return Array
 268          * @since 1.18 (changed)
 269          */
 270         function getQuery( $filteredTerm, $fulltext ) {
 271                 $query = array(
 272                         'tables' => array(),
 273                         'fields' => array(),
 274                         'conds' => array(),
 275                         'options' => array(),
 276                         'joins' => array(),
 277                 );
 278
 279                 $this->queryMain( $query, $filteredTerm, $fulltext );
 280                 $this->queryFeatures( $query );
 281                 $this->queryNamespaces( $query );
 282                 $this->limitResult( $query );
 283
 284                 return $query;
 285         }
 286
 287         /**
 288          * Picks which field to index on, depending on what type of query.
 289          * @param $fulltext Boolean
 290          * @return String
 291          */
 292         function getIndexField( $fulltext ) {
 293                 return $fulltext ? 'si_text' : 'si_title';
 294         }
 295
 296         /**
 297          * Get the base part of the search query.
 298          *
 299          * @param &$query array Search query array
 300          * @param $filteredTerm String
 301          * @param $fulltext Boolean
 302          * @since 1.18 (changed)
 303          */
 304         function queryMain( &$query, $filteredTerm, $fulltext ) {
 305                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 306                 $query['tables'][] = 'page';
 307                 $query['tables'][] = 'searchindex';
 308                 $query['fields'][] = 'page_id';
 309                 $query['fields'][] = 'page_namespace';
 310                 $query['fields'][] = 'page_title';
 311                 $query['conds'][] = 'page_id=si_page';
 312                 $query['conds'][] = $match;
 313         }
 314
 315         /**
 316          * @since 1.18 (changed)
 317          * @return array
 318          */
 319         function getCountQuery( $filteredTerm, $fulltext ) {
 320                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 321
 322                 $query = array(
 323                         'tables' => array( 'page', 'searchindex' ),
 324                         'fields' => array( 'COUNT(*) as c' ),
 325                         'conds' => array( 'page_id=si_page', $match ),
 326                         'options' => array(),
 327                         'joins' => array(),
 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 $id Integer
 341          * @param $title String
 342          * @param $text String
 343          */
 344         function update( $id, $title, $text ) {
 345                 $dbw = wfGetDB( DB_MASTER );
 346                 $dbw->replace( 'searchindex',
 347                         array( 'si_page' ),
 348                         array(
 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 $id Integer
 360          * @param $title String
 361          */
 362         function updateTitle( $id, $title ) {
 363                 $dbw = wfGetDB( DB_MASTER );
 364
 365                 $dbw->update( 'searchindex',
 366                         array( 'si_title' => $this->normalizeText( $title ) ),
 367                         array( 'si_page' => $id ),
 368                         __METHOD__,
 369                         array( $dbw->lowPriorityOption() ) );
 370         }
 371
 372         /**
 373          * Delete an indexed page
 374          * Title should be pre-processed.
 375          *
 376          * @param Integer $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', array( '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          * @return mixed|string
 389          */
 390         function normalizeText( $string ) {
 391                 global $wgContLang;
 392
 393                 wfProfileIn( __METHOD__ );
 394
 395                 $out = parent::normalizeText( $string );
 396
 397                 // MySQL fulltext index doesn't grok utf-8, so we
 398                 // need to fold cases and convert to hex
 399                 $out = preg_replace_callback(
 400                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 401                         array( $this, 'stripForSearchCallback' ),
 402                         $wgContLang->lc( $out ) );
 403
 404                 // And to add insult to injury, the default indexing
 405                 // ignores short words... Pad them so we can pass them
 406                 // through without reconfiguring the server...
 407                 $minLength = $this->minSearchLength();
 408                 if ( $minLength > 1 ) {
 409                         $n = $minLength - 1;
 410                         $out = preg_replace(
 411                                 "/\b(\w{1,$n})\b/",
 412                                 "$1u800",
 413                                 $out );
 414                 }
 415
 416                 // Periods within things like hostnames and IP addresses
 417                 // are also important -- we want a search for "example.com"
 418                 // or "192.168.1.1" to work sanely.
 419                 //
 420                 // MySQL's search seems to ignore them, so you'd match on
 421                 // "example.wikipedia.com" and "192.168.83.1" as well.
 422                 $out = preg_replace(
 423                         "/(\w)\.(\w|\*)/u",
 424                         "$1u82e$2",
 425                         $out );
 426
 427                 wfProfileOut( __METHOD__ );
 428
 429                 return $out;
 430         }
 431
 432         /**
 433          * Armor a case-folded UTF-8 string to get through MySQL's
 434          * fulltext search without being mucked up by funny charset
 435          * settings or anything else of the sort.
 436          * @return string
 437          */
 438         protected function stripForSearchCallback( $matches ) {
 439                 return 'u8' . bin2hex( $matches[1] );
 440         }
 441
 442         /**
 443          * Check MySQL server's ft_min_word_len setting so we know
 444          * if we need to pad short words...
 445          *
 446          * @return int
 447          */
 448         protected function minSearchLength() {
 449                 if ( is_null( self::$mMinSearchLength ) ) {
 450                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 451
 452                         $dbr = wfGetDB( DB_SLAVE );
 453                         $result = $dbr->query( $sql );
 454                         $row = $result->fetchObject();
 455                         $result->free();
 456
 457                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 458                                 self::$mMinSearchLength = intval( $row->Value );
 459                         } else {
 460                                 self::$mMinSearchLength = 0;
 461                         }
 462                 }
 463                 return self::$mMinSearchLength;
 464         }
 465 }
 466
 467 /**
 468  * @ingroup Search
 469  */
 470 class MySQLSearchResultSet extends SqlSearchResultSet {
 471         function __construct( $resultSet, $terms, $totalHits = null ) {
 472                 parent::__construct( $resultSet, $terms );
 473                 $this->mTotalHits = $totalHits;
 474         }
 475
 476         function getTotalHits() {
 477                 return $this->mTotalHits;
 478         }
 479 }