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 'title-suffix-filter':
 216                         return true;
 217                 default:
 218                         return parent::supports( $feature );
 219                 }
 220         }
 221
 222         /**
 223          * Add special conditions
 224          * @param $query Array
 225          * @since 1.18
 226          */
 227         protected function queryFeatures( &$query ) {
 228                 foreach ( $this->features as $feature => $value ) {
 229                         if ( $feature === 'list-redirects' && !$value ) {
 230                                 $query['conds']['page_is_redirect'] = 0;
 231                         } elseif ( $feature === 'title-suffix-filter' && $value ) {
 232                                 $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->anyString(), $value );
 233                         }
 234                 }
 235         }
 236
 237         /**
 238          * Add namespace conditions
 239          * @param $query Array
 240          * @since 1.18 (changed)
 241          */
 242         function queryNamespaces( &$query ) {
 243                 if ( is_array( $this->namespaces ) ) {
 244                         if ( count( $this->namespaces ) === 0 ) {
 245                                 $this->namespaces[] = '0';
 246                         }
 247                         $query['conds']['page_namespace'] = $this->namespaces;
 248                 }
 249         }
 250
 251         /**
 252          * Add limit options
 253          * @param $query Array
 254          * @since 1.18
 255          */
 256         protected function limitResult( &$query ) {
 257                 $query['options']['LIMIT'] = $this->limit;
 258                 $query['options']['OFFSET'] = $this->offset;
 259         }
 260
 261         /**
 262          * Construct the SQL query to do the search.
 263          * The guts shoulds be constructed in queryMain()
 264          * @param $filteredTerm String
 265          * @param $fulltext Boolean
 266          * @return Array
 267          * @since 1.18 (changed)
 268          */
 269         function getQuery( $filteredTerm, $fulltext ) {
 270                 $query = array(
 271                         'tables' => array(),
 272                         'fields' => array(),
 273                         'conds' => array(),
 274                         'options' => array(),
 275                         'joins' => array(),
 276                 );
 277
 278                 $this->queryMain( $query, $filteredTerm, $fulltext );
 279                 $this->queryFeatures( $query );
 280                 $this->queryNamespaces( $query );
 281                 $this->limitResult( $query );
 282
 283                 return $query;
 284         }
 285
 286         /**
 287          * Picks which field to index on, depending on what type of query.
 288          * @param $fulltext Boolean
 289          * @return String
 290          */
 291         function getIndexField( $fulltext ) {
 292                 return $fulltext ? 'si_text' : 'si_title';
 293         }
 294
 295         /**
 296          * Get the base part of the search query.
 297          *
 298          * @param &$query array Search query array
 299          * @param $filteredTerm String
 300          * @param $fulltext Boolean
 301          * @since 1.18 (changed)
 302          */
 303         function queryMain( &$query, $filteredTerm, $fulltext ) {
 304                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 305                 $query['tables'][] = 'page';
 306                 $query['tables'][] = 'searchindex';
 307                 $query['fields'][] = 'page_id';
 308                 $query['fields'][] = 'page_namespace';
 309                 $query['fields'][] = 'page_title';
 310                 $query['conds'][] = 'page_id=si_page';
 311                 $query['conds'][] = $match;
 312         }
 313
 314         /**
 315          * @since 1.18 (changed)
 316          * @return array
 317          */
 318         function getCountQuery( $filteredTerm, $fulltext ) {
 319                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 320
 321                 $query = array(
 322                         'tables' => array( 'page', 'searchindex' ),
 323                         'fields' => array( 'COUNT(*) as c' ),
 324                         'conds' => array( 'page_id=si_page', $match ),
 325                         'options' => array(),
 326                         'joins' => array(),
 327                 );
 328
 329                 $this->queryFeatures( $query );
 330                 $this->queryNamespaces( $query );
 331
 332                 return $query;
 333         }
 334
 335         /**
 336          * Create or update the search index record for the given page.
 337          * Title and text should be pre-processed.
 338          *
 339          * @param $id Integer
 340          * @param $title String
 341          * @param $text String
 342          */
 343         function update( $id, $title, $text ) {
 344                 $dbw = wfGetDB( DB_MASTER );
 345                 $dbw->replace( 'searchindex',
 346                         array( 'si_page' ),
 347                         array(
 348                                 'si_page' => $id,
 349                                 'si_title' => $this->normalizeText( $title ),
 350                                 'si_text' => $this->normalizeText( $text )
 351                         ), __METHOD__ );
 352         }
 353
 354         /**
 355          * Update a search index record's title only.
 356          * Title should be pre-processed.
 357          *
 358          * @param $id Integer
 359          * @param $title String
 360          */
 361         function updateTitle( $id, $title ) {
 362                 $dbw = wfGetDB( DB_MASTER );
 363
 364                 $dbw->update( 'searchindex',
 365                         array( 'si_title' => $this->normalizeText( $title ) ),
 366                         array( 'si_page' => $id ),
 367                         __METHOD__,
 368                         array( $dbw->lowPriorityOption() ) );
 369         }
 370
 371         /**
 372          * Delete an indexed page
 373          * Title should be pre-processed.
 374          *
 375          * @param Integer $id Page id that was deleted
 376          * @param String $title Title of page that was deleted
 377          */
 378         function delete( $id, $title ) {
 379                 $dbw = wfGetDB( DB_MASTER );
 380
 381                 $dbw->delete( 'searchindex', array( 'si_page' => $id ), __METHOD__ );
 382         }
 383
 384         /**
 385          * Converts some characters for MySQL's indexing to grok it correctly,
 386          * and pads short words to overcome limitations.
 387          * @return mixed|string
 388          */
 389         function normalizeText( $string ) {
 390                 global $wgContLang;
 391
 392                 wfProfileIn( __METHOD__ );
 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                         array( $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                 //
 419                 // MySQL's search seems to ignore them, so you'd match on
 420                 // "example.wikipedia.com" and "192.168.83.1" as well.
 421                 $out = preg_replace(
 422                         "/(\w)\.(\w|\*)/u",
 423                         "$1u82e$2",
 424                         $out );
 425
 426                 wfProfileOut( __METHOD__ );
 427
 428                 return $out;
 429         }
 430
 431         /**
 432          * Armor a case-folded UTF-8 string to get through MySQL's
 433          * fulltext search without being mucked up by funny charset
 434          * settings or anything else of the sort.
 435          * @return string
 436          */
 437         protected function stripForSearchCallback( $matches ) {
 438                 return 'u8' . bin2hex( $matches[1] );
 439         }
 440
 441         /**
 442          * Check MySQL server's ft_min_word_len setting so we know
 443          * if we need to pad short words...
 444          *
 445          * @return int
 446          */
 447         protected function minSearchLength() {
 448                 if ( is_null( self::$mMinSearchLength ) ) {
 449                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 450
 451                         $dbr = wfGetDB( DB_SLAVE );
 452                         $result = $dbr->query( $sql );
 453                         $row = $result->fetchObject();
 454                         $result->free();
 455
 456                         if ( $row && $row->Variable_name == 'ft_min_word_len' ) {
 457                                 self::$mMinSearchLength = intval( $row->Value );
 458                         } else {
 459                                 self::$mMinSearchLength = 0;
 460                         }
 461                 }
 462                 return self::$mMinSearchLength;
 463         }
 464 }
 465
 466 /**
 467  * @ingroup Search
 468  */
 469 class MySQLSearchResultSet extends SqlSearchResultSet {
 470         function __construct( $resultSet, $terms, $totalHits = null ) {
 471                 parent::__construct( $resultSet, $terms );
 472                 $this->mTotalHits = $totalHits;
 473         }
 474
 475         function getTotalHits() {
 476                 return $this->mTotalHits;
 477         }
 478 }