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 !== '' ) $searchon .= ' ';
  74                                 if( $this->strictMatching && ($modifier == '') ) {
  75                                         // If we leave this out, boolean op defaults to OR which is rarely helpful.
  76                                         $modifier = '+';
  77                                 }
  78
  79                                 // Some languages such as Serbian store the input form in the search index,
  80                                 // so we may need to search for matches in multiple writing system variants.
  81                                 $convertedVariants = $wgContLang->autoConvertToAllVariants( $term );
  82                                 if( is_array( $convertedVariants ) ) {
  83                                         $variants = array_unique( array_values( $convertedVariants ) );
  84                                 } else {
  85                                         $variants = array( $term );
  86                                 }
  87
  88                                 // The low-level search index does some processing on input to work
  89                                 // around problems with minimum lengths and encoding in MySQL's
  90                                 // fulltext engine.
  91                                 // For Chinese this also inserts spaces between adjacent Han characters.
  92                                 $strippedVariants = array_map(
  93                                         array( $wgContLang, 'normalizeForSearch' ),
  94                                         $variants );
  95
  96                                 // Some languages such as Chinese force all variants to a canonical
  97                                 // form when stripping to the low-level search index, so to be sure
  98                                 // let's check our variants list for unique items after stripping.
  99                                 $strippedVariants = array_unique( $strippedVariants );
 100
 101                                 $searchon .= $modifier;
 102                                 if( count( $strippedVariants) > 1 )
 103                                         $searchon .= '(';
 104                                 foreach( $strippedVariants as $stripped ) {
 105                                         $stripped = $this->normalizeText( $stripped );
 106                                         if( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
 107                                                 // Hack for Chinese: we need to toss in quotes for
 108                                                 // multiple-character phrases since normalizeForSearch()
 109                                                 // added spaces between them to make word breaks.
 110                                                 $stripped = '"' . trim( $stripped ) . '"';
 111                                         }
 112                                         $searchon .= "$quote$stripped$quote$wildcard ";
 113                                 }
 114                                 if( count( $strippedVariants) > 1 )
 115                                         $searchon .= ')';
 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->strencode( $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 $term String: raw search term
 160          * @return MySQLSearchResultSet
 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 $term String: raw search term
 170          * @return MySQLSearchResultSet
 171          */
 172         function searchTitle( $term ) {
 173                 return $this->searchInternal( $term, false );
 174         }
 175
 176         protected function searchInternal( $term, $fulltext ) {
 177                 global $wgCountTotalSearchHits;
 178
 179                 // This seems out of place, why is this called with empty term?
 180                 if ( trim( $term ) === '' ) return null;
 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                 if( $wgCountTotalSearchHits ) {
 191                         $query = $this->getCountQuery( $filteredTerm, $fulltext );
 192                         $totalResult = $this->db->select(
 193                                 $query['tables'], $query['fields'], $query['conds'],
 194                                 __METHOD__, $query['options'], $query['joins']
 195                         );
 196
 197                         $row = $totalResult->fetchObject();
 198                         if( $row ) {
 199                                 $total = intval( $row->c );
 200                         }
 201                         $totalResult->free();
 202                 }
 203
 204                 return new MySQLSearchResultSet( $resultSet, $this->searchTerms, $total );
 205         }
 206
 207         public function supports( $feature ) {
 208                 switch( $feature ) {
 209                 case 'list-redirects':
 210                 case 'title-suffix-filter':
 211                         return true;
 212                 default:
 213                         return false;
 214                 }
 215         }
 216
 217         /**
 218          * Add special conditions
 219          * @param $query Array
 220          * @since 1.18
 221          */
 222         protected function queryFeatures( &$query ) {
 223                 foreach ( $this->features as $feature => $value ) {
 224                         if ( $feature ===  'list-redirects' && !$value ) {
 225                                 $query['conds']['page_is_redirect'] = 0;
 226                         } elseif( $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 $query Array
 235          * @since 1.18 (changed)
 236          */
 237         function queryNamespaces( &$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 $query Array
 249          * @since 1.18
 250          */
 251         protected function limitResult( &$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 $filteredTerm String
 260          * @param $fulltext Boolean
 261          * @return Array
 262          * @since 1.18 (changed)
 263          */
 264         function getQuery( $filteredTerm, $fulltext ) {
 265                 $query = array(
 266                         'tables' => array(),
 267                         'fields' => array(),
 268                         'conds' => array(),
 269                         'options' => array(),
 270                         'joins' => array(),
 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 $fulltext Boolean
 284          * @return String
 285          */
 286         function getIndexField( $fulltext ) {
 287                 return $fulltext ? 'si_text' : 'si_title';
 288         }
 289
 290         /**
 291          * Get the base part of the search query.
 292          *
 293          * @param &$query array Search query array
 294          * @param $filteredTerm String
 295          * @param $fulltext Boolean
 296          * @since 1.18 (changed)
 297          */
 298         function queryMain( &$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;
 307         }
 308
 309         /**
 310          * @since 1.18 (changed)
 311          * @return array
 312          */
 313         function getCountQuery( $filteredTerm, $fulltext ) {
 314                 $match = $this->parseQuery( $filteredTerm, $fulltext );
 315
 316                 $query = array(
 317                         'tables' => array( 'page', 'searchindex' ),
 318                         'fields' => array( 'COUNT(*) as c' ),
 319                         'conds' => array( 'page_id=si_page', $match ),
 320                         'options' => array(),
 321                         'joins' => array(),
 322                 );
 323
 324                 $this->queryFeatures( $query );
 325                 $this->queryNamespaces( $query );
 326
 327                 return $query;
 328         }
 329
 330         /**
 331          * Create or update the search index record for the given page.
 332          * Title and text should be pre-processed.
 333          *
 334          * @param $id Integer
 335          * @param $title String
 336          * @param $text String
 337          */
 338         function update( $id, $title, $text ) {
 339                 $dbw = wfGetDB( DB_MASTER );
 340                 $dbw->replace( 'searchindex',
 341                         array( 'si_page' ),
 342                         array(
 343                                 'si_page' => $id,
 344                                 'si_title' => $this->normalizeText( $title ),
 345                                 'si_text' => $this->normalizeText( $text )
 346                         ), __METHOD__ );
 347         }
 348
 349         /**
 350          * Update a search index record's title only.
 351          * Title should be pre-processed.
 352          *
 353          * @param $id Integer
 354          * @param $title String
 355          */
 356         function updateTitle( $id, $title ) {
 357                 $dbw = wfGetDB( DB_MASTER );
 358
 359                 $dbw->update( 'searchindex',
 360                         array( 'si_title' => $this->normalizeText( $title ) ),
 361                         array( 'si_page'  => $id ),
 362                         __METHOD__,
 363                         array( $dbw->lowPriorityOption() ) );
 364         }
 365
 366         /**
 367          * Converts some characters for MySQL's indexing to grok it correctly,
 368          * and pads short words to overcome limitations.
 369          * @return mixed|string
 370          */
 371         function normalizeText( $string ) {
 372                 global $wgContLang;
 373
 374                 wfProfileIn( __METHOD__ );
 375
 376                 $out = parent::normalizeText( $string );
 377
 378                 // MySQL fulltext index doesn't grok utf-8, so we
 379                 // need to fold cases and convert to hex
 380                 $out = preg_replace_callback(
 381                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 382                         array( $this, 'stripForSearchCallback' ),
 383                         $wgContLang->lc( $out ) );
 384
 385                 // And to add insult to injury, the default indexing
 386                 // ignores short words... Pad them so we can pass them
 387                 // through without reconfiguring the server...
 388                 $minLength = $this->minSearchLength();
 389                 if( $minLength > 1 ) {
 390                         $n = $minLength - 1;
 391                         $out = preg_replace(
 392                                 "/\b(\w{1,$n})\b/",
 393                                 "$1u800",
 394                                 $out );
 395                 }
 396
 397                 // Periods within things like hostnames and IP addresses
 398                 // are also important -- we want a search for "example.com"
 399                 // or "192.168.1.1" to work sanely.
 400                 //
 401                 // MySQL's search seems to ignore them, so you'd match on
 402                 // "example.wikipedia.com" and "192.168.83.1" as well.
 403                 $out = preg_replace(
 404                         "/(\w)\.(\w|\*)/u",
 405                         "$1u82e$2",
 406                         $out );
 407
 408                 wfProfileOut( __METHOD__ );
 409
 410                 return $out;
 411         }
 412
 413         /**
 414          * Armor a case-folded UTF-8 string to get through MySQL's
 415          * fulltext search without being mucked up by funny charset
 416          * settings or anything else of the sort.
 417          * @return string
 418          */
 419         protected function stripForSearchCallback( $matches ) {
 420                 return 'u8' . bin2hex( $matches[1] );
 421         }
 422
 423         /**
 424          * Check MySQL server's ft_min_word_len setting so we know
 425          * if we need to pad short words...
 426          *
 427          * @return int
 428          */
 429         protected function minSearchLength() {
 430                 if( is_null( self::$mMinSearchLength ) ) {
 431                         $sql = "SHOW GLOBAL VARIABLES LIKE 'ft\\_min\\_word\\_len'";
 432
 433                         $dbr = wfGetDB( DB_SLAVE );
 434                         $result = $dbr->query( $sql );
 435                         $row = $result->fetchObject();
 436                         $result->free();
 437
 438                         if( $row && $row->Variable_name == 'ft_min_word_len' ) {
 439                                 self::$mMinSearchLength = intval( $row->Value );
 440                         } else {
 441                                 self::$mMinSearchLength = 0;
 442                         }
 443                 }
 444                 return self::$mMinSearchLength;
 445         }
 446 }
 447
 448 /**
 449  * @ingroup Search
 450  */
 451 class MySQLSearchResultSet extends SqlSearchResultSet {
 452         function __construct( $resultSet, $terms, $totalHits=null ) {
 453                 parent::__construct( $resultSet, $terms );
 454                 $this->mTotalHits = $totalHits;
 455         }
 456
 457         function getTotalHits() {
 458                 return $this->mTotalHits;
 459         }
 460 }