includes/search/SearchEngine.php

   1 <?php
   2 /**
   3  * Basic search engine
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Search
  22  */
  23
  24 /**
  25  * @defgroup Search Search
  26  */
  27
  28 /**
  29  * Contain a class for special pages
  30  * @ingroup Search
  31  */
  32 class SearchEngine {
  33         /** @var string */
  34         public $prefix = '';
  35
  36         /** @var int[] */
  37         public $namespaces = array( NS_MAIN );
  38
  39         /** @var int */
  40         protected $limit = 10;
  41
  42         /** @var int */
  43         protected $offset = 0;
  44
  45         /** @var array|string */
  46         protected $searchTerms = array();
  47
  48         /** @var bool */
  49         protected $showSuggestion = true;
  50
  51         /** @var array Feature values */
  52         protected $features = array();
  53
  54         /**
  55          * Perform a full text search query and return a result set.
  56          * If title searches are not supported or disabled, return null.
  57          * STUB
  58          *
  59          * @param string $term Raw search term
  60          * @return SearchResultSet|Status|null
  61          */
  62         function searchText( $term ) {
  63                 return null;
  64         }
  65
  66         /**
  67          * Perform a title-only search query and return a result set.
  68          * If title searches are not supported or disabled, return null.
  69          * STUB
  70          *
  71          * @param string $term Raw search term
  72          * @return SearchResultSet|null
  73          */
  74         function searchTitle( $term ) {
  75                 return null;
  76         }
  77
  78         /**
  79          * @since 1.18
  80          * @param string $feature
  81          * @return bool
  82          */
  83         public function supports( $feature ) {
  84                 switch ( $feature ) {
  85                 case 'search-update':
  86                         return true;
  87                 case 'title-suffix-filter':
  88                 default:
  89                         return false;
  90                 }
  91         }
  92
  93         /**
  94          * Way to pass custom data for engines
  95          * @since 1.18
  96          * @param string $feature
  97          * @param mixed $data
  98          * @return bool
  99          */
 100         public function setFeatureData( $feature, $data ) {
 101                 $this->features[$feature] = $data;
 102         }
 103
 104         /**
 105          * When overridden in derived class, performs database-specific conversions
 106          * on text to be used for searching or updating search index.
 107          * Default implementation does nothing (simply returns $string).
 108          *
 109          * @param string $string String to process
 110          * @return string
 111          */
 112         public function normalizeText( $string ) {
 113                 global $wgContLang;
 114
 115                 // Some languages such as Chinese require word segmentation
 116                 return $wgContLang->segmentByWord( $string );
 117         }
 118
 119         /**
 120          * Transform search term in cases when parts of the query came as different
 121          * GET params (when supported), e.g. for prefix queries:
 122          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 123          */
 124         function transformSearchTerm( $term ) {
 125                 return $term;
 126         }
 127
 128         /**
 129          * If an exact title match can be found, or a very slightly close match,
 130          * return the title. If no match, returns NULL.
 131          *
 132          * @param string $searchterm
 133          * @return Title
 134          */
 135         public static function getNearMatch( $searchterm ) {
 136                 $title = self::getNearMatchInternal( $searchterm );
 137
 138                 wfRunHooks( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
 139                 return $title;
 140         }
 141
 142         /**
 143          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 144          * SearchResultSet.
 145          *
 146          * @param string $searchterm
 147          * @return SearchResultSet
 148          */
 149         public static function getNearMatchResultSet( $searchterm ) {
 150                 return new SearchNearMatchResultSet( self::getNearMatch( $searchterm ) );
 151         }
 152
 153         /**
 154          * Really find the title match.
 155          * @return null|Title
 156          */
 157         private static function getNearMatchInternal( $searchterm ) {
 158                 global $wgContLang, $wgEnableSearchContributorsByIP;
 159
 160                 $allSearchTerms = array( $searchterm );
 161
 162                 if ( $wgContLang->hasVariants() ) {
 163                         $allSearchTerms = array_merge(
 164                                 $allSearchTerms,
 165                                 $wgContLang->autoConvertToAllVariants( $searchterm )
 166                         );
 167                 }
 168
 169                 $titleResult = null;
 170                 if ( !wfRunHooks( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
 171                         return $titleResult;
 172                 }
 173
 174                 foreach ( $allSearchTerms as $term ) {
 175
 176                         # Exact match? No need to look further.
 177                         $title = Title::newFromText( $term );
 178                         if ( is_null( $title ) ) {
 179                                 return null;
 180                         }
 181
 182                         # Try files if searching in the Media: namespace
 183                         if ( $title->getNamespace() == NS_MEDIA ) {
 184                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
 185                         }
 186
 187                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
 188                                 return $title;
 189                         }
 190
 191                         # See if it still otherwise has content is some sane sense
 192                         $page = WikiPage::factory( $title );
 193                         if ( $page->hasViewableContent() ) {
 194                                 return $title;
 195                         }
 196
 197                         if ( !wfRunHooks( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
 198                                 return $title;
 199                         }
 200
 201                         # Now try all lower case (i.e. first letter capitalized)
 202                         $title = Title::newFromText( $wgContLang->lc( $term ) );
 203                         if ( $title && $title->exists() ) {
 204                                 return $title;
 205                         }
 206
 207                         # Now try capitalized string
 208                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
 209                         if ( $title && $title->exists() ) {
 210                                 return $title;
 211                         }
 212
 213                         # Now try all upper case
 214                         $title = Title::newFromText( $wgContLang->uc( $term ) );
 215                         if ( $title && $title->exists() ) {
 216                                 return $title;
 217                         }
 218
 219                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 220                         $title = Title::newFromText( $wgContLang->ucwordbreaks( $term ) );
 221                         if ( $title && $title->exists() ) {
 222                                 return $title;
 223                         }
 224
 225                         // Give hooks a chance at better match variants
 226                         $title = null;
 227                         if ( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 228                                 return $title;
 229                         }
 230                 }
 231
 232                 $title = Title::newFromText( $searchterm );
 233
 234                 # Entering an IP address goes to the contributions page
 235                 if ( $wgEnableSearchContributorsByIP ) {
 236                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 237                                 || User::isIP( trim( $searchterm ) ) ) {
 238                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 239                         }
 240                 }
 241
 242                 # Entering a user goes to the user page whether it's there or not
 243                 if ( $title->getNamespace() == NS_USER ) {
 244                         return $title;
 245                 }
 246
 247                 # Go to images that exist even if there's no local page.
 248                 # There may have been a funny upload, or it may be on a shared
 249                 # file repository such as Wikimedia Commons.
 250                 if ( $title->getNamespace() == NS_FILE ) {
 251                         $image = wfFindFile( $title );
 252                         if ( $image ) {
 253                                 return $title;
 254                         }
 255                 }
 256
 257                 # MediaWiki namespace? Page may be "implied" if not customized.
 258                 # Just return it, with caps forced as the message system likes it.
 259                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 260                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 261                 }
 262
 263                 # Quoted term? Try without the quotes...
 264                 $matches = array();
 265                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 266                         return SearchEngine::getNearMatch( $matches[1] );
 267                 }
 268
 269                 return null;
 270         }
 271
 272         public static function legalSearchChars() {
 273                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 274         }
 275
 276         /**
 277          * Set the maximum number of results to return
 278          * and how many to skip before returning the first.
 279          *
 280          * @param int $limit
 281          * @param int $offset
 282          */
 283         function setLimitOffset( $limit, $offset = 0 ) {
 284                 $this->limit = intval( $limit );
 285                 $this->offset = intval( $offset );
 286         }
 287
 288         /**
 289          * Set which namespaces the search should include.
 290          * Give an array of namespace index numbers.
 291          *
 292          * @param array $namespaces
 293          */
 294         function setNamespaces( $namespaces ) {
 295                 $this->namespaces = $namespaces;
 296         }
 297
 298         /**
 299          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 300          * don't support building a suggestion in the first place and others don't respect
 301          * this flag.
 302          *
 303          * @param bool $showSuggestion Should the searcher try to build suggestions
 304          */
 305         function setShowSuggestion( $showSuggestion ) {
 306                 $this->showSuggestion = $showSuggestion;
 307         }
 308
 309         /**
 310          * Parse some common prefixes: all (search everything)
 311          * or namespace names
 312          *
 313          * @param string $query
 314          * @return string
 315          */
 316         function replacePrefixes( $query ) {
 317                 global $wgContLang;
 318
 319                 $parsed = $query;
 320                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 321                         return $parsed;
 322                 }
 323
 324                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 325                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 326                         $this->namespaces = null;
 327                         $parsed = substr( $query, strlen( $allkeyword ) );
 328                 } elseif ( strpos( $query, ':' ) !== false ) {
 329                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 330                         $index = $wgContLang->getNsIndex( $prefix );
 331                         if ( $index !== false ) {
 332                                 $this->namespaces = array( $index );
 333                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 334                         }
 335                 }
 336                 if ( trim( $parsed ) == '' ) {
 337                         $parsed = $query; // prefix was the whole query
 338                 }
 339
 340                 return $parsed;
 341         }
 342
 343         /**
 344          * Make a list of searchable namespaces and their canonical names.
 345          * @return array
 346          */
 347         public static function searchableNamespaces() {
 348                 global $wgContLang;
 349                 $arr = array();
 350                 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
 351                         if ( $ns >= NS_MAIN ) {
 352                                 $arr[$ns] = $name;
 353                         }
 354                 }
 355
 356                 wfRunHooks( 'SearchableNamespaces', array( &$arr ) );
 357                 return $arr;
 358         }
 359
 360         /**
 361          * Extract default namespaces to search from the given user's
 362          * settings, returning a list of index numbers.
 363          *
 364          * @param user $user
 365          * @return array
 366          */
 367         public static function userNamespaces( $user ) {
 368                 $arr = array();
 369                 foreach ( SearchEngine::searchableNamespaces() as $ns => $name ) {
 370                         if ( $user->getOption( 'searchNs' . $ns ) ) {
 371                                 $arr[] = $ns;
 372                         }
 373                 }
 374
 375                 return $arr;
 376         }
 377
 378         /**
 379          * Find snippet highlight settings for all users
 380          *
 381          * @return array Contextlines, contextchars
 382          */
 383         public static function userHighlightPrefs() {
 384                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 385                 $contextchars = 75; // same as above.... :P
 386                 return array( $contextlines, $contextchars );
 387         }
 388
 389         /**
 390          * An array of namespaces indexes to be searched by default
 391          *
 392          * @return array
 393          */
 394         public static function defaultNamespaces() {
 395                 global $wgNamespacesToBeSearchedDefault;
 396
 397                 return array_keys( $wgNamespacesToBeSearchedDefault, true );
 398         }
 399
 400         /**
 401          * Get a list of namespace names useful for showing in tooltips
 402          * and preferences
 403          *
 404          * @param array $namespaces
 405          * @return array
 406          */
 407         public static function namespacesAsText( $namespaces ) {
 408                 global $wgContLang;
 409
 410                 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
 411                 foreach ( $formatted as $key => $ns ) {
 412                         if ( empty( $ns ) ) {
 413                                 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
 414                         }
 415                 }
 416                 return $formatted;
 417         }
 418
 419         /**
 420          * Return a 'cleaned up' search string
 421          *
 422          * @param string $text
 423          * @return string
 424          */
 425         function filter( $text ) {
 426                 $lc = $this->legalSearchChars();
 427                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 428         }
 429
 430         /**
 431          * Load up the appropriate search engine class for the currently
 432          * active database backend, and return a configured instance.
 433          *
 434          * @param string $type Type of search backend, if not the default
 435          * @return SearchEngine
 436          */
 437         public static function create( $type = null ) {
 438                 global $wgSearchType;
 439                 $dbr = null;
 440
 441                 $alternatives = self::getSearchTypes();
 442
 443                 if ( $type && in_array( $type, $alternatives ) ) {
 444                         $class = $type;
 445                 } elseif ( $wgSearchType !== null ) {
 446                         $class = $wgSearchType;
 447                 } else {
 448                         $dbr = wfGetDB( DB_SLAVE );
 449                         $class = $dbr->getSearchEngine();
 450                 }
 451
 452                 $search = new $class( $dbr );
 453                 return $search;
 454         }
 455
 456         /**
 457          * Return the search engines we support. If only $wgSearchType
 458          * is set, it'll be an array of just that one item.
 459          *
 460          * @return array
 461          */
 462         public static function getSearchTypes() {
 463                 global $wgSearchType, $wgSearchTypeAlternatives;
 464
 465                 $alternatives = $wgSearchTypeAlternatives ?: array();
 466                 array_unshift( $alternatives, $wgSearchType );
 467
 468                 return $alternatives;
 469         }
 470
 471         /**
 472          * Create or update the search index record for the given page.
 473          * Title and text should be pre-processed.
 474          * STUB
 475          *
 476          * @param int $id
 477          * @param string $title
 478          * @param string $text
 479          */
 480         function update( $id, $title, $text ) {
 481                 // no-op
 482         }
 483
 484         /**
 485          * Update a search index record's title only.
 486          * Title should be pre-processed.
 487          * STUB
 488          *
 489          * @param int $id
 490          * @param string $title
 491          */
 492         function updateTitle( $id, $title ) {
 493                 // no-op
 494         }
 495
 496         /**
 497          * Delete an indexed page
 498          * Title should be pre-processed.
 499          * STUB
 500          *
 501          * @param int $id Page id that was deleted
 502          * @param string $title Title of page that was deleted
 503          */
 504         function delete( $id, $title ) {
 505                 // no-op
 506         }
 507
 508         /**
 509          * Get OpenSearch suggestion template
 510          *
 511          * @return string
 512          */
 513         public static function getOpenSearchTemplate() {
 514                 global $wgOpenSearchTemplate, $wgCanonicalServer;
 515
 516                 if ( $wgOpenSearchTemplate ) {
 517                         return $wgOpenSearchTemplate;
 518                 } else {
 519                         $ns = implode( '|', SearchEngine::defaultNamespaces() );
 520                         if ( !$ns ) {
 521                                 $ns = "0";
 522                         }
 523
 524                         return $wgCanonicalServer . wfScript( 'api' )
 525                                 . '?action=opensearch&search={searchTerms}&namespace=' . $ns;
 526                 }
 527         }
 528
 529         /**
 530          * Get the raw text for updating the index from a content object
 531          * Nicer search backends could possibly do something cooler than
 532          * just returning raw text
 533          *
 534          * @todo This isn't ideal, we'd really like to have content-specific handling here
 535          * @param Title $t Title we're indexing
 536          * @param Content $c Content of the page to index
 537          * @return string
 538          */
 539         public function getTextFromContent( Title $t, Content $c = null ) {
 540                 return $c ? $c->getTextForSearchIndex() : '';
 541         }
 542
 543         /**
 544          * If an implementation of SearchEngine handles all of its own text processing
 545          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 546          * rather silly handling, it should return true here instead.
 547          *
 548          * @return bool
 549          */
 550         public function textAlreadyUpdatedForIndex() {
 551                 return false;
 552         }
 553 }
 554
 555 /**
 556  * Dummy class to be used when non-supported Database engine is present.
 557  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 558  * such as a LIKE search through titles.
 559  * @ingroup Search
 560  */
 561 class SearchEngineDummy extends SearchEngine {
 562         // no-op
 563 }