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          * @param string $term
 124          * @return string
 125          */
 126         function transformSearchTerm( $term ) {
 127                 return $term;
 128         }
 129
 130         /**
 131          * If an exact title match can be found, or a very slightly close match,
 132          * return the title. If no match, returns NULL.
 133          *
 134          * @param string $searchterm
 135          * @return Title
 136          */
 137         public static function getNearMatch( $searchterm ) {
 138                 $title = self::getNearMatchInternal( $searchterm );
 139
 140                 Hooks::run( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
 141                 return $title;
 142         }
 143
 144         /**
 145          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 146          * SearchResultSet.
 147          *
 148          * @param string $searchterm
 149          * @return SearchResultSet
 150          */
 151         public static function getNearMatchResultSet( $searchterm ) {
 152                 return new SearchNearMatchResultSet( self::getNearMatch( $searchterm ) );
 153         }
 154
 155         /**
 156          * Really find the title match.
 157          * @param string $searchterm
 158          * @return null|Title
 159          */
 160         private static function getNearMatchInternal( $searchterm ) {
 161                 global $wgContLang, $wgEnableSearchContributorsByIP;
 162
 163                 $allSearchTerms = array( $searchterm );
 164
 165                 if ( $wgContLang->hasVariants() ) {
 166                         $allSearchTerms = array_merge(
 167                                 $allSearchTerms,
 168                                 $wgContLang->autoConvertToAllVariants( $searchterm )
 169                         );
 170                 }
 171
 172                 $titleResult = null;
 173                 if ( !Hooks::run( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
 174                         return $titleResult;
 175                 }
 176
 177                 foreach ( $allSearchTerms as $term ) {
 178
 179                         # Exact match? No need to look further.
 180                         $title = Title::newFromText( $term );
 181                         if ( is_null( $title ) ) {
 182                                 return null;
 183                         }
 184
 185                         # Try files if searching in the Media: namespace
 186                         if ( $title->getNamespace() == NS_MEDIA ) {
 187                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
 188                         }
 189
 190                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
 191                                 return $title;
 192                         }
 193
 194                         # See if it still otherwise has content is some sane sense
 195                         $page = WikiPage::factory( $title );
 196                         if ( $page->hasViewableContent() ) {
 197                                 return $title;
 198                         }
 199
 200                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
 201                                 return $title;
 202                         }
 203
 204                         # Now try all lower case (i.e. first letter capitalized)
 205                         $title = Title::newFromText( $wgContLang->lc( $term ) );
 206                         if ( $title && $title->exists() ) {
 207                                 return $title;
 208                         }
 209
 210                         # Now try capitalized string
 211                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
 212                         if ( $title && $title->exists() ) {
 213                                 return $title;
 214                         }
 215
 216                         # Now try all upper case
 217                         $title = Title::newFromText( $wgContLang->uc( $term ) );
 218                         if ( $title && $title->exists() ) {
 219                                 return $title;
 220                         }
 221
 222                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 223                         $title = Title::newFromText( $wgContLang->ucwordbreaks( $term ) );
 224                         if ( $title && $title->exists() ) {
 225                                 return $title;
 226                         }
 227
 228                         // Give hooks a chance at better match variants
 229                         $title = null;
 230                         if ( !Hooks::run( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 231                                 return $title;
 232                         }
 233                 }
 234
 235                 $title = Title::newFromText( $searchterm );
 236
 237                 # Entering an IP address goes to the contributions page
 238                 if ( $wgEnableSearchContributorsByIP ) {
 239                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 240                                 || User::isIP( trim( $searchterm ) ) ) {
 241                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 242                         }
 243                 }
 244
 245                 # Entering a user goes to the user page whether it's there or not
 246                 if ( $title->getNamespace() == NS_USER ) {
 247                         return $title;
 248                 }
 249
 250                 # Go to images that exist even if there's no local page.
 251                 # There may have been a funny upload, or it may be on a shared
 252                 # file repository such as Wikimedia Commons.
 253                 if ( $title->getNamespace() == NS_FILE ) {
 254                         $image = wfFindFile( $title );
 255                         if ( $image ) {
 256                                 return $title;
 257                         }
 258                 }
 259
 260                 # MediaWiki namespace? Page may be "implied" if not customized.
 261                 # Just return it, with caps forced as the message system likes it.
 262                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 263                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 264                 }
 265
 266                 # Quoted term? Try without the quotes...
 267                 $matches = array();
 268                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 269                         return SearchEngine::getNearMatch( $matches[1] );
 270                 }
 271
 272                 return null;
 273         }
 274
 275         public static function legalSearchChars() {
 276                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 277         }
 278
 279         /**
 280          * Set the maximum number of results to return
 281          * and how many to skip before returning the first.
 282          *
 283          * @param int $limit
 284          * @param int $offset
 285          */
 286         function setLimitOffset( $limit, $offset = 0 ) {
 287                 $this->limit = intval( $limit );
 288                 $this->offset = intval( $offset );
 289         }
 290
 291         /**
 292          * Set which namespaces the search should include.
 293          * Give an array of namespace index numbers.
 294          *
 295          * @param array $namespaces
 296          */
 297         function setNamespaces( $namespaces ) {
 298                 $this->namespaces = $namespaces;
 299         }
 300
 301         /**
 302          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 303          * don't support building a suggestion in the first place and others don't respect
 304          * this flag.
 305          *
 306          * @param bool $showSuggestion Should the searcher try to build suggestions
 307          */
 308         function setShowSuggestion( $showSuggestion ) {
 309                 $this->showSuggestion = $showSuggestion;
 310         }
 311
 312         /**
 313          * Parse some common prefixes: all (search everything)
 314          * or namespace names
 315          *
 316          * @param string $query
 317          * @return string
 318          */
 319         function replacePrefixes( $query ) {
 320                 global $wgContLang;
 321
 322                 $parsed = $query;
 323                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 324                         return $parsed;
 325                 }
 326
 327                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 328                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 329                         $this->namespaces = null;
 330                         $parsed = substr( $query, strlen( $allkeyword ) );
 331                 } elseif ( strpos( $query, ':' ) !== false ) {
 332                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 333                         $index = $wgContLang->getNsIndex( $prefix );
 334                         if ( $index !== false ) {
 335                                 $this->namespaces = array( $index );
 336                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 337                         }
 338                 }
 339                 if ( trim( $parsed ) == '' ) {
 340                         $parsed = $query; // prefix was the whole query
 341                 }
 342
 343                 return $parsed;
 344         }
 345
 346         /**
 347          * Make a list of searchable namespaces and their canonical names.
 348          * @return array
 349          */
 350         public static function searchableNamespaces() {
 351                 global $wgContLang;
 352                 $arr = array();
 353                 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
 354                         if ( $ns >= NS_MAIN ) {
 355                                 $arr[$ns] = $name;
 356                         }
 357                 }
 358
 359                 Hooks::run( 'SearchableNamespaces', array( &$arr ) );
 360                 return $arr;
 361         }
 362
 363         /**
 364          * Extract default namespaces to search from the given user's
 365          * settings, returning a list of index numbers.
 366          *
 367          * @param user $user
 368          * @return array
 369          */
 370         public static function userNamespaces( $user ) {
 371                 $arr = array();
 372                 foreach ( SearchEngine::searchableNamespaces() as $ns => $name ) {
 373                         if ( $user->getOption( 'searchNs' . $ns ) ) {
 374                                 $arr[] = $ns;
 375                         }
 376                 }
 377
 378                 return $arr;
 379         }
 380
 381         /**
 382          * Find snippet highlight settings for all users
 383          *
 384          * @return array Contextlines, contextchars
 385          */
 386         public static function userHighlightPrefs() {
 387                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 388                 $contextchars = 75; // same as above.... :P
 389                 return array( $contextlines, $contextchars );
 390         }
 391
 392         /**
 393          * An array of namespaces indexes to be searched by default
 394          *
 395          * @return array
 396          */
 397         public static function defaultNamespaces() {
 398                 global $wgNamespacesToBeSearchedDefault;
 399
 400                 return array_keys( $wgNamespacesToBeSearchedDefault, true );
 401         }
 402
 403         /**
 404          * Get a list of namespace names useful for showing in tooltips
 405          * and preferences
 406          *
 407          * @param array $namespaces
 408          * @return array
 409          */
 410         public static function namespacesAsText( $namespaces ) {
 411                 global $wgContLang;
 412
 413                 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
 414                 foreach ( $formatted as $key => $ns ) {
 415                         if ( empty( $ns ) ) {
 416                                 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
 417                         }
 418                 }
 419                 return $formatted;
 420         }
 421
 422         /**
 423          * Load up the appropriate search engine class for the currently
 424          * active database backend, and return a configured instance.
 425          *
 426          * @param string $type Type of search backend, if not the default
 427          * @return SearchEngine
 428          */
 429         public static function create( $type = null ) {
 430                 global $wgSearchType;
 431                 $dbr = null;
 432
 433                 $alternatives = self::getSearchTypes();
 434
 435                 if ( $type && in_array( $type, $alternatives ) ) {
 436                         $class = $type;
 437                 } elseif ( $wgSearchType !== null ) {
 438                         $class = $wgSearchType;
 439                 } else {
 440                         $dbr = wfGetDB( DB_SLAVE );
 441                         $class = $dbr->getSearchEngine();
 442                 }
 443
 444                 $search = new $class( $dbr );
 445                 return $search;
 446         }
 447
 448         /**
 449          * Return the search engines we support. If only $wgSearchType
 450          * is set, it'll be an array of just that one item.
 451          *
 452          * @return array
 453          */
 454         public static function getSearchTypes() {
 455                 global $wgSearchType, $wgSearchTypeAlternatives;
 456
 457                 $alternatives = $wgSearchTypeAlternatives ?: array();
 458                 array_unshift( $alternatives, $wgSearchType );
 459
 460                 return $alternatives;
 461         }
 462
 463         /**
 464          * Create or update the search index record for the given page.
 465          * Title and text should be pre-processed.
 466          * STUB
 467          *
 468          * @param int $id
 469          * @param string $title
 470          * @param string $text
 471          */
 472         function update( $id, $title, $text ) {
 473                 // no-op
 474         }
 475
 476         /**
 477          * Update a search index record's title only.
 478          * Title should be pre-processed.
 479          * STUB
 480          *
 481          * @param int $id
 482          * @param string $title
 483          */
 484         function updateTitle( $id, $title ) {
 485                 // no-op
 486         }
 487
 488         /**
 489          * Delete an indexed page
 490          * Title should be pre-processed.
 491          * STUB
 492          *
 493          * @param int $id Page id that was deleted
 494          * @param string $title Title of page that was deleted
 495          */
 496         function delete( $id, $title ) {
 497                 // no-op
 498         }
 499
 500         /**
 501          * Get OpenSearch suggestion template
 502          *
 503          * @deprecated since 1.25
 504          * @return string
 505          */
 506         public static function getOpenSearchTemplate() {
 507                 wfDeprecated( __METHOD__, '1.25' );
 508                 return ApiOpenSearch::getOpenSearchTemplate( 'application/x-suggestions+json' );
 509         }
 510
 511         /**
 512          * Get the raw text for updating the index from a content object
 513          * Nicer search backends could possibly do something cooler than
 514          * just returning raw text
 515          *
 516          * @todo This isn't ideal, we'd really like to have content-specific handling here
 517          * @param Title $t Title we're indexing
 518          * @param Content $c Content of the page to index
 519          * @return string
 520          */
 521         public function getTextFromContent( Title $t, Content $c = null ) {
 522                 return $c ? $c->getTextForSearchIndex() : '';
 523         }
 524
 525         /**
 526          * If an implementation of SearchEngine handles all of its own text processing
 527          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 528          * rather silly handling, it should return true here instead.
 529          *
 530          * @return bool
 531          */
 532         public function textAlreadyUpdatedForIndex() {
 533                 return false;
 534         }
 535 }
 536
 537 /**
 538  * Dummy class to be used when non-supported Database engine is present.
 539  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 540  * such as a LIKE search through titles.
 541  * @ingroup Search
 542  */
 543 class SearchEngineDummy extends SearchEngine {
 544         // no-op
 545 }