| Index: trunk/phase3/includes/SearchEngine.php |
| — | — | @@ -1,1154 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -/** |
| 4 | | - * @defgroup Search Search |
| 5 | | - * |
| 6 | | - * @file |
| 7 | | - * @ingroup Search |
| 8 | | - */ |
| 9 | | - |
| 10 | | -/** |
| 11 | | - * Contain a class for special pages |
| 12 | | - * @ingroup Search |
| 13 | | - */ |
| 14 | | -class SearchEngine { |
| 15 | | - var $limit = 10; |
| 16 | | - var $offset = 0; |
| 17 | | - var $searchTerms = array(); |
| 18 | | - var $namespaces = array( NS_MAIN ); |
| 19 | | - var $showRedirects = false; |
| 20 | | - |
| 21 | | - /** |
| 22 | | - * Perform a full text search query and return a result set. |
| 23 | | - * If title searches are not supported or disabled, return null. |
| 24 | | - * |
| 25 | | - * @param string $term - Raw search term |
| 26 | | - * @return SearchResultSet |
| 27 | | - * @access public |
| 28 | | - * @abstract |
| 29 | | - */ |
| 30 | | - function searchText( $term ) { |
| 31 | | - return null; |
| 32 | | - } |
| 33 | | - |
| 34 | | - /** |
| 35 | | - * Perform a title-only search query and return a result set. |
| 36 | | - * If title searches are not supported or disabled, return null. |
| 37 | | - * |
| 38 | | - * @param string $term - Raw search term |
| 39 | | - * @return SearchResultSet |
| 40 | | - * @access public |
| 41 | | - * @abstract |
| 42 | | - */ |
| 43 | | - function searchTitle( $term ) { |
| 44 | | - return null; |
| 45 | | - } |
| 46 | | - |
| 47 | | - /** |
| 48 | | - * If an exact title match can be find, or a very slightly close match, |
| 49 | | - * return the title. If no match, returns NULL. |
| 50 | | - * |
| 51 | | - * @param string $term |
| 52 | | - * @return Title |
| 53 | | - */ |
| 54 | | - public static function getNearMatch( $searchterm ) { |
| 55 | | - global $wgContLang; |
| 56 | | - |
| 57 | | - $allSearchTerms = array($searchterm); |
| 58 | | - |
| 59 | | - if($wgContLang->hasVariants()){ |
| 60 | | - $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm)); |
| 61 | | - } |
| 62 | | - |
| 63 | | - foreach($allSearchTerms as $term){ |
| 64 | | - |
| 65 | | - # Exact match? No need to look further. |
| 66 | | - $title = Title::newFromText( $term ); |
| 67 | | - if (is_null($title)) |
| 68 | | - return NULL; |
| 69 | | - |
| 70 | | - if ( $title->getNamespace() == NS_SPECIAL || $title->isExternal() |
| 71 | | - || $title->exists() ) { |
| 72 | | - return $title; |
| 73 | | - } |
| 74 | | - |
| 75 | | - # Now try all lower case (i.e. first letter capitalized) |
| 76 | | - # |
| 77 | | - $title = Title::newFromText( $wgContLang->lc( $term ) ); |
| 78 | | - if ( $title && $title->exists() ) { |
| 79 | | - return $title; |
| 80 | | - } |
| 81 | | - |
| 82 | | - # Now try capitalized string |
| 83 | | - # |
| 84 | | - $title = Title::newFromText( $wgContLang->ucwords( $term ) ); |
| 85 | | - if ( $title && $title->exists() ) { |
| 86 | | - return $title; |
| 87 | | - } |
| 88 | | - |
| 89 | | - # Now try all upper case |
| 90 | | - # |
| 91 | | - $title = Title::newFromText( $wgContLang->uc( $term ) ); |
| 92 | | - if ( $title && $title->exists() ) { |
| 93 | | - return $title; |
| 94 | | - } |
| 95 | | - |
| 96 | | - # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc |
| 97 | | - $title = Title::newFromText( $wgContLang->ucwordbreaks($term) ); |
| 98 | | - if ( $title && $title->exists() ) { |
| 99 | | - return $title; |
| 100 | | - } |
| 101 | | - |
| 102 | | - global $wgCapitalLinks, $wgContLang; |
| 103 | | - if( !$wgCapitalLinks ) { |
| 104 | | - // Catch differs-by-first-letter-case-only |
| 105 | | - $title = Title::newFromText( $wgContLang->ucfirst( $term ) ); |
| 106 | | - if ( $title && $title->exists() ) { |
| 107 | | - return $title; |
| 108 | | - } |
| 109 | | - $title = Title::newFromText( $wgContLang->lcfirst( $term ) ); |
| 110 | | - if ( $title && $title->exists() ) { |
| 111 | | - return $title; |
| 112 | | - } |
| 113 | | - } |
| 114 | | - |
| 115 | | - // Give hooks a chance at better match variants |
| 116 | | - $title = null; |
| 117 | | - if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) { |
| 118 | | - return $title; |
| 119 | | - } |
| 120 | | - } |
| 121 | | - |
| 122 | | - $title = Title::newFromText( $searchterm ); |
| 123 | | - |
| 124 | | - # Entering an IP address goes to the contributions page |
| 125 | | - if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) ) |
| 126 | | - || User::isIP( trim( $searchterm ) ) ) { |
| 127 | | - return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() ); |
| 128 | | - } |
| 129 | | - |
| 130 | | - |
| 131 | | - # Entering a user goes to the user page whether it's there or not |
| 132 | | - if ( $title->getNamespace() == NS_USER ) { |
| 133 | | - return $title; |
| 134 | | - } |
| 135 | | - |
| 136 | | - # Go to images that exist even if there's no local page. |
| 137 | | - # There may have been a funny upload, or it may be on a shared |
| 138 | | - # file repository such as Wikimedia Commons. |
| 139 | | - if( $title->getNamespace() == NS_IMAGE ) { |
| 140 | | - $image = wfFindFile( $title ); |
| 141 | | - if( $image ) { |
| 142 | | - return $title; |
| 143 | | - } |
| 144 | | - } |
| 145 | | - |
| 146 | | - # MediaWiki namespace? Page may be "implied" if not customized. |
| 147 | | - # Just return it, with caps forced as the message system likes it. |
| 148 | | - if( $title->getNamespace() == NS_MEDIAWIKI ) { |
| 149 | | - return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) ); |
| 150 | | - } |
| 151 | | - |
| 152 | | - # Quoted term? Try without the quotes... |
| 153 | | - $matches = array(); |
| 154 | | - if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) { |
| 155 | | - return SearchEngine::getNearMatch( $matches[1] ); |
| 156 | | - } |
| 157 | | - |
| 158 | | - return NULL; |
| 159 | | - } |
| 160 | | - |
| 161 | | - public static function legalSearchChars() { |
| 162 | | - return "A-Za-z_'0-9\\x80-\\xFF\\-"; |
| 163 | | - } |
| 164 | | - |
| 165 | | - /** |
| 166 | | - * Set the maximum number of results to return |
| 167 | | - * and how many to skip before returning the first. |
| 168 | | - * |
| 169 | | - * @param int $limit |
| 170 | | - * @param int $offset |
| 171 | | - * @access public |
| 172 | | - */ |
| 173 | | - function setLimitOffset( $limit, $offset = 0 ) { |
| 174 | | - $this->limit = intval( $limit ); |
| 175 | | - $this->offset = intval( $offset ); |
| 176 | | - } |
| 177 | | - |
| 178 | | - /** |
| 179 | | - * Set which namespaces the search should include. |
| 180 | | - * Give an array of namespace index numbers. |
| 181 | | - * |
| 182 | | - * @param array $namespaces |
| 183 | | - * @access public |
| 184 | | - */ |
| 185 | | - function setNamespaces( $namespaces ) { |
| 186 | | - $this->namespaces = $namespaces; |
| 187 | | - } |
| 188 | | - |
| 189 | | - /** |
| 190 | | - * Parse some common prefixes: all (search everything) |
| 191 | | - * or namespace names |
| 192 | | - * |
| 193 | | - * @param string $query |
| 194 | | - */ |
| 195 | | - function replacePrefixes( $query ){ |
| 196 | | - global $wgContLang; |
| 197 | | - |
| 198 | | - if( strpos($query,':') === false ) |
| 199 | | - return $query; // nothing to do |
| 200 | | - |
| 201 | | - $parsed = $query; |
| 202 | | - $allkeyword = wfMsgForContent('searchall').":"; |
| 203 | | - if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){ |
| 204 | | - $this->namespaces = null; |
| 205 | | - $parsed = substr($query,strlen($allkeyword)); |
| 206 | | - } else if( strpos($query,':') !== false ) { |
| 207 | | - $prefix = substr($query,0,strpos($query,':')); |
| 208 | | - $index = $wgContLang->getNsIndex($prefix); |
| 209 | | - if($index !== false){ |
| 210 | | - $this->namespaces = array($index); |
| 211 | | - $parsed = substr($query,strlen($prefix)+1); |
| 212 | | - } |
| 213 | | - } |
| 214 | | - if(trim($parsed) == '') |
| 215 | | - return $query; // prefix was the whole query |
| 216 | | - |
| 217 | | - return $parsed; |
| 218 | | - } |
| 219 | | - |
| 220 | | - /** |
| 221 | | - * Make a list of searchable namespaces and their canonical names. |
| 222 | | - * @return array |
| 223 | | - */ |
| 224 | | - public static function searchableNamespaces() { |
| 225 | | - global $wgContLang; |
| 226 | | - $arr = array(); |
| 227 | | - foreach( $wgContLang->getNamespaces() as $ns => $name ) { |
| 228 | | - if( $ns >= NS_MAIN ) { |
| 229 | | - $arr[$ns] = $name; |
| 230 | | - } |
| 231 | | - } |
| 232 | | - return $arr; |
| 233 | | - } |
| 234 | | - |
| 235 | | - /** |
| 236 | | - * Extract default namespaces to search from the given user's |
| 237 | | - * settings, returning a list of index numbers. |
| 238 | | - * |
| 239 | | - * @param User $user |
| 240 | | - * @return array |
| 241 | | - * @static |
| 242 | | - */ |
| 243 | | - public static function userNamespaces( &$user ) { |
| 244 | | - $arr = array(); |
| 245 | | - foreach( SearchEngine::searchableNamespaces() as $ns => $name ) { |
| 246 | | - if( $user->getOption( 'searchNs' . $ns ) ) { |
| 247 | | - $arr[] = $ns; |
| 248 | | - } |
| 249 | | - } |
| 250 | | - return $arr; |
| 251 | | - } |
| 252 | | - |
| 253 | | - /** |
| 254 | | - * Find snippet highlight settings for a given user |
| 255 | | - * |
| 256 | | - * @param User $user |
| 257 | | - * @return array contextlines, contextchars |
| 258 | | - * @static |
| 259 | | - */ |
| 260 | | - public static function userHighlightPrefs( &$user ){ |
| 261 | | - //$contextlines = $user->getOption( 'contextlines', 5 ); |
| 262 | | - //$contextchars = $user->getOption( 'contextchars', 50 ); |
| 263 | | - $contextlines = 2; // Hardcode this. Old defaults sucked. :) |
| 264 | | - $contextchars = 75; // same as above.... :P |
| 265 | | - return array($contextlines, $contextchars); |
| 266 | | - } |
| 267 | | - |
| 268 | | - /** |
| 269 | | - * An array of namespaces indexes to be searched by default |
| 270 | | - * |
| 271 | | - * @return array |
| 272 | | - * @static |
| 273 | | - */ |
| 274 | | - public static function defaultNamespaces(){ |
| 275 | | - global $wgNamespacesToBeSearchedDefault; |
| 276 | | - |
| 277 | | - return array_keys($wgNamespacesToBeSearchedDefault, true); |
| 278 | | - } |
| 279 | | - |
| 280 | | - /** |
| 281 | | - * Return a 'cleaned up' search string |
| 282 | | - * |
| 283 | | - * @return string |
| 284 | | - * @access public |
| 285 | | - */ |
| 286 | | - function filter( $text ) { |
| 287 | | - $lc = $this->legalSearchChars(); |
| 288 | | - return trim( preg_replace( "/[^{$lc}]/", " ", $text ) ); |
| 289 | | - } |
| 290 | | - /** |
| 291 | | - * Load up the appropriate search engine class for the currently |
| 292 | | - * active database backend, and return a configured instance. |
| 293 | | - * |
| 294 | | - * @return SearchEngine |
| 295 | | - */ |
| 296 | | - public static function create() { |
| 297 | | - global $wgDBtype, $wgSearchType; |
| 298 | | - if( $wgSearchType ) { |
| 299 | | - $class = $wgSearchType; |
| 300 | | - } elseif( $wgDBtype == 'mysql' ) { |
| 301 | | - $class = 'SearchMySQL'; |
| 302 | | - } else if ( $wgDBtype == 'postgres' ) { |
| 303 | | - $class = 'SearchPostgres'; |
| 304 | | - } else if ( $wgDBtype == 'oracle' ) { |
| 305 | | - $class = 'SearchOracle'; |
| 306 | | - } else { |
| 307 | | - $class = 'SearchEngineDummy'; |
| 308 | | - } |
| 309 | | - $search = new $class( wfGetDB( DB_SLAVE ) ); |
| 310 | | - $search->setLimitOffset(0,0); |
| 311 | | - return $search; |
| 312 | | - } |
| 313 | | - |
| 314 | | - /** |
| 315 | | - * Create or update the search index record for the given page. |
| 316 | | - * Title and text should be pre-processed. |
| 317 | | - * |
| 318 | | - * @param int $id |
| 319 | | - * @param string $title |
| 320 | | - * @param string $text |
| 321 | | - * @abstract |
| 322 | | - */ |
| 323 | | - function update( $id, $title, $text ) { |
| 324 | | - // no-op |
| 325 | | - } |
| 326 | | - |
| 327 | | - /** |
| 328 | | - * Update a search index record's title only. |
| 329 | | - * Title should be pre-processed. |
| 330 | | - * |
| 331 | | - * @param int $id |
| 332 | | - * @param string $title |
| 333 | | - * @abstract |
| 334 | | - */ |
| 335 | | - function updateTitle( $id, $title ) { |
| 336 | | - // no-op |
| 337 | | - } |
| 338 | | - |
| 339 | | - /** |
| 340 | | - * Get OpenSearch suggestion template |
| 341 | | - * |
| 342 | | - * @return string |
| 343 | | - * @static |
| 344 | | - */ |
| 345 | | - public static function getOpenSearchTemplate() { |
| 346 | | - global $wgOpenSearchTemplate, $wgServer, $wgScriptPath; |
| 347 | | - if($wgOpenSearchTemplate) |
| 348 | | - return $wgOpenSearchTemplate; |
| 349 | | - else{ |
| 350 | | - $ns = implode(',',SearchEngine::defaultNamespaces()); |
| 351 | | - if(!$ns) $ns = "0"; |
| 352 | | - return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns; |
| 353 | | - } |
| 354 | | - } |
| 355 | | - |
| 356 | | - /** |
| 357 | | - * Get internal MediaWiki Suggest template |
| 358 | | - * |
| 359 | | - * @return string |
| 360 | | - * @static |
| 361 | | - */ |
| 362 | | - public static function getMWSuggestTemplate() { |
| 363 | | - global $wgMWSuggestTemplate, $wgServer, $wgScriptPath; |
| 364 | | - if($wgMWSuggestTemplate) |
| 365 | | - return $wgMWSuggestTemplate; |
| 366 | | - else |
| 367 | | - return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}'; |
| 368 | | - } |
| 369 | | -} |
| 370 | | - |
| 371 | | -/** |
| 372 | | - * @ingroup Search |
| 373 | | - */ |
| 374 | | -class SearchResultSet { |
| 375 | | - /** |
| 376 | | - * Fetch an array of regular expression fragments for matching |
| 377 | | - * the search terms as parsed by this engine in a text extract. |
| 378 | | - * |
| 379 | | - * @return array |
| 380 | | - * @access public |
| 381 | | - * @abstract |
| 382 | | - */ |
| 383 | | - function termMatches() { |
| 384 | | - return array(); |
| 385 | | - } |
| 386 | | - |
| 387 | | - function numRows() { |
| 388 | | - return 0; |
| 389 | | - } |
| 390 | | - |
| 391 | | - /** |
| 392 | | - * Return true if results are included in this result set. |
| 393 | | - * @return bool |
| 394 | | - * @abstract |
| 395 | | - */ |
| 396 | | - function hasResults() { |
| 397 | | - return false; |
| 398 | | - } |
| 399 | | - |
| 400 | | - /** |
| 401 | | - * Some search modes return a total hit count for the query |
| 402 | | - * in the entire article database. This may include pages |
| 403 | | - * in namespaces that would not be matched on the given |
| 404 | | - * settings. |
| 405 | | - * |
| 406 | | - * Return null if no total hits number is supported. |
| 407 | | - * |
| 408 | | - * @return int |
| 409 | | - * @access public |
| 410 | | - */ |
| 411 | | - function getTotalHits() { |
| 412 | | - return null; |
| 413 | | - } |
| 414 | | - |
| 415 | | - /** |
| 416 | | - * Some search modes return a suggested alternate term if there are |
| 417 | | - * no exact hits. Returns true if there is one on this set. |
| 418 | | - * |
| 419 | | - * @return bool |
| 420 | | - * @access public |
| 421 | | - */ |
| 422 | | - function hasSuggestion() { |
| 423 | | - return false; |
| 424 | | - } |
| 425 | | - |
| 426 | | - /** |
| 427 | | - * @return string suggested query, null if none |
| 428 | | - */ |
| 429 | | - function getSuggestionQuery(){ |
| 430 | | - return null; |
| 431 | | - } |
| 432 | | - |
| 433 | | - /** |
| 434 | | - * @return string highlighted suggested query, '' if none |
| 435 | | - */ |
| 436 | | - function getSuggestionSnippet(){ |
| 437 | | - return ''; |
| 438 | | - } |
| 439 | | - |
| 440 | | - /** |
| 441 | | - * Return information about how and from where the results were fetched, |
| 442 | | - * should be useful for diagnostics and debugging |
| 443 | | - * |
| 444 | | - * @return string |
| 445 | | - */ |
| 446 | | - function getInfo() { |
| 447 | | - return null; |
| 448 | | - } |
| 449 | | - |
| 450 | | - /** |
| 451 | | - * Return a result set of hits on other (multiple) wikis associated with this one |
| 452 | | - * |
| 453 | | - * @return SearchResultSet |
| 454 | | - */ |
| 455 | | - function getInterwikiResults() { |
| 456 | | - return null; |
| 457 | | - } |
| 458 | | - |
| 459 | | - /** |
| 460 | | - * Check if there are results on other wikis |
| 461 | | - * |
| 462 | | - * @return boolean |
| 463 | | - */ |
| 464 | | - function hasInterwikiResults() { |
| 465 | | - return $this->getInterwikiResults() != null; |
| 466 | | - } |
| 467 | | - |
| 468 | | - |
| 469 | | - /** |
| 470 | | - * Fetches next search result, or false. |
| 471 | | - * @return SearchResult |
| 472 | | - * @access public |
| 473 | | - * @abstract |
| 474 | | - */ |
| 475 | | - function next() { |
| 476 | | - return false; |
| 477 | | - } |
| 478 | | - |
| 479 | | - /** |
| 480 | | - * Frees the result set, if applicable. |
| 481 | | - * @ access public |
| 482 | | - */ |
| 483 | | - function free() { |
| 484 | | - // ... |
| 485 | | - } |
| 486 | | -} |
| 487 | | - |
| 488 | | - |
| 489 | | -/** |
| 490 | | - * @ingroup Search |
| 491 | | - */ |
| 492 | | -class SearchResultTooMany { |
| 493 | | - ## Some search engines may bail out if too many matches are found |
| 494 | | -} |
| 495 | | - |
| 496 | | - |
| 497 | | -/** |
| 498 | | - * @ingroup Search |
| 499 | | - */ |
| 500 | | -class SearchResult { |
| 501 | | - var $mRevision = null; |
| 502 | | - |
| 503 | | - function SearchResult( $row ) { |
| 504 | | - $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| 505 | | - if( !is_null($this->mTitle) ) |
| 506 | | - $this->mRevision = Revision::newFromTitle( $this->mTitle ); |
| 507 | | - } |
| 508 | | - |
| 509 | | - /** |
| 510 | | - * Check if this is result points to an invalid title |
| 511 | | - * |
| 512 | | - * @return boolean |
| 513 | | - * @access public |
| 514 | | - */ |
| 515 | | - function isBrokenTitle(){ |
| 516 | | - if( is_null($this->mTitle) ) |
| 517 | | - return true; |
| 518 | | - return false; |
| 519 | | - } |
| 520 | | - |
| 521 | | - /** |
| 522 | | - * Check if target page is missing, happens when index is out of date |
| 523 | | - * |
| 524 | | - * @return boolean |
| 525 | | - * @access public |
| 526 | | - */ |
| 527 | | - function isMissingRevision(){ |
| 528 | | - if( !$this->mRevision ) |
| 529 | | - return true; |
| 530 | | - return false; |
| 531 | | - } |
| 532 | | - |
| 533 | | - /** |
| 534 | | - * @return Title |
| 535 | | - * @access public |
| 536 | | - */ |
| 537 | | - function getTitle() { |
| 538 | | - return $this->mTitle; |
| 539 | | - } |
| 540 | | - |
| 541 | | - /** |
| 542 | | - * @return double or null if not supported |
| 543 | | - */ |
| 544 | | - function getScore() { |
| 545 | | - return null; |
| 546 | | - } |
| 547 | | - |
| 548 | | - /** |
| 549 | | - * Lazy initialization of article text from DB |
| 550 | | - */ |
| 551 | | - protected function initText(){ |
| 552 | | - if( !isset($this->mText) ){ |
| 553 | | - $this->mText = $this->mRevision->getText(); |
| 554 | | - } |
| 555 | | - } |
| 556 | | - |
| 557 | | - /** |
| 558 | | - * @param array $terms terms to highlight |
| 559 | | - * @return string highlighted text snippet, null (and not '') if not supported |
| 560 | | - */ |
| 561 | | - function getTextSnippet($terms){ |
| 562 | | - global $wgUser, $wgAdvancedSearchHighlighting; |
| 563 | | - $this->initText(); |
| 564 | | - list($contextlines,$contextchars) = SearchEngine::userHighlightPrefs($wgUser); |
| 565 | | - $h = new SearchHighlighter(); |
| 566 | | - if( $wgAdvancedSearchHighlighting ) |
| 567 | | - return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars ); |
| 568 | | - else |
| 569 | | - return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars ); |
| 570 | | - } |
| 571 | | - |
| 572 | | - /** |
| 573 | | - * @param array $terms terms to highlight |
| 574 | | - * @return string highlighted title, '' if not supported |
| 575 | | - */ |
| 576 | | - function getTitleSnippet($terms){ |
| 577 | | - return ''; |
| 578 | | - } |
| 579 | | - |
| 580 | | - /** |
| 581 | | - * @param array $terms terms to highlight |
| 582 | | - * @return string highlighted redirect name (redirect to this page), '' if none or not supported |
| 583 | | - */ |
| 584 | | - function getRedirectSnippet($terms){ |
| 585 | | - return ''; |
| 586 | | - } |
| 587 | | - |
| 588 | | - /** |
| 589 | | - * @return Title object for the redirect to this page, null if none or not supported |
| 590 | | - */ |
| 591 | | - function getRedirectTitle(){ |
| 592 | | - return null; |
| 593 | | - } |
| 594 | | - |
| 595 | | - /** |
| 596 | | - * @return string highlighted relevant section name, null if none or not supported |
| 597 | | - */ |
| 598 | | - function getSectionSnippet(){ |
| 599 | | - return ''; |
| 600 | | - } |
| 601 | | - |
| 602 | | - /** |
| 603 | | - * @return Title object (pagename+fragment) for the section, null if none or not supported |
| 604 | | - */ |
| 605 | | - function getSectionTitle(){ |
| 606 | | - return null; |
| 607 | | - } |
| 608 | | - |
| 609 | | - /** |
| 610 | | - * @return string timestamp |
| 611 | | - */ |
| 612 | | - function getTimestamp(){ |
| 613 | | - return $this->mRevision->getTimestamp(); |
| 614 | | - } |
| 615 | | - |
| 616 | | - /** |
| 617 | | - * @return int number of words |
| 618 | | - */ |
| 619 | | - function getWordCount(){ |
| 620 | | - $this->initText(); |
| 621 | | - return str_word_count( $this->mText ); |
| 622 | | - } |
| 623 | | - |
| 624 | | - /** |
| 625 | | - * @return int size in bytes |
| 626 | | - */ |
| 627 | | - function getByteSize(){ |
| 628 | | - $this->initText(); |
| 629 | | - return strlen( $this->mText ); |
| 630 | | - } |
| 631 | | - |
| 632 | | - /** |
| 633 | | - * @return boolean if hit has related articles |
| 634 | | - */ |
| 635 | | - function hasRelated(){ |
| 636 | | - return false; |
| 637 | | - } |
| 638 | | - |
| 639 | | - /** |
| 640 | | - * @return interwiki prefix of the title (return iw even if title is broken) |
| 641 | | - */ |
| 642 | | - function getInterwikiPrefix(){ |
| 643 | | - return ''; |
| 644 | | - } |
| 645 | | -} |
| 646 | | - |
| 647 | | -/** |
| 648 | | - * Highlight bits of wikitext |
| 649 | | - * |
| 650 | | - * @ingroup Search |
| 651 | | - */ |
| 652 | | -class SearchHighlighter { |
| 653 | | - var $mCleanWikitext = true; |
| 654 | | - |
| 655 | | - function SearchHighlighter($cleanupWikitext = true){ |
| 656 | | - $this->mCleanWikitext = $cleanupWikitext; |
| 657 | | - } |
| 658 | | - |
| 659 | | - /** |
| 660 | | - * Default implementation of wikitext highlighting |
| 661 | | - * |
| 662 | | - * @param string $text |
| 663 | | - * @param array $terms Terms to highlight (unescaped) |
| 664 | | - * @param int $contextlines |
| 665 | | - * @param int $contextchars |
| 666 | | - * @return string |
| 667 | | - */ |
| 668 | | - public function highlightText( $text, $terms, $contextlines, $contextchars ) { |
| 669 | | - global $wgLang, $wgContLang; |
| 670 | | - global $wgSearchHighlightBoundaries; |
| 671 | | - $fname = __METHOD__; |
| 672 | | - |
| 673 | | - if($text == '') |
| 674 | | - return ''; |
| 675 | | - |
| 676 | | - // spli text into text + templates/links/tables |
| 677 | | - $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)"; |
| 678 | | - // first capture group is for detecting nested templates/links/tables/references |
| 679 | | - $endPatterns = array( |
| 680 | | - 1 => '/(\{\{)|(\}\})/', // template |
| 681 | | - 2 => '/(\[\[)|(\]\])/', // image |
| 682 | | - 3 => "/(\n\\{\\|)|(\n\\|\\})/"); // table |
| 683 | | - |
| 684 | | - // FIXME: this should prolly be a hook or something |
| 685 | | - if(function_exists('wfCite')){ |
| 686 | | - $spat .= '|(<ref>)'; // references via cite extension |
| 687 | | - $endPatterns[4] = '/(<ref>)|(<\/ref>)/'; |
| 688 | | - } |
| 689 | | - $spat .= '/'; |
| 690 | | - $textExt = array(); // text extracts |
| 691 | | - $otherExt = array(); // other extracts |
| 692 | | - wfProfileIn( "$fname-split" ); |
| 693 | | - $start = 0; |
| 694 | | - $textLen = strlen($text); |
| 695 | | - $count = 0; // sequence number to maintain ordering |
| 696 | | - while( $start < $textLen ){ |
| 697 | | - // find start of template/image/table |
| 698 | | - if( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ){ |
| 699 | | - $epat = ''; |
| 700 | | - foreach($matches as $key => $val){ |
| 701 | | - if($key > 0 && $val[1] != -1){ |
| 702 | | - if($key == 2){ |
| 703 | | - // see if this is an image link |
| 704 | | - $ns = substr($val[0],2,-1); |
| 705 | | - if( $wgContLang->getNsIndex($ns) != NS_IMAGE ) |
| 706 | | - break; |
| 707 | | - |
| 708 | | - } |
| 709 | | - $epat = $endPatterns[$key]; |
| 710 | | - $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) ); |
| 711 | | - $start = $val[1]; |
| 712 | | - break; |
| 713 | | - } |
| 714 | | - } |
| 715 | | - if( $epat ){ |
| 716 | | - // find end (and detect any nested elements) |
| 717 | | - $level = 0; |
| 718 | | - $offset = $start + 1; |
| 719 | | - $found = false; |
| 720 | | - while( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ){ |
| 721 | | - if( array_key_exists(2,$endMatches) ){ |
| 722 | | - // found end |
| 723 | | - if($level == 0){ |
| 724 | | - $len = strlen($endMatches[2][0]); |
| 725 | | - $off = $endMatches[2][1]; |
| 726 | | - $this->splitAndAdd( $otherExt, $count, |
| 727 | | - substr( $text, $start, $off + $len - $start ) ); |
| 728 | | - $start = $off + $len; |
| 729 | | - $found = true; |
| 730 | | - break; |
| 731 | | - } else{ |
| 732 | | - // end of nested element |
| 733 | | - $level -= 1; |
| 734 | | - } |
| 735 | | - } else{ |
| 736 | | - // nested |
| 737 | | - $level += 1; |
| 738 | | - } |
| 739 | | - $offset = $endMatches[0][1] + strlen($endMatches[0][0]); |
| 740 | | - } |
| 741 | | - if( ! $found ){ |
| 742 | | - // couldn't find appropriate closing tag, skip |
| 743 | | - $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen($matches[0][0]) ) ); |
| 744 | | - $start += strlen($matches[0][0]); |
| 745 | | - } |
| 746 | | - continue; |
| 747 | | - } |
| 748 | | - } |
| 749 | | - // else: add as text extract |
| 750 | | - $this->splitAndAdd( $textExt, $count, substr($text,$start) ); |
| 751 | | - break; |
| 752 | | - } |
| 753 | | - |
| 754 | | - $all = $textExt + $otherExt; // these have disjunct key sets |
| 755 | | - |
| 756 | | - wfProfileOut( "$fname-split" ); |
| 757 | | - |
| 758 | | - // prepare regexps |
| 759 | | - foreach( $terms as $index => $term ) { |
| 760 | | - $terms[$index] = preg_quote( $term, '/' ); |
| 761 | | - // manually do upper/lowercase stuff for utf-8 since PHP won't do it |
| 762 | | - if(preg_match('/[\x80-\xff]/', $term) ){ |
| 763 | | - $terms[$index] = preg_replace_callback('/./us',array($this,'caseCallback'),$terms[$index]); |
| 764 | | - } |
| 765 | | - |
| 766 | | - |
| 767 | | - } |
| 768 | | - $anyterm = implode( '|', $terms ); |
| 769 | | - $phrase = implode("$wgSearchHighlightBoundaries+", $terms ); |
| 770 | | - |
| 771 | | - // FIXME: a hack to scale contextchars, a correct solution |
| 772 | | - // would be to have contextchars actually be char and not byte |
| 773 | | - // length, and do proper utf-8 substrings and lengths everywhere, |
| 774 | | - // but PHP is making that very hard and unclean to implement :( |
| 775 | | - $scale = strlen($anyterm) / mb_strlen($anyterm); |
| 776 | | - $contextchars = intval( $contextchars * $scale ); |
| 777 | | - |
| 778 | | - $patPre = "(^|$wgSearchHighlightBoundaries)"; |
| 779 | | - $patPost = "($wgSearchHighlightBoundaries|$)"; |
| 780 | | - |
| 781 | | - $pat1 = "/(".$phrase.")/ui"; |
| 782 | | - $pat2 = "/$patPre(".$anyterm.")$patPost/ui"; |
| 783 | | - |
| 784 | | - wfProfileIn( "$fname-extract" ); |
| 785 | | - |
| 786 | | - $left = $contextlines; |
| 787 | | - |
| 788 | | - $snippets = array(); |
| 789 | | - $offsets = array(); |
| 790 | | - |
| 791 | | - // show beginning only if it contains all words |
| 792 | | - $first = 0; |
| 793 | | - $firstText = ''; |
| 794 | | - foreach($textExt as $index => $line){ |
| 795 | | - if(strlen($line)>0 && $line[0] != ';' && $line[0] != ':'){ |
| 796 | | - $firstText = $this->extract( $line, 0, $contextchars * $contextlines ); |
| 797 | | - $first = $index; |
| 798 | | - break; |
| 799 | | - } |
| 800 | | - } |
| 801 | | - if( $firstText ){ |
| 802 | | - $succ = true; |
| 803 | | - // check if first text contains all terms |
| 804 | | - foreach($terms as $term){ |
| 805 | | - if( ! preg_match("/$patPre".$term."$patPost/ui", $firstText) ){ |
| 806 | | - $succ = false; |
| 807 | | - break; |
| 808 | | - } |
| 809 | | - } |
| 810 | | - if( $succ ){ |
| 811 | | - $snippets[$first] = $firstText; |
| 812 | | - $offsets[$first] = 0; |
| 813 | | - } |
| 814 | | - } |
| 815 | | - if( ! $snippets ) { |
| 816 | | - // match whole query on text |
| 817 | | - $this->process($pat1, $textExt, $left, $contextchars, $snippets, $offsets); |
| 818 | | - // match whole query on templates/tables/images |
| 819 | | - $this->process($pat1, $otherExt, $left, $contextchars, $snippets, $offsets); |
| 820 | | - // match any words on text |
| 821 | | - $this->process($pat2, $textExt, $left, $contextchars, $snippets, $offsets); |
| 822 | | - // match any words on templates/tables/images |
| 823 | | - $this->process($pat2, $otherExt, $left, $contextchars, $snippets, $offsets); |
| 824 | | - |
| 825 | | - ksort($snippets); |
| 826 | | - } |
| 827 | | - |
| 828 | | - // add extra chars to each snippet to make snippets constant size |
| 829 | | - $extended = array(); |
| 830 | | - if( count( $snippets ) == 0){ |
| 831 | | - // couldn't find the target words, just show beginning of article |
| 832 | | - $targetchars = $contextchars * $contextlines; |
| 833 | | - $snippets[$first] = ''; |
| 834 | | - $offsets[$first] = 0; |
| 835 | | - } else{ |
| 836 | | - // if begin of the article contains the whole phrase, show only that !! |
| 837 | | - if( array_key_exists($first,$snippets) && preg_match($pat1,$snippets[$first]) |
| 838 | | - && $offsets[$first] < $contextchars * 2 ){ |
| 839 | | - $snippets = array ($first => $snippets[$first]); |
| 840 | | - } |
| 841 | | - |
| 842 | | - // calc by how much to extend existing snippets |
| 843 | | - $targetchars = intval( ($contextchars * $contextlines) / count ( $snippets ) ); |
| 844 | | - } |
| 845 | | - |
| 846 | | - foreach($snippets as $index => $line){ |
| 847 | | - $extended[$index] = $line; |
| 848 | | - $len = strlen($line); |
| 849 | | - if( $len < $targetchars - 20 ){ |
| 850 | | - // complete this line |
| 851 | | - if($len < strlen( $all[$index] )){ |
| 852 | | - $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index]+$targetchars, $offsets[$index]); |
| 853 | | - $len = strlen( $extended[$index] ); |
| 854 | | - } |
| 855 | | - |
| 856 | | - // add more lines |
| 857 | | - $add = $index + 1; |
| 858 | | - while( $len < $targetchars - 20 |
| 859 | | - && array_key_exists($add,$all) |
| 860 | | - && !array_key_exists($add,$snippets) ){ |
| 861 | | - $offsets[$add] = 0; |
| 862 | | - $tt = "\n".$this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] ); |
| 863 | | - $extended[$add] = $tt; |
| 864 | | - $len += strlen( $tt ); |
| 865 | | - $add++; |
| 866 | | - } |
| 867 | | - } |
| 868 | | - } |
| 869 | | - |
| 870 | | - //$snippets = array_map('htmlspecialchars', $extended); |
| 871 | | - $snippets = $extended; |
| 872 | | - $last = -1; |
| 873 | | - $extract = ''; |
| 874 | | - foreach($snippets as $index => $line){ |
| 875 | | - if($last == -1) |
| 876 | | - $extract .= $line; // first line |
| 877 | | - elseif($last+1 == $index && $offsets[$last]+strlen($snippets[$last]) >= strlen($all[$last])) |
| 878 | | - $extract .= " ".$line; // continous lines |
| 879 | | - else |
| 880 | | - $extract .= '<b> ... </b>' . $line; |
| 881 | | - |
| 882 | | - $last = $index; |
| 883 | | - } |
| 884 | | - if( $extract ) |
| 885 | | - $extract .= '<b> ... </b>'; |
| 886 | | - |
| 887 | | - $processed = array(); |
| 888 | | - foreach($terms as $term){ |
| 889 | | - if( ! isset($processed[$term]) ){ |
| 890 | | - $pat3 = "/$patPre(".$term.")$patPost/ui"; // highlight word |
| 891 | | - $extract = preg_replace( $pat3, |
| 892 | | - "\\1<span class='searchmatch'>\\2</span>\\3", $extract ); |
| 893 | | - $processed[$term] = true; |
| 894 | | - } |
| 895 | | - } |
| 896 | | - |
| 897 | | - wfProfileOut( "$fname-extract" ); |
| 898 | | - |
| 899 | | - return $extract; |
| 900 | | - } |
| 901 | | - |
| 902 | | - /** |
| 903 | | - * Split text into lines and add it to extracts array |
| 904 | | - * |
| 905 | | - * @param array $extracts index -> $line |
| 906 | | - * @param int $count |
| 907 | | - * @param string $text |
| 908 | | - */ |
| 909 | | - function splitAndAdd(&$extracts, &$count, $text){ |
| 910 | | - $split = explode( "\n", $this->mCleanWikitext? $this->removeWiki($text) : $text ); |
| 911 | | - foreach($split as $line){ |
| 912 | | - $tt = trim($line); |
| 913 | | - if( $tt ) |
| 914 | | - $extracts[$count++] = $tt; |
| 915 | | - } |
| 916 | | - } |
| 917 | | - |
| 918 | | - /** |
| 919 | | - * Do manual case conversion for non-ascii chars |
| 920 | | - * |
| 921 | | - * @param unknown_type $matches |
| 922 | | - */ |
| 923 | | - function caseCallback($matches){ |
| 924 | | - global $wgContLang; |
| 925 | | - if( strlen($matches[0]) > 1 ){ |
| 926 | | - return '['.$wgContLang->lc($matches[0]).$wgContLang->uc($matches[0]).']'; |
| 927 | | - } else |
| 928 | | - return $matches[0]; |
| 929 | | - } |
| 930 | | - |
| 931 | | - /** |
| 932 | | - * Extract part of the text from start to end, but by |
| 933 | | - * not chopping up words |
| 934 | | - * @param string $text |
| 935 | | - * @param int $start |
| 936 | | - * @param int $end |
| 937 | | - * @param int $posStart (out) actual start position |
| 938 | | - * @param int $posEnd (out) actual end position |
| 939 | | - * @return string |
| 940 | | - */ |
| 941 | | - function extract($text, $start, $end, &$posStart = null, &$posEnd = null ){ |
| 942 | | - global $wgContLang; |
| 943 | | - |
| 944 | | - if( $start != 0) |
| 945 | | - $start = $this->position( $text, $start, 1 ); |
| 946 | | - if( $end >= strlen($text) ) |
| 947 | | - $end = strlen($text); |
| 948 | | - else |
| 949 | | - $end = $this->position( $text, $end ); |
| 950 | | - |
| 951 | | - if(!is_null($posStart)) |
| 952 | | - $posStart = $start; |
| 953 | | - if(!is_null($posEnd)) |
| 954 | | - $posEnd = $end; |
| 955 | | - |
| 956 | | - if($end > $start) |
| 957 | | - return substr($text, $start, $end-$start); |
| 958 | | - else |
| 959 | | - return ''; |
| 960 | | - } |
| 961 | | - |
| 962 | | - /** |
| 963 | | - * Find a nonletter near a point (index) in the text |
| 964 | | - * |
| 965 | | - * @param string $text |
| 966 | | - * @param int $point |
| 967 | | - * @param int $offset to found index |
| 968 | | - * @return int nearest nonletter index, or beginning of utf8 char if none |
| 969 | | - */ |
| 970 | | - function position($text, $point, $offset=0 ){ |
| 971 | | - $tolerance = 10; |
| 972 | | - $s = max( 0, $point - $tolerance ); |
| 973 | | - $l = min( strlen($text), $point + $tolerance ) - $s; |
| 974 | | - $m = array(); |
| 975 | | - if( preg_match('/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr($text,$s,$l), $m, PREG_OFFSET_CAPTURE ) ){ |
| 976 | | - return $m[0][1] + $s + $offset; |
| 977 | | - } else{ |
| 978 | | - // check if point is on a valid first UTF8 char |
| 979 | | - $char = ord( $text[$point] ); |
| 980 | | - while( $char >= 0x80 && $char < 0xc0 ) { |
| 981 | | - // skip trailing bytes |
| 982 | | - $point++; |
| 983 | | - if($point >= strlen($text)) |
| 984 | | - return strlen($text); |
| 985 | | - $char = ord( $text[$point] ); |
| 986 | | - } |
| 987 | | - return $point; |
| 988 | | - |
| 989 | | - } |
| 990 | | - } |
| 991 | | - |
| 992 | | - /** |
| 993 | | - * Search extracts for a pattern, and return snippets |
| 994 | | - * |
| 995 | | - * @param string $pattern regexp for matching lines |
| 996 | | - * @param array $extracts extracts to search |
| 997 | | - * @param int $linesleft number of extracts to make |
| 998 | | - * @param int $contextchars length of snippet |
| 999 | | - * @param array $out map for highlighted snippets |
| 1000 | | - * @param array $offsets map of starting points of snippets |
| 1001 | | - * @protected |
| 1002 | | - */ |
| 1003 | | - function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){ |
| 1004 | | - if($linesleft == 0) |
| 1005 | | - return; // nothing to do |
| 1006 | | - foreach($extracts as $index => $line){ |
| 1007 | | - if( array_key_exists($index,$out) ) |
| 1008 | | - continue; // this line already highlighted |
| 1009 | | - |
| 1010 | | - $m = array(); |
| 1011 | | - if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) |
| 1012 | | - continue; |
| 1013 | | - |
| 1014 | | - $offset = $m[0][1]; |
| 1015 | | - $len = strlen($m[0][0]); |
| 1016 | | - if($offset + $len < $contextchars) |
| 1017 | | - $begin = 0; |
| 1018 | | - elseif( $len > $contextchars) |
| 1019 | | - $begin = $offset; |
| 1020 | | - else |
| 1021 | | - $begin = $offset + intval( ($len - $contextchars) / 2 ); |
| 1022 | | - |
| 1023 | | - $end = $begin + $contextchars; |
| 1024 | | - |
| 1025 | | - $posBegin = $begin; |
| 1026 | | - // basic snippet from this line |
| 1027 | | - $out[$index] = $this->extract($line,$begin,$end,$posBegin); |
| 1028 | | - $offsets[$index] = $posBegin; |
| 1029 | | - $linesleft--; |
| 1030 | | - if($linesleft == 0) |
| 1031 | | - return; |
| 1032 | | - } |
| 1033 | | - } |
| 1034 | | - |
| 1035 | | - /** |
| 1036 | | - * Basic wikitext removal |
| 1037 | | - * @protected |
| 1038 | | - */ |
| 1039 | | - function removeWiki($text) { |
| 1040 | | - $fname = __METHOD__; |
| 1041 | | - wfProfileIn( $fname ); |
| 1042 | | - |
| 1043 | | - //$text = preg_replace("/'{2,5}/", "", $text); |
| 1044 | | - //$text = preg_replace("/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text); |
| 1045 | | - //$text = preg_replace("/\[\[([^]|]+)\]\]/", "\\1", $text); |
| 1046 | | - //$text = preg_replace("/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text); |
| 1047 | | - //$text = preg_replace("/\\{\\|(.*?)\\|\\}/", "", $text); |
| 1048 | | - //$text = preg_replace("/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text); |
| 1049 | | - $text = preg_replace("/\\{\\{([^|]+?)\\}\\}/", "", $text); |
| 1050 | | - $text = preg_replace("/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text); |
| 1051 | | - $text = preg_replace("/\\[\\[([^|]+?)\\]\\]/", "\\1", $text); |
| 1052 | | - $text = preg_replace_callback("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array($this,'linkReplace'), $text); |
| 1053 | | - //$text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text); |
| 1054 | | - $text = preg_replace("/<\/?[^>]+>/", "", $text); |
| 1055 | | - $text = preg_replace("/'''''/", "", $text); |
| 1056 | | - $text = preg_replace("/('''|<\/?[iIuUbB]>)/", "", $text); |
| 1057 | | - $text = preg_replace("/''/", "", $text); |
| 1058 | | - |
| 1059 | | - wfProfileOut( $fname ); |
| 1060 | | - return $text; |
| 1061 | | - } |
| 1062 | | - |
| 1063 | | - /** |
| 1064 | | - * callback to replace [[target|caption]] kind of links, if |
| 1065 | | - * the target is category or image, leave it |
| 1066 | | - * |
| 1067 | | - * @param array $matches |
| 1068 | | - */ |
| 1069 | | - function linkReplace($matches){ |
| 1070 | | - $colon = strpos( $matches[1], ':' ); |
| 1071 | | - if( $colon === false ) |
| 1072 | | - return $matches[2]; // replace with caption |
| 1073 | | - global $wgContLang; |
| 1074 | | - $ns = substr( $matches[1], 0, $colon ); |
| 1075 | | - $index = $wgContLang->getNsIndex($ns); |
| 1076 | | - if( $index !== false && ($index == NS_IMAGE || $index == NS_CATEGORY) ) |
| 1077 | | - return $matches[0]; // return the whole thing |
| 1078 | | - else |
| 1079 | | - return $matches[2]; |
| 1080 | | - |
| 1081 | | - } |
| 1082 | | - |
| 1083 | | - /** |
| 1084 | | - * Simple & fast snippet extraction, but gives completely unrelevant |
| 1085 | | - * snippets |
| 1086 | | - * |
| 1087 | | - * @param string $text |
| 1088 | | - * @param array $terms |
| 1089 | | - * @param int $contextlines |
| 1090 | | - * @param int $contextchars |
| 1091 | | - * @return string |
| 1092 | | - */ |
| 1093 | | - public function highlightSimple( $text, $terms, $contextlines, $contextchars ) { |
| 1094 | | - global $wgLang, $wgContLang; |
| 1095 | | - $fname = __METHOD__; |
| 1096 | | - |
| 1097 | | - $lines = explode( "\n", $text ); |
| 1098 | | - |
| 1099 | | - $terms = implode( '|', $terms ); |
| 1100 | | - $terms = str_replace( '/', "\\/", $terms); |
| 1101 | | - $max = intval( $contextchars ) + 1; |
| 1102 | | - $pat1 = "/(.*)($terms)(.{0,$max})/i"; |
| 1103 | | - |
| 1104 | | - $lineno = 0; |
| 1105 | | - |
| 1106 | | - $extract = ""; |
| 1107 | | - wfProfileIn( "$fname-extract" ); |
| 1108 | | - foreach ( $lines as $line ) { |
| 1109 | | - if ( 0 == $contextlines ) { |
| 1110 | | - break; |
| 1111 | | - } |
| 1112 | | - ++$lineno; |
| 1113 | | - $m = array(); |
| 1114 | | - if ( ! preg_match( $pat1, $line, $m ) ) { |
| 1115 | | - continue; |
| 1116 | | - } |
| 1117 | | - --$contextlines; |
| 1118 | | - $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' ); |
| 1119 | | - |
| 1120 | | - if ( count( $m ) < 3 ) { |
| 1121 | | - $post = ''; |
| 1122 | | - } else { |
| 1123 | | - $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' ); |
| 1124 | | - } |
| 1125 | | - |
| 1126 | | - $found = $m[2]; |
| 1127 | | - |
| 1128 | | - $line = htmlspecialchars( $pre . $found . $post ); |
| 1129 | | - $pat2 = '/(' . $terms . ")/i"; |
| 1130 | | - $line = preg_replace( $pat2, |
| 1131 | | - "<span class='searchmatch'>\\1</span>", $line ); |
| 1132 | | - |
| 1133 | | - $extract .= "${line}\n"; |
| 1134 | | - } |
| 1135 | | - wfProfileOut( "$fname-extract" ); |
| 1136 | | - |
| 1137 | | - return $extract; |
| 1138 | | - } |
| 1139 | | - |
| 1140 | | -} |
| 1141 | | - |
| 1142 | | -/** |
| 1143 | | - * @ingroup Search |
| 1144 | | - */ |
| 1145 | | -class SearchEngineDummy { |
| 1146 | | - function search( $term ) { |
| 1147 | | - return null; |
| 1148 | | - } |
| 1149 | | - function setLimitOffset($l, $o) {} |
| 1150 | | - function legalSearchChars() {} |
| 1151 | | - function update() {} |
| 1152 | | - function setnamespaces() {} |
| 1153 | | - function searchtitle() {} |
| 1154 | | - function searchtext() {} |
| 1155 | | -} |
| Index: trunk/phase3/includes/SearchPostgres.php |
| — | — | @@ -1,255 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -# Copyright (C) 2006-2007 Greg Sabino Mullane <greg@turnstep.com> |
| 4 | | -# http://www.mediawiki.org/ |
| 5 | | -# |
| 6 | | -# This program is free software; you can redistribute it and/or modify |
| 7 | | -# it under the terms of the GNU General Public License as published by |
| 8 | | -# the Free Software Foundation; either version 2 of the License, or |
| 9 | | -# (at your option) any later version. |
| 10 | | -# |
| 11 | | -# This program is distributed in the hope that it will be useful, |
| 12 | | -# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | | -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | | -# GNU General Public License for more details. |
| 15 | | -# |
| 16 | | -# You should have received a copy of the GNU General Public License along |
| 17 | | -# with this program; if not, write to the Free Software Foundation, Inc., |
| 18 | | -# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 19 | | -# http://www.gnu.org/copyleft/gpl.html |
| 20 | | - |
| 21 | | -/** |
| 22 | | - * @file |
| 23 | | - * @ingroup Search |
| 24 | | - */ |
| 25 | | - |
| 26 | | -/** |
| 27 | | - * Search engine hook base class for Postgres |
| 28 | | - * @ingroup Search |
| 29 | | - */ |
| 30 | | -class SearchPostgres extends SearchEngine { |
| 31 | | - |
| 32 | | - function SearchPostgres( $db ) { |
| 33 | | - $this->db = $db; |
| 34 | | - } |
| 35 | | - |
| 36 | | - /** |
| 37 | | - * Perform a full text search query via tsearch2 and return a result set. |
| 38 | | - * Currently searches a page's current title (page.page_title) and |
| 39 | | - * latest revision article text (pagecontent.old_text) |
| 40 | | - * |
| 41 | | - * @param string $term - Raw search term |
| 42 | | - * @return PostgresSearchResultSet |
| 43 | | - * @access public |
| 44 | | - */ |
| 45 | | - function searchTitle( $term ) { |
| 46 | | - $q = $this->searchQuery( $term , 'titlevector', 'page_title' ); |
| 47 | | - $olderror = error_reporting(E_ERROR); |
| 48 | | - $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); |
| 49 | | - error_reporting($olderror); |
| 50 | | - if (!$resultSet) { |
| 51 | | - // Needed for "Query requires full scan, GIN doesn't support it" |
| 52 | | - return new SearchResultTooMany(); |
| 53 | | - } |
| 54 | | - return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); |
| 55 | | - } |
| 56 | | - function searchText( $term ) { |
| 57 | | - $q = $this->searchQuery( $term, 'textvector', 'old_text' ); |
| 58 | | - $olderror = error_reporting(E_ERROR); |
| 59 | | - $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); |
| 60 | | - error_reporting($olderror); |
| 61 | | - if (!$resultSet) { |
| 62 | | - return new SearchResultTooMany(); |
| 63 | | - } |
| 64 | | - return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); |
| 65 | | - } |
| 66 | | - |
| 67 | | - |
| 68 | | - /* |
| 69 | | - * Transform the user's search string into a better form for tsearch2 |
| 70 | | - */ |
| 71 | | - function parseQuery( $term ) { |
| 72 | | - |
| 73 | | - wfDebug( "parseQuery received: $term" ); |
| 74 | | - |
| 75 | | - ## No backslashes allowed |
| 76 | | - $term = preg_replace('/\\\/', '', $term); |
| 77 | | - |
| 78 | | - ## Collapse parens into nearby words: |
| 79 | | - $term = preg_replace('/\s*\(\s*/', ' (', $term); |
| 80 | | - $term = preg_replace('/\s*\)\s*/', ') ', $term); |
| 81 | | - |
| 82 | | - ## Treat colons as word separators: |
| 83 | | - $term = preg_replace('/:/', ' ', $term); |
| 84 | | - |
| 85 | | - $searchstring = ''; |
| 86 | | - $m = array(); |
| 87 | | - if( preg_match_all('/([-!]?)(\S+)\s*/', $term, $m, PREG_SET_ORDER ) ) { |
| 88 | | - foreach( $m as $terms ) { |
| 89 | | - if (strlen($terms[1])) { |
| 90 | | - $searchstring .= ' & !'; |
| 91 | | - } |
| 92 | | - if (strtolower($terms[2]) === 'and') { |
| 93 | | - $searchstring .= ' & '; |
| 94 | | - } |
| 95 | | - else if (strtolower($terms[2]) === 'or' or $terms[2] === '|') { |
| 96 | | - $searchstring .= ' | '; |
| 97 | | - } |
| 98 | | - else if (strtolower($terms[2]) === 'not') { |
| 99 | | - $searchstring .= ' & !'; |
| 100 | | - } |
| 101 | | - else { |
| 102 | | - $searchstring .= " & $terms[2]"; |
| 103 | | - } |
| 104 | | - } |
| 105 | | - } |
| 106 | | - |
| 107 | | - ## Strip out leading junk |
| 108 | | - $searchstring = preg_replace('/^[\s\&\|]+/', '', $searchstring); |
| 109 | | - |
| 110 | | - ## Remove any doubled-up operators |
| 111 | | - $searchstring = preg_replace('/([\!\&\|]) +(?:[\&\|] +)+/', "$1 ", $searchstring); |
| 112 | | - |
| 113 | | - ## Remove any non-spaced operators (e.g. "Zounds!") |
| 114 | | - $searchstring = preg_replace('/([^ ])[\!\&\|]/', "$1", $searchstring); |
| 115 | | - |
| 116 | | - ## Remove any trailing whitespace or operators |
| 117 | | - $searchstring = preg_replace('/[\s\!\&\|]+$/', '', $searchstring); |
| 118 | | - |
| 119 | | - ## Remove unnecessary quotes around everything |
| 120 | | - $searchstring = preg_replace('/^[\'"](.*)[\'"]$/', "$1", $searchstring); |
| 121 | | - |
| 122 | | - ## Quote the whole thing |
| 123 | | - $searchstring = $this->db->addQuotes($searchstring); |
| 124 | | - |
| 125 | | - wfDebug( "parseQuery returned: $searchstring" ); |
| 126 | | - |
| 127 | | - return $searchstring; |
| 128 | | - |
| 129 | | - } |
| 130 | | - |
| 131 | | - /** |
| 132 | | - * Construct the full SQL query to do the search. |
| 133 | | - * @param string $filteredTerm |
| 134 | | - * @param string $fulltext |
| 135 | | - * @private |
| 136 | | - */ |
| 137 | | - function searchQuery( $term, $fulltext, $colname ) { |
| 138 | | - global $wgDBversion; |
| 139 | | - |
| 140 | | - if ( !isset( $wgDBversion ) ) { |
| 141 | | - $this->db->getServerVersion(); |
| 142 | | - $wgDBversion = $this->db->numeric_version; |
| 143 | | - } |
| 144 | | - $prefix = $wgDBversion < 8.3 ? "'default'," : ''; |
| 145 | | - |
| 146 | | - $searchstring = $this->parseQuery( $term ); |
| 147 | | - |
| 148 | | - ## We need a separate query here so gin does not complain about empty searches |
| 149 | | - $SQL = "SELECT to_tsquery($prefix $searchstring)"; |
| 150 | | - $res = $this->db->doQuery($SQL); |
| 151 | | - if (!$res) { |
| 152 | | - ## TODO: Better output (example to catch: one 'two) |
| 153 | | - die ("Sorry, that was not a valid search string. Please go back and try again"); |
| 154 | | - } |
| 155 | | - $top = pg_fetch_result($res,0,0); |
| 156 | | - |
| 157 | | - if ($top === "") { ## e.g. if only stopwords are used XXX return something better |
| 158 | | - $query = "SELECT page_id, page_namespace, page_title, 0 AS score ". |
| 159 | | - "FROM page p, revision r, pagecontent c WHERE p.page_latest = r.rev_id " . |
| 160 | | - "AND r.rev_text_id = c.old_id AND 1=0"; |
| 161 | | - } |
| 162 | | - else { |
| 163 | | - $m = array(); |
| 164 | | - if( preg_match_all("/'([^']+)'/", $top, $m, PREG_SET_ORDER ) ) { |
| 165 | | - foreach( $m as $terms ) { |
| 166 | | - $this->searchTerms[$terms[1]] = $terms[1]; |
| 167 | | - } |
| 168 | | - } |
| 169 | | - |
| 170 | | - $rankscore = $wgDBversion > 8.2 ? 5 : 1; |
| 171 | | - $rank = $wgDBversion < 8.3 ? 'rank' : 'ts_rank'; |
| 172 | | - $query = "SELECT page_id, page_namespace, page_title, ". |
| 173 | | - "$rank($fulltext, to_tsquery($prefix $searchstring), $rankscore) AS score ". |
| 174 | | - "FROM page p, revision r, pagecontent c WHERE p.page_latest = r.rev_id " . |
| 175 | | - "AND r.rev_text_id = c.old_id AND $fulltext @@ to_tsquery($prefix $searchstring)"; |
| 176 | | - } |
| 177 | | - |
| 178 | | - ## Redirects |
| 179 | | - if (! $this->showRedirects) |
| 180 | | - $query .= ' AND page_is_redirect = 0'; |
| 181 | | - |
| 182 | | - ## Namespaces - defaults to 0 |
| 183 | | - if( !is_null($this->namespaces) ){ // null -> search all |
| 184 | | - if ( count($this->namespaces) < 1) |
| 185 | | - $query .= ' AND page_namespace = 0'; |
| 186 | | - else { |
| 187 | | - $namespaces = implode( ',', $this->namespaces ); |
| 188 | | - $query .= " AND page_namespace IN ($namespaces)"; |
| 189 | | - } |
| 190 | | - } |
| 191 | | - |
| 192 | | - $query .= " ORDER BY score DESC, page_id DESC"; |
| 193 | | - |
| 194 | | - $query .= $this->db->limitResult( '', $this->limit, $this->offset ); |
| 195 | | - |
| 196 | | - wfDebug( "searchQuery returned: $query" ); |
| 197 | | - |
| 198 | | - return $query; |
| 199 | | - } |
| 200 | | - |
| 201 | | - ## Most of the work of these two functions are done automatically via triggers |
| 202 | | - |
| 203 | | - function update( $pageid, $title, $text ) { |
| 204 | | - ## We don't want to index older revisions |
| 205 | | - $SQL = "UPDATE pagecontent SET textvector = NULL WHERE old_id = ". |
| 206 | | - "(SELECT rev_text_id FROM revision WHERE rev_page = $pageid ". |
| 207 | | - "ORDER BY rev_text_id DESC LIMIT 1 OFFSET 1)"; |
| 208 | | - $this->db->doQuery($SQL); |
| 209 | | - return true; |
| 210 | | - } |
| 211 | | - |
| 212 | | - function updateTitle( $id, $title ) { |
| 213 | | - return true; |
| 214 | | - } |
| 215 | | - |
| 216 | | -} ## end of the SearchPostgres class |
| 217 | | - |
| 218 | | -/** |
| 219 | | - * @ingroup Search |
| 220 | | - */ |
| 221 | | -class PostgresSearchResult extends SearchResult { |
| 222 | | - function PostgresSearchResult( $row ) { |
| 223 | | - $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| 224 | | - $this->score = $row->score; |
| 225 | | - } |
| 226 | | - function getScore() { |
| 227 | | - return $this->score; |
| 228 | | - } |
| 229 | | -} |
| 230 | | - |
| 231 | | -/** |
| 232 | | - * @ingroup Search |
| 233 | | - */ |
| 234 | | -class PostgresSearchResultSet extends SearchResultSet { |
| 235 | | - function PostgresSearchResultSet( $resultSet, $terms ) { |
| 236 | | - $this->mResultSet = $resultSet; |
| 237 | | - $this->mTerms = $terms; |
| 238 | | - } |
| 239 | | - |
| 240 | | - function termMatches() { |
| 241 | | - return $this->mTerms; |
| 242 | | - } |
| 243 | | - |
| 244 | | - function numRows() { |
| 245 | | - return $this->mResultSet->numRows(); |
| 246 | | - } |
| 247 | | - |
| 248 | | - function next() { |
| 249 | | - $row = $this->mResultSet->fetchObject(); |
| 250 | | - if( $row === false ) { |
| 251 | | - return false; |
| 252 | | - } else { |
| 253 | | - return new PostgresSearchResult( $row ); |
| 254 | | - } |
| 255 | | - } |
| 256 | | -} |
| Index: trunk/phase3/includes/SearchUpdate.php |
| — | — | @@ -1,113 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -/** |
| 4 | | - * See deferred.txt |
| 5 | | - * @ingroup Search |
| 6 | | - */ |
| 7 | | -class SearchUpdate { |
| 8 | | - |
| 9 | | - /* private */ var $mId = 0, $mNamespace, $mTitle, $mText; |
| 10 | | - /* private */ var $mTitleWords; |
| 11 | | - |
| 12 | | - function SearchUpdate( $id, $title, $text = false ) { |
| 13 | | - $nt = Title::newFromText( $title ); |
| 14 | | - if( $nt ) { |
| 15 | | - $this->mId = $id; |
| 16 | | - $this->mText = $text; |
| 17 | | - |
| 18 | | - $this->mNamespace = $nt->getNamespace(); |
| 19 | | - $this->mTitle = $nt->getText(); # Discard namespace |
| 20 | | - |
| 21 | | - $this->mTitleWords = $this->mTextWords = array(); |
| 22 | | - } else { |
| 23 | | - wfDebug( "SearchUpdate object created with invalid title '$title'\n" ); |
| 24 | | - } |
| 25 | | - } |
| 26 | | - |
| 27 | | - function doUpdate() { |
| 28 | | - global $wgContLang, $wgDisableSearchUpdate; |
| 29 | | - |
| 30 | | - if( $wgDisableSearchUpdate || !$this->mId ) { |
| 31 | | - return false; |
| 32 | | - } |
| 33 | | - $fname = 'SearchUpdate::doUpdate'; |
| 34 | | - wfProfileIn( $fname ); |
| 35 | | - |
| 36 | | - $search = SearchEngine::create(); |
| 37 | | - $lc = SearchEngine::legalSearchChars() . '&#;'; |
| 38 | | - |
| 39 | | - if( $this->mText === false ) { |
| 40 | | - $search->updateTitle($this->mId, |
| 41 | | - Title::indexTitle( $this->mNamespace, $this->mTitle )); |
| 42 | | - wfProfileOut( $fname ); |
| 43 | | - return; |
| 44 | | - } |
| 45 | | - |
| 46 | | - # Language-specific strip/conversion |
| 47 | | - $text = $wgContLang->stripForSearch( $this->mText ); |
| 48 | | - |
| 49 | | - wfProfileIn( $fname.'-regexps' ); |
| 50 | | - $text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/", |
| 51 | | - ' ', strtolower( " " . $text /*$this->mText*/ . " " ) ); # Strip HTML markup |
| 52 | | - $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD", |
| 53 | | - "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings |
| 54 | | - |
| 55 | | - # Strip external URLs |
| 56 | | - $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\xA0-\\xFF"; |
| 57 | | - $protos = "http|https|ftp|mailto|news|gopher"; |
| 58 | | - $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/"; |
| 59 | | - $text = preg_replace( $pat, "\\1 \\3", $text ); |
| 60 | | - |
| 61 | | - $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/"; |
| 62 | | - $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/"; |
| 63 | | - $text = preg_replace( $p1, "\\1 ", $text ); |
| 64 | | - $text = preg_replace( $p2, "\\1 \\3 ", $text ); |
| 65 | | - |
| 66 | | - # Internal image links |
| 67 | | - $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i"; |
| 68 | | - $text = preg_replace( $pat2, " \\1 \\3", $text ); |
| 69 | | - |
| 70 | | - $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/", |
| 71 | | - "\\1\\2 \\2\\3", $text ); # Handle [[game]]s |
| 72 | | - |
| 73 | | - # Strip all remaining non-search characters |
| 74 | | - $text = preg_replace( "/[^{$lc}]+/", " ", $text ); |
| 75 | | - |
| 76 | | - # Handle 's, s' |
| 77 | | - # |
| 78 | | - # $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text ); |
| 79 | | - # $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text ); |
| 80 | | - # |
| 81 | | - # These tail-anchored regexps are insanely slow. The worst case comes |
| 82 | | - # when Japanese or Chinese text (ie, no word spacing) is written on |
| 83 | | - # a wiki configured for Western UTF-8 mode. The Unicode characters are |
| 84 | | - # expanded to hex codes and the "words" are very long paragraph-length |
| 85 | | - # monstrosities. On a large page the above regexps may take over 20 |
| 86 | | - # seconds *each* on a 1GHz-level processor. |
| 87 | | - # |
| 88 | | - # Following are reversed versions which are consistently fast |
| 89 | | - # (about 3 milliseconds on 1GHz-level processor). |
| 90 | | - # |
| 91 | | - $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) ); |
| 92 | | - $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) ); |
| 93 | | - |
| 94 | | - # Strip wiki '' and ''' |
| 95 | | - $text = preg_replace( "/''[']*/", " ", $text ); |
| 96 | | - wfProfileOut( "$fname-regexps" ); |
| 97 | | - |
| 98 | | - wfRunHooks( 'SearchUpdate', array( $this->mId, $this->mNamespace, $this->mTitle, &$text ) ); |
| 99 | | - |
| 100 | | - # Perform the actual update |
| 101 | | - $search->update($this->mId, Title::indexTitle( $this->mNamespace, $this->mTitle ), |
| 102 | | - $text); |
| 103 | | - |
| 104 | | - wfProfileOut( $fname ); |
| 105 | | - } |
| 106 | | -} |
| 107 | | - |
| 108 | | -/** |
| 109 | | - * Placeholder class |
| 110 | | - * @ingroup Search |
| 111 | | - */ |
| 112 | | -class SearchUpdateMyISAM extends SearchUpdate { |
| 113 | | - # Inherits everything |
| 114 | | -} |
| Index: trunk/phase3/includes/SearchOracle.php |
| — | — | @@ -1,240 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| 4 | | -# http://www.mediawiki.org/ |
| 5 | | -# |
| 6 | | -# This program is free software; you can redistribute it and/or modify |
| 7 | | -# it under the terms of the GNU General Public License as published by |
| 8 | | -# the Free Software Foundation; either version 2 of the License, or |
| 9 | | -# (at your option) any later version. |
| 10 | | -# |
| 11 | | -# This program is distributed in the hope that it will be useful, |
| 12 | | -# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | | -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | | -# GNU General Public License for more details. |
| 15 | | -# |
| 16 | | -# You should have received a copy of the GNU General Public License along |
| 17 | | -# with this program; if not, write to the Free Software Foundation, Inc., |
| 18 | | -# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 19 | | -# http://www.gnu.org/copyleft/gpl.html |
| 20 | | - |
| 21 | | -/** |
| 22 | | - * @file |
| 23 | | - * @ingroup Search |
| 24 | | - */ |
| 25 | | - |
| 26 | | -/** |
| 27 | | - * Search engine hook base class for Oracle (ConText). |
| 28 | | - * @ingroup Search |
| 29 | | - */ |
| 30 | | -class SearchOracle extends SearchEngine { |
| 31 | | - function __construct($db) { |
| 32 | | - $this->db = $db; |
| 33 | | - } |
| 34 | | - |
| 35 | | - /** |
| 36 | | - * Perform a full text search query and return a result set. |
| 37 | | - * |
| 38 | | - * @param string $term - Raw search term |
| 39 | | - * @return OracleSearchResultSet |
| 40 | | - * @access public |
| 41 | | - */ |
| 42 | | - function searchText( $term ) { |
| 43 | | - $resultSet = $this->db->resultObject($this->db->query($this->getQuery($this->filter($term), true))); |
| 44 | | - return new OracleSearchResultSet($resultSet, $this->searchTerms); |
| 45 | | - } |
| 46 | | - |
| 47 | | - /** |
| 48 | | - * Perform a title-only search query and return a result set. |
| 49 | | - * |
| 50 | | - * @param string $term - Raw search term |
| 51 | | - * @return ORacleSearchResultSet |
| 52 | | - * @access public |
| 53 | | - */ |
| 54 | | - function searchTitle($term) { |
| 55 | | - $resultSet = $this->db->resultObject($this->db->query($this->getQuery($this->filter($term), false))); |
| 56 | | - return new MySQLSearchResultSet($resultSet, $this->searchTerms); |
| 57 | | - } |
| 58 | | - |
| 59 | | - |
| 60 | | - /** |
| 61 | | - * Return a partial WHERE clause to exclude redirects, if so set |
| 62 | | - * @return string |
| 63 | | - * @private |
| 64 | | - */ |
| 65 | | - function queryRedirect() { |
| 66 | | - if ($this->showRedirects) { |
| 67 | | - return ''; |
| 68 | | - } else { |
| 69 | | - return 'AND page_is_redirect=0'; |
| 70 | | - } |
| 71 | | - } |
| 72 | | - |
| 73 | | - /** |
| 74 | | - * Return a partial WHERE clause to limit the search to the given namespaces |
| 75 | | - * @return string |
| 76 | | - * @private |
| 77 | | - */ |
| 78 | | - function queryNamespaces() { |
| 79 | | - if( is_null($this->namespaces) ) |
| 80 | | - return ''; |
| 81 | | - $namespaces = implode(',', $this->namespaces); |
| 82 | | - if ($namespaces == '') { |
| 83 | | - $namespaces = '0'; |
| 84 | | - } |
| 85 | | - return 'AND page_namespace IN (' . $namespaces . ')'; |
| 86 | | - } |
| 87 | | - |
| 88 | | - /** |
| 89 | | - * Return a LIMIT clause to limit results on the query. |
| 90 | | - * @return string |
| 91 | | - * @private |
| 92 | | - */ |
| 93 | | - function queryLimit($sql) { |
| 94 | | - return $this->db->limitResult($sql, $this->limit, $this->offset); |
| 95 | | - } |
| 96 | | - |
| 97 | | - /** |
| 98 | | - * Does not do anything for generic search engine |
| 99 | | - * subclasses may define this though |
| 100 | | - * @return string |
| 101 | | - * @private |
| 102 | | - */ |
| 103 | | - function queryRanking($filteredTerm, $fulltext) { |
| 104 | | - return ' ORDER BY score(1)'; |
| 105 | | - } |
| 106 | | - |
| 107 | | - /** |
| 108 | | - * Construct the full SQL query to do the search. |
| 109 | | - * The guts shoulds be constructed in queryMain() |
| 110 | | - * @param string $filteredTerm |
| 111 | | - * @param bool $fulltext |
| 112 | | - * @private |
| 113 | | - */ |
| 114 | | - function getQuery( $filteredTerm, $fulltext ) { |
| 115 | | - return $this->queryLimit($this->queryMain($filteredTerm, $fulltext) . ' ' . |
| 116 | | - $this->queryRedirect() . ' ' . |
| 117 | | - $this->queryNamespaces() . ' ' . |
| 118 | | - $this->queryRanking( $filteredTerm, $fulltext ) . ' '); |
| 119 | | - } |
| 120 | | - |
| 121 | | - |
| 122 | | - /** |
| 123 | | - * Picks which field to index on, depending on what type of query. |
| 124 | | - * @param bool $fulltext |
| 125 | | - * @return string |
| 126 | | - */ |
| 127 | | - function getIndexField($fulltext) { |
| 128 | | - return $fulltext ? 'si_text' : 'si_title'; |
| 129 | | - } |
| 130 | | - |
| 131 | | - /** |
| 132 | | - * Get the base part of the search query. |
| 133 | | - * |
| 134 | | - * @param string $filteredTerm |
| 135 | | - * @param bool $fulltext |
| 136 | | - * @return string |
| 137 | | - * @private |
| 138 | | - */ |
| 139 | | - function queryMain( $filteredTerm, $fulltext ) { |
| 140 | | - $match = $this->parseQuery($filteredTerm, $fulltext); |
| 141 | | - $page = $this->db->tableName('page'); |
| 142 | | - $searchindex = $this->db->tableName('searchindex'); |
| 143 | | - return 'SELECT page_id, page_namespace, page_title ' . |
| 144 | | - "FROM $page,$searchindex " . |
| 145 | | - 'WHERE page_id=si_page AND ' . $match; |
| 146 | | - } |
| 147 | | - |
| 148 | | - /** @todo document */ |
| 149 | | - function parseQuery($filteredText, $fulltext) { |
| 150 | | - global $wgContLang; |
| 151 | | - $lc = SearchEngine::legalSearchChars(); |
| 152 | | - $this->searchTerms = array(); |
| 153 | | - |
| 154 | | - # FIXME: This doesn't handle parenthetical expressions. |
| 155 | | - $m = array(); |
| 156 | | - $q = array(); |
| 157 | | - |
| 158 | | - if (preg_match_all('/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| 159 | | - $filteredText, $m, PREG_SET_ORDER)) { |
| 160 | | - foreach($m as $terms) { |
| 161 | | - $q[] = $terms[1] . $wgContLang->stripForSearch($terms[2]); |
| 162 | | - |
| 163 | | - if (!empty($terms[3])) { |
| 164 | | - $regexp = preg_quote( $terms[3], '/' ); |
| 165 | | - if ($terms[4]) |
| 166 | | - $regexp .= "[0-9A-Za-z_]+"; |
| 167 | | - } else { |
| 168 | | - $regexp = preg_quote(str_replace('"', '', $terms[2]), '/'); |
| 169 | | - } |
| 170 | | - $this->searchTerms[] = $regexp; |
| 171 | | - } |
| 172 | | - } |
| 173 | | - |
| 174 | | - $searchon = $this->db->strencode(join(',', $q)); |
| 175 | | - $field = $this->getIndexField($fulltext); |
| 176 | | - return " CONTAINS($field, '$searchon', 1) > 0 "; |
| 177 | | - } |
| 178 | | - |
| 179 | | - /** |
| 180 | | - * Create or update the search index record for the given page. |
| 181 | | - * Title and text should be pre-processed. |
| 182 | | - * |
| 183 | | - * @param int $id |
| 184 | | - * @param string $title |
| 185 | | - * @param string $text |
| 186 | | - */ |
| 187 | | - function update($id, $title, $text) { |
| 188 | | - $dbw = wfGetDB(DB_MASTER); |
| 189 | | - $dbw->replace('searchindex', |
| 190 | | - array('si_page'), |
| 191 | | - array( |
| 192 | | - 'si_page' => $id, |
| 193 | | - 'si_title' => $title, |
| 194 | | - 'si_text' => $text |
| 195 | | - ), 'SearchOracle::update' ); |
| 196 | | - $dbw->query("CALL ctx_ddl.sync_index('si_text_idx')"); |
| 197 | | - $dbw->query("CALL ctx_ddl.sync_index('si_title_idx')"); |
| 198 | | - } |
| 199 | | - |
| 200 | | - /** |
| 201 | | - * Update a search index record's title only. |
| 202 | | - * Title should be pre-processed. |
| 203 | | - * |
| 204 | | - * @param int $id |
| 205 | | - * @param string $title |
| 206 | | - */ |
| 207 | | - function updateTitle($id, $title) { |
| 208 | | - $dbw = wfGetDB(DB_MASTER); |
| 209 | | - |
| 210 | | - $dbw->update('searchindex', |
| 211 | | - array('si_title' => $title), |
| 212 | | - array('si_page' => $id), |
| 213 | | - 'SearchOracle::updateTitle', |
| 214 | | - array()); |
| 215 | | - } |
| 216 | | -} |
| 217 | | - |
| 218 | | -/** |
| 219 | | - * @ingroup Search |
| 220 | | - */ |
| 221 | | -class OracleSearchResultSet extends SearchResultSet { |
| 222 | | - function __construct($resultSet, $terms) { |
| 223 | | - $this->mResultSet = $resultSet; |
| 224 | | - $this->mTerms = $terms; |
| 225 | | - } |
| 226 | | - |
| 227 | | - function termMatches() { |
| 228 | | - return $this->mTerms; |
| 229 | | - } |
| 230 | | - |
| 231 | | - function numRows() { |
| 232 | | - return $this->mResultSet->numRows(); |
| 233 | | - } |
| 234 | | - |
| 235 | | - function next() { |
| 236 | | - $row = $this->mResultSet->fetchObject(); |
| 237 | | - if ($row === false) |
| 238 | | - return false; |
| 239 | | - return new SearchResult($row); |
| 240 | | - } |
| 241 | | -} |
| Index: trunk/phase3/includes/SearchTsearch2.php |
| — | — | @@ -1,120 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -# Copyright (C) 2004 Brion Vibber <brion@pobox.com>, Domas Mituzas <domas.mituzas@gmail.com> |
| 4 | | -# http://www.mediawiki.org/ |
| 5 | | -# |
| 6 | | -# This program is free software; you can redistribute it and/or modify |
| 7 | | -# it under the terms of the GNU General Public License as published by |
| 8 | | -# the Free Software Foundation; either version 2 of the License, or |
| 9 | | -# (at your option) any later version. |
| 10 | | -# |
| 11 | | -# This program is distributed in the hope that it will be useful, |
| 12 | | -# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | | -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | | -# GNU General Public License for more details. |
| 15 | | -# |
| 16 | | -# You should have received a copy of the GNU General Public License along |
| 17 | | -# with this program; if not, write to the Free Software Foundation, Inc., |
| 18 | | -# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 19 | | -# http://www.gnu.org/copyleft/gpl.html |
| 20 | | - |
| 21 | | -/** |
| 22 | | - * Search engine hook for PostgreSQL / Tsearch2 |
| 23 | | - * @file |
| 24 | | - * @ingroup Search |
| 25 | | - */ |
| 26 | | - |
| 27 | | -/** |
| 28 | | - * @todo document |
| 29 | | - * @ingroup Search |
| 30 | | - */ |
| 31 | | -class SearchTsearch2 extends SearchEngine { |
| 32 | | - var $strictMatching = false; |
| 33 | | - |
| 34 | | - function __construct( $db ) { |
| 35 | | - $this->db = $db; |
| 36 | | - $this->mRanking = true; |
| 37 | | - } |
| 38 | | - |
| 39 | | - function getIndexField( $fulltext ) { |
| 40 | | - return $fulltext ? 'si_text' : 'si_title'; |
| 41 | | - } |
| 42 | | - |
| 43 | | - function parseQuery( $filteredText, $fulltext ) { |
| 44 | | - global $wgContLang; |
| 45 | | - $lc = SearchEngine::legalSearchChars(); |
| 46 | | - $searchon = ''; |
| 47 | | - $this->searchTerms = array(); |
| 48 | | - |
| 49 | | - # FIXME: This doesn't handle parenthetical expressions. |
| 50 | | - $m = array(); |
| 51 | | - if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| 52 | | - $filteredText, $m, PREG_SET_ORDER ) ) { |
| 53 | | - foreach( $m as $terms ) { |
| 54 | | - if( $searchon !== '' ) $searchon .= ' '; |
| 55 | | - if( $this->strictMatching && ($terms[1] == '') ) { |
| 56 | | - $terms[1] = '+'; |
| 57 | | - } |
| 58 | | - $searchon .= $terms[1] . $wgContLang->stripForSearch( $terms[2] ); |
| 59 | | - if( !empty( $terms[3] ) ) { |
| 60 | | - $regexp = preg_quote( $terms[3], '/' ); |
| 61 | | - if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+"; |
| 62 | | - } else { |
| 63 | | - $regexp = preg_quote( str_replace( '"', '', $terms[2] ), '/' ); |
| 64 | | - } |
| 65 | | - $this->searchTerms[] = $regexp; |
| 66 | | - } |
| 67 | | - wfDebug( "Would search with '$searchon'\n" ); |
| 68 | | - wfDebug( 'Match with /\b' . implode( '\b|\b', $this->searchTerms ) . "\b/\n" ); |
| 69 | | - } else { |
| 70 | | - wfDebug( "Can't understand search query '{$this->filteredText}'\n" ); |
| 71 | | - } |
| 72 | | - |
| 73 | | - $searchon = preg_replace( '/(\s+)/', '&', $searchon ); |
| 74 | | - $searchon = $this->db->strencode( $searchon ); |
| 75 | | - return $searchon; |
| 76 | | - } |
| 77 | | - |
| 78 | | - function queryRanking( $filteredTerm, $fulltext ) { |
| 79 | | - $field = $this->getIndexField( $fulltext ); |
| 80 | | - $searchon = $this->parseQuery( $filteredTerm, $fulltext ); |
| 81 | | - if ($this->mRanking) |
| 82 | | - return " ORDER BY rank($field,to_tsquery('$searchon')) DESC"; |
| 83 | | - else |
| 84 | | - return ""; |
| 85 | | - } |
| 86 | | - |
| 87 | | - |
| 88 | | - function queryMain( $filteredTerm, $fulltext ) { |
| 89 | | - $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 90 | | - $field = $this->getIndexField( $fulltext ); |
| 91 | | - $cur = $this->db->tableName( 'cur' ); |
| 92 | | - $searchindex = $this->db->tableName( 'searchindex' ); |
| 93 | | - return 'SELECT cur_id, cur_namespace, cur_title, cur_text ' . |
| 94 | | - "FROM $cur,$searchindex " . |
| 95 | | - 'WHERE cur_id=si_page AND ' . |
| 96 | | - " $field @@ to_tsquery ('$match') " ; |
| 97 | | - } |
| 98 | | - |
| 99 | | - function update( $id, $title, $text ) { |
| 100 | | - $dbw = wfGetDB( DB_MASTER ); |
| 101 | | - $searchindex = $dbw->tableName( 'searchindex' ); |
| 102 | | - $sql = "DELETE FROM $searchindex WHERE si_page={$id}"; |
| 103 | | - $dbw->query( $sql, __METHOD__ ); |
| 104 | | - $sql = "INSERT INTO $searchindex (si_page,si_title,si_text) ". |
| 105 | | - " VALUES ( $id, to_tsvector('". |
| 106 | | - $dbw->strencode($title). |
| 107 | | - "'),to_tsvector('". |
| 108 | | - $dbw->strencode( $text)."')) "; |
| 109 | | - $dbw->query($sql, __METHOD__ ); |
| 110 | | - } |
| 111 | | - |
| 112 | | - function updateTitle($id,$title) { |
| 113 | | - $dbw = wfGetDB(DB_MASTER); |
| 114 | | - $searchindex = $dbw->tableName( 'searchindex' ); |
| 115 | | - $sql = "UPDATE $searchindex SET si_title=to_tsvector('" . |
| 116 | | - $dbw->strencode( $title ) . |
| 117 | | - "') WHERE si_page={$id}"; |
| 118 | | - |
| 119 | | - $dbw->query( $sql, __METHOD__ ); |
| 120 | | - } |
| 121 | | -} |
| Index: trunk/phase3/includes/SearchMySQL4.php |
| — | — | @@ -1,34 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| 4 | | -# http://www.mediawiki.org/ |
| 5 | | -# |
| 6 | | -# This program is free software; you can redistribute it and/or modify |
| 7 | | -# it under the terms of the GNU General Public License as published by |
| 8 | | -# the Free Software Foundation; either version 2 of the License, or |
| 9 | | -# (at your option) any later version. |
| 10 | | -# |
| 11 | | -# This program is distributed in the hope that it will be useful, |
| 12 | | -# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | | -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | | -# GNU General Public License for more details. |
| 15 | | -# |
| 16 | | -# You should have received a copy of the GNU General Public License along |
| 17 | | -# with this program; if not, write to the Free Software Foundation, Inc., |
| 18 | | -# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 19 | | -# http://www.gnu.org/copyleft/gpl.html |
| 20 | | - |
| 21 | | -/** |
| 22 | | - * @file |
| 23 | | - * @ingroup Search |
| 24 | | - */ |
| 25 | | - |
| 26 | | -/** |
| 27 | | - * Search engine hook for MySQL 4+ |
| 28 | | - * This class retained for backwards compatibility... |
| 29 | | - * The meat's been moved to SearchMySQL, since the 3.x variety is gone. |
| 30 | | - * @ingroup Search |
| 31 | | - * @deprecated |
| 32 | | - */ |
| 33 | | -class SearchMySQL4 extends SearchMySQL { |
| 34 | | - /* whee */ |
| 35 | | -} |
| Index: trunk/phase3/includes/SearchMySQL.php |
| — | — | @@ -1,262 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| 4 | | -# http://www.mediawiki.org/ |
| 5 | | -# |
| 6 | | -# This program is free software; you can redistribute it and/or modify |
| 7 | | -# it under the terms of the GNU General Public License as published by |
| 8 | | -# the Free Software Foundation; either version 2 of the License, or |
| 9 | | -# (at your option) any later version. |
| 10 | | -# |
| 11 | | -# This program is distributed in the hope that it will be useful, |
| 12 | | -# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | | -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | | -# GNU General Public License for more details. |
| 15 | | -# |
| 16 | | -# You should have received a copy of the GNU General Public License along |
| 17 | | -# with this program; if not, write to the Free Software Foundation, Inc., |
| 18 | | -# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 19 | | -# http://www.gnu.org/copyleft/gpl.html |
| 20 | | - |
| 21 | | -/** |
| 22 | | - * @file |
| 23 | | - * @ingroup Search |
| 24 | | - */ |
| 25 | | - |
| 26 | | -/** |
| 27 | | - * Search engine hook for MySQL 4+ |
| 28 | | - * @ingroup Search |
| 29 | | - */ |
| 30 | | -class SearchMySQL extends SearchEngine { |
| 31 | | - var $strictMatching = true; |
| 32 | | - |
| 33 | | - /** @todo document */ |
| 34 | | - function __construct( $db ) { |
| 35 | | - $this->db = $db; |
| 36 | | - } |
| 37 | | - |
| 38 | | - /** @todo document */ |
| 39 | | - function parseQuery( $filteredText, $fulltext ) { |
| 40 | | - global $wgContLang; |
| 41 | | - $lc = SearchEngine::legalSearchChars(); // Minus format chars |
| 42 | | - $searchon = ''; |
| 43 | | - $this->searchTerms = array(); |
| 44 | | - |
| 45 | | - # FIXME: This doesn't handle parenthetical expressions. |
| 46 | | - $m = array(); |
| 47 | | - if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| 48 | | - $filteredText, $m, PREG_SET_ORDER ) ) { |
| 49 | | - foreach( $m as $terms ) { |
| 50 | | - if( $searchon !== '' ) $searchon .= ' '; |
| 51 | | - if( $this->strictMatching && ($terms[1] == '') ) { |
| 52 | | - $terms[1] = '+'; |
| 53 | | - } |
| 54 | | - $searchon .= $terms[1] . $wgContLang->stripForSearch( $terms[2] ); |
| 55 | | - if( !empty( $terms[3] ) ) { |
| 56 | | - // Match individual terms in result highlighting... |
| 57 | | - $regexp = preg_quote( $terms[3], '/' ); |
| 58 | | - if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+"; |
| 59 | | - } else { |
| 60 | | - // Match the quoted term in result highlighting... |
| 61 | | - $regexp = preg_quote( str_replace( '"', '', $terms[2] ), '/' ); |
| 62 | | - } |
| 63 | | - $this->searchTerms[] = $regexp; |
| 64 | | - } |
| 65 | | - wfDebug( "Would search with '$searchon'\n" ); |
| 66 | | - wfDebug( 'Match with /' . implode( '|', $this->searchTerms ) . "/\n" ); |
| 67 | | - } else { |
| 68 | | - wfDebug( "Can't understand search query '{$filteredText}'\n" ); |
| 69 | | - } |
| 70 | | - |
| 71 | | - $searchon = $this->db->strencode( $searchon ); |
| 72 | | - $field = $this->getIndexField( $fulltext ); |
| 73 | | - return " MATCH($field) AGAINST('$searchon' IN BOOLEAN MODE) "; |
| 74 | | - } |
| 75 | | - |
| 76 | | - public static function legalSearchChars() { |
| 77 | | - return "\"*" . parent::legalSearchChars(); |
| 78 | | - } |
| 79 | | - |
| 80 | | - /** |
| 81 | | - * Perform a full text search query and return a result set. |
| 82 | | - * |
| 83 | | - * @param string $term - Raw search term |
| 84 | | - * @return MySQLSearchResultSet |
| 85 | | - * @access public |
| 86 | | - */ |
| 87 | | - function searchText( $term ) { |
| 88 | | - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) ); |
| 89 | | - return new MySQLSearchResultSet( $resultSet, $this->searchTerms ); |
| 90 | | - } |
| 91 | | - |
| 92 | | - /** |
| 93 | | - * Perform a title-only search query and return a result set. |
| 94 | | - * |
| 95 | | - * @param string $term - Raw search term |
| 96 | | - * @return MySQLSearchResultSet |
| 97 | | - * @access public |
| 98 | | - */ |
| 99 | | - function searchTitle( $term ) { |
| 100 | | - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) ); |
| 101 | | - return new MySQLSearchResultSet( $resultSet, $this->searchTerms ); |
| 102 | | - } |
| 103 | | - |
| 104 | | - |
| 105 | | - /** |
| 106 | | - * Return a partial WHERE clause to exclude redirects, if so set |
| 107 | | - * @return string |
| 108 | | - * @private |
| 109 | | - */ |
| 110 | | - function queryRedirect() { |
| 111 | | - if( $this->showRedirects ) { |
| 112 | | - return ''; |
| 113 | | - } else { |
| 114 | | - return 'AND page_is_redirect=0'; |
| 115 | | - } |
| 116 | | - } |
| 117 | | - |
| 118 | | - /** |
| 119 | | - * Return a partial WHERE clause to limit the search to the given namespaces |
| 120 | | - * @return string |
| 121 | | - * @private |
| 122 | | - */ |
| 123 | | - function queryNamespaces() { |
| 124 | | - if( is_null($this->namespaces) ) |
| 125 | | - return ''; # search all |
| 126 | | - $namespaces = implode( ',', $this->namespaces ); |
| 127 | | - if ($namespaces == '') { |
| 128 | | - $namespaces = '0'; |
| 129 | | - } |
| 130 | | - return 'AND page_namespace IN (' . $namespaces . ')'; |
| 131 | | - } |
| 132 | | - |
| 133 | | - /** |
| 134 | | - * Return a LIMIT clause to limit results on the query. |
| 135 | | - * @return string |
| 136 | | - * @private |
| 137 | | - */ |
| 138 | | - function queryLimit() { |
| 139 | | - return $this->db->limitResult( '', $this->limit, $this->offset ); |
| 140 | | - } |
| 141 | | - |
| 142 | | - /** |
| 143 | | - * Does not do anything for generic search engine |
| 144 | | - * subclasses may define this though |
| 145 | | - * @return string |
| 146 | | - * @private |
| 147 | | - */ |
| 148 | | - function queryRanking( $filteredTerm, $fulltext ) { |
| 149 | | - return ''; |
| 150 | | - } |
| 151 | | - |
| 152 | | - /** |
| 153 | | - * Construct the full SQL query to do the search. |
| 154 | | - * The guts shoulds be constructed in queryMain() |
| 155 | | - * @param string $filteredTerm |
| 156 | | - * @param bool $fulltext |
| 157 | | - * @private |
| 158 | | - */ |
| 159 | | - function getQuery( $filteredTerm, $fulltext ) { |
| 160 | | - return $this->queryMain( $filteredTerm, $fulltext ) . ' ' . |
| 161 | | - $this->queryRedirect() . ' ' . |
| 162 | | - $this->queryNamespaces() . ' ' . |
| 163 | | - $this->queryRanking( $filteredTerm, $fulltext ) . ' ' . |
| 164 | | - $this->queryLimit(); |
| 165 | | - } |
| 166 | | - |
| 167 | | - |
| 168 | | - /** |
| 169 | | - * Picks which field to index on, depending on what type of query. |
| 170 | | - * @param bool $fulltext |
| 171 | | - * @return string |
| 172 | | - */ |
| 173 | | - function getIndexField( $fulltext ) { |
| 174 | | - return $fulltext ? 'si_text' : 'si_title'; |
| 175 | | - } |
| 176 | | - |
| 177 | | - /** |
| 178 | | - * Get the base part of the search query. |
| 179 | | - * The actual match syntax will depend on the server |
| 180 | | - * version; MySQL 3 and MySQL 4 have different capabilities |
| 181 | | - * in their fulltext search indexes. |
| 182 | | - * |
| 183 | | - * @param string $filteredTerm |
| 184 | | - * @param bool $fulltext |
| 185 | | - * @return string |
| 186 | | - * @private |
| 187 | | - */ |
| 188 | | - function queryMain( $filteredTerm, $fulltext ) { |
| 189 | | - $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| 190 | | - $page = $this->db->tableName( 'page' ); |
| 191 | | - $searchindex = $this->db->tableName( 'searchindex' ); |
| 192 | | - return 'SELECT page_id, page_namespace, page_title ' . |
| 193 | | - "FROM $page,$searchindex " . |
| 194 | | - 'WHERE page_id=si_page AND ' . $match; |
| 195 | | - } |
| 196 | | - |
| 197 | | - /** |
| 198 | | - * Create or update the search index record for the given page. |
| 199 | | - * Title and text should be pre-processed. |
| 200 | | - * |
| 201 | | - * @param int $id |
| 202 | | - * @param string $title |
| 203 | | - * @param string $text |
| 204 | | - */ |
| 205 | | - function update( $id, $title, $text ) { |
| 206 | | - $dbw = wfGetDB( DB_MASTER ); |
| 207 | | - $dbw->replace( 'searchindex', |
| 208 | | - array( 'si_page' ), |
| 209 | | - array( |
| 210 | | - 'si_page' => $id, |
| 211 | | - 'si_title' => $title, |
| 212 | | - 'si_text' => $text |
| 213 | | - ), __METHOD__ ); |
| 214 | | - } |
| 215 | | - |
| 216 | | - /** |
| 217 | | - * Update a search index record's title only. |
| 218 | | - * Title should be pre-processed. |
| 219 | | - * |
| 220 | | - * @param int $id |
| 221 | | - * @param string $title |
| 222 | | - */ |
| 223 | | - function updateTitle( $id, $title ) { |
| 224 | | - $dbw = wfGetDB( DB_MASTER ); |
| 225 | | - |
| 226 | | - $dbw->update( 'searchindex', |
| 227 | | - array( 'si_title' => $title ), |
| 228 | | - array( 'si_page' => $id ), |
| 229 | | - __METHOD__, |
| 230 | | - array( $dbw->lowPriorityOption() ) ); |
| 231 | | - } |
| 232 | | -} |
| 233 | | - |
| 234 | | -/** |
| 235 | | - * @ingroup Search |
| 236 | | - */ |
| 237 | | -class MySQLSearchResultSet extends SearchResultSet { |
| 238 | | - function MySQLSearchResultSet( $resultSet, $terms ) { |
| 239 | | - $this->mResultSet = $resultSet; |
| 240 | | - $this->mTerms = $terms; |
| 241 | | - } |
| 242 | | - |
| 243 | | - function termMatches() { |
| 244 | | - return $this->mTerms; |
| 245 | | - } |
| 246 | | - |
| 247 | | - function numRows() { |
| 248 | | - return $this->mResultSet->numRows(); |
| 249 | | - } |
| 250 | | - |
| 251 | | - function next() { |
| 252 | | - $row = $this->mResultSet->fetchObject(); |
| 253 | | - if( $row === false ) { |
| 254 | | - return false; |
| 255 | | - } else { |
| 256 | | - return new SearchResult( $row ); |
| 257 | | - } |
| 258 | | - } |
| 259 | | - |
| 260 | | - function free() { |
| 261 | | - $this->mResultSet->free(); |
| 262 | | - } |
| 263 | | -} |
| Index: trunk/phase3/includes/search/MySQL4.php |
| — | — | @@ -0,0 +1,34 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | + |
| | 21 | +/** |
| | 22 | + * @file |
| | 23 | + * @ingroup Search |
| | 24 | + */ |
| | 25 | + |
| | 26 | +/** |
| | 27 | + * Search engine hook for MySQL 4+ |
| | 28 | + * This class retained for backwards compatibility... |
| | 29 | + * The meat's been moved to SearchMySQL, since the 3.x variety is gone. |
| | 30 | + * @ingroup Search |
| | 31 | + * @deprecated |
| | 32 | + */ |
| | 33 | +class SearchMySQL4 extends SearchMySQL { |
| | 34 | + /* whee */ |
| | 35 | +} |
| Property changes on: trunk/phase3/includes/search/MySQL4.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 36 | + native |
| Index: trunk/phase3/includes/search/Tsearch2.php |
| — | — | @@ -0,0 +1,120 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2004 Brion Vibber <brion@pobox.com>, Domas Mituzas <domas.mituzas@gmail.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | + |
| | 21 | +/** |
| | 22 | + * Search engine hook for PostgreSQL / Tsearch2 |
| | 23 | + * @file |
| | 24 | + * @ingroup Search |
| | 25 | + */ |
| | 26 | + |
| | 27 | +/** |
| | 28 | + * @todo document |
| | 29 | + * @ingroup Search |
| | 30 | + */ |
| | 31 | +class SearchTsearch2 extends SearchEngine { |
| | 32 | + var $strictMatching = false; |
| | 33 | + |
| | 34 | + function __construct( $db ) { |
| | 35 | + $this->db = $db; |
| | 36 | + $this->mRanking = true; |
| | 37 | + } |
| | 38 | + |
| | 39 | + function getIndexField( $fulltext ) { |
| | 40 | + return $fulltext ? 'si_text' : 'si_title'; |
| | 41 | + } |
| | 42 | + |
| | 43 | + function parseQuery( $filteredText, $fulltext ) { |
| | 44 | + global $wgContLang; |
| | 45 | + $lc = SearchEngine::legalSearchChars(); |
| | 46 | + $searchon = ''; |
| | 47 | + $this->searchTerms = array(); |
| | 48 | + |
| | 49 | + # FIXME: This doesn't handle parenthetical expressions. |
| | 50 | + $m = array(); |
| | 51 | + if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| | 52 | + $filteredText, $m, PREG_SET_ORDER ) ) { |
| | 53 | + foreach( $m as $terms ) { |
| | 54 | + if( $searchon !== '' ) $searchon .= ' '; |
| | 55 | + if( $this->strictMatching && ($terms[1] == '') ) { |
| | 56 | + $terms[1] = '+'; |
| | 57 | + } |
| | 58 | + $searchon .= $terms[1] . $wgContLang->stripForSearch( $terms[2] ); |
| | 59 | + if( !empty( $terms[3] ) ) { |
| | 60 | + $regexp = preg_quote( $terms[3], '/' ); |
| | 61 | + if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+"; |
| | 62 | + } else { |
| | 63 | + $regexp = preg_quote( str_replace( '"', '', $terms[2] ), '/' ); |
| | 64 | + } |
| | 65 | + $this->searchTerms[] = $regexp; |
| | 66 | + } |
| | 67 | + wfDebug( "Would search with '$searchon'\n" ); |
| | 68 | + wfDebug( 'Match with /\b' . implode( '\b|\b', $this->searchTerms ) . "\b/\n" ); |
| | 69 | + } else { |
| | 70 | + wfDebug( "Can't understand search query '{$this->filteredText}'\n" ); |
| | 71 | + } |
| | 72 | + |
| | 73 | + $searchon = preg_replace( '/(\s+)/', '&', $searchon ); |
| | 74 | + $searchon = $this->db->strencode( $searchon ); |
| | 75 | + return $searchon; |
| | 76 | + } |
| | 77 | + |
| | 78 | + function queryRanking( $filteredTerm, $fulltext ) { |
| | 79 | + $field = $this->getIndexField( $fulltext ); |
| | 80 | + $searchon = $this->parseQuery( $filteredTerm, $fulltext ); |
| | 81 | + if ($this->mRanking) |
| | 82 | + return " ORDER BY rank($field,to_tsquery('$searchon')) DESC"; |
| | 83 | + else |
| | 84 | + return ""; |
| | 85 | + } |
| | 86 | + |
| | 87 | + |
| | 88 | + function queryMain( $filteredTerm, $fulltext ) { |
| | 89 | + $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| | 90 | + $field = $this->getIndexField( $fulltext ); |
| | 91 | + $cur = $this->db->tableName( 'cur' ); |
| | 92 | + $searchindex = $this->db->tableName( 'searchindex' ); |
| | 93 | + return 'SELECT cur_id, cur_namespace, cur_title, cur_text ' . |
| | 94 | + "FROM $cur,$searchindex " . |
| | 95 | + 'WHERE cur_id=si_page AND ' . |
| | 96 | + " $field @@ to_tsquery ('$match') " ; |
| | 97 | + } |
| | 98 | + |
| | 99 | + function update( $id, $title, $text ) { |
| | 100 | + $dbw = wfGetDB( DB_MASTER ); |
| | 101 | + $searchindex = $dbw->tableName( 'searchindex' ); |
| | 102 | + $sql = "DELETE FROM $searchindex WHERE si_page={$id}"; |
| | 103 | + $dbw->query( $sql, __METHOD__ ); |
| | 104 | + $sql = "INSERT INTO $searchindex (si_page,si_title,si_text) ". |
| | 105 | + " VALUES ( $id, to_tsvector('". |
| | 106 | + $dbw->strencode($title). |
| | 107 | + "'),to_tsvector('". |
| | 108 | + $dbw->strencode( $text)."')) "; |
| | 109 | + $dbw->query($sql, __METHOD__ ); |
| | 110 | + } |
| | 111 | + |
| | 112 | + function updateTitle($id,$title) { |
| | 113 | + $dbw = wfGetDB(DB_MASTER); |
| | 114 | + $searchindex = $dbw->tableName( 'searchindex' ); |
| | 115 | + $sql = "UPDATE $searchindex SET si_title=to_tsvector('" . |
| | 116 | + $dbw->strencode( $title ) . |
| | 117 | + "') WHERE si_page={$id}"; |
| | 118 | + |
| | 119 | + $dbw->query( $sql, __METHOD__ ); |
| | 120 | + } |
| | 121 | +} |
| Property changes on: trunk/phase3/includes/search/Tsearch2.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 122 | + native |
| Index: trunk/phase3/includes/search/Update.php |
| — | — | @@ -0,0 +1,113 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | + * See deferred.txt |
| | 5 | + * @ingroup Search |
| | 6 | + */ |
| | 7 | +class SearchUpdate { |
| | 8 | + |
| | 9 | + /* private */ var $mId = 0, $mNamespace, $mTitle, $mText; |
| | 10 | + /* private */ var $mTitleWords; |
| | 11 | + |
| | 12 | + function SearchUpdate( $id, $title, $text = false ) { |
| | 13 | + $nt = Title::newFromText( $title ); |
| | 14 | + if( $nt ) { |
| | 15 | + $this->mId = $id; |
| | 16 | + $this->mText = $text; |
| | 17 | + |
| | 18 | + $this->mNamespace = $nt->getNamespace(); |
| | 19 | + $this->mTitle = $nt->getText(); # Discard namespace |
| | 20 | + |
| | 21 | + $this->mTitleWords = $this->mTextWords = array(); |
| | 22 | + } else { |
| | 23 | + wfDebug( "SearchUpdate object created with invalid title '$title'\n" ); |
| | 24 | + } |
| | 25 | + } |
| | 26 | + |
| | 27 | + function doUpdate() { |
| | 28 | + global $wgContLang, $wgDisableSearchUpdate; |
| | 29 | + |
| | 30 | + if( $wgDisableSearchUpdate || !$this->mId ) { |
| | 31 | + return false; |
| | 32 | + } |
| | 33 | + $fname = 'SearchUpdate::doUpdate'; |
| | 34 | + wfProfileIn( $fname ); |
| | 35 | + |
| | 36 | + $search = SearchEngine::create(); |
| | 37 | + $lc = SearchEngine::legalSearchChars() . '&#;'; |
| | 38 | + |
| | 39 | + if( $this->mText === false ) { |
| | 40 | + $search->updateTitle($this->mId, |
| | 41 | + Title::indexTitle( $this->mNamespace, $this->mTitle )); |
| | 42 | + wfProfileOut( $fname ); |
| | 43 | + return; |
| | 44 | + } |
| | 45 | + |
| | 46 | + # Language-specific strip/conversion |
| | 47 | + $text = $wgContLang->stripForSearch( $this->mText ); |
| | 48 | + |
| | 49 | + wfProfileIn( $fname.'-regexps' ); |
| | 50 | + $text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/", |
| | 51 | + ' ', strtolower( " " . $text /*$this->mText*/ . " " ) ); # Strip HTML markup |
| | 52 | + $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD", |
| | 53 | + "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings |
| | 54 | + |
| | 55 | + # Strip external URLs |
| | 56 | + $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\xA0-\\xFF"; |
| | 57 | + $protos = "http|https|ftp|mailto|news|gopher"; |
| | 58 | + $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/"; |
| | 59 | + $text = preg_replace( $pat, "\\1 \\3", $text ); |
| | 60 | + |
| | 61 | + $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/"; |
| | 62 | + $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/"; |
| | 63 | + $text = preg_replace( $p1, "\\1 ", $text ); |
| | 64 | + $text = preg_replace( $p2, "\\1 \\3 ", $text ); |
| | 65 | + |
| | 66 | + # Internal image links |
| | 67 | + $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i"; |
| | 68 | + $text = preg_replace( $pat2, " \\1 \\3", $text ); |
| | 69 | + |
| | 70 | + $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/", |
| | 71 | + "\\1\\2 \\2\\3", $text ); # Handle [[game]]s |
| | 72 | + |
| | 73 | + # Strip all remaining non-search characters |
| | 74 | + $text = preg_replace( "/[^{$lc}]+/", " ", $text ); |
| | 75 | + |
| | 76 | + # Handle 's, s' |
| | 77 | + # |
| | 78 | + # $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text ); |
| | 79 | + # $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text ); |
| | 80 | + # |
| | 81 | + # These tail-anchored regexps are insanely slow. The worst case comes |
| | 82 | + # when Japanese or Chinese text (ie, no word spacing) is written on |
| | 83 | + # a wiki configured for Western UTF-8 mode. The Unicode characters are |
| | 84 | + # expanded to hex codes and the "words" are very long paragraph-length |
| | 85 | + # monstrosities. On a large page the above regexps may take over 20 |
| | 86 | + # seconds *each* on a 1GHz-level processor. |
| | 87 | + # |
| | 88 | + # Following are reversed versions which are consistently fast |
| | 89 | + # (about 3 milliseconds on 1GHz-level processor). |
| | 90 | + # |
| | 91 | + $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) ); |
| | 92 | + $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) ); |
| | 93 | + |
| | 94 | + # Strip wiki '' and ''' |
| | 95 | + $text = preg_replace( "/''[']*/", " ", $text ); |
| | 96 | + wfProfileOut( "$fname-regexps" ); |
| | 97 | + |
| | 98 | + wfRunHooks( 'SearchUpdate', array( $this->mId, $this->mNamespace, $this->mTitle, &$text ) ); |
| | 99 | + |
| | 100 | + # Perform the actual update |
| | 101 | + $search->update($this->mId, Title::indexTitle( $this->mNamespace, $this->mTitle ), |
| | 102 | + $text); |
| | 103 | + |
| | 104 | + wfProfileOut( $fname ); |
| | 105 | + } |
| | 106 | +} |
| | 107 | + |
| | 108 | +/** |
| | 109 | + * Placeholder class |
| | 110 | + * @ingroup Search |
| | 111 | + */ |
| | 112 | +class SearchUpdateMyISAM extends SearchUpdate { |
| | 113 | + # Inherits everything |
| | 114 | +} |
| Property changes on: trunk/phase3/includes/search/Update.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 115 | + native |
| Index: trunk/phase3/includes/search/Engine.php |
| — | — | @@ -0,0 +1,1154 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | + * @defgroup Search Search |
| | 5 | + * |
| | 6 | + * @file |
| | 7 | + * @ingroup Search |
| | 8 | + */ |
| | 9 | + |
| | 10 | +/** |
| | 11 | + * Contain a class for special pages |
| | 12 | + * @ingroup Search |
| | 13 | + */ |
| | 14 | +class SearchEngine { |
| | 15 | + var $limit = 10; |
| | 16 | + var $offset = 0; |
| | 17 | + var $searchTerms = array(); |
| | 18 | + var $namespaces = array( NS_MAIN ); |
| | 19 | + var $showRedirects = false; |
| | 20 | + |
| | 21 | + /** |
| | 22 | + * Perform a full text search query and return a result set. |
| | 23 | + * If title searches are not supported or disabled, return null. |
| | 24 | + * |
| | 25 | + * @param string $term - Raw search term |
| | 26 | + * @return SearchResultSet |
| | 27 | + * @access public |
| | 28 | + * @abstract |
| | 29 | + */ |
| | 30 | + function searchText( $term ) { |
| | 31 | + return null; |
| | 32 | + } |
| | 33 | + |
| | 34 | + /** |
| | 35 | + * Perform a title-only search query and return a result set. |
| | 36 | + * If title searches are not supported or disabled, return null. |
| | 37 | + * |
| | 38 | + * @param string $term - Raw search term |
| | 39 | + * @return SearchResultSet |
| | 40 | + * @access public |
| | 41 | + * @abstract |
| | 42 | + */ |
| | 43 | + function searchTitle( $term ) { |
| | 44 | + return null; |
| | 45 | + } |
| | 46 | + |
| | 47 | + /** |
| | 48 | + * If an exact title match can be find, or a very slightly close match, |
| | 49 | + * return the title. If no match, returns NULL. |
| | 50 | + * |
| | 51 | + * @param string $term |
| | 52 | + * @return Title |
| | 53 | + */ |
| | 54 | + public static function getNearMatch( $searchterm ) { |
| | 55 | + global $wgContLang; |
| | 56 | + |
| | 57 | + $allSearchTerms = array($searchterm); |
| | 58 | + |
| | 59 | + if($wgContLang->hasVariants()){ |
| | 60 | + $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm)); |
| | 61 | + } |
| | 62 | + |
| | 63 | + foreach($allSearchTerms as $term){ |
| | 64 | + |
| | 65 | + # Exact match? No need to look further. |
| | 66 | + $title = Title::newFromText( $term ); |
| | 67 | + if (is_null($title)) |
| | 68 | + return NULL; |
| | 69 | + |
| | 70 | + if ( $title->getNamespace() == NS_SPECIAL || $title->isExternal() |
| | 71 | + || $title->exists() ) { |
| | 72 | + return $title; |
| | 73 | + } |
| | 74 | + |
| | 75 | + # Now try all lower case (i.e. first letter capitalized) |
| | 76 | + # |
| | 77 | + $title = Title::newFromText( $wgContLang->lc( $term ) ); |
| | 78 | + if ( $title && $title->exists() ) { |
| | 79 | + return $title; |
| | 80 | + } |
| | 81 | + |
| | 82 | + # Now try capitalized string |
| | 83 | + # |
| | 84 | + $title = Title::newFromText( $wgContLang->ucwords( $term ) ); |
| | 85 | + if ( $title && $title->exists() ) { |
| | 86 | + return $title; |
| | 87 | + } |
| | 88 | + |
| | 89 | + # Now try all upper case |
| | 90 | + # |
| | 91 | + $title = Title::newFromText( $wgContLang->uc( $term ) ); |
| | 92 | + if ( $title && $title->exists() ) { |
| | 93 | + return $title; |
| | 94 | + } |
| | 95 | + |
| | 96 | + # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc |
| | 97 | + $title = Title::newFromText( $wgContLang->ucwordbreaks($term) ); |
| | 98 | + if ( $title && $title->exists() ) { |
| | 99 | + return $title; |
| | 100 | + } |
| | 101 | + |
| | 102 | + global $wgCapitalLinks, $wgContLang; |
| | 103 | + if( !$wgCapitalLinks ) { |
| | 104 | + // Catch differs-by-first-letter-case-only |
| | 105 | + $title = Title::newFromText( $wgContLang->ucfirst( $term ) ); |
| | 106 | + if ( $title && $title->exists() ) { |
| | 107 | + return $title; |
| | 108 | + } |
| | 109 | + $title = Title::newFromText( $wgContLang->lcfirst( $term ) ); |
| | 110 | + if ( $title && $title->exists() ) { |
| | 111 | + return $title; |
| | 112 | + } |
| | 113 | + } |
| | 114 | + |
| | 115 | + // Give hooks a chance at better match variants |
| | 116 | + $title = null; |
| | 117 | + if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) { |
| | 118 | + return $title; |
| | 119 | + } |
| | 120 | + } |
| | 121 | + |
| | 122 | + $title = Title::newFromText( $searchterm ); |
| | 123 | + |
| | 124 | + # Entering an IP address goes to the contributions page |
| | 125 | + if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) ) |
| | 126 | + || User::isIP( trim( $searchterm ) ) ) { |
| | 127 | + return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() ); |
| | 128 | + } |
| | 129 | + |
| | 130 | + |
| | 131 | + # Entering a user goes to the user page whether it's there or not |
| | 132 | + if ( $title->getNamespace() == NS_USER ) { |
| | 133 | + return $title; |
| | 134 | + } |
| | 135 | + |
| | 136 | + # Go to images that exist even if there's no local page. |
| | 137 | + # There may have been a funny upload, or it may be on a shared |
| | 138 | + # file repository such as Wikimedia Commons. |
| | 139 | + if( $title->getNamespace() == NS_IMAGE ) { |
| | 140 | + $image = wfFindFile( $title ); |
| | 141 | + if( $image ) { |
| | 142 | + return $title; |
| | 143 | + } |
| | 144 | + } |
| | 145 | + |
| | 146 | + # MediaWiki namespace? Page may be "implied" if not customized. |
| | 147 | + # Just return it, with caps forced as the message system likes it. |
| | 148 | + if( $title->getNamespace() == NS_MEDIAWIKI ) { |
| | 149 | + return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) ); |
| | 150 | + } |
| | 151 | + |
| | 152 | + # Quoted term? Try without the quotes... |
| | 153 | + $matches = array(); |
| | 154 | + if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) { |
| | 155 | + return SearchEngine::getNearMatch( $matches[1] ); |
| | 156 | + } |
| | 157 | + |
| | 158 | + return NULL; |
| | 159 | + } |
| | 160 | + |
| | 161 | + public static function legalSearchChars() { |
| | 162 | + return "A-Za-z_'0-9\\x80-\\xFF\\-"; |
| | 163 | + } |
| | 164 | + |
| | 165 | + /** |
| | 166 | + * Set the maximum number of results to return |
| | 167 | + * and how many to skip before returning the first. |
| | 168 | + * |
| | 169 | + * @param int $limit |
| | 170 | + * @param int $offset |
| | 171 | + * @access public |
| | 172 | + */ |
| | 173 | + function setLimitOffset( $limit, $offset = 0 ) { |
| | 174 | + $this->limit = intval( $limit ); |
| | 175 | + $this->offset = intval( $offset ); |
| | 176 | + } |
| | 177 | + |
| | 178 | + /** |
| | 179 | + * Set which namespaces the search should include. |
| | 180 | + * Give an array of namespace index numbers. |
| | 181 | + * |
| | 182 | + * @param array $namespaces |
| | 183 | + * @access public |
| | 184 | + */ |
| | 185 | + function setNamespaces( $namespaces ) { |
| | 186 | + $this->namespaces = $namespaces; |
| | 187 | + } |
| | 188 | + |
| | 189 | + /** |
| | 190 | + * Parse some common prefixes: all (search everything) |
| | 191 | + * or namespace names |
| | 192 | + * |
| | 193 | + * @param string $query |
| | 194 | + */ |
| | 195 | + function replacePrefixes( $query ){ |
| | 196 | + global $wgContLang; |
| | 197 | + |
| | 198 | + if( strpos($query,':') === false ) |
| | 199 | + return $query; // nothing to do |
| | 200 | + |
| | 201 | + $parsed = $query; |
| | 202 | + $allkeyword = wfMsgForContent('searchall').":"; |
| | 203 | + if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){ |
| | 204 | + $this->namespaces = null; |
| | 205 | + $parsed = substr($query,strlen($allkeyword)); |
| | 206 | + } else if( strpos($query,':') !== false ) { |
| | 207 | + $prefix = substr($query,0,strpos($query,':')); |
| | 208 | + $index = $wgContLang->getNsIndex($prefix); |
| | 209 | + if($index !== false){ |
| | 210 | + $this->namespaces = array($index); |
| | 211 | + $parsed = substr($query,strlen($prefix)+1); |
| | 212 | + } |
| | 213 | + } |
| | 214 | + if(trim($parsed) == '') |
| | 215 | + return $query; // prefix was the whole query |
| | 216 | + |
| | 217 | + return $parsed; |
| | 218 | + } |
| | 219 | + |
| | 220 | + /** |
| | 221 | + * Make a list of searchable namespaces and their canonical names. |
| | 222 | + * @return array |
| | 223 | + */ |
| | 224 | + public static function searchableNamespaces() { |
| | 225 | + global $wgContLang; |
| | 226 | + $arr = array(); |
| | 227 | + foreach( $wgContLang->getNamespaces() as $ns => $name ) { |
| | 228 | + if( $ns >= NS_MAIN ) { |
| | 229 | + $arr[$ns] = $name; |
| | 230 | + } |
| | 231 | + } |
| | 232 | + return $arr; |
| | 233 | + } |
| | 234 | + |
| | 235 | + /** |
| | 236 | + * Extract default namespaces to search from the given user's |
| | 237 | + * settings, returning a list of index numbers. |
| | 238 | + * |
| | 239 | + * @param User $user |
| | 240 | + * @return array |
| | 241 | + * @static |
| | 242 | + */ |
| | 243 | + public static function userNamespaces( &$user ) { |
| | 244 | + $arr = array(); |
| | 245 | + foreach( SearchEngine::searchableNamespaces() as $ns => $name ) { |
| | 246 | + if( $user->getOption( 'searchNs' . $ns ) ) { |
| | 247 | + $arr[] = $ns; |
| | 248 | + } |
| | 249 | + } |
| | 250 | + return $arr; |
| | 251 | + } |
| | 252 | + |
| | 253 | + /** |
| | 254 | + * Find snippet highlight settings for a given user |
| | 255 | + * |
| | 256 | + * @param User $user |
| | 257 | + * @return array contextlines, contextchars |
| | 258 | + * @static |
| | 259 | + */ |
| | 260 | + public static function userHighlightPrefs( &$user ){ |
| | 261 | + //$contextlines = $user->getOption( 'contextlines', 5 ); |
| | 262 | + //$contextchars = $user->getOption( 'contextchars', 50 ); |
| | 263 | + $contextlines = 2; // Hardcode this. Old defaults sucked. :) |
| | 264 | + $contextchars = 75; // same as above.... :P |
| | 265 | + return array($contextlines, $contextchars); |
| | 266 | + } |
| | 267 | + |
| | 268 | + /** |
| | 269 | + * An array of namespaces indexes to be searched by default |
| | 270 | + * |
| | 271 | + * @return array |
| | 272 | + * @static |
| | 273 | + */ |
| | 274 | + public static function defaultNamespaces(){ |
| | 275 | + global $wgNamespacesToBeSearchedDefault; |
| | 276 | + |
| | 277 | + return array_keys($wgNamespacesToBeSearchedDefault, true); |
| | 278 | + } |
| | 279 | + |
| | 280 | + /** |
| | 281 | + * Return a 'cleaned up' search string |
| | 282 | + * |
| | 283 | + * @return string |
| | 284 | + * @access public |
| | 285 | + */ |
| | 286 | + function filter( $text ) { |
| | 287 | + $lc = $this->legalSearchChars(); |
| | 288 | + return trim( preg_replace( "/[^{$lc}]/", " ", $text ) ); |
| | 289 | + } |
| | 290 | + /** |
| | 291 | + * Load up the appropriate search engine class for the currently |
| | 292 | + * active database backend, and return a configured instance. |
| | 293 | + * |
| | 294 | + * @return SearchEngine |
| | 295 | + */ |
| | 296 | + public static function create() { |
| | 297 | + global $wgDBtype, $wgSearchType; |
| | 298 | + if( $wgSearchType ) { |
| | 299 | + $class = $wgSearchType; |
| | 300 | + } elseif( $wgDBtype == 'mysql' ) { |
| | 301 | + $class = 'SearchMySQL'; |
| | 302 | + } else if ( $wgDBtype == 'postgres' ) { |
| | 303 | + $class = 'SearchPostgres'; |
| | 304 | + } else if ( $wgDBtype == 'oracle' ) { |
| | 305 | + $class = 'SearchOracle'; |
| | 306 | + } else { |
| | 307 | + $class = 'SearchEngineDummy'; |
| | 308 | + } |
| | 309 | + $search = new $class( wfGetDB( DB_SLAVE ) ); |
| | 310 | + $search->setLimitOffset(0,0); |
| | 311 | + return $search; |
| | 312 | + } |
| | 313 | + |
| | 314 | + /** |
| | 315 | + * Create or update the search index record for the given page. |
| | 316 | + * Title and text should be pre-processed. |
| | 317 | + * |
| | 318 | + * @param int $id |
| | 319 | + * @param string $title |
| | 320 | + * @param string $text |
| | 321 | + * @abstract |
| | 322 | + */ |
| | 323 | + function update( $id, $title, $text ) { |
| | 324 | + // no-op |
| | 325 | + } |
| | 326 | + |
| | 327 | + /** |
| | 328 | + * Update a search index record's title only. |
| | 329 | + * Title should be pre-processed. |
| | 330 | + * |
| | 331 | + * @param int $id |
| | 332 | + * @param string $title |
| | 333 | + * @abstract |
| | 334 | + */ |
| | 335 | + function updateTitle( $id, $title ) { |
| | 336 | + // no-op |
| | 337 | + } |
| | 338 | + |
| | 339 | + /** |
| | 340 | + * Get OpenSearch suggestion template |
| | 341 | + * |
| | 342 | + * @return string |
| | 343 | + * @static |
| | 344 | + */ |
| | 345 | + public static function getOpenSearchTemplate() { |
| | 346 | + global $wgOpenSearchTemplate, $wgServer, $wgScriptPath; |
| | 347 | + if($wgOpenSearchTemplate) |
| | 348 | + return $wgOpenSearchTemplate; |
| | 349 | + else{ |
| | 350 | + $ns = implode(',',SearchEngine::defaultNamespaces()); |
| | 351 | + if(!$ns) $ns = "0"; |
| | 352 | + return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns; |
| | 353 | + } |
| | 354 | + } |
| | 355 | + |
| | 356 | + /** |
| | 357 | + * Get internal MediaWiki Suggest template |
| | 358 | + * |
| | 359 | + * @return string |
| | 360 | + * @static |
| | 361 | + */ |
| | 362 | + public static function getMWSuggestTemplate() { |
| | 363 | + global $wgMWSuggestTemplate, $wgServer, $wgScriptPath; |
| | 364 | + if($wgMWSuggestTemplate) |
| | 365 | + return $wgMWSuggestTemplate; |
| | 366 | + else |
| | 367 | + return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}'; |
| | 368 | + } |
| | 369 | +} |
| | 370 | + |
| | 371 | +/** |
| | 372 | + * @ingroup Search |
| | 373 | + */ |
| | 374 | +class SearchResultSet { |
| | 375 | + /** |
| | 376 | + * Fetch an array of regular expression fragments for matching |
| | 377 | + * the search terms as parsed by this engine in a text extract. |
| | 378 | + * |
| | 379 | + * @return array |
| | 380 | + * @access public |
| | 381 | + * @abstract |
| | 382 | + */ |
| | 383 | + function termMatches() { |
| | 384 | + return array(); |
| | 385 | + } |
| | 386 | + |
| | 387 | + function numRows() { |
| | 388 | + return 0; |
| | 389 | + } |
| | 390 | + |
| | 391 | + /** |
| | 392 | + * Return true if results are included in this result set. |
| | 393 | + * @return bool |
| | 394 | + * @abstract |
| | 395 | + */ |
| | 396 | + function hasResults() { |
| | 397 | + return false; |
| | 398 | + } |
| | 399 | + |
| | 400 | + /** |
| | 401 | + * Some search modes return a total hit count for the query |
| | 402 | + * in the entire article database. This may include pages |
| | 403 | + * in namespaces that would not be matched on the given |
| | 404 | + * settings. |
| | 405 | + * |
| | 406 | + * Return null if no total hits number is supported. |
| | 407 | + * |
| | 408 | + * @return int |
| | 409 | + * @access public |
| | 410 | + */ |
| | 411 | + function getTotalHits() { |
| | 412 | + return null; |
| | 413 | + } |
| | 414 | + |
| | 415 | + /** |
| | 416 | + * Some search modes return a suggested alternate term if there are |
| | 417 | + * no exact hits. Returns true if there is one on this set. |
| | 418 | + * |
| | 419 | + * @return bool |
| | 420 | + * @access public |
| | 421 | + */ |
| | 422 | + function hasSuggestion() { |
| | 423 | + return false; |
| | 424 | + } |
| | 425 | + |
| | 426 | + /** |
| | 427 | + * @return string suggested query, null if none |
| | 428 | + */ |
| | 429 | + function getSuggestionQuery(){ |
| | 430 | + return null; |
| | 431 | + } |
| | 432 | + |
| | 433 | + /** |
| | 434 | + * @return string highlighted suggested query, '' if none |
| | 435 | + */ |
| | 436 | + function getSuggestionSnippet(){ |
| | 437 | + return ''; |
| | 438 | + } |
| | 439 | + |
| | 440 | + /** |
| | 441 | + * Return information about how and from where the results were fetched, |
| | 442 | + * should be useful for diagnostics and debugging |
| | 443 | + * |
| | 444 | + * @return string |
| | 445 | + */ |
| | 446 | + function getInfo() { |
| | 447 | + return null; |
| | 448 | + } |
| | 449 | + |
| | 450 | + /** |
| | 451 | + * Return a result set of hits on other (multiple) wikis associated with this one |
| | 452 | + * |
| | 453 | + * @return SearchResultSet |
| | 454 | + */ |
| | 455 | + function getInterwikiResults() { |
| | 456 | + return null; |
| | 457 | + } |
| | 458 | + |
| | 459 | + /** |
| | 460 | + * Check if there are results on other wikis |
| | 461 | + * |
| | 462 | + * @return boolean |
| | 463 | + */ |
| | 464 | + function hasInterwikiResults() { |
| | 465 | + return $this->getInterwikiResults() != null; |
| | 466 | + } |
| | 467 | + |
| | 468 | + |
| | 469 | + /** |
| | 470 | + * Fetches next search result, or false. |
| | 471 | + * @return SearchResult |
| | 472 | + * @access public |
| | 473 | + * @abstract |
| | 474 | + */ |
| | 475 | + function next() { |
| | 476 | + return false; |
| | 477 | + } |
| | 478 | + |
| | 479 | + /** |
| | 480 | + * Frees the result set, if applicable. |
| | 481 | + * @ access public |
| | 482 | + */ |
| | 483 | + function free() { |
| | 484 | + // ... |
| | 485 | + } |
| | 486 | +} |
| | 487 | + |
| | 488 | + |
| | 489 | +/** |
| | 490 | + * @ingroup Search |
| | 491 | + */ |
| | 492 | +class SearchResultTooMany { |
| | 493 | + ## Some search engines may bail out if too many matches are found |
| | 494 | +} |
| | 495 | + |
| | 496 | + |
| | 497 | +/** |
| | 498 | + * @ingroup Search |
| | 499 | + */ |
| | 500 | +class SearchResult { |
| | 501 | + var $mRevision = null; |
| | 502 | + |
| | 503 | + function SearchResult( $row ) { |
| | 504 | + $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| | 505 | + if( !is_null($this->mTitle) ) |
| | 506 | + $this->mRevision = Revision::newFromTitle( $this->mTitle ); |
| | 507 | + } |
| | 508 | + |
| | 509 | + /** |
| | 510 | + * Check if this is result points to an invalid title |
| | 511 | + * |
| | 512 | + * @return boolean |
| | 513 | + * @access public |
| | 514 | + */ |
| | 515 | + function isBrokenTitle(){ |
| | 516 | + if( is_null($this->mTitle) ) |
| | 517 | + return true; |
| | 518 | + return false; |
| | 519 | + } |
| | 520 | + |
| | 521 | + /** |
| | 522 | + * Check if target page is missing, happens when index is out of date |
| | 523 | + * |
| | 524 | + * @return boolean |
| | 525 | + * @access public |
| | 526 | + */ |
| | 527 | + function isMissingRevision(){ |
| | 528 | + if( !$this->mRevision ) |
| | 529 | + return true; |
| | 530 | + return false; |
| | 531 | + } |
| | 532 | + |
| | 533 | + /** |
| | 534 | + * @return Title |
| | 535 | + * @access public |
| | 536 | + */ |
| | 537 | + function getTitle() { |
| | 538 | + return $this->mTitle; |
| | 539 | + } |
| | 540 | + |
| | 541 | + /** |
| | 542 | + * @return double or null if not supported |
| | 543 | + */ |
| | 544 | + function getScore() { |
| | 545 | + return null; |
| | 546 | + } |
| | 547 | + |
| | 548 | + /** |
| | 549 | + * Lazy initialization of article text from DB |
| | 550 | + */ |
| | 551 | + protected function initText(){ |
| | 552 | + if( !isset($this->mText) ){ |
| | 553 | + $this->mText = $this->mRevision->getText(); |
| | 554 | + } |
| | 555 | + } |
| | 556 | + |
| | 557 | + /** |
| | 558 | + * @param array $terms terms to highlight |
| | 559 | + * @return string highlighted text snippet, null (and not '') if not supported |
| | 560 | + */ |
| | 561 | + function getTextSnippet($terms){ |
| | 562 | + global $wgUser, $wgAdvancedSearchHighlighting; |
| | 563 | + $this->initText(); |
| | 564 | + list($contextlines,$contextchars) = SearchEngine::userHighlightPrefs($wgUser); |
| | 565 | + $h = new SearchHighlighter(); |
| | 566 | + if( $wgAdvancedSearchHighlighting ) |
| | 567 | + return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars ); |
| | 568 | + else |
| | 569 | + return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars ); |
| | 570 | + } |
| | 571 | + |
| | 572 | + /** |
| | 573 | + * @param array $terms terms to highlight |
| | 574 | + * @return string highlighted title, '' if not supported |
| | 575 | + */ |
| | 576 | + function getTitleSnippet($terms){ |
| | 577 | + return ''; |
| | 578 | + } |
| | 579 | + |
| | 580 | + /** |
| | 581 | + * @param array $terms terms to highlight |
| | 582 | + * @return string highlighted redirect name (redirect to this page), '' if none or not supported |
| | 583 | + */ |
| | 584 | + function getRedirectSnippet($terms){ |
| | 585 | + return ''; |
| | 586 | + } |
| | 587 | + |
| | 588 | + /** |
| | 589 | + * @return Title object for the redirect to this page, null if none or not supported |
| | 590 | + */ |
| | 591 | + function getRedirectTitle(){ |
| | 592 | + return null; |
| | 593 | + } |
| | 594 | + |
| | 595 | + /** |
| | 596 | + * @return string highlighted relevant section name, null if none or not supported |
| | 597 | + */ |
| | 598 | + function getSectionSnippet(){ |
| | 599 | + return ''; |
| | 600 | + } |
| | 601 | + |
| | 602 | + /** |
| | 603 | + * @return Title object (pagename+fragment) for the section, null if none or not supported |
| | 604 | + */ |
| | 605 | + function getSectionTitle(){ |
| | 606 | + return null; |
| | 607 | + } |
| | 608 | + |
| | 609 | + /** |
| | 610 | + * @return string timestamp |
| | 611 | + */ |
| | 612 | + function getTimestamp(){ |
| | 613 | + return $this->mRevision->getTimestamp(); |
| | 614 | + } |
| | 615 | + |
| | 616 | + /** |
| | 617 | + * @return int number of words |
| | 618 | + */ |
| | 619 | + function getWordCount(){ |
| | 620 | + $this->initText(); |
| | 621 | + return str_word_count( $this->mText ); |
| | 622 | + } |
| | 623 | + |
| | 624 | + /** |
| | 625 | + * @return int size in bytes |
| | 626 | + */ |
| | 627 | + function getByteSize(){ |
| | 628 | + $this->initText(); |
| | 629 | + return strlen( $this->mText ); |
| | 630 | + } |
| | 631 | + |
| | 632 | + /** |
| | 633 | + * @return boolean if hit has related articles |
| | 634 | + */ |
| | 635 | + function hasRelated(){ |
| | 636 | + return false; |
| | 637 | + } |
| | 638 | + |
| | 639 | + /** |
| | 640 | + * @return interwiki prefix of the title (return iw even if title is broken) |
| | 641 | + */ |
| | 642 | + function getInterwikiPrefix(){ |
| | 643 | + return ''; |
| | 644 | + } |
| | 645 | +} |
| | 646 | + |
| | 647 | +/** |
| | 648 | + * Highlight bits of wikitext |
| | 649 | + * |
| | 650 | + * @ingroup Search |
| | 651 | + */ |
| | 652 | +class SearchHighlighter { |
| | 653 | + var $mCleanWikitext = true; |
| | 654 | + |
| | 655 | + function SearchHighlighter($cleanupWikitext = true){ |
| | 656 | + $this->mCleanWikitext = $cleanupWikitext; |
| | 657 | + } |
| | 658 | + |
| | 659 | + /** |
| | 660 | + * Default implementation of wikitext highlighting |
| | 661 | + * |
| | 662 | + * @param string $text |
| | 663 | + * @param array $terms Terms to highlight (unescaped) |
| | 664 | + * @param int $contextlines |
| | 665 | + * @param int $contextchars |
| | 666 | + * @return string |
| | 667 | + */ |
| | 668 | + public function highlightText( $text, $terms, $contextlines, $contextchars ) { |
| | 669 | + global $wgLang, $wgContLang; |
| | 670 | + global $wgSearchHighlightBoundaries; |
| | 671 | + $fname = __METHOD__; |
| | 672 | + |
| | 673 | + if($text == '') |
| | 674 | + return ''; |
| | 675 | + |
| | 676 | + // spli text into text + templates/links/tables |
| | 677 | + $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)"; |
| | 678 | + // first capture group is for detecting nested templates/links/tables/references |
| | 679 | + $endPatterns = array( |
| | 680 | + 1 => '/(\{\{)|(\}\})/', // template |
| | 681 | + 2 => '/(\[\[)|(\]\])/', // image |
| | 682 | + 3 => "/(\n\\{\\|)|(\n\\|\\})/"); // table |
| | 683 | + |
| | 684 | + // FIXME: this should prolly be a hook or something |
| | 685 | + if(function_exists('wfCite')){ |
| | 686 | + $spat .= '|(<ref>)'; // references via cite extension |
| | 687 | + $endPatterns[4] = '/(<ref>)|(<\/ref>)/'; |
| | 688 | + } |
| | 689 | + $spat .= '/'; |
| | 690 | + $textExt = array(); // text extracts |
| | 691 | + $otherExt = array(); // other extracts |
| | 692 | + wfProfileIn( "$fname-split" ); |
| | 693 | + $start = 0; |
| | 694 | + $textLen = strlen($text); |
| | 695 | + $count = 0; // sequence number to maintain ordering |
| | 696 | + while( $start < $textLen ){ |
| | 697 | + // find start of template/image/table |
| | 698 | + if( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ){ |
| | 699 | + $epat = ''; |
| | 700 | + foreach($matches as $key => $val){ |
| | 701 | + if($key > 0 && $val[1] != -1){ |
| | 702 | + if($key == 2){ |
| | 703 | + // see if this is an image link |
| | 704 | + $ns = substr($val[0],2,-1); |
| | 705 | + if( $wgContLang->getNsIndex($ns) != NS_IMAGE ) |
| | 706 | + break; |
| | 707 | + |
| | 708 | + } |
| | 709 | + $epat = $endPatterns[$key]; |
| | 710 | + $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) ); |
| | 711 | + $start = $val[1]; |
| | 712 | + break; |
| | 713 | + } |
| | 714 | + } |
| | 715 | + if( $epat ){ |
| | 716 | + // find end (and detect any nested elements) |
| | 717 | + $level = 0; |
| | 718 | + $offset = $start + 1; |
| | 719 | + $found = false; |
| | 720 | + while( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ){ |
| | 721 | + if( array_key_exists(2,$endMatches) ){ |
| | 722 | + // found end |
| | 723 | + if($level == 0){ |
| | 724 | + $len = strlen($endMatches[2][0]); |
| | 725 | + $off = $endMatches[2][1]; |
| | 726 | + $this->splitAndAdd( $otherExt, $count, |
| | 727 | + substr( $text, $start, $off + $len - $start ) ); |
| | 728 | + $start = $off + $len; |
| | 729 | + $found = true; |
| | 730 | + break; |
| | 731 | + } else{ |
| | 732 | + // end of nested element |
| | 733 | + $level -= 1; |
| | 734 | + } |
| | 735 | + } else{ |
| | 736 | + // nested |
| | 737 | + $level += 1; |
| | 738 | + } |
| | 739 | + $offset = $endMatches[0][1] + strlen($endMatches[0][0]); |
| | 740 | + } |
| | 741 | + if( ! $found ){ |
| | 742 | + // couldn't find appropriate closing tag, skip |
| | 743 | + $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen($matches[0][0]) ) ); |
| | 744 | + $start += strlen($matches[0][0]); |
| | 745 | + } |
| | 746 | + continue; |
| | 747 | + } |
| | 748 | + } |
| | 749 | + // else: add as text extract |
| | 750 | + $this->splitAndAdd( $textExt, $count, substr($text,$start) ); |
| | 751 | + break; |
| | 752 | + } |
| | 753 | + |
| | 754 | + $all = $textExt + $otherExt; // these have disjunct key sets |
| | 755 | + |
| | 756 | + wfProfileOut( "$fname-split" ); |
| | 757 | + |
| | 758 | + // prepare regexps |
| | 759 | + foreach( $terms as $index => $term ) { |
| | 760 | + $terms[$index] = preg_quote( $term, '/' ); |
| | 761 | + // manually do upper/lowercase stuff for utf-8 since PHP won't do it |
| | 762 | + if(preg_match('/[\x80-\xff]/', $term) ){ |
| | 763 | + $terms[$index] = preg_replace_callback('/./us',array($this,'caseCallback'),$terms[$index]); |
| | 764 | + } |
| | 765 | + |
| | 766 | + |
| | 767 | + } |
| | 768 | + $anyterm = implode( '|', $terms ); |
| | 769 | + $phrase = implode("$wgSearchHighlightBoundaries+", $terms ); |
| | 770 | + |
| | 771 | + // FIXME: a hack to scale contextchars, a correct solution |
| | 772 | + // would be to have contextchars actually be char and not byte |
| | 773 | + // length, and do proper utf-8 substrings and lengths everywhere, |
| | 774 | + // but PHP is making that very hard and unclean to implement :( |
| | 775 | + $scale = strlen($anyterm) / mb_strlen($anyterm); |
| | 776 | + $contextchars = intval( $contextchars * $scale ); |
| | 777 | + |
| | 778 | + $patPre = "(^|$wgSearchHighlightBoundaries)"; |
| | 779 | + $patPost = "($wgSearchHighlightBoundaries|$)"; |
| | 780 | + |
| | 781 | + $pat1 = "/(".$phrase.")/ui"; |
| | 782 | + $pat2 = "/$patPre(".$anyterm.")$patPost/ui"; |
| | 783 | + |
| | 784 | + wfProfileIn( "$fname-extract" ); |
| | 785 | + |
| | 786 | + $left = $contextlines; |
| | 787 | + |
| | 788 | + $snippets = array(); |
| | 789 | + $offsets = array(); |
| | 790 | + |
| | 791 | + // show beginning only if it contains all words |
| | 792 | + $first = 0; |
| | 793 | + $firstText = ''; |
| | 794 | + foreach($textExt as $index => $line){ |
| | 795 | + if(strlen($line)>0 && $line[0] != ';' && $line[0] != ':'){ |
| | 796 | + $firstText = $this->extract( $line, 0, $contextchars * $contextlines ); |
| | 797 | + $first = $index; |
| | 798 | + break; |
| | 799 | + } |
| | 800 | + } |
| | 801 | + if( $firstText ){ |
| | 802 | + $succ = true; |
| | 803 | + // check if first text contains all terms |
| | 804 | + foreach($terms as $term){ |
| | 805 | + if( ! preg_match("/$patPre".$term."$patPost/ui", $firstText) ){ |
| | 806 | + $succ = false; |
| | 807 | + break; |
| | 808 | + } |
| | 809 | + } |
| | 810 | + if( $succ ){ |
| | 811 | + $snippets[$first] = $firstText; |
| | 812 | + $offsets[$first] = 0; |
| | 813 | + } |
| | 814 | + } |
| | 815 | + if( ! $snippets ) { |
| | 816 | + // match whole query on text |
| | 817 | + $this->process($pat1, $textExt, $left, $contextchars, $snippets, $offsets); |
| | 818 | + // match whole query on templates/tables/images |
| | 819 | + $this->process($pat1, $otherExt, $left, $contextchars, $snippets, $offsets); |
| | 820 | + // match any words on text |
| | 821 | + $this->process($pat2, $textExt, $left, $contextchars, $snippets, $offsets); |
| | 822 | + // match any words on templates/tables/images |
| | 823 | + $this->process($pat2, $otherExt, $left, $contextchars, $snippets, $offsets); |
| | 824 | + |
| | 825 | + ksort($snippets); |
| | 826 | + } |
| | 827 | + |
| | 828 | + // add extra chars to each snippet to make snippets constant size |
| | 829 | + $extended = array(); |
| | 830 | + if( count( $snippets ) == 0){ |
| | 831 | + // couldn't find the target words, just show beginning of article |
| | 832 | + $targetchars = $contextchars * $contextlines; |
| | 833 | + $snippets[$first] = ''; |
| | 834 | + $offsets[$first] = 0; |
| | 835 | + } else{ |
| | 836 | + // if begin of the article contains the whole phrase, show only that !! |
| | 837 | + if( array_key_exists($first,$snippets) && preg_match($pat1,$snippets[$first]) |
| | 838 | + && $offsets[$first] < $contextchars * 2 ){ |
| | 839 | + $snippets = array ($first => $snippets[$first]); |
| | 840 | + } |
| | 841 | + |
| | 842 | + // calc by how much to extend existing snippets |
| | 843 | + $targetchars = intval( ($contextchars * $contextlines) / count ( $snippets ) ); |
| | 844 | + } |
| | 845 | + |
| | 846 | + foreach($snippets as $index => $line){ |
| | 847 | + $extended[$index] = $line; |
| | 848 | + $len = strlen($line); |
| | 849 | + if( $len < $targetchars - 20 ){ |
| | 850 | + // complete this line |
| | 851 | + if($len < strlen( $all[$index] )){ |
| | 852 | + $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index]+$targetchars, $offsets[$index]); |
| | 853 | + $len = strlen( $extended[$index] ); |
| | 854 | + } |
| | 855 | + |
| | 856 | + // add more lines |
| | 857 | + $add = $index + 1; |
| | 858 | + while( $len < $targetchars - 20 |
| | 859 | + && array_key_exists($add,$all) |
| | 860 | + && !array_key_exists($add,$snippets) ){ |
| | 861 | + $offsets[$add] = 0; |
| | 862 | + $tt = "\n".$this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] ); |
| | 863 | + $extended[$add] = $tt; |
| | 864 | + $len += strlen( $tt ); |
| | 865 | + $add++; |
| | 866 | + } |
| | 867 | + } |
| | 868 | + } |
| | 869 | + |
| | 870 | + //$snippets = array_map('htmlspecialchars', $extended); |
| | 871 | + $snippets = $extended; |
| | 872 | + $last = -1; |
| | 873 | + $extract = ''; |
| | 874 | + foreach($snippets as $index => $line){ |
| | 875 | + if($last == -1) |
| | 876 | + $extract .= $line; // first line |
| | 877 | + elseif($last+1 == $index && $offsets[$last]+strlen($snippets[$last]) >= strlen($all[$last])) |
| | 878 | + $extract .= " ".$line; // continous lines |
| | 879 | + else |
| | 880 | + $extract .= '<b> ... </b>' . $line; |
| | 881 | + |
| | 882 | + $last = $index; |
| | 883 | + } |
| | 884 | + if( $extract ) |
| | 885 | + $extract .= '<b> ... </b>'; |
| | 886 | + |
| | 887 | + $processed = array(); |
| | 888 | + foreach($terms as $term){ |
| | 889 | + if( ! isset($processed[$term]) ){ |
| | 890 | + $pat3 = "/$patPre(".$term.")$patPost/ui"; // highlight word |
| | 891 | + $extract = preg_replace( $pat3, |
| | 892 | + "\\1<span class='searchmatch'>\\2</span>\\3", $extract ); |
| | 893 | + $processed[$term] = true; |
| | 894 | + } |
| | 895 | + } |
| | 896 | + |
| | 897 | + wfProfileOut( "$fname-extract" ); |
| | 898 | + |
| | 899 | + return $extract; |
| | 900 | + } |
| | 901 | + |
| | 902 | + /** |
| | 903 | + * Split text into lines and add it to extracts array |
| | 904 | + * |
| | 905 | + * @param array $extracts index -> $line |
| | 906 | + * @param int $count |
| | 907 | + * @param string $text |
| | 908 | + */ |
| | 909 | + function splitAndAdd(&$extracts, &$count, $text){ |
| | 910 | + $split = explode( "\n", $this->mCleanWikitext? $this->removeWiki($text) : $text ); |
| | 911 | + foreach($split as $line){ |
| | 912 | + $tt = trim($line); |
| | 913 | + if( $tt ) |
| | 914 | + $extracts[$count++] = $tt; |
| | 915 | + } |
| | 916 | + } |
| | 917 | + |
| | 918 | + /** |
| | 919 | + * Do manual case conversion for non-ascii chars |
| | 920 | + * |
| | 921 | + * @param unknown_type $matches |
| | 922 | + */ |
| | 923 | + function caseCallback($matches){ |
| | 924 | + global $wgContLang; |
| | 925 | + if( strlen($matches[0]) > 1 ){ |
| | 926 | + return '['.$wgContLang->lc($matches[0]).$wgContLang->uc($matches[0]).']'; |
| | 927 | + } else |
| | 928 | + return $matches[0]; |
| | 929 | + } |
| | 930 | + |
| | 931 | + /** |
| | 932 | + * Extract part of the text from start to end, but by |
| | 933 | + * not chopping up words |
| | 934 | + * @param string $text |
| | 935 | + * @param int $start |
| | 936 | + * @param int $end |
| | 937 | + * @param int $posStart (out) actual start position |
| | 938 | + * @param int $posEnd (out) actual end position |
| | 939 | + * @return string |
| | 940 | + */ |
| | 941 | + function extract($text, $start, $end, &$posStart = null, &$posEnd = null ){ |
| | 942 | + global $wgContLang; |
| | 943 | + |
| | 944 | + if( $start != 0) |
| | 945 | + $start = $this->position( $text, $start, 1 ); |
| | 946 | + if( $end >= strlen($text) ) |
| | 947 | + $end = strlen($text); |
| | 948 | + else |
| | 949 | + $end = $this->position( $text, $end ); |
| | 950 | + |
| | 951 | + if(!is_null($posStart)) |
| | 952 | + $posStart = $start; |
| | 953 | + if(!is_null($posEnd)) |
| | 954 | + $posEnd = $end; |
| | 955 | + |
| | 956 | + if($end > $start) |
| | 957 | + return substr($text, $start, $end-$start); |
| | 958 | + else |
| | 959 | + return ''; |
| | 960 | + } |
| | 961 | + |
| | 962 | + /** |
| | 963 | + * Find a nonletter near a point (index) in the text |
| | 964 | + * |
| | 965 | + * @param string $text |
| | 966 | + * @param int $point |
| | 967 | + * @param int $offset to found index |
| | 968 | + * @return int nearest nonletter index, or beginning of utf8 char if none |
| | 969 | + */ |
| | 970 | + function position($text, $point, $offset=0 ){ |
| | 971 | + $tolerance = 10; |
| | 972 | + $s = max( 0, $point - $tolerance ); |
| | 973 | + $l = min( strlen($text), $point + $tolerance ) - $s; |
| | 974 | + $m = array(); |
| | 975 | + if( preg_match('/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr($text,$s,$l), $m, PREG_OFFSET_CAPTURE ) ){ |
| | 976 | + return $m[0][1] + $s + $offset; |
| | 977 | + } else{ |
| | 978 | + // check if point is on a valid first UTF8 char |
| | 979 | + $char = ord( $text[$point] ); |
| | 980 | + while( $char >= 0x80 && $char < 0xc0 ) { |
| | 981 | + // skip trailing bytes |
| | 982 | + $point++; |
| | 983 | + if($point >= strlen($text)) |
| | 984 | + return strlen($text); |
| | 985 | + $char = ord( $text[$point] ); |
| | 986 | + } |
| | 987 | + return $point; |
| | 988 | + |
| | 989 | + } |
| | 990 | + } |
| | 991 | + |
| | 992 | + /** |
| | 993 | + * Search extracts for a pattern, and return snippets |
| | 994 | + * |
| | 995 | + * @param string $pattern regexp for matching lines |
| | 996 | + * @param array $extracts extracts to search |
| | 997 | + * @param int $linesleft number of extracts to make |
| | 998 | + * @param int $contextchars length of snippet |
| | 999 | + * @param array $out map for highlighted snippets |
| | 1000 | + * @param array $offsets map of starting points of snippets |
| | 1001 | + * @protected |
| | 1002 | + */ |
| | 1003 | + function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){ |
| | 1004 | + if($linesleft == 0) |
| | 1005 | + return; // nothing to do |
| | 1006 | + foreach($extracts as $index => $line){ |
| | 1007 | + if( array_key_exists($index,$out) ) |
| | 1008 | + continue; // this line already highlighted |
| | 1009 | + |
| | 1010 | + $m = array(); |
| | 1011 | + if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) |
| | 1012 | + continue; |
| | 1013 | + |
| | 1014 | + $offset = $m[0][1]; |
| | 1015 | + $len = strlen($m[0][0]); |
| | 1016 | + if($offset + $len < $contextchars) |
| | 1017 | + $begin = 0; |
| | 1018 | + elseif( $len > $contextchars) |
| | 1019 | + $begin = $offset; |
| | 1020 | + else |
| | 1021 | + $begin = $offset + intval( ($len - $contextchars) / 2 ); |
| | 1022 | + |
| | 1023 | + $end = $begin + $contextchars; |
| | 1024 | + |
| | 1025 | + $posBegin = $begin; |
| | 1026 | + // basic snippet from this line |
| | 1027 | + $out[$index] = $this->extract($line,$begin,$end,$posBegin); |
| | 1028 | + $offsets[$index] = $posBegin; |
| | 1029 | + $linesleft--; |
| | 1030 | + if($linesleft == 0) |
| | 1031 | + return; |
| | 1032 | + } |
| | 1033 | + } |
| | 1034 | + |
| | 1035 | + /** |
| | 1036 | + * Basic wikitext removal |
| | 1037 | + * @protected |
| | 1038 | + */ |
| | 1039 | + function removeWiki($text) { |
| | 1040 | + $fname = __METHOD__; |
| | 1041 | + wfProfileIn( $fname ); |
| | 1042 | + |
| | 1043 | + //$text = preg_replace("/'{2,5}/", "", $text); |
| | 1044 | + //$text = preg_replace("/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text); |
| | 1045 | + //$text = preg_replace("/\[\[([^]|]+)\]\]/", "\\1", $text); |
| | 1046 | + //$text = preg_replace("/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text); |
| | 1047 | + //$text = preg_replace("/\\{\\|(.*?)\\|\\}/", "", $text); |
| | 1048 | + //$text = preg_replace("/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text); |
| | 1049 | + $text = preg_replace("/\\{\\{([^|]+?)\\}\\}/", "", $text); |
| | 1050 | + $text = preg_replace("/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text); |
| | 1051 | + $text = preg_replace("/\\[\\[([^|]+?)\\]\\]/", "\\1", $text); |
| | 1052 | + $text = preg_replace_callback("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array($this,'linkReplace'), $text); |
| | 1053 | + //$text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text); |
| | 1054 | + $text = preg_replace("/<\/?[^>]+>/", "", $text); |
| | 1055 | + $text = preg_replace("/'''''/", "", $text); |
| | 1056 | + $text = preg_replace("/('''|<\/?[iIuUbB]>)/", "", $text); |
| | 1057 | + $text = preg_replace("/''/", "", $text); |
| | 1058 | + |
| | 1059 | + wfProfileOut( $fname ); |
| | 1060 | + return $text; |
| | 1061 | + } |
| | 1062 | + |
| | 1063 | + /** |
| | 1064 | + * callback to replace [[target|caption]] kind of links, if |
| | 1065 | + * the target is category or image, leave it |
| | 1066 | + * |
| | 1067 | + * @param array $matches |
| | 1068 | + */ |
| | 1069 | + function linkReplace($matches){ |
| | 1070 | + $colon = strpos( $matches[1], ':' ); |
| | 1071 | + if( $colon === false ) |
| | 1072 | + return $matches[2]; // replace with caption |
| | 1073 | + global $wgContLang; |
| | 1074 | + $ns = substr( $matches[1], 0, $colon ); |
| | 1075 | + $index = $wgContLang->getNsIndex($ns); |
| | 1076 | + if( $index !== false && ($index == NS_IMAGE || $index == NS_CATEGORY) ) |
| | 1077 | + return $matches[0]; // return the whole thing |
| | 1078 | + else |
| | 1079 | + return $matches[2]; |
| | 1080 | + |
| | 1081 | + } |
| | 1082 | + |
| | 1083 | + /** |
| | 1084 | + * Simple & fast snippet extraction, but gives completely unrelevant |
| | 1085 | + * snippets |
| | 1086 | + * |
| | 1087 | + * @param string $text |
| | 1088 | + * @param array $terms |
| | 1089 | + * @param int $contextlines |
| | 1090 | + * @param int $contextchars |
| | 1091 | + * @return string |
| | 1092 | + */ |
| | 1093 | + public function highlightSimple( $text, $terms, $contextlines, $contextchars ) { |
| | 1094 | + global $wgLang, $wgContLang; |
| | 1095 | + $fname = __METHOD__; |
| | 1096 | + |
| | 1097 | + $lines = explode( "\n", $text ); |
| | 1098 | + |
| | 1099 | + $terms = implode( '|', $terms ); |
| | 1100 | + $terms = str_replace( '/', "\\/", $terms); |
| | 1101 | + $max = intval( $contextchars ) + 1; |
| | 1102 | + $pat1 = "/(.*)($terms)(.{0,$max})/i"; |
| | 1103 | + |
| | 1104 | + $lineno = 0; |
| | 1105 | + |
| | 1106 | + $extract = ""; |
| | 1107 | + wfProfileIn( "$fname-extract" ); |
| | 1108 | + foreach ( $lines as $line ) { |
| | 1109 | + if ( 0 == $contextlines ) { |
| | 1110 | + break; |
| | 1111 | + } |
| | 1112 | + ++$lineno; |
| | 1113 | + $m = array(); |
| | 1114 | + if ( ! preg_match( $pat1, $line, $m ) ) { |
| | 1115 | + continue; |
| | 1116 | + } |
| | 1117 | + --$contextlines; |
| | 1118 | + $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' ); |
| | 1119 | + |
| | 1120 | + if ( count( $m ) < 3 ) { |
| | 1121 | + $post = ''; |
| | 1122 | + } else { |
| | 1123 | + $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' ); |
| | 1124 | + } |
| | 1125 | + |
| | 1126 | + $found = $m[2]; |
| | 1127 | + |
| | 1128 | + $line = htmlspecialchars( $pre . $found . $post ); |
| | 1129 | + $pat2 = '/(' . $terms . ")/i"; |
| | 1130 | + $line = preg_replace( $pat2, |
| | 1131 | + "<span class='searchmatch'>\\1</span>", $line ); |
| | 1132 | + |
| | 1133 | + $extract .= "${line}\n"; |
| | 1134 | + } |
| | 1135 | + wfProfileOut( "$fname-extract" ); |
| | 1136 | + |
| | 1137 | + return $extract; |
| | 1138 | + } |
| | 1139 | + |
| | 1140 | +} |
| | 1141 | + |
| | 1142 | +/** |
| | 1143 | + * @ingroup Search |
| | 1144 | + */ |
| | 1145 | +class SearchEngineDummy { |
| | 1146 | + function search( $term ) { |
| | 1147 | + return null; |
| | 1148 | + } |
| | 1149 | + function setLimitOffset($l, $o) {} |
| | 1150 | + function legalSearchChars() {} |
| | 1151 | + function update() {} |
| | 1152 | + function setnamespaces() {} |
| | 1153 | + function searchtitle() {} |
| | 1154 | + function searchtext() {} |
| | 1155 | +} |
| Property changes on: trunk/phase3/includes/search/Engine.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 1156 | + native |
| Index: trunk/phase3/includes/search/MySQL.php |
| — | — | @@ -0,0 +1,262 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | + |
| | 21 | +/** |
| | 22 | + * @file |
| | 23 | + * @ingroup Search |
| | 24 | + */ |
| | 25 | + |
| | 26 | +/** |
| | 27 | + * Search engine hook for MySQL 4+ |
| | 28 | + * @ingroup Search |
| | 29 | + */ |
| | 30 | +class SearchMySQL extends SearchEngine { |
| | 31 | + var $strictMatching = true; |
| | 32 | + |
| | 33 | + /** @todo document */ |
| | 34 | + function __construct( $db ) { |
| | 35 | + $this->db = $db; |
| | 36 | + } |
| | 37 | + |
| | 38 | + /** @todo document */ |
| | 39 | + function parseQuery( $filteredText, $fulltext ) { |
| | 40 | + global $wgContLang; |
| | 41 | + $lc = SearchEngine::legalSearchChars(); // Minus format chars |
| | 42 | + $searchon = ''; |
| | 43 | + $this->searchTerms = array(); |
| | 44 | + |
| | 45 | + # FIXME: This doesn't handle parenthetical expressions. |
| | 46 | + $m = array(); |
| | 47 | + if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| | 48 | + $filteredText, $m, PREG_SET_ORDER ) ) { |
| | 49 | + foreach( $m as $terms ) { |
| | 50 | + if( $searchon !== '' ) $searchon .= ' '; |
| | 51 | + if( $this->strictMatching && ($terms[1] == '') ) { |
| | 52 | + $terms[1] = '+'; |
| | 53 | + } |
| | 54 | + $searchon .= $terms[1] . $wgContLang->stripForSearch( $terms[2] ); |
| | 55 | + if( !empty( $terms[3] ) ) { |
| | 56 | + // Match individual terms in result highlighting... |
| | 57 | + $regexp = preg_quote( $terms[3], '/' ); |
| | 58 | + if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+"; |
| | 59 | + } else { |
| | 60 | + // Match the quoted term in result highlighting... |
| | 61 | + $regexp = preg_quote( str_replace( '"', '', $terms[2] ), '/' ); |
| | 62 | + } |
| | 63 | + $this->searchTerms[] = $regexp; |
| | 64 | + } |
| | 65 | + wfDebug( "Would search with '$searchon'\n" ); |
| | 66 | + wfDebug( 'Match with /' . implode( '|', $this->searchTerms ) . "/\n" ); |
| | 67 | + } else { |
| | 68 | + wfDebug( "Can't understand search query '{$filteredText}'\n" ); |
| | 69 | + } |
| | 70 | + |
| | 71 | + $searchon = $this->db->strencode( $searchon ); |
| | 72 | + $field = $this->getIndexField( $fulltext ); |
| | 73 | + return " MATCH($field) AGAINST('$searchon' IN BOOLEAN MODE) "; |
| | 74 | + } |
| | 75 | + |
| | 76 | + public static function legalSearchChars() { |
| | 77 | + return "\"*" . parent::legalSearchChars(); |
| | 78 | + } |
| | 79 | + |
| | 80 | + /** |
| | 81 | + * Perform a full text search query and return a result set. |
| | 82 | + * |
| | 83 | + * @param string $term - Raw search term |
| | 84 | + * @return MySQLSearchResultSet |
| | 85 | + * @access public |
| | 86 | + */ |
| | 87 | + function searchText( $term ) { |
| | 88 | + $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) ); |
| | 89 | + return new MySQLSearchResultSet( $resultSet, $this->searchTerms ); |
| | 90 | + } |
| | 91 | + |
| | 92 | + /** |
| | 93 | + * Perform a title-only search query and return a result set. |
| | 94 | + * |
| | 95 | + * @param string $term - Raw search term |
| | 96 | + * @return MySQLSearchResultSet |
| | 97 | + * @access public |
| | 98 | + */ |
| | 99 | + function searchTitle( $term ) { |
| | 100 | + $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) ); |
| | 101 | + return new MySQLSearchResultSet( $resultSet, $this->searchTerms ); |
| | 102 | + } |
| | 103 | + |
| | 104 | + |
| | 105 | + /** |
| | 106 | + * Return a partial WHERE clause to exclude redirects, if so set |
| | 107 | + * @return string |
| | 108 | + * @private |
| | 109 | + */ |
| | 110 | + function queryRedirect() { |
| | 111 | + if( $this->showRedirects ) { |
| | 112 | + return ''; |
| | 113 | + } else { |
| | 114 | + return 'AND page_is_redirect=0'; |
| | 115 | + } |
| | 116 | + } |
| | 117 | + |
| | 118 | + /** |
| | 119 | + * Return a partial WHERE clause to limit the search to the given namespaces |
| | 120 | + * @return string |
| | 121 | + * @private |
| | 122 | + */ |
| | 123 | + function queryNamespaces() { |
| | 124 | + if( is_null($this->namespaces) ) |
| | 125 | + return ''; # search all |
| | 126 | + $namespaces = implode( ',', $this->namespaces ); |
| | 127 | + if ($namespaces == '') { |
| | 128 | + $namespaces = '0'; |
| | 129 | + } |
| | 130 | + return 'AND page_namespace IN (' . $namespaces . ')'; |
| | 131 | + } |
| | 132 | + |
| | 133 | + /** |
| | 134 | + * Return a LIMIT clause to limit results on the query. |
| | 135 | + * @return string |
| | 136 | + * @private |
| | 137 | + */ |
| | 138 | + function queryLimit() { |
| | 139 | + return $this->db->limitResult( '', $this->limit, $this->offset ); |
| | 140 | + } |
| | 141 | + |
| | 142 | + /** |
| | 143 | + * Does not do anything for generic search engine |
| | 144 | + * subclasses may define this though |
| | 145 | + * @return string |
| | 146 | + * @private |
| | 147 | + */ |
| | 148 | + function queryRanking( $filteredTerm, $fulltext ) { |
| | 149 | + return ''; |
| | 150 | + } |
| | 151 | + |
| | 152 | + /** |
| | 153 | + * Construct the full SQL query to do the search. |
| | 154 | + * The guts shoulds be constructed in queryMain() |
| | 155 | + * @param string $filteredTerm |
| | 156 | + * @param bool $fulltext |
| | 157 | + * @private |
| | 158 | + */ |
| | 159 | + function getQuery( $filteredTerm, $fulltext ) { |
| | 160 | + return $this->queryMain( $filteredTerm, $fulltext ) . ' ' . |
| | 161 | + $this->queryRedirect() . ' ' . |
| | 162 | + $this->queryNamespaces() . ' ' . |
| | 163 | + $this->queryRanking( $filteredTerm, $fulltext ) . ' ' . |
| | 164 | + $this->queryLimit(); |
| | 165 | + } |
| | 166 | + |
| | 167 | + |
| | 168 | + /** |
| | 169 | + * Picks which field to index on, depending on what type of query. |
| | 170 | + * @param bool $fulltext |
| | 171 | + * @return string |
| | 172 | + */ |
| | 173 | + function getIndexField( $fulltext ) { |
| | 174 | + return $fulltext ? 'si_text' : 'si_title'; |
| | 175 | + } |
| | 176 | + |
| | 177 | + /** |
| | 178 | + * Get the base part of the search query. |
| | 179 | + * The actual match syntax will depend on the server |
| | 180 | + * version; MySQL 3 and MySQL 4 have different capabilities |
| | 181 | + * in their fulltext search indexes. |
| | 182 | + * |
| | 183 | + * @param string $filteredTerm |
| | 184 | + * @param bool $fulltext |
| | 185 | + * @return string |
| | 186 | + * @private |
| | 187 | + */ |
| | 188 | + function queryMain( $filteredTerm, $fulltext ) { |
| | 189 | + $match = $this->parseQuery( $filteredTerm, $fulltext ); |
| | 190 | + $page = $this->db->tableName( 'page' ); |
| | 191 | + $searchindex = $this->db->tableName( 'searchindex' ); |
| | 192 | + return 'SELECT page_id, page_namespace, page_title ' . |
| | 193 | + "FROM $page,$searchindex " . |
| | 194 | + 'WHERE page_id=si_page AND ' . $match; |
| | 195 | + } |
| | 196 | + |
| | 197 | + /** |
| | 198 | + * Create or update the search index record for the given page. |
| | 199 | + * Title and text should be pre-processed. |
| | 200 | + * |
| | 201 | + * @param int $id |
| | 202 | + * @param string $title |
| | 203 | + * @param string $text |
| | 204 | + */ |
| | 205 | + function update( $id, $title, $text ) { |
| | 206 | + $dbw = wfGetDB( DB_MASTER ); |
| | 207 | + $dbw->replace( 'searchindex', |
| | 208 | + array( 'si_page' ), |
| | 209 | + array( |
| | 210 | + 'si_page' => $id, |
| | 211 | + 'si_title' => $title, |
| | 212 | + 'si_text' => $text |
| | 213 | + ), __METHOD__ ); |
| | 214 | + } |
| | 215 | + |
| | 216 | + /** |
| | 217 | + * Update a search index record's title only. |
| | 218 | + * Title should be pre-processed. |
| | 219 | + * |
| | 220 | + * @param int $id |
| | 221 | + * @param string $title |
| | 222 | + */ |
| | 223 | + function updateTitle( $id, $title ) { |
| | 224 | + $dbw = wfGetDB( DB_MASTER ); |
| | 225 | + |
| | 226 | + $dbw->update( 'searchindex', |
| | 227 | + array( 'si_title' => $title ), |
| | 228 | + array( 'si_page' => $id ), |
| | 229 | + __METHOD__, |
| | 230 | + array( $dbw->lowPriorityOption() ) ); |
| | 231 | + } |
| | 232 | +} |
| | 233 | + |
| | 234 | +/** |
| | 235 | + * @ingroup Search |
| | 236 | + */ |
| | 237 | +class MySQLSearchResultSet extends SearchResultSet { |
| | 238 | + function MySQLSearchResultSet( $resultSet, $terms ) { |
| | 239 | + $this->mResultSet = $resultSet; |
| | 240 | + $this->mTerms = $terms; |
| | 241 | + } |
| | 242 | + |
| | 243 | + function termMatches() { |
| | 244 | + return $this->mTerms; |
| | 245 | + } |
| | 246 | + |
| | 247 | + function numRows() { |
| | 248 | + return $this->mResultSet->numRows(); |
| | 249 | + } |
| | 250 | + |
| | 251 | + function next() { |
| | 252 | + $row = $this->mResultSet->fetchObject(); |
| | 253 | + if( $row === false ) { |
| | 254 | + return false; |
| | 255 | + } else { |
| | 256 | + return new SearchResult( $row ); |
| | 257 | + } |
| | 258 | + } |
| | 259 | + |
| | 260 | + function free() { |
| | 261 | + $this->mResultSet->free(); |
| | 262 | + } |
| | 263 | +} |
| Property changes on: trunk/phase3/includes/search/MySQL.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 264 | + native |
| Index: trunk/phase3/includes/search/Oracle.php |
| — | — | @@ -0,0 +1,240 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2004 Brion Vibber <brion@pobox.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | + |
| | 21 | +/** |
| | 22 | + * @file |
| | 23 | + * @ingroup Search |
| | 24 | + */ |
| | 25 | + |
| | 26 | +/** |
| | 27 | + * Search engine hook base class for Oracle (ConText). |
| | 28 | + * @ingroup Search |
| | 29 | + */ |
| | 30 | +class SearchOracle extends SearchEngine { |
| | 31 | + function __construct($db) { |
| | 32 | + $this->db = $db; |
| | 33 | + } |
| | 34 | + |
| | 35 | + /** |
| | 36 | + * Perform a full text search query and return a result set. |
| | 37 | + * |
| | 38 | + * @param string $term - Raw search term |
| | 39 | + * @return OracleSearchResultSet |
| | 40 | + * @access public |
| | 41 | + */ |
| | 42 | + function searchText( $term ) { |
| | 43 | + $resultSet = $this->db->resultObject($this->db->query($this->getQuery($this->filter($term), true))); |
| | 44 | + return new OracleSearchResultSet($resultSet, $this->searchTerms); |
| | 45 | + } |
| | 46 | + |
| | 47 | + /** |
| | 48 | + * Perform a title-only search query and return a result set. |
| | 49 | + * |
| | 50 | + * @param string $term - Raw search term |
| | 51 | + * @return ORacleSearchResultSet |
| | 52 | + * @access public |
| | 53 | + */ |
| | 54 | + function searchTitle($term) { |
| | 55 | + $resultSet = $this->db->resultObject($this->db->query($this->getQuery($this->filter($term), false))); |
| | 56 | + return new MySQLSearchResultSet($resultSet, $this->searchTerms); |
| | 57 | + } |
| | 58 | + |
| | 59 | + |
| | 60 | + /** |
| | 61 | + * Return a partial WHERE clause to exclude redirects, if so set |
| | 62 | + * @return string |
| | 63 | + * @private |
| | 64 | + */ |
| | 65 | + function queryRedirect() { |
| | 66 | + if ($this->showRedirects) { |
| | 67 | + return ''; |
| | 68 | + } else { |
| | 69 | + return 'AND page_is_redirect=0'; |
| | 70 | + } |
| | 71 | + } |
| | 72 | + |
| | 73 | + /** |
| | 74 | + * Return a partial WHERE clause to limit the search to the given namespaces |
| | 75 | + * @return string |
| | 76 | + * @private |
| | 77 | + */ |
| | 78 | + function queryNamespaces() { |
| | 79 | + if( is_null($this->namespaces) ) |
| | 80 | + return ''; |
| | 81 | + $namespaces = implode(',', $this->namespaces); |
| | 82 | + if ($namespaces == '') { |
| | 83 | + $namespaces = '0'; |
| | 84 | + } |
| | 85 | + return 'AND page_namespace IN (' . $namespaces . ')'; |
| | 86 | + } |
| | 87 | + |
| | 88 | + /** |
| | 89 | + * Return a LIMIT clause to limit results on the query. |
| | 90 | + * @return string |
| | 91 | + * @private |
| | 92 | + */ |
| | 93 | + function queryLimit($sql) { |
| | 94 | + return $this->db->limitResult($sql, $this->limit, $this->offset); |
| | 95 | + } |
| | 96 | + |
| | 97 | + /** |
| | 98 | + * Does not do anything for generic search engine |
| | 99 | + * subclasses may define this though |
| | 100 | + * @return string |
| | 101 | + * @private |
| | 102 | + */ |
| | 103 | + function queryRanking($filteredTerm, $fulltext) { |
| | 104 | + return ' ORDER BY score(1)'; |
| | 105 | + } |
| | 106 | + |
| | 107 | + /** |
| | 108 | + * Construct the full SQL query to do the search. |
| | 109 | + * The guts shoulds be constructed in queryMain() |
| | 110 | + * @param string $filteredTerm |
| | 111 | + * @param bool $fulltext |
| | 112 | + * @private |
| | 113 | + */ |
| | 114 | + function getQuery( $filteredTerm, $fulltext ) { |
| | 115 | + return $this->queryLimit($this->queryMain($filteredTerm, $fulltext) . ' ' . |
| | 116 | + $this->queryRedirect() . ' ' . |
| | 117 | + $this->queryNamespaces() . ' ' . |
| | 118 | + $this->queryRanking( $filteredTerm, $fulltext ) . ' '); |
| | 119 | + } |
| | 120 | + |
| | 121 | + |
| | 122 | + /** |
| | 123 | + * Picks which field to index on, depending on what type of query. |
| | 124 | + * @param bool $fulltext |
| | 125 | + * @return string |
| | 126 | + */ |
| | 127 | + function getIndexField($fulltext) { |
| | 128 | + return $fulltext ? 'si_text' : 'si_title'; |
| | 129 | + } |
| | 130 | + |
| | 131 | + /** |
| | 132 | + * Get the base part of the search query. |
| | 133 | + * |
| | 134 | + * @param string $filteredTerm |
| | 135 | + * @param bool $fulltext |
| | 136 | + * @return string |
| | 137 | + * @private |
| | 138 | + */ |
| | 139 | + function queryMain( $filteredTerm, $fulltext ) { |
| | 140 | + $match = $this->parseQuery($filteredTerm, $fulltext); |
| | 141 | + $page = $this->db->tableName('page'); |
| | 142 | + $searchindex = $this->db->tableName('searchindex'); |
| | 143 | + return 'SELECT page_id, page_namespace, page_title ' . |
| | 144 | + "FROM $page,$searchindex " . |
| | 145 | + 'WHERE page_id=si_page AND ' . $match; |
| | 146 | + } |
| | 147 | + |
| | 148 | + /** @todo document */ |
| | 149 | + function parseQuery($filteredText, $fulltext) { |
| | 150 | + global $wgContLang; |
| | 151 | + $lc = SearchEngine::legalSearchChars(); |
| | 152 | + $this->searchTerms = array(); |
| | 153 | + |
| | 154 | + # FIXME: This doesn't handle parenthetical expressions. |
| | 155 | + $m = array(); |
| | 156 | + $q = array(); |
| | 157 | + |
| | 158 | + if (preg_match_all('/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', |
| | 159 | + $filteredText, $m, PREG_SET_ORDER)) { |
| | 160 | + foreach($m as $terms) { |
| | 161 | + $q[] = $terms[1] . $wgContLang->stripForSearch($terms[2]); |
| | 162 | + |
| | 163 | + if (!empty($terms[3])) { |
| | 164 | + $regexp = preg_quote( $terms[3], '/' ); |
| | 165 | + if ($terms[4]) |
| | 166 | + $regexp .= "[0-9A-Za-z_]+"; |
| | 167 | + } else { |
| | 168 | + $regexp = preg_quote(str_replace('"', '', $terms[2]), '/'); |
| | 169 | + } |
| | 170 | + $this->searchTerms[] = $regexp; |
| | 171 | + } |
| | 172 | + } |
| | 173 | + |
| | 174 | + $searchon = $this->db->strencode(join(',', $q)); |
| | 175 | + $field = $this->getIndexField($fulltext); |
| | 176 | + return " CONTAINS($field, '$searchon', 1) > 0 "; |
| | 177 | + } |
| | 178 | + |
| | 179 | + /** |
| | 180 | + * Create or update the search index record for the given page. |
| | 181 | + * Title and text should be pre-processed. |
| | 182 | + * |
| | 183 | + * @param int $id |
| | 184 | + * @param string $title |
| | 185 | + * @param string $text |
| | 186 | + */ |
| | 187 | + function update($id, $title, $text) { |
| | 188 | + $dbw = wfGetDB(DB_MASTER); |
| | 189 | + $dbw->replace('searchindex', |
| | 190 | + array('si_page'), |
| | 191 | + array( |
| | 192 | + 'si_page' => $id, |
| | 193 | + 'si_title' => $title, |
| | 194 | + 'si_text' => $text |
| | 195 | + ), 'SearchOracle::update' ); |
| | 196 | + $dbw->query("CALL ctx_ddl.sync_index('si_text_idx')"); |
| | 197 | + $dbw->query("CALL ctx_ddl.sync_index('si_title_idx')"); |
| | 198 | + } |
| | 199 | + |
| | 200 | + /** |
| | 201 | + * Update a search index record's title only. |
| | 202 | + * Title should be pre-processed. |
| | 203 | + * |
| | 204 | + * @param int $id |
| | 205 | + * @param string $title |
| | 206 | + */ |
| | 207 | + function updateTitle($id, $title) { |
| | 208 | + $dbw = wfGetDB(DB_MASTER); |
| | 209 | + |
| | 210 | + $dbw->update('searchindex', |
| | 211 | + array('si_title' => $title), |
| | 212 | + array('si_page' => $id), |
| | 213 | + 'SearchOracle::updateTitle', |
| | 214 | + array()); |
| | 215 | + } |
| | 216 | +} |
| | 217 | + |
| | 218 | +/** |
| | 219 | + * @ingroup Search |
| | 220 | + */ |
| | 221 | +class OracleSearchResultSet extends SearchResultSet { |
| | 222 | + function __construct($resultSet, $terms) { |
| | 223 | + $this->mResultSet = $resultSet; |
| | 224 | + $this->mTerms = $terms; |
| | 225 | + } |
| | 226 | + |
| | 227 | + function termMatches() { |
| | 228 | + return $this->mTerms; |
| | 229 | + } |
| | 230 | + |
| | 231 | + function numRows() { |
| | 232 | + return $this->mResultSet->numRows(); |
| | 233 | + } |
| | 234 | + |
| | 235 | + function next() { |
| | 236 | + $row = $this->mResultSet->fetchObject(); |
| | 237 | + if ($row === false) |
| | 238 | + return false; |
| | 239 | + return new SearchResult($row); |
| | 240 | + } |
| | 241 | +} |
| Property changes on: trunk/phase3/includes/search/Oracle.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 242 | + native |
| Index: trunk/phase3/includes/search/Postgres.php |
| — | — | @@ -0,0 +1,255 @@ |
| | 2 | +<?php |
| | 3 | +# Copyright (C) 2006-2007 Greg Sabino Mullane <greg@turnstep.com> |
| | 4 | +# http://www.mediawiki.org/ |
| | 5 | +# |
| | 6 | +# This program is free software; you can redistribute it and/or modify |
| | 7 | +# it under the terms of the GNU General Public License as published by |
| | 8 | +# the Free Software Foundation; either version 2 of the License, or |
| | 9 | +# (at your option) any later version. |
| | 10 | +# |
| | 11 | +# This program is distributed in the hope that it will be useful, |
| | 12 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| | 13 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| | 14 | +# GNU General Public License for more details. |
| | 15 | +# |
| | 16 | +# You should have received a copy of the GNU General Public License along |
| | 17 | +# with this program; if not, write to the Free Software Foundation, Inc., |
| | 18 | +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| | 19 | +# http://www.gnu.org/copyleft/gpl.html |
| | 20 | + |
| | 21 | +/** |
| | 22 | + * @file |
| | 23 | + * @ingroup Search |
| | 24 | + */ |
| | 25 | + |
| | 26 | +/** |
| | 27 | + * Search engine hook base class for Postgres |
| | 28 | + * @ingroup Search |
| | 29 | + */ |
| | 30 | +class SearchPostgres extends SearchEngine { |
| | 31 | + |
| | 32 | + function SearchPostgres( $db ) { |
| | 33 | + $this->db = $db; |
| | 34 | + } |
| | 35 | + |
| | 36 | + /** |
| | 37 | + * Perform a full text search query via tsearch2 and return a result set. |
| | 38 | + * Currently searches a page's current title (page.page_title) and |
| | 39 | + * latest revision article text (pagecontent.old_text) |
| | 40 | + * |
| | 41 | + * @param string $term - Raw search term |
| | 42 | + * @return PostgresSearchResultSet |
| | 43 | + * @access public |
| | 44 | + */ |
| | 45 | + function searchTitle( $term ) { |
| | 46 | + $q = $this->searchQuery( $term , 'titlevector', 'page_title' ); |
| | 47 | + $olderror = error_reporting(E_ERROR); |
| | 48 | + $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); |
| | 49 | + error_reporting($olderror); |
| | 50 | + if (!$resultSet) { |
| | 51 | + // Needed for "Query requires full scan, GIN doesn't support it" |
| | 52 | + return new SearchResultTooMany(); |
| | 53 | + } |
| | 54 | + return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); |
| | 55 | + } |
| | 56 | + function searchText( $term ) { |
| | 57 | + $q = $this->searchQuery( $term, 'textvector', 'old_text' ); |
| | 58 | + $olderror = error_reporting(E_ERROR); |
| | 59 | + $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); |
| | 60 | + error_reporting($olderror); |
| | 61 | + if (!$resultSet) { |
| | 62 | + return new SearchResultTooMany(); |
| | 63 | + } |
| | 64 | + return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); |
| | 65 | + } |
| | 66 | + |
| | 67 | + |
| | 68 | + /* |
| | 69 | + * Transform the user's search string into a better form for tsearch2 |
| | 70 | + */ |
| | 71 | + function parseQuery( $term ) { |
| | 72 | + |
| | 73 | + wfDebug( "parseQuery received: $term" ); |
| | 74 | + |
| | 75 | + ## No backslashes allowed |
| | 76 | + $term = preg_replace('/\\\/', '', $term); |
| | 77 | + |
| | 78 | + ## Collapse parens into nearby words: |
| | 79 | + $term = preg_replace('/\s*\(\s*/', ' (', $term); |
| | 80 | + $term = preg_replace('/\s*\)\s*/', ') ', $term); |
| | 81 | + |
| | 82 | + ## Treat colons as word separators: |
| | 83 | + $term = preg_replace('/:/', ' ', $term); |
| | 84 | + |
| | 85 | + $searchstring = ''; |
| | 86 | + $m = array(); |
| | 87 | + if( preg_match_all('/([-!]?)(\S+)\s*/', $term, $m, PREG_SET_ORDER ) ) { |
| | 88 | + foreach( $m as $terms ) { |
| | 89 | + if (strlen($terms[1])) { |
| | 90 | + $searchstring .= ' & !'; |
| | 91 | + } |
| | 92 | + if (strtolower($terms[2]) === 'and') { |
| | 93 | + $searchstring .= ' & '; |
| | 94 | + } |
| | 95 | + else if (strtolower($terms[2]) === 'or' or $terms[2] === '|') { |
| | 96 | + $searchstring .= ' | '; |
| | 97 | + } |
| | 98 | + else if (strtolower($terms[2]) === 'not') { |
| | 99 | + $searchstring .= ' & !'; |
| | 100 | + } |
| | 101 | + else { |
| | 102 | + $searchstring .= " & $terms[2]"; |
| | 103 | + } |
| | 104 | + } |
| | 105 | + } |
| | 106 | + |
| | 107 | + ## Strip out leading junk |
| | 108 | + $searchstring = preg_replace('/^[\s\&\|]+/', '', $searchstring); |
| | 109 | + |
| | 110 | + ## Remove any doubled-up operators |
| | 111 | + $searchstring = preg_replace('/([\!\&\|]) +(?:[\&\|] +)+/', "$1 ", $searchstring); |
| | 112 | + |
| | 113 | + ## Remove any non-spaced operators (e.g. "Zounds!") |
| | 114 | + $searchstring = preg_replace('/([^ ])[\!\&\|]/', "$1", $searchstring); |
| | 115 | + |
| | 116 | + ## Remove any trailing whitespace or operators |
| | 117 | + $searchstring = preg_replace('/[\s\!\&\|]+$/', '', $searchstring); |
| | 118 | + |
| | 119 | + ## Remove unnecessary quotes around everything |
| | 120 | + $searchstring = preg_replace('/^[\'"](.*)[\'"]$/', "$1", $searchstring); |
| | 121 | + |
| | 122 | + ## Quote the whole thing |
| | 123 | + $searchstring = $this->db->addQuotes($searchstring); |
| | 124 | + |
| | 125 | + wfDebug( "parseQuery returned: $searchstring" ); |
| | 126 | + |
| | 127 | + return $searchstring; |
| | 128 | + |
| | 129 | + } |
| | 130 | + |
| | 131 | + /** |
| | 132 | + * Construct the full SQL query to do the search. |
| | 133 | + * @param string $filteredTerm |
| | 134 | + * @param string $fulltext |
| | 135 | + * @private |
| | 136 | + */ |
| | 137 | + function searchQuery( $term, $fulltext, $colname ) { |
| | 138 | + global $wgDBversion; |
| | 139 | + |
| | 140 | + if ( !isset( $wgDBversion ) ) { |
| | 141 | + $this->db->getServerVersion(); |
| | 142 | + $wgDBversion = $this->db->numeric_version; |
| | 143 | + } |
| | 144 | + $prefix = $wgDBversion < 8.3 ? "'default'," : ''; |
| | 145 | + |
| | 146 | + $searchstring = $this->parseQuery( $term ); |
| | 147 | + |
| | 148 | + ## We need a separate query here so gin does not complain about empty searches |
| | 149 | + $SQL = "SELECT to_tsquery($prefix $searchstring)"; |
| | 150 | + $res = $this->db->doQuery($SQL); |
| | 151 | + if (!$res) { |
| | 152 | + ## TODO: Better output (example to catch: one 'two) |
| | 153 | + die ("Sorry, that was not a valid search string. Please go back and try again"); |
| | 154 | + } |
| | 155 | + $top = pg_fetch_result($res,0,0); |
| | 156 | + |
| | 157 | + if ($top === "") { ## e.g. if only stopwords are used XXX return something better |
| | 158 | + $query = "SELECT page_id, page_namespace, page_title, 0 AS score ". |
| | 159 | + "FROM page p, revision r, pagecontent c WHERE p.page_latest = r.rev_id " . |
| | 160 | + "AND r.rev_text_id = c.old_id AND 1=0"; |
| | 161 | + } |
| | 162 | + else { |
| | 163 | + $m = array(); |
| | 164 | + if( preg_match_all("/'([^']+)'/", $top, $m, PREG_SET_ORDER ) ) { |
| | 165 | + foreach( $m as $terms ) { |
| | 166 | + $this->searchTerms[$terms[1]] = $terms[1]; |
| | 167 | + } |
| | 168 | + } |
| | 169 | + |
| | 170 | + $rankscore = $wgDBversion > 8.2 ? 5 : 1; |
| | 171 | + $rank = $wgDBversion < 8.3 ? 'rank' : 'ts_rank'; |
| | 172 | + $query = "SELECT page_id, page_namespace, page_title, ". |
| | 173 | + "$rank($fulltext, to_tsquery($prefix $searchstring), $rankscore) AS score ". |
| | 174 | + "FROM page p, revision r, pagecontent c WHERE p.page_latest = r.rev_id " . |
| | 175 | + "AND r.rev_text_id = c.old_id AND $fulltext @@ to_tsquery($prefix $searchstring)"; |
| | 176 | + } |
| | 177 | + |
| | 178 | + ## Redirects |
| | 179 | + if (! $this->showRedirects) |
| | 180 | + $query .= ' AND page_is_redirect = 0'; |
| | 181 | + |
| | 182 | + ## Namespaces - defaults to 0 |
| | 183 | + if( !is_null($this->namespaces) ){ // null -> search all |
| | 184 | + if ( count($this->namespaces) < 1) |
| | 185 | + $query .= ' AND page_namespace = 0'; |
| | 186 | + else { |
| | 187 | + $namespaces = implode( ',', $this->namespaces ); |
| | 188 | + $query .= " AND page_namespace IN ($namespaces)"; |
| | 189 | + } |
| | 190 | + } |
| | 191 | + |
| | 192 | + $query .= " ORDER BY score DESC, page_id DESC"; |
| | 193 | + |
| | 194 | + $query .= $this->db->limitResult( '', $this->limit, $this->offset ); |
| | 195 | + |
| | 196 | + wfDebug( "searchQuery returned: $query" ); |
| | 197 | + |
| | 198 | + return $query; |
| | 199 | + } |
| | 200 | + |
| | 201 | + ## Most of the work of these two functions are done automatically via triggers |
| | 202 | + |
| | 203 | + function update( $pageid, $title, $text ) { |
| | 204 | + ## We don't want to index older revisions |
| | 205 | + $SQL = "UPDATE pagecontent SET textvector = NULL WHERE old_id = ". |
| | 206 | + "(SELECT rev_text_id FROM revision WHERE rev_page = $pageid ". |
| | 207 | + "ORDER BY rev_text_id DESC LIMIT 1 OFFSET 1)"; |
| | 208 | + $this->db->doQuery($SQL); |
| | 209 | + return true; |
| | 210 | + } |
| | 211 | + |
| | 212 | + function updateTitle( $id, $title ) { |
| | 213 | + return true; |
| | 214 | + } |
| | 215 | + |
| | 216 | +} ## end of the SearchPostgres class |
| | 217 | + |
| | 218 | +/** |
| | 219 | + * @ingroup Search |
| | 220 | + */ |
| | 221 | +class PostgresSearchResult extends SearchResult { |
| | 222 | + function PostgresSearchResult( $row ) { |
| | 223 | + $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| | 224 | + $this->score = $row->score; |
| | 225 | + } |
| | 226 | + function getScore() { |
| | 227 | + return $this->score; |
| | 228 | + } |
| | 229 | +} |
| | 230 | + |
| | 231 | +/** |
| | 232 | + * @ingroup Search |
| | 233 | + */ |
| | 234 | +class PostgresSearchResultSet extends SearchResultSet { |
| | 235 | + function PostgresSearchResultSet( $resultSet, $terms ) { |
| | 236 | + $this->mResultSet = $resultSet; |
| | 237 | + $this->mTerms = $terms; |
| | 238 | + } |
| | 239 | + |
| | 240 | + function termMatches() { |
| | 241 | + return $this->mTerms; |
| | 242 | + } |
| | 243 | + |
| | 244 | + function numRows() { |
| | 245 | + return $this->mResultSet->numRows(); |
| | 246 | + } |
| | 247 | + |
| | 248 | + function next() { |
| | 249 | + $row = $this->mResultSet->fetchObject(); |
| | 250 | + if( $row === false ) { |
| | 251 | + return false; |
| | 252 | + } else { |
| | 253 | + return new PostgresSearchResult( $row ); |
| | 254 | + } |
| | 255 | + } |
| | 256 | +} |
| Property changes on: trunk/phase3/includes/search/Postgres.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 257 | + native |
| Index: trunk/phase3/includes/AutoLoader.php |
| — | — | @@ -126,10 +126,8 @@ |
| 127 | 127 | 'MimeMagic' => 'includes/MimeMagic.php', |
| 128 | 128 | 'MWException' => 'includes/Exception.php', |
| 129 | 129 | 'MWNamespace' => 'includes/Namespace.php', |
| 130 | | - 'MySQLSearchResultSet' => 'includes/SearchMySQL.php', |
| 131 | 130 | 'Namespace' => 'includes/NamespaceCompat.php', // Compat |
| 132 | 131 | 'OldChangesList' => 'includes/ChangesList.php', |
| 133 | | - 'OracleSearchResultSet' => 'includes/SearchOracle.php', |
| 134 | 132 | 'OutputPage' => 'includes/OutputPage.php', |
| 135 | 133 | 'PageHistory' => 'includes/PageHistory.php', |
| 136 | 134 | 'PageHistoryPager' => 'includes/PageHistory.php', |
| — | — | @@ -137,8 +135,6 @@ |
| 138 | 136 | 'Pager' => 'includes/Pager.php', |
| 139 | 137 | 'PasswordError' => 'includes/User.php', |
| 140 | 138 | 'PatrolLog' => 'includes/PatrolLog.php', |
| 141 | | - 'PostgresSearchResult' => 'includes/SearchPostgres.php', |
| 142 | | - 'PostgresSearchResultSet' => 'includes/SearchPostgres.php', |
| 143 | 139 | 'PrefixSearch' => 'includes/PrefixSearch.php', |
| 144 | 140 | 'Profiler' => 'includes/Profiler.php', |
| 145 | 141 | 'ProfilerSimple' => 'includes/ProfilerSimple.php', |
| — | — | @@ -158,18 +154,6 @@ |
| 159 | 155 | 'Revision' => 'includes/Revision.php', |
| 160 | 156 | 'RSSFeed' => 'includes/Feed.php', |
| 161 | 157 | 'Sanitizer' => 'includes/Sanitizer.php', |
| 162 | | - 'SearchEngineDummy' => 'includes/SearchEngine.php', |
| 163 | | - 'SearchEngine' => 'includes/SearchEngine.php', |
| 164 | | - 'SearchHighlighter' => 'includes/SearchEngine.php', |
| 165 | | - 'SearchMySQL4' => 'includes/SearchMySQL4.php', |
| 166 | | - 'SearchMySQL' => 'includes/SearchMySQL.php', |
| 167 | | - 'SearchOracle' => 'includes/SearchOracle.php', |
| 168 | | - 'SearchPostgres' => 'includes/SearchPostgres.php', |
| 169 | | - 'SearchResult' => 'includes/SearchEngine.php', |
| 170 | | - 'SearchResultSet' => 'includes/SearchEngine.php', |
| 171 | | - 'SearchResultTooMany' => 'includes/SearchEngine.php', |
| 172 | | - 'SearchUpdate' => 'includes/SearchUpdate.php', |
| 173 | | - 'SearchUpdateMyISAM' => 'includes/SearchUpdate.php', |
| 174 | 158 | 'SiteConfiguration' => 'includes/SiteConfiguration.php', |
| 175 | 159 | 'SiteStats' => 'includes/SiteStats.php', |
| 176 | 160 | 'SiteStatsUpdate' => 'includes/SiteStats.php', |
| — | — | @@ -376,6 +360,24 @@ |
| 377 | 361 | 'Preprocessor_Hash' => 'includes/parser/Preprocessor_Hash.php', |
| 378 | 362 | 'StripState' => 'includes/parser/Parser.php', |
| 379 | 363 | |
| | 364 | + # includes/search |
| | 365 | + 'OracleSearchResultSet' => 'includes/search/Oracle.php', |
| | 366 | + 'PostgresSearchResult' => 'includes/search/Postgres.php', |
| | 367 | + 'PostgresSearchResultSet' => 'includes/search/Postgres.php', |
| | 368 | + 'MySQLSearchResultSet' => 'includes/Search/MySQL.php', |
| | 369 | + 'SearchEngineDummy' => 'includes/search/Engine.php', |
| | 370 | + 'SearchEngine' => 'includes/search/Engine.php', |
| | 371 | + 'SearchHighlighter' => 'includes/search/Engine.php', |
| | 372 | + 'SearchMySQL4' => 'includes/search/MySQL4.php', |
| | 373 | + 'SearchMySQL' => 'includes/search/MySQL.php', |
| | 374 | + 'SearchOracle' => 'includes/search/Oracle.php', |
| | 375 | + 'SearchPostgres' => 'includes/search/Postgres.php', |
| | 376 | + 'SearchResult' => 'includes/search/Engine.php', |
| | 377 | + 'SearchResultSet' => 'includes/search/Engine.php', |
| | 378 | + 'SearchResultTooMany' => 'includes/search/Engine.php', |
| | 379 | + 'SearchUpdate' => 'includes/search/Update.php', |
| | 380 | + 'SearchUpdateMyISAM' => 'includes/search/Update.php', |
| | 381 | + |
| 380 | 382 | # includes/specials |
| 381 | 383 | 'AncientPagesPage' => 'includes/specials/Ancientpages.php', |
| 382 | 384 | 'BrokenRedirectsPage' => 'includes/specials/BrokenRedirects.php', |