| Index: trunk/extensions/MWSearch/MWSearch.php |
| — | — | @@ -59,8 +59,7 @@ |
| 60 | 60 | $wgLuceneSearchCacheExpiry = 0; |
| 61 | 61 | |
| 62 | 62 | # Not a valid entry point, skip unless MEDIAWIKI is defined |
| 63 | | -if (defined('MEDIAWIKI')) { |
| 64 | | -$wgExtensionFunctions[] = "wfLuceneSearch"; |
| | 63 | +if( defined('MEDIAWIKI') ){ |
| 65 | 64 | |
| 66 | 65 | $wgExtensionCredits['other'][] = array( |
| 67 | 66 | 'name' => 'MWSearch', |
| — | — | @@ -70,719 +69,17 @@ |
| 71 | 70 | 'descriptionmsg' => 'mwsearch-desc', |
| 72 | 71 | 'url' => 'http://www.mediawiki.org/wiki/Extension:MWSearch', |
| 73 | 72 | ); |
| 74 | | -$wgExtensionMessagesFiles['MWSearch'] = dirname(__FILE__) . '/MWSearch.i18n.php'; |
| 75 | 73 | |
| | 74 | +$dir = dirname(__FILE__) . '/'; |
| | 75 | + |
| | 76 | +$wgExtensionMessagesFiles['MWSearch'] = $dir . 'MWSearch.i18n.php'; |
| | 77 | + |
| 76 | 78 | if($wgLuceneSearchVersion >= 2.1 && $wgEnableLucenePrefixSearch) |
| 77 | 79 | $wgHooks['PrefixSearchBackend'][] = 'LuceneSearch::prefixSearch'; |
| 78 | 80 | |
| 79 | | -function wfLuceneSearch() { |
| | 81 | +$wgAutoloadClasses['LuceneSearch'] = $dir . 'MWSearch_body.php'; |
| | 82 | +$wgAutoloadClasses['LuceneResult'] = $dir . 'MWSearch_body.php'; |
| | 83 | +$wgAutoloadClasses['LuceneSearchSet'] = $dir . 'MWSearch_body.php'; |
| 80 | 84 | |
| 81 | | -require_once( 'search/Engine.php' ); |
| 82 | | - |
| 83 | | -class LuceneSearch extends SearchEngine { |
| 84 | | - /** |
| 85 | | - * Perform a full text search query and return a result set. |
| 86 | | - * |
| 87 | | - * @param string $term - Raw search term |
| 88 | | - * @return LuceneSearchSet |
| 89 | | - * @access public |
| 90 | | - */ |
| 91 | | - function searchText( $term ) { |
| 92 | | - return LuceneSearchSet::newFromQuery( isset($this->related)? 'related' : 'search', |
| 93 | | - $term, $this->namespaces, $this->limit, $this->offset ); |
| 94 | | - } |
| 95 | | - |
| 96 | | - /** |
| 97 | | - * Perform a title-only search query and return a result set. |
| 98 | | - * |
| 99 | | - * @param string $term - Raw search term |
| 100 | | - * @return LuceneSearchSet |
| 101 | | - * @access public |
| 102 | | - */ |
| 103 | | - function searchTitle( $term ) { |
| 104 | | - return null; |
| 105 | | - } |
| 106 | | - |
| 107 | | - /** |
| 108 | | - * PrefixSearchBackend override for OpenSearch results |
| 109 | | - */ |
| 110 | | - static function prefixSearch( $ns, $search, $limit, &$results ) { |
| 111 | | - $it = LuceneSearchSet::newFromQuery( 'prefix', $search, $ns, $limit, 0 ); |
| 112 | | - $results = array(); |
| 113 | | - while( $res = $it->next() ) { |
| 114 | | - $results[] = $res->getTitle()->getPrefixedText(); |
| 115 | | - } |
| 116 | | - |
| 117 | | - return false; |
| 118 | | - } |
| 119 | | - |
| 120 | | - /** |
| 121 | | - * Prepare query for the lucene-search daemon: |
| 122 | | - * |
| 123 | | - * 1) rewrite namespaces into standardized form |
| 124 | | - * e.g. image:clouds -> [6]:clouds |
| 125 | | - * e.g. help,wp:npov -> [12,4]:npov |
| 126 | | - * |
| 127 | | - * 2) rewrite localizations of "search everything" keyword |
| 128 | | - * e.g. alle:heidegger -> all:heidegger |
| 129 | | - * |
| 130 | | - * @param string query |
| 131 | | - * @return string rewritten query |
| 132 | | - * @access private |
| 133 | | - */ |
| 134 | | - function replacePrefixes( $query ) { |
| 135 | | - global $wgContLang, $wgLuceneUseRelated; |
| 136 | | - $fname = 'LuceneSearch::replacePrefixes'; |
| 137 | | - wfProfileIn($fname); |
| 138 | | - $qlen = strlen($query); |
| 139 | | - $start = 0; $len = 0; // token start pos and length |
| 140 | | - $rewritten = ''; // rewritten query |
| 141 | | - $rindex = 0; // point to last rewritten character |
| 142 | | - $inquotes = false; |
| 143 | | - |
| 144 | | - // quick check, most of the time we don't need any rewriting |
| 145 | | - if(strpos($query,':')===false){ |
| 146 | | - wfProfileOut($fname); |
| 147 | | - return $query; |
| 148 | | - } |
| 149 | | - |
| 150 | | - // check if this is query for related articles |
| 151 | | - $relatedkey = wfMsgForContent('searchrelated').':'; |
| 152 | | - if($wgLuceneUseRelated && strncmp($query, $relatedkey, strlen($relatedkey)) == 0){ |
| 153 | | - $this->related = true; |
| 154 | | - list($dummy,$ret) = explode(":",$query,2); |
| 155 | | - wfProfileOut($fname); |
| 156 | | - return trim($ret); |
| 157 | | - } |
| 158 | | - |
| 159 | | - // "search everything" |
| 160 | | - // might not be at the beginning for complex queries |
| 161 | | - $allkeyword = wfMsgForContent('searchall'); |
| 162 | | - |
| 163 | | - for($i = 0 ; $i < $qlen ; $i++){ |
| 164 | | - $c = $query[$i]; |
| 165 | | - |
| 166 | | - // ignore chars in quotes |
| 167 | | - if($inquotes && $c!='"'); |
| 168 | | - // check if $c is valid prefix character |
| 169 | | - else if(($c >= 'a' && $c <= 'z') || |
| 170 | | - ($c >= 'A' && $c <= 'Z') || |
| 171 | | - $c == '_' || $c == '-' || $c ==','){ |
| 172 | | - if($len == 0){ |
| 173 | | - $start = $i; // begin of token |
| 174 | | - $len = 1; |
| 175 | | - } else |
| 176 | | - $len++; |
| 177 | | - // check for utf-8 chars |
| 178 | | - } else if(($c >= "\xc0" && $c <= "\xff")){ |
| 179 | | - $utf8len = 1; |
| 180 | | - for($j = $i+1; $j < $qlen; $j++){ // fetch extra utf-8 bytes |
| 181 | | - if($query[$j] >= "\x80" && $query[$j] <= "\xbf") |
| 182 | | - $utf8len++; |
| 183 | | - else |
| 184 | | - break; |
| 185 | | - } |
| 186 | | - if($len == 0){ |
| 187 | | - $start = $i; |
| 188 | | - $len = $utf8len; |
| 189 | | - } else |
| 190 | | - $len += $utf8len; |
| 191 | | - $i = $j - 1; // we consumed the chars |
| 192 | | - // check for end of prefix (i.e. semicolon) |
| 193 | | - } else if($c == ':' && $len !=0){ |
| 194 | | - $rewrite = array(); // here we collect namespaces |
| 195 | | - $prefixes = explode(',',substr($query,$start,$len)); |
| 196 | | - // iterate thru comma-separated list of prefixes |
| 197 | | - foreach($prefixes as $prefix){ |
| 198 | | - $index = $wgContLang->getNsIndex($prefix); |
| 199 | | - |
| 200 | | - // check for special prefixes all/incategory |
| 201 | | - if($prefix == $allkeyword){ |
| 202 | | - $rewrite = 'all'; |
| 203 | | - break; |
| 204 | | - // check for localized names of namespaces |
| 205 | | - } else if($index !== false) |
| 206 | | - $rewrite[] = $index; |
| 207 | | - } |
| 208 | | - $translated = null; |
| 209 | | - if($rewrite === 'all') |
| 210 | | - $translated = $rewrite; |
| 211 | | - else if(count($rewrite) != 0) |
| 212 | | - $translated = '['.implode(',',array_unique($rewrite)).']'; |
| 213 | | - |
| 214 | | - if(isset($translated)){ |
| 215 | | - // append text before the prefix, and then the prefix |
| 216 | | - $rewritten .= substr($query,$rindex,$start-$rindex); |
| 217 | | - $rewritten .= $translated . ':'; |
| 218 | | - $rindex = $i+1; |
| 219 | | - } |
| 220 | | - |
| 221 | | - $len = 0; |
| 222 | | - } else{ // end of token |
| 223 | | - if($c == '"') // get in/out of quotes |
| 224 | | - $inquotes = !$inquotes; |
| 225 | | - |
| 226 | | - $len = 0; |
| 227 | | - } |
| 228 | | - |
| 229 | | - } |
| 230 | | - // add rest of the original query that doesn't need rewritting |
| 231 | | - $rewritten .= substr($query,$rindex,$qlen-$rindex); |
| 232 | | - wfProfileOut($fname); |
| 233 | | - return $rewritten; |
| 234 | | - } |
| 235 | | -} |
| 236 | | - |
| 237 | | -class LuceneResult extends SearchResult { |
| 238 | | - /** |
| 239 | | - * Construct a result object from single result line |
| 240 | | - * |
| 241 | | - * @param array $lines |
| 242 | | - * @return array (float, Title) |
| 243 | | - * @access private |
| 244 | | - */ |
| 245 | | - function LuceneResult( $lines ) { |
| 246 | | - global $wgContLang; |
| 247 | | - |
| 248 | | - $score = null; |
| 249 | | - $interwiki = null; |
| 250 | | - $namespace = null; |
| 251 | | - $title = null; |
| 252 | | - |
| 253 | | - $line = $lines['result']; |
| 254 | | - wfDebug( "Lucene line: '$line'\n" ); |
| 255 | | - |
| 256 | | - # detect format |
| 257 | | - $parts = explode(' ', $line); |
| 258 | | - if(count($parts) == 3) |
| 259 | | - list( $score, $namespace, $title ) = $parts; |
| 260 | | - else |
| 261 | | - list( $score, $interwiki, $namespace, $nsText, $title ) = $parts; |
| 262 | | - |
| 263 | | - $score = floatval( $score ); |
| 264 | | - $namespace = intval( $namespace ); |
| 265 | | - $title = urldecode( $title ); |
| 266 | | - if(!isset($nsText)) |
| 267 | | - $nsText = $wgContLang->getNsText($namespace); |
| 268 | | - else |
| 269 | | - $nsText = urldecode($nsText); |
| 270 | | - |
| 271 | | - $this->mInterwiki = ''; |
| 272 | | - // make title |
| 273 | | - if( is_null($interwiki)){ |
| 274 | | - $this->mTitle = Title::makeTitle( $namespace, $title ); |
| 275 | | - } else{ |
| 276 | | - $interwiki = urldecode( $interwiki ); |
| 277 | | - // there might be a better way to make an interwiki link |
| 278 | | - $t = $interwiki.':'.$nsText.':'.str_replace( '_', ' ', $title ); |
| 279 | | - $this->mTitle = Title::newFromText( $t ); |
| 280 | | - $this->mInterwiki = $interwiki; |
| 281 | | - } |
| 282 | | - |
| 283 | | - $this->mScore = $score; |
| 284 | | - |
| 285 | | - $this->mWordCount = null; |
| 286 | | - if(array_key_exists("#h.wordcount",$lines)) |
| 287 | | - $this->mWordCount = intval($lines["#h.wordcount"][0]); |
| 288 | | - |
| 289 | | - $this->mSize = null; |
| 290 | | - if(array_key_exists("#h.size",$lines)) |
| 291 | | - $this->mSize = intval($lines["#h.size"][0]); |
| 292 | | - |
| 293 | | - $this->mDate = null; |
| 294 | | - if(array_key_exists("#h.date",$lines)) |
| 295 | | - $this->mDate = $lines["#h.date"][0]; |
| 296 | | - |
| 297 | | - // various snippets |
| 298 | | - list( $this->mHighlightTitle, $dummy ) = $this->extractSnippet($lines,$nsText,"#h.title"); |
| 299 | | - if( is_null($this->mHighlightTitle) && $this->isInterwiki() ){ |
| 300 | | - // construct highlighted interwiki title without the interwiki part |
| 301 | | - $this->mHighlightTitle = ($nsText==''? '' : $nsText.':') . str_replace( '_', ' ', $title ); |
| 302 | | - } |
| 303 | | - |
| 304 | | - list( $this->mHighlightText, $dummy ) = $this->extractSnippet($lines,'',"#h.text",true); |
| 305 | | - |
| 306 | | - list( $this->mHighlightRedirect, $redirect ) = $this->extractSnippet($lines,$nsText,"#h.redirect"); |
| 307 | | - $this->mRedirectTitle = null; |
| 308 | | - if( !is_null($redirect)){ |
| 309 | | - # build redirect Title object |
| 310 | | - if($interwiki != ''){ |
| 311 | | - $t = $interwiki.':'.$redirect; |
| 312 | | - $this->mRedirectTitle = Title::newFromText( $t ); |
| 313 | | - } else{ |
| 314 | | - $parts = explode(":",$redirect,2); |
| 315 | | - $redirectNs = intval($parts[0]); |
| 316 | | - $redirectText = str_replace('_', ' ', $parts[1]); |
| 317 | | - $this->mRedirectTitle = Title::makeTitle($redirectNs,$redirectText); |
| 318 | | - } |
| 319 | | - } |
| 320 | | - |
| 321 | | - list( $this->mHighlightSection, $section) = $this->extractSnippet($lines,'',"#h.section"); |
| 322 | | - $this->mSectionTitle = null; |
| 323 | | - if( !is_null($section)){ |
| 324 | | - # build title + fragment Title object |
| 325 | | - $t = $nsText.':'.str_replace( '_', ' ', $title ).'#'.$section; |
| 326 | | - $this->mSectionTitle = Title::newFromText($t); |
| 327 | | - } |
| 328 | | - |
| 329 | | - if($this->mInterwiki == '') |
| 330 | | - $this->mRevision = Revision::newFromTitle( $this->mTitle ); |
| 331 | | - } |
| 332 | | - |
| 333 | | - /** |
| 334 | | - * Get the pair [highlighted snippet, unmodified text] for highlighted text |
| 335 | | - * |
| 336 | | - * @param string $lines |
| 337 | | - * @param string $nsText textual form of namespace |
| 338 | | - * @param string $type |
| 339 | | - * @param boolean $useFinalSeparator |
| 340 | | - * @return array (highlighted, unmodified text) |
| 341 | | - */ |
| 342 | | - function extractSnippet($lines, $nsText, $type, $useFinalSeparator=false){ |
| 343 | | - if(!array_key_exists($type,$lines)) |
| 344 | | - return array(null,null); |
| 345 | | - $ret = ""; |
| 346 | | - $original = null; |
| 347 | | - foreach($lines[$type] as $h){ |
| 348 | | - list($s,$o) = $this->extractSnippetLine($h,$useFinalSeparator); |
| 349 | | - $ret .= $s; |
| 350 | | - $original = $o; |
| 351 | | - } |
| 352 | | - if($nsText!='') |
| 353 | | - $ret = $nsText.':'.$ret; |
| 354 | | - return array($ret,$original); |
| 355 | | - } |
| 356 | | - |
| 357 | | - /** |
| 358 | | - * Parse one line of a snippet |
| 359 | | - * |
| 360 | | - * @param string $line |
| 361 | | - * @param boolean $useFinalSeparator if "..." is to be appended to the end of snippet |
| 362 | | - * @access protected |
| 363 | | - * @return array(snippet,unmodified text) |
| 364 | | - */ |
| 365 | | - function extractSnippetLine($line, $useFinalSeparator){ |
| 366 | | - $parts = explode(" ",$line); |
| 367 | | - if(count($parts)!=4 && count($parts)!=5){ |
| 368 | | - wfDebug("Bad result line:".$line."\n"); |
| 369 | | - return ""; |
| 370 | | - } |
| 371 | | - $splits = $this->stripBracketsSplit($parts[0]); |
| 372 | | - $highlight = $this->stripBracketsSplit($parts[1]); |
| 373 | | - $suffix = urldecode($this->stripBrackets($parts[2])); |
| 374 | | - $text = urldecode($parts[3]); |
| 375 | | - $original = null; |
| 376 | | - if(count($parts) > 4) |
| 377 | | - $original = urldecode($parts[4]); |
| 378 | | - |
| 379 | | - $splits[] = strlen($text); |
| 380 | | - $start = 0; |
| 381 | | - $snippet = ""; |
| 382 | | - $hi = 0; |
| 383 | | - |
| 384 | | - foreach($splits as $sp){ |
| 385 | | - $sp = intval($sp); |
| 386 | | - // highlight words! |
| 387 | | - while($hi < count($highlight) && intval($highlight[$hi]) < $sp){ |
| 388 | | - $s = intval($highlight[$hi]); |
| 389 | | - $e = intval($highlight[$hi+1]); |
| 390 | | - $snippet .= substr($text,$start,$s-$start)."<span class='searchmatch'>".substr($text,$s,$e-$s)."</span>"; |
| 391 | | - $start = $e; |
| 392 | | - $hi += 2; |
| 393 | | - } |
| 394 | | - // copy till split point |
| 395 | | - $snippet .= substr($text,$start,$sp-$start); |
| 396 | | - if($sp == strlen($text) && $suffix != '') |
| 397 | | - $snippet .= $suffix; |
| 398 | | - else if($useFinalSeparator) |
| 399 | | - $snippet .= " <b>...</b> "; |
| 400 | | - |
| 401 | | - $start = $sp; |
| 402 | | - } |
| 403 | | - return array($snippet,$original); |
| 404 | | - } |
| 405 | | - |
| 406 | | - |
| 407 | | - /** |
| 408 | | - * @access private |
| 409 | | - */ |
| 410 | | - function stripBrackets($str){ |
| 411 | | - if($str == '[]') |
| 412 | | - return ''; |
| 413 | | - return substr($str,1,strlen($str)-2); |
| 414 | | - } |
| 415 | | - |
| 416 | | - /** |
| 417 | | - * @access private |
| 418 | | - * @return array |
| 419 | | - */ |
| 420 | | - function stripBracketsSplit($str){ |
| 421 | | - $strip = $this->stripBrackets($str); |
| 422 | | - if($strip == '') |
| 423 | | - return array(); |
| 424 | | - else |
| 425 | | - return explode(",",$strip); |
| 426 | | - } |
| 427 | | - |
| 428 | | - function getTitle() { |
| 429 | | - return $this->mTitle; |
| 430 | | - } |
| 431 | | - |
| 432 | | - function getScore() { |
| 433 | | - return null; // lucene scores are meaningless to the user... |
| 434 | | - } |
| 435 | | - |
| 436 | | - function getTitleSnippet($terms){ |
| 437 | | - if( is_null($this->mHighlightTitle) ) |
| 438 | | - return ''; |
| 439 | | - return $this->mHighlightTitle; |
| 440 | | - } |
| 441 | | - |
| 442 | | - function getTextSnippet($terms) { |
| 443 | | - if( is_null($this->mHighlightText) ) |
| 444 | | - return parent::getTextSnippet($terms); |
| 445 | | - return $this->mHighlightText; |
| 446 | | - } |
| 447 | | - |
| 448 | | - function getRedirectSnippet($terms) { |
| 449 | | - if( is_null($this->mHighlightRedirect) ) |
| 450 | | - return ''; |
| 451 | | - return $this->mHighlightRedirect; |
| 452 | | - } |
| 453 | | - |
| 454 | | - function getRedirectTitle(){ |
| 455 | | - return $this->mRedirectTitle; |
| 456 | | - } |
| 457 | | - |
| 458 | | - function getSectionSnippet(){ |
| 459 | | - if( is_null($this->mHighlightSection) ) |
| 460 | | - return ''; |
| 461 | | - return $this->mHighlightSection; |
| 462 | | - } |
| 463 | | - |
| 464 | | - function getSectionTitle(){ |
| 465 | | - return $this->mSectionTitle; |
| 466 | | - } |
| 467 | | - |
| 468 | | - function getInterwikiPrefix(){ |
| 469 | | - return $this->mInterwiki; |
| 470 | | - } |
| 471 | | - |
| 472 | | - function isInterwiki(){ |
| 473 | | - return $this->mInterwiki != ''; |
| 474 | | - } |
| 475 | | - |
| 476 | | - function getTimestamp(){ |
| 477 | | - if( is_null($this->mDate) ) |
| 478 | | - return parent::getTimestamp(); |
| 479 | | - return $this->mDate; |
| 480 | | - } |
| 481 | | - |
| 482 | | - function getWordCount(){ |
| 483 | | - if( is_null($this->mWordCount) ) |
| 484 | | - return parent::getWordCount(); |
| 485 | | - return $this->mWordCount; |
| 486 | | - } |
| 487 | | - |
| 488 | | - function getByteSize(){ |
| 489 | | - if( is_null($this->mSize) ) |
| 490 | | - return parent::getByteSize(); |
| 491 | | - return $this->mSize; |
| 492 | | - } |
| 493 | | - |
| 494 | | - function hasRelated(){ |
| 495 | | - global $wgLuceneSearchVersion, $wgLuceneUseRelated; |
| 496 | | - return $wgLuceneSearchVersion >= 2.1 && $wgLuceneUseRelated; |
| 497 | | - } |
| 498 | | -} |
| 499 | | - |
| 500 | | -class LuceneSearchSet extends SearchResultSet { |
| 501 | | - /** |
| 502 | | - * Contact the MWDaemon search server and return a wrapper |
| 503 | | - * object with the set of results. Results may be cached. |
| 504 | | - * |
| 505 | | - * @param string $method The protocol verb to use |
| 506 | | - * @param string $query |
| 507 | | - * @param int $limit |
| 508 | | - * @return array |
| 509 | | - * @access public |
| 510 | | - * @static |
| 511 | | - */ |
| 512 | | - function newFromQuery( $method, $query, $namespaces = array(), $limit = 20, $offset = 0 ) { |
| 513 | | - $fname = 'LuceneSearchSet::newFromQuery'; |
| 514 | | - wfProfileIn( $fname ); |
| 515 | | - |
| 516 | | - global $wgLuceneHost, $wgLucenePort, $wgDBname, $wgMemc; |
| 517 | | - global $wgLuceneSearchVersion, $wgLuceneSearchCacheExpiry; |
| 518 | | - |
| 519 | | - if( is_array( $wgLuceneHost ) ) { |
| 520 | | - $pick = mt_rand( 0, count( $wgLuceneHost ) - 1 ); |
| 521 | | - $host = $wgLuceneHost[$pick]; |
| 522 | | - } else { |
| 523 | | - $host = $wgLuceneHost; |
| 524 | | - } |
| 525 | | - |
| 526 | | - $enctext = rawurlencode( trim( $query ) ); |
| 527 | | - $searchUrl = "http://$host:$wgLucenePort/$method/$wgDBname/$enctext?" . |
| 528 | | - wfArrayToCGI( array( |
| 529 | | - 'namespaces' => implode( ',', $namespaces ), |
| 530 | | - 'offset' => $offset, |
| 531 | | - 'limit' => $limit, |
| 532 | | - 'version' => $wgLuceneSearchVersion, |
| 533 | | - 'iwlimit' => 10, |
| 534 | | - ) ); |
| 535 | | - |
| 536 | | - // try to fetch cached if caching is turned on |
| 537 | | - if($wgLuceneSearchCacheExpiry > 0){ |
| 538 | | - $key = "$wgDBname:lucene:" . md5( $searchUrl ); |
| 539 | | - $resultSet = $wgMemc->get( $key ); |
| 540 | | - if( is_object( $resultSet ) ) { |
| 541 | | - wfDebug( "$fname: got cached lucene results for key $key\n" ); |
| 542 | | - wfProfileOut( $fname ); |
| 543 | | - return $resultSet; |
| 544 | | - } |
| 545 | | - } |
| 546 | | - |
| 547 | | - wfDebug( "Fetching search data from $searchUrl\n" ); |
| 548 | | - wfSuppressWarnings(); |
| 549 | | - wfProfileIn( $fname.'-contact-'.$host ); |
| 550 | | - $data = Http::get( $searchUrl ); |
| 551 | | - wfProfileOut( $fname.'-contact-'.$host ); |
| 552 | | - wfRestoreWarnings(); |
| 553 | | - if( $data === false ) { |
| 554 | | - // Network error or server error |
| 555 | | - wfProfileOut( $fname ); |
| 556 | | - return null; |
| 557 | | - } else { |
| 558 | | - $inputLines = explode( "\n", trim( $data ) ); |
| 559 | | - $resultLines = array_map( 'trim', $inputLines ); |
| 560 | | - } |
| 561 | | - |
| 562 | | - $suggestion = null; |
| 563 | | - $totalHits = null; |
| 564 | | - $info = null; |
| 565 | | - $interwiki = null; |
| 566 | | - |
| 567 | | - # All methods have same syntax... |
| 568 | | - $totalHits = array_shift( $resultLines ); |
| 569 | | - if( $totalHits === false ) { |
| 570 | | - # I/O error? this shouldn't happen |
| 571 | | - wfDebug( "Couldn't read summary line...\n" ); |
| 572 | | - } else { |
| 573 | | - $totalHits = intval( $totalHits ); |
| 574 | | - wfDebug( "total [$totalHits] hits\n" ); |
| 575 | | - if($wgLuceneSearchVersion >= 2.1){ |
| 576 | | - # second line is info |
| 577 | | - list($dummy,$info) = explode(' ',array_shift($resultLines),2); |
| 578 | | - # third line is suggestions |
| 579 | | - $s = array_shift($resultLines); |
| 580 | | - if(self::startsWith($s,'#suggest ')) |
| 581 | | - $suggestion = $s; |
| 582 | | - |
| 583 | | - # fifth line is interwiki info line |
| 584 | | - $iwHeading = array_shift($resultLines); |
| 585 | | - list($dummy,$iwCount,$iwTotal) = explode(' ',$iwHeading); |
| 586 | | - if($iwCount>0){ |
| 587 | | - # pack interwiki lines into a separate result set |
| 588 | | - $interwikiLen = 0; |
| 589 | | - while(!self::startsWith($resultLines[$interwikiLen],"#results")) |
| 590 | | - $interwikiLen++; |
| 591 | | - $interwikiLines = array_splice($resultLines,0,$interwikiLen); |
| 592 | | - $interwiki = new LuceneSearchSet( $query, $interwikiLines, intval($iwCount), intval($iwTotal) ); |
| 593 | | - } |
| 594 | | - |
| 595 | | - # how many results we got |
| 596 | | - list($dummy,$resultCount) = explode(" ",array_shift($resultLines)); |
| 597 | | - $resultCount = intval($resultCount); |
| 598 | | - } else{ |
| 599 | | - $resultCount = count($resultLines); |
| 600 | | - } |
| 601 | | - } |
| 602 | | - |
| 603 | | - |
| 604 | | - $resultSet = new LuceneSearchSet( $query, $resultLines, $resultCount, $totalHits, |
| 605 | | - $suggestion, $info, $interwiki ); |
| 606 | | - |
| 607 | | - if($wgLuceneSearchCacheExpiry > 0){ |
| 608 | | - wfDebug( "$fname: caching lucene results for key $key\n" ); |
| 609 | | - $wgMemc->add( $key, $resultSet, $wgLuceneSearchCacheExpiry ); |
| 610 | | - } |
| 611 | | - |
| 612 | | - wfProfileOut( $fname ); |
| 613 | | - return $resultSet; |
| 614 | | - } |
| 615 | | - |
| 616 | | - static function startsWith($source, $prefix){ |
| 617 | | - return strncmp($source, $prefix, strlen($prefix)) == 0; |
| 618 | | - } |
| 619 | | - |
| 620 | | - /** |
| 621 | | - * Private constructor. Use LuceneSearchSet::newFromQuery(). |
| 622 | | - * |
| 623 | | - * @param string $query |
| 624 | | - * @param array $lines |
| 625 | | - * @param int $resultCount |
| 626 | | - * @param int $totalHits |
| 627 | | - * @param string $suggestion |
| 628 | | - * @param string $info |
| 629 | | - * @access private |
| 630 | | - */ |
| 631 | | - function LuceneSearchSet( $query, $lines, $resultCount, $totalHits = null, $suggestion = null, $info = null, $interwiki = null ) { |
| 632 | | - $this->mQuery = $query; |
| 633 | | - $this->mTotalHits = $totalHits; |
| 634 | | - $this->mResults = $lines; |
| 635 | | - $this->mResultCount = $resultCount; |
| 636 | | - $this->mPos = 0; |
| 637 | | - $this->mSuggestionQuery = null; |
| 638 | | - $this->mSuggestionSnippet = ''; |
| 639 | | - $this->parseSuggestion($suggestion); |
| 640 | | - $this->mInfo = $info; |
| 641 | | - $this->mInterwiki = $interwiki; |
| 642 | | - } |
| 643 | | - |
| 644 | | - /** Get suggestions from a suggestion result line */ |
| 645 | | - function parseSuggestion($suggestion){ |
| 646 | | - if( is_null($suggestion) ) |
| 647 | | - return; |
| 648 | | - // parse split points and highlight changes |
| 649 | | - list($dummy,$points,$sug) = explode(" ",$suggestion); |
| 650 | | - $sug = urldecode($sug); |
| 651 | | - $points = explode(",",substr($points,1,-1)); |
| 652 | | - array_unshift($points,0); |
| 653 | | - $suggestText = ""; |
| 654 | | - for($i=1;$i<count($points);$i+=2){ |
| 655 | | - $suggestText .= substr($sug,$points[$i-1],$points[$i]-$points[$i-1]); |
| 656 | | - $suggestText .= "<i>".substr($sug,$points[$i],$points[$i+1]-$points[$i])."</i>"; |
| 657 | | - } |
| 658 | | - $suggestText .= substr($sug,end($points)); |
| 659 | | - |
| 660 | | - $this->mSuggestionQuery = $this->replaceGenericPrefixes($sug); |
| 661 | | - $this->mSuggestionSnippet = $this->replaceGenericPrefixes($suggestText); |
| 662 | | - } |
| 663 | | - |
| 664 | | - /** replace prefixes like [2]: that are not in phrases */ |
| 665 | | - function replaceGenericPrefixes($text){ |
| 666 | | - $out = ""; |
| 667 | | - $phrases = explode('"',$text); |
| 668 | | - for($i=0;$i<count($phrases);$i+=2){ |
| 669 | | - $out .= preg_replace_callback('/\[([0-9]+)\]:/', array($this,'genericPrefixCallback'), $phrases[$i]); |
| 670 | | - if($i+1 < count($phrases)) |
| 671 | | - $out .= '"'.$phrases[$i+1].'"'; // phrase text |
| 672 | | - } |
| 673 | | - return $out; |
| 674 | | - } |
| 675 | | - |
| 676 | | - function genericPrefixCallback($matches){ |
| 677 | | - global $wgContLang; |
| 678 | | - return $wgContLang->getFormattedNsText($matches[1]).":"; |
| 679 | | - } |
| 680 | | - |
| 681 | | - function numRows() { |
| 682 | | - return $this->mResultCount; |
| 683 | | - } |
| 684 | | - |
| 685 | | - function termMatches() { |
| 686 | | - $resq = preg_replace( "/\\[.*?\\]:/", " ", $this->mQuery ); # generic prefixes |
| 687 | | - $resq = preg_replace( "/all:/", " ", $resq ); |
| 688 | | - $resq = trim( preg_replace( "/[ |\\[\\]()\"{}+\\-_@!?%&*=\\|:;><,.\\/]+/", " ", $resq ) ); |
| 689 | | - $terms = array_map( array( &$this, 'regexQuote' ), |
| 690 | | - explode( ' ', $resq ) ); |
| 691 | | - return $terms; |
| 692 | | - } |
| 693 | | - |
| 694 | | - /** |
| 695 | | - * Stupid hack around PHP's limited lambda support |
| 696 | | - * @access private |
| 697 | | - */ |
| 698 | | - function regexQuote( $term ) { |
| 699 | | - return preg_quote( $term, '/' ); |
| 700 | | - } |
| 701 | | - |
| 702 | | - function hasResults() { |
| 703 | | - return count( $this->mResults ) > 0; |
| 704 | | - } |
| 705 | | - |
| 706 | | - /** |
| 707 | | - * Some search modes return a total hit count for the query |
| 708 | | - * in the entire article database. This may include pages |
| 709 | | - * in namespaces that would not be matched on the given |
| 710 | | - * settings. |
| 711 | | - * |
| 712 | | - * @return int |
| 713 | | - * @access public |
| 714 | | - */ |
| 715 | | - function getTotalHits() { |
| 716 | | - return $this->mTotalHits; |
| 717 | | - } |
| 718 | | - |
| 719 | | - /** |
| 720 | | - * Return information about how and from where the results were fetched, |
| 721 | | - * should be useful for diagnostics and debugging |
| 722 | | - * |
| 723 | | - * @return string |
| 724 | | - */ |
| 725 | | - function getInfo() { |
| 726 | | - if( is_null($this->mInfo) ) |
| 727 | | - return null; |
| 728 | | - return "Search results fetched via ".$this->mInfo; |
| 729 | | - } |
| 730 | | - |
| 731 | | - /** |
| 732 | | - * Return a result set of hits on other (multiple) wikis associated with this one |
| 733 | | - * |
| 734 | | - * @return SearchResultSet |
| 735 | | - */ |
| 736 | | - function getInterwikiResults() { |
| 737 | | - return $this->mInterwiki; |
| 738 | | - } |
| 739 | | - |
| 740 | | - /** |
| 741 | | - * Some search modes return a suggested alternate term if there are |
| 742 | | - * no exact hits. Returns true if there is one on this set. |
| 743 | | - * |
| 744 | | - * @return bool |
| 745 | | - * @access public |
| 746 | | - */ |
| 747 | | - function hasSuggestion() { |
| 748 | | - return is_string( $this->mSuggestionQuery ) && $this->mSuggestionQuery != ''; |
| 749 | | - } |
| 750 | | - |
| 751 | | - function getSuggestionQuery(){ |
| 752 | | - return $this->mSuggestionQuery; |
| 753 | | - } |
| 754 | | - |
| 755 | | - function getSuggestionSnippet(){ |
| 756 | | - return $this->mSuggestionSnippet; |
| 757 | | - } |
| 758 | | - |
| 759 | | - /** |
| 760 | | - * Fetches next search result, or false. |
| 761 | | - * @return LuceneResult |
| 762 | | - * @access public |
| 763 | | - * @abstract |
| 764 | | - */ |
| 765 | | - function next() { |
| 766 | | - # Group together lines belonging to one hit |
| 767 | | - $group = array(); |
| 768 | | - |
| 769 | | - for(;$this->mPos < count($this->mResults);$this->mPos++){ |
| 770 | | - $l = trim($this->mResults[$this->mPos]); |
| 771 | | - if(count($group) == 0) // main line |
| 772 | | - $group['result'] = $l; |
| 773 | | - else if($l[0] == '#'){ // additional meta |
| 774 | | - list($meta,$value) = explode(" ",$l,2); |
| 775 | | - $group[$meta][] = $value; |
| 776 | | - } else |
| 777 | | - break; |
| 778 | | - } |
| 779 | | - if($group == false) |
| 780 | | - return false; |
| 781 | | - else |
| 782 | | - return new LuceneResult( $group ); |
| 783 | | - } |
| 784 | | - |
| 785 | | -} |
| 786 | | - |
| 787 | | -} # End of extension function |
| 788 | 85 | } # End of invocation guard |
| 789 | 86 | |
| Index: trunk/extensions/MWSearch/MWSearch_body.php |
| — | — | @@ -0,0 +1,704 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +class LuceneSearch extends SearchEngine { |
| | 5 | + /** |
| | 6 | + * Perform a full text search query and return a result set. |
| | 7 | + * |
| | 8 | + * @param string $term - Raw search term |
| | 9 | + * @return LuceneSearchSet |
| | 10 | + * @access public |
| | 11 | + */ |
| | 12 | + function searchText( $term ) { |
| | 13 | + return LuceneSearchSet::newFromQuery( isset($this->related)? 'related' : 'search', |
| | 14 | + $term, $this->namespaces, $this->limit, $this->offset ); |
| | 15 | + } |
| | 16 | + |
| | 17 | + /** |
| | 18 | + * Perform a title-only search query and return a result set. |
| | 19 | + * |
| | 20 | + * @param string $term - Raw search term |
| | 21 | + * @return LuceneSearchSet |
| | 22 | + * @access public |
| | 23 | + */ |
| | 24 | + function searchTitle( $term ) { |
| | 25 | + return null; |
| | 26 | + } |
| | 27 | + |
| | 28 | + /** |
| | 29 | + * PrefixSearchBackend override for OpenSearch results |
| | 30 | + */ |
| | 31 | + static function prefixSearch( $ns, $search, $limit, &$results ) { |
| | 32 | + $it = LuceneSearchSet::newFromQuery( 'prefix', $search, $ns, $limit, 0 ); |
| | 33 | + $results = array(); |
| | 34 | + while( $res = $it->next() ) { |
| | 35 | + $results[] = $res->getTitle()->getPrefixedText(); |
| | 36 | + } |
| | 37 | + |
| | 38 | + return false; |
| | 39 | + } |
| | 40 | + |
| | 41 | + /** |
| | 42 | + * Prepare query for the lucene-search daemon: |
| | 43 | + * |
| | 44 | + * 1) rewrite namespaces into standardized form |
| | 45 | + * e.g. image:clouds -> [6]:clouds |
| | 46 | + * e.g. help,wp:npov -> [12,4]:npov |
| | 47 | + * |
| | 48 | + * 2) rewrite localizations of "search everything" keyword |
| | 49 | + * e.g. alle:heidegger -> all:heidegger |
| | 50 | + * |
| | 51 | + * @param string query |
| | 52 | + * @return string rewritten query |
| | 53 | + * @access private |
| | 54 | + */ |
| | 55 | + function replacePrefixes( $query ) { |
| | 56 | + global $wgContLang, $wgLuceneUseRelated; |
| | 57 | + $fname = 'LuceneSearch::replacePrefixes'; |
| | 58 | + wfProfileIn($fname); |
| | 59 | + $qlen = strlen($query); |
| | 60 | + $start = 0; $len = 0; // token start pos and length |
| | 61 | + $rewritten = ''; // rewritten query |
| | 62 | + $rindex = 0; // point to last rewritten character |
| | 63 | + $inquotes = false; |
| | 64 | + |
| | 65 | + // quick check, most of the time we don't need any rewriting |
| | 66 | + if(strpos($query,':')===false){ |
| | 67 | + wfProfileOut($fname); |
| | 68 | + return $query; |
| | 69 | + } |
| | 70 | + |
| | 71 | + // check if this is query for related articles |
| | 72 | + $relatedkey = wfMsgForContent('searchrelated').':'; |
| | 73 | + if($wgLuceneUseRelated && strncmp($query, $relatedkey, strlen($relatedkey)) == 0){ |
| | 74 | + $this->related = true; |
| | 75 | + list($dummy,$ret) = explode(":",$query,2); |
| | 76 | + wfProfileOut($fname); |
| | 77 | + return trim($ret); |
| | 78 | + } |
| | 79 | + |
| | 80 | + // "search everything" |
| | 81 | + // might not be at the beginning for complex queries |
| | 82 | + $allkeyword = wfMsgForContent('searchall'); |
| | 83 | + |
| | 84 | + for($i = 0 ; $i < $qlen ; $i++){ |
| | 85 | + $c = $query[$i]; |
| | 86 | + |
| | 87 | + // ignore chars in quotes |
| | 88 | + if($inquotes && $c!='"'); |
| | 89 | + // check if $c is valid prefix character |
| | 90 | + else if(($c >= 'a' && $c <= 'z') || |
| | 91 | + ($c >= 'A' && $c <= 'Z') || |
| | 92 | + $c == '_' || $c == '-' || $c ==','){ |
| | 93 | + if($len == 0){ |
| | 94 | + $start = $i; // begin of token |
| | 95 | + $len = 1; |
| | 96 | + } else |
| | 97 | + $len++; |
| | 98 | + // check for utf-8 chars |
| | 99 | + } else if(($c >= "\xc0" && $c <= "\xff")){ |
| | 100 | + $utf8len = 1; |
| | 101 | + for($j = $i+1; $j < $qlen; $j++){ // fetch extra utf-8 bytes |
| | 102 | + if($query[$j] >= "\x80" && $query[$j] <= "\xbf") |
| | 103 | + $utf8len++; |
| | 104 | + else |
| | 105 | + break; |
| | 106 | + } |
| | 107 | + if($len == 0){ |
| | 108 | + $start = $i; |
| | 109 | + $len = $utf8len; |
| | 110 | + } else |
| | 111 | + $len += $utf8len; |
| | 112 | + $i = $j - 1; // we consumed the chars |
| | 113 | + // check for end of prefix (i.e. semicolon) |
| | 114 | + } else if($c == ':' && $len !=0){ |
| | 115 | + $rewrite = array(); // here we collect namespaces |
| | 116 | + $prefixes = explode(',',substr($query,$start,$len)); |
| | 117 | + // iterate thru comma-separated list of prefixes |
| | 118 | + foreach($prefixes as $prefix){ |
| | 119 | + $index = $wgContLang->getNsIndex($prefix); |
| | 120 | + |
| | 121 | + // check for special prefixes all/incategory |
| | 122 | + if($prefix == $allkeyword){ |
| | 123 | + $rewrite = 'all'; |
| | 124 | + break; |
| | 125 | + // check for localized names of namespaces |
| | 126 | + } else if($index !== false) |
| | 127 | + $rewrite[] = $index; |
| | 128 | + } |
| | 129 | + $translated = null; |
| | 130 | + if($rewrite === 'all') |
| | 131 | + $translated = $rewrite; |
| | 132 | + else if(count($rewrite) != 0) |
| | 133 | + $translated = '['.implode(',',array_unique($rewrite)).']'; |
| | 134 | + |
| | 135 | + if(isset($translated)){ |
| | 136 | + // append text before the prefix, and then the prefix |
| | 137 | + $rewritten .= substr($query,$rindex,$start-$rindex); |
| | 138 | + $rewritten .= $translated . ':'; |
| | 139 | + $rindex = $i+1; |
| | 140 | + } |
| | 141 | + |
| | 142 | + $len = 0; |
| | 143 | + } else{ // end of token |
| | 144 | + if($c == '"') // get in/out of quotes |
| | 145 | + $inquotes = !$inquotes; |
| | 146 | + |
| | 147 | + $len = 0; |
| | 148 | + } |
| | 149 | + |
| | 150 | + } |
| | 151 | + // add rest of the original query that doesn't need rewritting |
| | 152 | + $rewritten .= substr($query,$rindex,$qlen-$rindex); |
| | 153 | + wfProfileOut($fname); |
| | 154 | + return $rewritten; |
| | 155 | + } |
| | 156 | +} |
| | 157 | + |
| | 158 | +class LuceneResult extends SearchResult { |
| | 159 | + /** |
| | 160 | + * Construct a result object from single result line |
| | 161 | + * |
| | 162 | + * @param array $lines |
| | 163 | + * @return array (float, Title) |
| | 164 | + * @access private |
| | 165 | + */ |
| | 166 | + function LuceneResult( $lines ) { |
| | 167 | + global $wgContLang; |
| | 168 | + |
| | 169 | + $score = null; |
| | 170 | + $interwiki = null; |
| | 171 | + $namespace = null; |
| | 172 | + $title = null; |
| | 173 | + |
| | 174 | + $line = $lines['result']; |
| | 175 | + wfDebug( "Lucene line: '$line'\n" ); |
| | 176 | + |
| | 177 | + # detect format |
| | 178 | + $parts = explode(' ', $line); |
| | 179 | + if(count($parts) == 3) |
| | 180 | + list( $score, $namespace, $title ) = $parts; |
| | 181 | + else |
| | 182 | + list( $score, $interwiki, $namespace, $nsText, $title ) = $parts; |
| | 183 | + |
| | 184 | + $score = floatval( $score ); |
| | 185 | + $namespace = intval( $namespace ); |
| | 186 | + $title = urldecode( $title ); |
| | 187 | + if(!isset($nsText)) |
| | 188 | + $nsText = $wgContLang->getNsText($namespace); |
| | 189 | + else |
| | 190 | + $nsText = urldecode($nsText); |
| | 191 | + |
| | 192 | + $this->mInterwiki = ''; |
| | 193 | + // make title |
| | 194 | + if( is_null($interwiki)){ |
| | 195 | + $this->mTitle = Title::makeTitle( $namespace, $title ); |
| | 196 | + } else{ |
| | 197 | + $interwiki = urldecode( $interwiki ); |
| | 198 | + // there might be a better way to make an interwiki link |
| | 199 | + $t = $interwiki.':'.$nsText.':'.str_replace( '_', ' ', $title ); |
| | 200 | + $this->mTitle = Title::newFromText( $t ); |
| | 201 | + $this->mInterwiki = $interwiki; |
| | 202 | + } |
| | 203 | + |
| | 204 | + $this->mScore = $score; |
| | 205 | + |
| | 206 | + $this->mWordCount = null; |
| | 207 | + if(array_key_exists("#h.wordcount",$lines)) |
| | 208 | + $this->mWordCount = intval($lines["#h.wordcount"][0]); |
| | 209 | + |
| | 210 | + $this->mSize = null; |
| | 211 | + if(array_key_exists("#h.size",$lines)) |
| | 212 | + $this->mSize = intval($lines["#h.size"][0]); |
| | 213 | + |
| | 214 | + $this->mDate = null; |
| | 215 | + if(array_key_exists("#h.date",$lines)) |
| | 216 | + $this->mDate = $lines["#h.date"][0]; |
| | 217 | + |
| | 218 | + // various snippets |
| | 219 | + list( $this->mHighlightTitle, $dummy ) = $this->extractSnippet($lines,$nsText,"#h.title"); |
| | 220 | + if( is_null($this->mHighlightTitle) && $this->isInterwiki() ){ |
| | 221 | + // construct highlighted interwiki title without the interwiki part |
| | 222 | + $this->mHighlightTitle = ($nsText==''? '' : $nsText.':') . str_replace( '_', ' ', $title ); |
| | 223 | + } |
| | 224 | + |
| | 225 | + list( $this->mHighlightText, $dummy ) = $this->extractSnippet($lines,'',"#h.text",true); |
| | 226 | + |
| | 227 | + list( $this->mHighlightRedirect, $redirect ) = $this->extractSnippet($lines,$nsText,"#h.redirect"); |
| | 228 | + $this->mRedirectTitle = null; |
| | 229 | + if( !is_null($redirect)){ |
| | 230 | + # build redirect Title object |
| | 231 | + if($interwiki != ''){ |
| | 232 | + $t = $interwiki.':'.$redirect; |
| | 233 | + $this->mRedirectTitle = Title::newFromText( $t ); |
| | 234 | + } else{ |
| | 235 | + $parts = explode(":",$redirect,2); |
| | 236 | + $redirectNs = intval($parts[0]); |
| | 237 | + $redirectText = str_replace('_', ' ', $parts[1]); |
| | 238 | + $this->mRedirectTitle = Title::makeTitle($redirectNs,$redirectText); |
| | 239 | + } |
| | 240 | + } |
| | 241 | + |
| | 242 | + list( $this->mHighlightSection, $section) = $this->extractSnippet($lines,'',"#h.section"); |
| | 243 | + $this->mSectionTitle = null; |
| | 244 | + if( !is_null($section)){ |
| | 245 | + # build title + fragment Title object |
| | 246 | + $t = $nsText.':'.str_replace( '_', ' ', $title ).'#'.$section; |
| | 247 | + $this->mSectionTitle = Title::newFromText($t); |
| | 248 | + } |
| | 249 | + |
| | 250 | + if($this->mInterwiki == '') |
| | 251 | + $this->mRevision = Revision::newFromTitle( $this->mTitle ); |
| | 252 | + } |
| | 253 | + |
| | 254 | + /** |
| | 255 | + * Get the pair [highlighted snippet, unmodified text] for highlighted text |
| | 256 | + * |
| | 257 | + * @param string $lines |
| | 258 | + * @param string $nsText textual form of namespace |
| | 259 | + * @param string $type |
| | 260 | + * @param boolean $useFinalSeparator |
| | 261 | + * @return array (highlighted, unmodified text) |
| | 262 | + */ |
| | 263 | + function extractSnippet($lines, $nsText, $type, $useFinalSeparator=false){ |
| | 264 | + if(!array_key_exists($type,$lines)) |
| | 265 | + return array(null,null); |
| | 266 | + $ret = ""; |
| | 267 | + $original = null; |
| | 268 | + foreach($lines[$type] as $h){ |
| | 269 | + list($s,$o) = $this->extractSnippetLine($h,$useFinalSeparator); |
| | 270 | + $ret .= $s; |
| | 271 | + $original = $o; |
| | 272 | + } |
| | 273 | + if($nsText!='') |
| | 274 | + $ret = $nsText.':'.$ret; |
| | 275 | + return array($ret,$original); |
| | 276 | + } |
| | 277 | + |
| | 278 | + /** |
| | 279 | + * Parse one line of a snippet |
| | 280 | + * |
| | 281 | + * @param string $line |
| | 282 | + * @param boolean $useFinalSeparator if "..." is to be appended to the end of snippet |
| | 283 | + * @access protected |
| | 284 | + * @return array(snippet,unmodified text) |
| | 285 | + */ |
| | 286 | + function extractSnippetLine($line, $useFinalSeparator){ |
| | 287 | + $parts = explode(" ",$line); |
| | 288 | + if(count($parts)!=4 && count($parts)!=5){ |
| | 289 | + wfDebug("Bad result line:".$line."\n"); |
| | 290 | + return ""; |
| | 291 | + } |
| | 292 | + $splits = $this->stripBracketsSplit($parts[0]); |
| | 293 | + $highlight = $this->stripBracketsSplit($parts[1]); |
| | 294 | + $suffix = urldecode($this->stripBrackets($parts[2])); |
| | 295 | + $text = urldecode($parts[3]); |
| | 296 | + $original = null; |
| | 297 | + if(count($parts) > 4) |
| | 298 | + $original = urldecode($parts[4]); |
| | 299 | + |
| | 300 | + $splits[] = strlen($text); |
| | 301 | + $start = 0; |
| | 302 | + $snippet = ""; |
| | 303 | + $hi = 0; |
| | 304 | + |
| | 305 | + foreach($splits as $sp){ |
| | 306 | + $sp = intval($sp); |
| | 307 | + // highlight words! |
| | 308 | + while($hi < count($highlight) && intval($highlight[$hi]) < $sp){ |
| | 309 | + $s = intval($highlight[$hi]); |
| | 310 | + $e = intval($highlight[$hi+1]); |
| | 311 | + $snippet .= substr($text,$start,$s-$start)."<span class='searchmatch'>".substr($text,$s,$e-$s)."</span>"; |
| | 312 | + $start = $e; |
| | 313 | + $hi += 2; |
| | 314 | + } |
| | 315 | + // copy till split point |
| | 316 | + $snippet .= substr($text,$start,$sp-$start); |
| | 317 | + if($sp == strlen($text) && $suffix != '') |
| | 318 | + $snippet .= $suffix; |
| | 319 | + else if($useFinalSeparator) |
| | 320 | + $snippet .= " <b>...</b> "; |
| | 321 | + |
| | 322 | + $start = $sp; |
| | 323 | + } |
| | 324 | + return array($snippet,$original); |
| | 325 | + } |
| | 326 | + |
| | 327 | + |
| | 328 | + /** |
| | 329 | + * @access private |
| | 330 | + */ |
| | 331 | + function stripBrackets($str){ |
| | 332 | + if($str == '[]') |
| | 333 | + return ''; |
| | 334 | + return substr($str,1,strlen($str)-2); |
| | 335 | + } |
| | 336 | + |
| | 337 | + /** |
| | 338 | + * @access private |
| | 339 | + * @return array |
| | 340 | + */ |
| | 341 | + function stripBracketsSplit($str){ |
| | 342 | + $strip = $this->stripBrackets($str); |
| | 343 | + if($strip == '') |
| | 344 | + return array(); |
| | 345 | + else |
| | 346 | + return explode(",",$strip); |
| | 347 | + } |
| | 348 | + |
| | 349 | + function getTitle() { |
| | 350 | + return $this->mTitle; |
| | 351 | + } |
| | 352 | + |
| | 353 | + function getScore() { |
| | 354 | + return null; // lucene scores are meaningless to the user... |
| | 355 | + } |
| | 356 | + |
| | 357 | + function getTitleSnippet($terms){ |
| | 358 | + if( is_null($this->mHighlightTitle) ) |
| | 359 | + return ''; |
| | 360 | + return $this->mHighlightTitle; |
| | 361 | + } |
| | 362 | + |
| | 363 | + function getTextSnippet($terms) { |
| | 364 | + if( is_null($this->mHighlightText) ) |
| | 365 | + return parent::getTextSnippet($terms); |
| | 366 | + return $this->mHighlightText; |
| | 367 | + } |
| | 368 | + |
| | 369 | + function getRedirectSnippet($terms) { |
| | 370 | + if( is_null($this->mHighlightRedirect) ) |
| | 371 | + return ''; |
| | 372 | + return $this->mHighlightRedirect; |
| | 373 | + } |
| | 374 | + |
| | 375 | + function getRedirectTitle(){ |
| | 376 | + return $this->mRedirectTitle; |
| | 377 | + } |
| | 378 | + |
| | 379 | + function getSectionSnippet(){ |
| | 380 | + if( is_null($this->mHighlightSection) ) |
| | 381 | + return ''; |
| | 382 | + return $this->mHighlightSection; |
| | 383 | + } |
| | 384 | + |
| | 385 | + function getSectionTitle(){ |
| | 386 | + return $this->mSectionTitle; |
| | 387 | + } |
| | 388 | + |
| | 389 | + function getInterwikiPrefix(){ |
| | 390 | + return $this->mInterwiki; |
| | 391 | + } |
| | 392 | + |
| | 393 | + function isInterwiki(){ |
| | 394 | + return $this->mInterwiki != ''; |
| | 395 | + } |
| | 396 | + |
| | 397 | + function getTimestamp(){ |
| | 398 | + if( is_null($this->mDate) ) |
| | 399 | + return parent::getTimestamp(); |
| | 400 | + return $this->mDate; |
| | 401 | + } |
| | 402 | + |
| | 403 | + function getWordCount(){ |
| | 404 | + if( is_null($this->mWordCount) ) |
| | 405 | + return parent::getWordCount(); |
| | 406 | + return $this->mWordCount; |
| | 407 | + } |
| | 408 | + |
| | 409 | + function getByteSize(){ |
| | 410 | + if( is_null($this->mSize) ) |
| | 411 | + return parent::getByteSize(); |
| | 412 | + return $this->mSize; |
| | 413 | + } |
| | 414 | + |
| | 415 | + function hasRelated(){ |
| | 416 | + global $wgLuceneSearchVersion, $wgLuceneUseRelated; |
| | 417 | + return $wgLuceneSearchVersion >= 2.1 && $wgLuceneUseRelated; |
| | 418 | + } |
| | 419 | +} |
| | 420 | + |
| | 421 | +class LuceneSearchSet extends SearchResultSet { |
| | 422 | + /** |
| | 423 | + * Contact the MWDaemon search server and return a wrapper |
| | 424 | + * object with the set of results. Results may be cached. |
| | 425 | + * |
| | 426 | + * @param string $method The protocol verb to use |
| | 427 | + * @param string $query |
| | 428 | + * @param int $limit |
| | 429 | + * @return array |
| | 430 | + * @access public |
| | 431 | + */ |
| | 432 | + static function newFromQuery( $method, $query, $namespaces = array(), $limit = 20, $offset = 0 ) { |
| | 433 | + $fname = 'LuceneSearchSet::newFromQuery'; |
| | 434 | + wfProfileIn( $fname ); |
| | 435 | + |
| | 436 | + global $wgLuceneHost, $wgLucenePort, $wgDBname, $wgMemc; |
| | 437 | + global $wgLuceneSearchVersion, $wgLuceneSearchCacheExpiry; |
| | 438 | + |
| | 439 | + if( is_array( $wgLuceneHost ) ) { |
| | 440 | + $pick = mt_rand( 0, count( $wgLuceneHost ) - 1 ); |
| | 441 | + $host = $wgLuceneHost[$pick]; |
| | 442 | + } else { |
| | 443 | + $host = $wgLuceneHost; |
| | 444 | + } |
| | 445 | + |
| | 446 | + $enctext = rawurlencode( trim( $query ) ); |
| | 447 | + $searchUrl = "http://$host:$wgLucenePort/$method/$wgDBname/$enctext?" . |
| | 448 | + wfArrayToCGI( array( |
| | 449 | + 'namespaces' => implode( ',', $namespaces ), |
| | 450 | + 'offset' => $offset, |
| | 451 | + 'limit' => $limit, |
| | 452 | + 'version' => $wgLuceneSearchVersion, |
| | 453 | + 'iwlimit' => 10, |
| | 454 | + ) ); |
| | 455 | + |
| | 456 | + // try to fetch cached if caching is turned on |
| | 457 | + if($wgLuceneSearchCacheExpiry > 0){ |
| | 458 | + $key = "$wgDBname:lucene:" . md5( $searchUrl ); |
| | 459 | + $resultSet = $wgMemc->get( $key ); |
| | 460 | + if( is_object( $resultSet ) ) { |
| | 461 | + wfDebug( "$fname: got cached lucene results for key $key\n" ); |
| | 462 | + wfProfileOut( $fname ); |
| | 463 | + return $resultSet; |
| | 464 | + } |
| | 465 | + } |
| | 466 | + |
| | 467 | + wfDebug( "Fetching search data from $searchUrl\n" ); |
| | 468 | + wfSuppressWarnings(); |
| | 469 | + wfProfileIn( $fname.'-contact-'.$host ); |
| | 470 | + $data = Http::get( $searchUrl ); |
| | 471 | + wfProfileOut( $fname.'-contact-'.$host ); |
| | 472 | + wfRestoreWarnings(); |
| | 473 | + if( $data === false ) { |
| | 474 | + // Network error or server error |
| | 475 | + wfProfileOut( $fname ); |
| | 476 | + return null; |
| | 477 | + } else { |
| | 478 | + $inputLines = explode( "\n", trim( $data ) ); |
| | 479 | + $resultLines = array_map( 'trim', $inputLines ); |
| | 480 | + } |
| | 481 | + |
| | 482 | + $suggestion = null; |
| | 483 | + $totalHits = null; |
| | 484 | + $info = null; |
| | 485 | + $interwiki = null; |
| | 486 | + |
| | 487 | + # All methods have same syntax... |
| | 488 | + $totalHits = array_shift( $resultLines ); |
| | 489 | + if( $totalHits === false ) { |
| | 490 | + # I/O error? this shouldn't happen |
| | 491 | + wfDebug( "Couldn't read summary line...\n" ); |
| | 492 | + } else { |
| | 493 | + $totalHits = intval( $totalHits ); |
| | 494 | + wfDebug( "total [$totalHits] hits\n" ); |
| | 495 | + if($wgLuceneSearchVersion >= 2.1){ |
| | 496 | + # second line is info |
| | 497 | + list($dummy,$info) = explode(' ',array_shift($resultLines),2); |
| | 498 | + # third line is suggestions |
| | 499 | + $s = array_shift($resultLines); |
| | 500 | + if(self::startsWith($s,'#suggest ')) |
| | 501 | + $suggestion = $s; |
| | 502 | + |
| | 503 | + # fifth line is interwiki info line |
| | 504 | + $iwHeading = array_shift($resultLines); |
| | 505 | + list($dummy,$iwCount,$iwTotal) = explode(' ',$iwHeading); |
| | 506 | + if($iwCount>0){ |
| | 507 | + # pack interwiki lines into a separate result set |
| | 508 | + $interwikiLen = 0; |
| | 509 | + while(!self::startsWith($resultLines[$interwikiLen],"#results")) |
| | 510 | + $interwikiLen++; |
| | 511 | + $interwikiLines = array_splice($resultLines,0,$interwikiLen); |
| | 512 | + $interwiki = new LuceneSearchSet( $query, $interwikiLines, intval($iwCount), intval($iwTotal) ); |
| | 513 | + } |
| | 514 | + |
| | 515 | + # how many results we got |
| | 516 | + list($dummy,$resultCount) = explode(" ",array_shift($resultLines)); |
| | 517 | + $resultCount = intval($resultCount); |
| | 518 | + } else{ |
| | 519 | + $resultCount = count($resultLines); |
| | 520 | + } |
| | 521 | + } |
| | 522 | + |
| | 523 | + |
| | 524 | + $resultSet = new LuceneSearchSet( $query, $resultLines, $resultCount, $totalHits, |
| | 525 | + $suggestion, $info, $interwiki ); |
| | 526 | + |
| | 527 | + if($wgLuceneSearchCacheExpiry > 0){ |
| | 528 | + wfDebug( "$fname: caching lucene results for key $key\n" ); |
| | 529 | + $wgMemc->add( $key, $resultSet, $wgLuceneSearchCacheExpiry ); |
| | 530 | + } |
| | 531 | + |
| | 532 | + wfProfileOut( $fname ); |
| | 533 | + return $resultSet; |
| | 534 | + } |
| | 535 | + |
| | 536 | + static function startsWith($source, $prefix){ |
| | 537 | + return strncmp($source, $prefix, strlen($prefix)) == 0; |
| | 538 | + } |
| | 539 | + |
| | 540 | + /** |
| | 541 | + * Private constructor. Use LuceneSearchSet::newFromQuery(). |
| | 542 | + * |
| | 543 | + * @param string $query |
| | 544 | + * @param array $lines |
| | 545 | + * @param int $resultCount |
| | 546 | + * @param int $totalHits |
| | 547 | + * @param string $suggestion |
| | 548 | + * @param string $info |
| | 549 | + * @access private |
| | 550 | + */ |
| | 551 | + function LuceneSearchSet( $query, $lines, $resultCount, $totalHits = null, $suggestion = null, $info = null, $interwiki = null ) { |
| | 552 | + $this->mQuery = $query; |
| | 553 | + $this->mTotalHits = $totalHits; |
| | 554 | + $this->mResults = $lines; |
| | 555 | + $this->mResultCount = $resultCount; |
| | 556 | + $this->mPos = 0; |
| | 557 | + $this->mSuggestionQuery = null; |
| | 558 | + $this->mSuggestionSnippet = ''; |
| | 559 | + $this->parseSuggestion($suggestion); |
| | 560 | + $this->mInfo = $info; |
| | 561 | + $this->mInterwiki = $interwiki; |
| | 562 | + } |
| | 563 | + |
| | 564 | + /** Get suggestions from a suggestion result line */ |
| | 565 | + function parseSuggestion($suggestion){ |
| | 566 | + if( is_null($suggestion) ) |
| | 567 | + return; |
| | 568 | + // parse split points and highlight changes |
| | 569 | + list($dummy,$points,$sug) = explode(" ",$suggestion); |
| | 570 | + $sug = urldecode($sug); |
| | 571 | + $points = explode(",",substr($points,1,-1)); |
| | 572 | + array_unshift($points,0); |
| | 573 | + $suggestText = ""; |
| | 574 | + for($i=1;$i<count($points);$i+=2){ |
| | 575 | + $suggestText .= substr($sug,$points[$i-1],$points[$i]-$points[$i-1]); |
| | 576 | + $suggestText .= "<i>".substr($sug,$points[$i],$points[$i+1]-$points[$i])."</i>"; |
| | 577 | + } |
| | 578 | + $suggestText .= substr($sug,end($points)); |
| | 579 | + |
| | 580 | + $this->mSuggestionQuery = $this->replaceGenericPrefixes($sug); |
| | 581 | + $this->mSuggestionSnippet = $this->replaceGenericPrefixes($suggestText); |
| | 582 | + } |
| | 583 | + |
| | 584 | + /** replace prefixes like [2]: that are not in phrases */ |
| | 585 | + function replaceGenericPrefixes($text){ |
| | 586 | + $out = ""; |
| | 587 | + $phrases = explode('"',$text); |
| | 588 | + for($i=0;$i<count($phrases);$i+=2){ |
| | 589 | + $out .= preg_replace_callback('/\[([0-9]+)\]:/', array($this,'genericPrefixCallback'), $phrases[$i]); |
| | 590 | + if($i+1 < count($phrases)) |
| | 591 | + $out .= '"'.$phrases[$i+1].'"'; // phrase text |
| | 592 | + } |
| | 593 | + return $out; |
| | 594 | + } |
| | 595 | + |
| | 596 | + function genericPrefixCallback($matches){ |
| | 597 | + global $wgContLang; |
| | 598 | + return $wgContLang->getFormattedNsText($matches[1]).":"; |
| | 599 | + } |
| | 600 | + |
| | 601 | + function numRows() { |
| | 602 | + return $this->mResultCount; |
| | 603 | + } |
| | 604 | + |
| | 605 | + function termMatches() { |
| | 606 | + $resq = preg_replace( "/\\[.*?\\]:/", " ", $this->mQuery ); # generic prefixes |
| | 607 | + $resq = preg_replace( "/all:/", " ", $resq ); |
| | 608 | + $resq = trim( preg_replace( "/[ |\\[\\]()\"{}+\\-_@!?%&*=\\|:;><,.\\/]+/", " ", $resq ) ); |
| | 609 | + $terms = array_map( array( &$this, 'regexQuote' ), |
| | 610 | + explode( ' ', $resq ) ); |
| | 611 | + return $terms; |
| | 612 | + } |
| | 613 | + |
| | 614 | + /** |
| | 615 | + * Stupid hack around PHP's limited lambda support |
| | 616 | + * @access private |
| | 617 | + */ |
| | 618 | + function regexQuote( $term ) { |
| | 619 | + return preg_quote( $term, '/' ); |
| | 620 | + } |
| | 621 | + |
| | 622 | + function hasResults() { |
| | 623 | + return count( $this->mResults ) > 0; |
| | 624 | + } |
| | 625 | + |
| | 626 | + /** |
| | 627 | + * Some search modes return a total hit count for the query |
| | 628 | + * in the entire article database. This may include pages |
| | 629 | + * in namespaces that would not be matched on the given |
| | 630 | + * settings. |
| | 631 | + * |
| | 632 | + * @return int |
| | 633 | + * @access public |
| | 634 | + */ |
| | 635 | + function getTotalHits() { |
| | 636 | + return $this->mTotalHits; |
| | 637 | + } |
| | 638 | + |
| | 639 | + /** |
| | 640 | + * Return information about how and from where the results were fetched, |
| | 641 | + * should be useful for diagnostics and debugging |
| | 642 | + * |
| | 643 | + * @return string |
| | 644 | + */ |
| | 645 | + function getInfo() { |
| | 646 | + if( is_null($this->mInfo) ) |
| | 647 | + return null; |
| | 648 | + return "Search results fetched via ".$this->mInfo; |
| | 649 | + } |
| | 650 | + |
| | 651 | + /** |
| | 652 | + * Return a result set of hits on other (multiple) wikis associated with this one |
| | 653 | + * |
| | 654 | + * @return SearchResultSet |
| | 655 | + */ |
| | 656 | + function getInterwikiResults() { |
| | 657 | + return $this->mInterwiki; |
| | 658 | + } |
| | 659 | + |
| | 660 | + /** |
| | 661 | + * Some search modes return a suggested alternate term if there are |
| | 662 | + * no exact hits. Returns true if there is one on this set. |
| | 663 | + * |
| | 664 | + * @return bool |
| | 665 | + * @access public |
| | 666 | + */ |
| | 667 | + function hasSuggestion() { |
| | 668 | + return is_string( $this->mSuggestionQuery ) && $this->mSuggestionQuery != ''; |
| | 669 | + } |
| | 670 | + |
| | 671 | + function getSuggestionQuery(){ |
| | 672 | + return $this->mSuggestionQuery; |
| | 673 | + } |
| | 674 | + |
| | 675 | + function getSuggestionSnippet(){ |
| | 676 | + return $this->mSuggestionSnippet; |
| | 677 | + } |
| | 678 | + |
| | 679 | + /** |
| | 680 | + * Fetches next search result, or false. |
| | 681 | + * @return LuceneResult |
| | 682 | + * @access public |
| | 683 | + * @abstract |
| | 684 | + */ |
| | 685 | + function next() { |
| | 686 | + # Group together lines belonging to one hit |
| | 687 | + $group = array(); |
| | 688 | + |
| | 689 | + for(;$this->mPos < count($this->mResults);$this->mPos++){ |
| | 690 | + $l = trim($this->mResults[$this->mPos]); |
| | 691 | + if(count($group) == 0) // main line |
| | 692 | + $group['result'] = $l; |
| | 693 | + else if($l[0] == '#'){ // additional meta |
| | 694 | + list($meta,$value) = explode(" ",$l,2); |
| | 695 | + $group[$meta][] = $value; |
| | 696 | + } else |
| | 697 | + break; |
| | 698 | + } |
| | 699 | + if($group == false) |
| | 700 | + return false; |
| | 701 | + else |
| | 702 | + return new LuceneResult( $group ); |
| | 703 | + } |
| | 704 | + |
| | 705 | +} |
| Property changes on: trunk/extensions/MWSearch/MWSearch_body.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 706 | + native |