| Index: trunk/extensions/Collection/Collection.php |
| — | — | @@ -101,6 +101,10 @@ |
| 102 | 102 | |
| 103 | 103 | $wgCollectionMaxSuggestions = 10; |
| 104 | 104 | |
| | 105 | +$wgCollectionSuggestCheapWeightThreshhold = 50; |
| | 106 | + |
| | 107 | +$wgCollectionSuggestThreshhold = 100; |
| | 108 | + |
| 105 | 109 | # ============================================================================== |
| 106 | 110 | |
| 107 | 111 | $wgExtensionCredits['specialpage'][] = array( |
| Index: trunk/extensions/Collection/Collection.suggest.php |
| — | — | @@ -158,6 +158,7 @@ |
| 159 | 159 | private static function getCollectionSuggestTemplate( $mode, $param ) { |
| 160 | 160 | global $wgCollectionMaxSuggestions; |
| 161 | 161 | |
| | 162 | + |
| 162 | 163 | switch($mode) { |
| 163 | 164 | case 'add': |
| 164 | 165 | SpecialCollection::addArticleFromName(NS_MAIN, $param); |
| — | — | @@ -326,10 +327,14 @@ |
| 327 | 328 | |
| 328 | 329 | // Check if all articles form the book are in $mLinkList |
| 329 | 330 | private function addCollectionArticles() { |
| 330 | | - global $wgCollectionSuggestCountWordsThreshold; |
| | 331 | + global $wgCollectionSuggestThreshhold; |
| 331 | 332 | |
| 332 | 333 | $numItems = count( $this->mColl['items'] ); |
| 333 | 334 | |
| | 335 | + if ( $numItems > $wgCollectionSuggestThreshhold ) { |
| | 336 | + return; |
| | 337 | + } |
| | 338 | + |
| 334 | 339 | foreach( $this->mColl['items'] as $item ) { |
| 335 | 340 | if ( $this->searchEntry( $item['title'], $this->mLinkList ) === false |
| 336 | 341 | && $item['type'] == 'article' |
| — | — | @@ -344,7 +349,7 @@ |
| 345 | 350 | |
| 346 | 351 | $this->mLinkList[] = array( |
| 347 | 352 | 'name' => $articleName, |
| 348 | | - 'links' => $this->getWeightedLinks( $article->getContent() ), |
| | 353 | + 'links' => $this->getWeightedLinks( $numItems, $article->getContent() ), |
| 349 | 354 | ); |
| 350 | 355 | } |
| 351 | 356 | } |
| — | — | @@ -380,7 +385,9 @@ |
| 381 | 386 | * @param wikitext: article text |
| 382 | 387 | * @return an array with links and their weights |
| 383 | 388 | */ |
| 384 | | - private function getWeightedLinks( $wikitext ) { |
| | 389 | + private function getWeightedLinks( $num_articles, $wikitext ) { |
| | 390 | + global $wgCollectionSuggestCheapWeightThreshhold; |
| | 391 | + |
| 385 | 392 | $allLinks = array(); |
| 386 | 393 | preg_match_all( |
| 387 | 394 | '/\[\[(.+?)\]\]/', |
| — | — | @@ -428,43 +435,56 @@ |
| 429 | 436 | } |
| 430 | 437 | |
| 431 | 438 | $linkcount = array(); |
| 432 | | - foreach ( $linkmap as $alias => $linked ) { |
| 433 | | - $matches = array(); |
| 434 | | - preg_match_all( |
| 435 | | - '/\W' . preg_quote( $alias, '/' ) . '\W/i', |
| 436 | | - $wikitext, |
| 437 | | - $matches |
| 438 | | - ); |
| 439 | | - $num = count( $matches[0] ); |
| | 439 | + if ( $num_articles < $wgCollectionSuggestCheapWeightThreshhold ) { |
| | 440 | + // more expensive algorithm: count words |
| | 441 | + foreach ( $linkmap as $alias => $linked ) { |
| | 442 | + $matches = array(); |
| | 443 | + preg_match_all( |
| | 444 | + '/\W' . preg_quote( $alias, '/' ) . '\W/i', |
| | 445 | + $wikitext, |
| | 446 | + $matches |
| | 447 | + ); |
| | 448 | + $num = count( $matches[0] ); |
| 440 | 449 | |
| 441 | | - foreach ( $linked as $link => $dummy ) { |
| 442 | | - if ( isset( $linkcount[$link] ) ) { |
| 443 | | - $linkcount[$link] += $num; |
| 444 | | - } else { |
| 445 | | - $linkcount[$link] = $num; |
| | 450 | + foreach ( $linked as $link => $dummy ) { |
| | 451 | + if ( isset( $linkcount[$link] ) ) { |
| | 452 | + $linkcount[$link] += $num; |
| | 453 | + } else { |
| | 454 | + $linkcount[$link] = $num; |
| | 455 | + } |
| 446 | 456 | } |
| 447 | 457 | } |
| 448 | | - } |
| | 458 | + |
| | 459 | + if ( count( $linkcount ) == 0 ) { |
| | 460 | + return array(); |
| | 461 | + } |
| 449 | 462 | |
| 450 | | - if ( count( $linkcount ) == 0 ) { |
| 451 | | - return array(); |
| 452 | | - } |
| | 463 | + // normalize: |
| | 464 | + $lc_max = 0; |
| | 465 | + foreach ( $linkcount as $link => $count ) { |
| | 466 | + if ( $count > $lc_max) { |
| | 467 | + $lc_max = $count; |
| | 468 | + } |
| | 469 | + } |
| | 470 | + $norm = log( $lc_max ); |
| | 471 | + $result = array(); |
| | 472 | + if ( $norm > 0 ) { |
| | 473 | + foreach ( $linkcount as $link => $count ) { |
| | 474 | + $result[$link] = 1 + 0.5*log($count)/$norm; |
| | 475 | + } |
| | 476 | + } |
| 453 | 477 | |
| 454 | | - // normalize: |
| 455 | | - $lc_max = 0; |
| 456 | | - foreach ( $linkcount as $link => $count ) { |
| 457 | | - if ( $num > $lc_max) { |
| 458 | | - $lc_max = $count; |
| | 478 | + return $result; |
| | 479 | + } else { |
| | 480 | + // cheaper algorithm: just count links |
| | 481 | + foreach ( $linkmap as $alias => $linked ) { |
| | 482 | + foreach ( $linked as $link => $dummy) { |
| | 483 | + $linkcount[$link] = 1; |
| | 484 | + } |
| 459 | 485 | } |
| | 486 | + |
| | 487 | + return $linkcount; |
| 460 | 488 | } |
| 461 | | - $norm = log( $lc_max ); |
| 462 | | - $result = array(); |
| 463 | | - if ( $norm > 0 ) { |
| 464 | | - foreach ( $linkcount as $link => $count ) { |
| 465 | | - $result[$link] = 1 + 0.5*log($count)/$norm; |
| 466 | | - } |
| 467 | | - } |
| 468 | | - return $result; |
| 469 | 489 | } |
| 470 | 490 | |
| 471 | 491 | // Calculate the $mPropList from $mLinkList and $mBanList |