/** * Extract parts of the text - opening, main and auxiliary. */ private function extractWikitextParts() { if (!is_null($this->allText)) { return; } $this->parserOutput->setEditSectionTokens(false); $this->parserOutput->setTOCEnabled(false); $text = $this->parserOutput->getText(); if (strlen($text) == 0) { $this->allText = ""; // empty text - nothing to seek here return; } $opening = null; $this->openingText = $this->extractHeadingBeforeFirstHeading($text); // Add extra spacing around break tags so text crammed together like<br>this // doesn't make one word. $text = str_replace('<br', "\n<br", $text); $formatter = new HtmlFormatter($text); // Strip elements from the page that we never want in the search text. $formatter->remove($this->excludedElementSelectors); $formatter->filterContent(); // Strip elements from the page that are auxiliary text. These will still be // searched but matches will be ranked lower and non-auxiliary matches will be // preferred in highlighting. $formatter->remove($this->auxiliaryElementSelectors); $auxiliaryElements = $formatter->filterContent(); $this->allText = trim(Sanitizer::stripAllTags($formatter->getText())); foreach ($auxiliaryElements as $auxiliaryElement) { $this->auxText[] = trim(Sanitizer::stripAllTags($formatter->getText($auxiliaryElement))); } }
/** * OutputPageParserOutput hook handler * @param OutputPage $out * @param ParserOutput $parserOutput * @return type */ public static function outputPageParserOutput(OutputPage &$out, ParserOutput $parserOutput) { $out->addModuleStyles('ext.articleEmblems'); if (isset($parserOutput->articleEmblems)) { $emblems = array(); foreach ($parserOutput->articleEmblems as $emblem) { $emblems[] = '<li class="articleEmblem">' . $emblem . '</li>'; } $parserOutput->setText('<ul id="articleEmblems" class="noprint">' . implode($emblems) . '</ul>' . $parserOutput->getText()); } return true; }
/** * OutputPageParserOutput hook handler * Disables TOC in output before it grabs HTML * @see https://www.mediawiki.org/wiki/Manual:Hooks/OutputPageParserOutput * * @param OutputPage $outputPage * @param ParserOutput $po * @return bool */ public static function onOutputPageParserOutput($outputPage, ParserOutput $po) { global $wgMFWikibaseImageCategory; $context = MobileContext::singleton(); $isBeta = $context->isBetaGroupMember(); $mfUseWikibaseDescription = $context->getMFConfig()->get('MFUseWikibaseDescription'); if ($context->shouldDisplayMobileView()) { $outputPage->enableTOC(false); $outputPage->setProperty('MinervaTOC', $po->getTOCHTML() !== ''); if ($mfUseWikibaseDescription && $isBeta) { $item = $po->getProperty('wikibase_item'); if ($item) { $desc = ExtMobileFrontend::getWikibaseDescription($item); $category = ExtMobileFrontend::getWikibasePropertyValue($item, $wgMFWikibaseImageCategory); if ($desc) { $outputPage->setProperty('wgMFDescription', $desc); } if ($category) { $outputPage->setProperty('wgMFImagesCategory', $category); } } } // Enable wrapped sections $po->setText(ExtMobileFrontend::DOMParse($outputPage, $po->getText(), $isBeta)); } return true; }
/** * Add the HTML associated with a ParserOutput object, without any metadata. * * @since 1.24 * @param ParserOutput $parserOutput */ public function addParserOutputText($parserOutput) { $text = $parserOutput->getText(); Hooks::run('OutputPageBeforeHTML', array(&$this, &$text)); $this->addHTML($text); }
/** * Visualize <add> and <del> tags that are nested in themselves correctly. Remove tags that are not available in the editor for visualization. * These tags will still be visible in the editor. */ private function visualiseStrayTagsAndRemoveNotSupportedTags(ParserOutput $parser_output) { $text = $parser_output->getText(); //look for stray </add> tags, and replace them with a tei-add span element $text = preg_replace('/<\\/span><\\/span>(.*?)<\\/add>/', '</span></span><span class="tei-add">$1</span>', $text); //look for stray </del> tags, and replace them with a tei-del span element $text = preg_replace('/<\\/span><\\/span>(.*?)<\\/del>/', '</span></span><span class="tei-del">$1</span>', $text); $text = preg_replace('/<\\/span><\\/span>(.*?)<\\/hi>/', '</span></span><span class="tei-hi superscript">$1</span>', $text); //look for any other escaped tags, and remove them $text = preg_replace('/<(.*?)>/s', '', $text); $parser_output->setText($text); return true; }
public function execute() { // The data is hot but user-dependent, like page views, so we set vary cookies $this->getMain()->setCacheMode('anon-public-user-private'); // Get parameters $params = $this->extractRequestParams(); $text = $params['text']; $title = $params['title']; $page = $params['page']; $pageid = $params['pageid']; $oldid = $params['oldid']; if (!is_null($page) && (!is_null($text) || $title != 'API')) { $this->dieUsage('The page parameter cannot be used together with the text and title parameters', 'params'); } $prop = array_flip($params['prop']); if (isset($params['section'])) { $this->section = $params['section']; } else { $this->section = false; } // The parser needs $wgTitle to be set, apparently the // $title parameter in Parser::parse isn't enough *sigh* // TODO: Does this still need $wgTitle? global $wgParser, $wgTitle, $wgLang; // Currently unnecessary, code to act as a safeguard against any change in current behaviour of uselang breaks $oldLang = null; if (isset($params['uselang']) && $params['uselang'] != $wgLang->getCode()) { $oldLang = $wgLang; // Backup wgLang $wgLang = Language::factory($params['uselang']); } $popts = ParserOptions::newFromContext($this->getContext()); $popts->setTidy(true); $popts->enableLimitReport(!$params['disablepp']); $redirValues = null; // Return result $result = $this->getResult(); if (!is_null($oldid) || !is_null($pageid) || !is_null($page)) { if (!is_null($oldid)) { // Don't use the parser cache $rev = Revision::newFromID($oldid); if (!$rev) { $this->dieUsage("There is no revision ID {$oldid}", 'missingrev'); } if (!$rev->userCan(Revision::DELETED_TEXT, $this->getUser())) { $this->dieUsage("You don't have permission to view deleted revisions", 'permissiondenied'); } $titleObj = $rev->getTitle(); $wgTitle = $titleObj; // If for some reason the "oldid" is actually the current revision, it may be cached if ($titleObj->getLatestRevID() === intval($oldid)) { // May get from/save to parser cache $p_result = $this->getParsedSectionOrText($titleObj, $popts, $pageid, isset($prop['wikitext'])); } else { // This is an old revision, so get the text differently $this->text = $rev->getText(Revision::FOR_THIS_USER, $this->getUser()); if ($this->section !== false) { $this->text = $this->getSectionText($this->text, 'r' . $rev->getId()); } // Should we save old revision parses to the parser cache? $p_result = $wgParser->parse($this->text, $titleObj, $popts); } } else { // Not $oldid, but $pageid or $page if ($params['redirects']) { $reqParams = array('action' => 'query', 'redirects' => ''); if (!is_null($pageid)) { $reqParams['pageids'] = $pageid; } else { // $page $reqParams['titles'] = $page; } $req = new FauxRequest($reqParams); $main = new ApiMain($req); $main->execute(); $data = $main->getResultData(); $redirValues = isset($data['query']['redirects']) ? $data['query']['redirects'] : array(); $to = $page; foreach ((array) $redirValues as $r) { $to = $r['to']; } $titleObj = Title::newFromText($to); } else { if (!is_null($pageid)) { $reqParams['pageids'] = $pageid; $titleObj = Title::newFromID($pageid); } else { // $page $to = $page; $titleObj = Title::newFromText($to); } } if (!is_null($pageid)) { if (!$titleObj) { // Still throw nosuchpageid error if pageid was provided $this->dieUsageMsg(array('nosuchpageid', $pageid)); } } elseif (!$titleObj || !$titleObj->exists()) { $this->dieUsage("The page you specified doesn't exist", 'missingtitle'); } $wgTitle = $titleObj; if (isset($prop['revid'])) { $oldid = $titleObj->getLatestRevID(); } // Potentially cached $p_result = $this->getParsedSectionOrText($titleObj, $popts, $pageid, isset($prop['wikitext'])); } } else { // Not $oldid, $pageid, $page. Hence based on $text if (is_null($text)) { $this->dieUsage('The text parameter should be passed with the title parameter. Should you be using the "page" parameter instead?', 'params'); } $this->text = $text; $titleObj = Title::newFromText($title); if (!$titleObj) { $this->dieUsageMsg(array('invalidtitle', $title)); } $wgTitle = $titleObj; if ($this->section !== false) { $this->text = $this->getSectionText($this->text, $titleObj->getText()); } if ($params['pst'] || $params['onlypst']) { $this->pstText = $wgParser->preSaveTransform($this->text, $titleObj, $this->getUser(), $popts); } if ($params['onlypst']) { // Build a result and bail out $result_array = array(); $result_array['text'] = array(); $result->setContent($result_array['text'], $this->pstText); if (isset($prop['wikitext'])) { $result_array['wikitext'] = array(); $result->setContent($result_array['wikitext'], $this->text); } $result->addValue(null, $this->getModuleName(), $result_array); return; } // Not cached (save or load) $p_result = $wgParser->parse($params['pst'] ? $this->pstText : $this->text, $titleObj, $popts); } $result_array = array(); $result_array['title'] = $titleObj->getPrefixedText(); if (!is_null($oldid)) { $result_array['revid'] = intval($oldid); } if ($params['redirects'] && !is_null($redirValues)) { $result_array['redirects'] = $redirValues; } /* Wikia Change Starts */ if (!$p_result instanceof ParserOutput) { \Wikia\Logger\WikiaLogger::instance()->error('ApiParse no ParserOutput', ['pageTitle' => $title]); // generate empty parser output to support code working on it $p_result = new ParserOutput(''); } /* Wikia Change Ends */ if (isset($prop['text'])) { $result_array['text'] = array(); $result->setContent($result_array['text'], $p_result->getText()); } if (!is_null($params['summary'])) { $result_array['parsedsummary'] = array(); $result->setContent($result_array['parsedsummary'], Linker::formatComment($params['summary'], $titleObj)); } if (isset($prop['langlinks'])) { $result_array['langlinks'] = $this->formatLangLinks($p_result->getLanguageLinks()); } if (isset($prop['languageshtml'])) { $languagesHtml = $this->languagesHtml($p_result->getLanguageLinks()); $result_array['languageshtml'] = array(); $result->setContent($result_array['languageshtml'], $languagesHtml); } if (isset($prop['categories'])) { $result_array['categories'] = $this->formatCategoryLinks($p_result->getCategories()); } if (isset($prop['categorieshtml'])) { $categoriesHtml = $this->categoriesHtml($p_result->getCategories()); $result_array['categorieshtml'] = array(); $result->setContent($result_array['categorieshtml'], $categoriesHtml); } if (isset($prop['links'])) { $result_array['links'] = $this->formatLinks($p_result->getLinks()); } if (isset($prop['templates'])) { $result_array['templates'] = $this->formatLinks($p_result->getTemplates()); } if (isset($prop['images'])) { $result_array['images'] = array_keys($p_result->getImages()); } if (isset($prop['externallinks'])) { $result_array['externallinks'] = array_keys($p_result->getExternalLinks()); } if (isset($prop['sections'])) { $result_array['sections'] = $p_result->getSections(); } if (isset($prop['displaytitle'])) { $result_array['displaytitle'] = $p_result->getDisplayTitle() ? $p_result->getDisplayTitle() : $titleObj->getPrefixedText(); } if (isset($prop['headitems']) || isset($prop['headhtml'])) { $context = $this->getContext(); $context->setTitle($titleObj); $context->getOutput()->addParserOutputNoText($p_result); if (isset($prop['headitems'])) { $headItems = $this->formatHeadItems($p_result->getHeadItems()); $css = $this->formatCss($context->getOutput()->buildCssLinksArray()); $scripts = array($context->getOutput()->getHeadScripts()); $result_array['headitems'] = array_merge($headItems, $css, $scripts); } if (isset($prop['headhtml'])) { $result_array['headhtml'] = array(); $result->setContent($result_array['headhtml'], $context->getOutput()->headElement($context->getSkin())); } } if (isset($prop['iwlinks'])) { $result_array['iwlinks'] = $this->formatIWLinks($p_result->getInterwikiLinks()); } if (isset($prop['wikitext'])) { $result_array['wikitext'] = array(); $result->setContent($result_array['wikitext'], $this->text); if (!is_null($this->pstText)) { $result_array['psttext'] = array(); $result->setContent($result_array['psttext'], $this->pstText); } } $result_mapping = array('redirects' => 'r', 'langlinks' => 'll', 'categories' => 'cl', 'links' => 'pl', 'templates' => 'tl', 'images' => 'img', 'externallinks' => 'el', 'iwlinks' => 'iw', 'sections' => 's', 'headitems' => 'hi'); $this->setIndexedTagNames($result_array, $result_mapping); $result->addValue(null, $this->getModuleName(), $result_array); if (!is_null($oldLang)) { $wgLang = $oldLang; // Reset $wgLang to $oldLang } }