parserOutput = $parserOutput; } /** * Get headings on the page. * @return string[] * First strip out things that look like references. We can't use HTML filtering because * the references come back as tags without a class. To keep from breaking stuff like * ==Applicability of the strict mass–energy equivalence formula, ''E'' = ''mc''2== * we don't remove the whole tag. We also don't want to strip the tag and remove * everything that looks like [2] because, I dunno, maybe there is a band named Word [2] Foo * or something. Whatever. So we only strip things that look like tags wrapping a * reference. And since the data looks like: * Reference in heading [1][2] * we can not really use HtmlFormatter as we have no suitable selector. */ public function headings() { $headings = []; $ignoredHeadings = $this->getIgnoredHeadings(); foreach ( $this->parserOutput->getSections() as $heading ) { $heading = $heading[ 'line' ]; // Some wikis wrap the brackets in a span: // https://en.wikipedia.org/wiki/MediaWiki:Cite_reference_link $heading = preg_replace( '/<\/?span>/', '', $heading ); // Normalize [] so the following regexp would work. $heading = preg_replace( [ '/[/', '/]/' ], [ '[', ']' ], $heading ); $heading = preg_replace( '/\s*\[\s*\d+\s*\]\s*<\/sup>/is', '', $heading ); // Strip tags from the heading or else we'll display them (escaped) in search results $heading = trim( Sanitizer::stripAllTags( $heading ) ); // Note that we don't take the level of the heading into account - all headings are equal. // Except the ones we ignore. if ( !in_array( $heading, $ignoredHeadings ) ) { $headings[] = $heading; } } return $headings; } /** * Parse a message content into an array. This function is generally used to * parse settings stored as i18n messages (see search-ignored-headings). * * @param string $message * @return string[] */ public static function parseSettingsInMessage( $message ) { $lines = explode( "\n", $message ); $lines = preg_replace( '/#.*$/', '', $lines ); // Remove comments $lines = array_map( 'trim', $lines ); // Remove extra spaces $lines = array_filter( $lines ); // Remove empty lines return $lines; } /** * Get list of heading to ignore. * @return string[] */ private function getIgnoredHeadings() { static $ignoredHeadings = null; if ( $ignoredHeadings === null ) { $ignoredHeadings = []; $source = wfMessage( 'search-ignored-headings' )->inContentLanguage(); if ( $source->isBlank() ) { // Try old version too, just in case $source = wfMessage( 'cirrussearch-ignored-headings' )->inContentLanguage(); } if ( !$source->isDisabled() ) { $lines = self::parseSettingsInMessage( $source->plain() ); $ignoredHeadings = $lines; // Now we just have headings! } } return $ignoredHeadings; } /** * Extract parts of the text - opening, main and auxiliary. */ private function extractWikitextParts() { if ( $this->allText !== null ) { return; } $text = $this->parserOutput->getText( [ 'enableSectionEditTokens' => false, 'allowTOC' => false, ] ); if ( $text === '' ) { $this->allText = ""; // empty text - nothing to seek here return; } $opening = null; $this->openingText = $this->extractHeadingBeforeFirstHeading( $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 ) ) ); } } /** * Get text before first heading. * @param string $text * @return string|null */ private function extractHeadingBeforeFirstHeading( $text ) { $matches = []; if ( !preg_match( '//', $text, $matches, PREG_OFFSET_CAPTURE ) ) { // There isn't a first heading so we interpret this as the article // being entirely without heading. return null; } $text = substr( $text, 0, $matches[ 0 ][ 1 ] ); if ( !$text ) { // There isn't any text before the first heading so we declare there isn't // a first heading. return null; } $formatter = new HtmlFormatter( $text ); $formatter->remove( $this->excludedElementSelectors ); $formatter->remove( $this->auxiliaryElementSelectors ); $formatter->filterContent(); $text = trim( Sanitizer::stripAllTags( $formatter->getText() ) ); if ( !$text ) { // There isn't any text after filtering before the first heading so we declare // that there isn't a first heading. return null; } return $text; } /** * Get opening text * @return string */ public function getOpeningText() { $this->extractWikitextParts(); return $this->openingText; } /** * Get main text * @return string */ public function getMainText() { $this->extractWikitextParts(); return $this->allText; } /** * Get auxiliary text * @return string[] */ public function getAuxiliaryText() { $this->extractWikitextParts(); return $this->auxText; } /** * Get the defaultsort property * @return string|null */ public function getDefaultSort() { return $this->parserOutput->getProperty( 'defaultsort' ); } }