wfDplLst..). * So any version of LabeledSectionTransclusion can be installed together with DPL * * Enhancements were made to * - allow inclusion of templates ("template swapping") * - reduce the size of the transcluded text to a limit of characters * * * Thanks to Steve for his great work! * -- Algorithmix */ namespace MediaWiki\Extension\DynamicPageList3; use MediaWiki\Extension\DynamicPageList3\Lister\Lister; use MediaWiki\MediaWikiServices; use Parser; use Title; class LST { # ############################################################# # To do transclusion from an extension, we need to interact with the parser # at a low level. This is the general transclusion functionality ############################################################## /** * Register what we're working on in the parser, so we don't fall into a trap. * * @param Parser $parser * @param $part1 * @return bool * * @suppress PhanUndeclaredProperty Use of Parser::mTemplatePath */ public static function open( $parser, $part1 ) { // Infinite loop test if ( isset( $parser->mTemplatePath[$part1] ) ) { wfDebug( __METHOD__ . ": template loop broken at '$part1'\n" ); return false; } else { $parser->mTemplatePath[$part1] = 1; return true; } } /** * Finish processing the function. * * @param Parser $parser * @param $part1 * * @suppress PhanUndeclaredProperty Use of Parser::mTemplatePath */ public static function close( $parser, $part1 ) { // Infinite loop test if ( isset( $parser->mTemplatePath[$part1] ) ) { unset( $parser->mTemplatePath[$part1] ); } else { wfDebug( __METHOD__ . ": close unopened template loop at '$part1'\n" ); } } /** * Handle recursive substitution here, so we can break cycles, and set up * return values so that edit sections will resolve correctly. */ private static function parse( $parser, $text, $part1, $skiphead = 0, $recursionCheck = true, $maxLength = -1, $link = '', $trim = false, $skipPattern = [] ) { // if someone tries something like

lst only

// text, may as well do the right thing. $text = str_replace( '', '', $text ); // if desired we remove portions of the text, esp. template calls foreach ( $skipPattern as $skipPat ) { $text = preg_replace( $skipPat, '', $text ); } if ( self::open( $parser, $part1 ) ) { // Handle recursion here, so we can break cycles. if ( $recursionCheck == false ) { $text = $parser->preprocess( $text, $parser->getTitle(), $parser->getOptions() ); self::close( $parser, $part1 ); } if ( $maxLength > 0 ) { $text = self::limitTranscludedText( $text, $maxLength, $link ); } if ( $trim ) { return trim( $text ); } else { return $text; } } else { return "[[" . $parser->getTitle()->getPrefixedText() . "]]" . ""; } } # ############################################################# # And now, the labeled section transclusion ############################################################## /** * Parser tag hook for

. * The section markers aren't paired, so we only need to remove them. * * @param string $in * @param array $assocArgs * @param Parser $parser * @return string */ private static function noop( $in, $assocArgs = [], $parser = null ) { return ''; } // Generate a regex to match the section(s) we're interested in. private static function createSectionPattern( $sec, $to, &$any ) { $any = false; $to_sec = ( $to == '' ) ? $sec : $to; if ( $sec[0] == '*' ) { $any = true; if ( $sec == '**' ) { $sec = '[^\/>"' . "']+"; } else { $sec = str_replace( '/', '\/', substr( $sec, 1 ) ); } } else { $sec = preg_quote( $sec, '/' ); } if ( $to_sec[0] == '*' ) { if ( $to_sec == '**' ) { $to_sec = '[^\/>"' . "']+"; } else { $to_sec = str_replace( '/', '\/', substr( $to_sec, 1 ) ); } } else { $to_sec = preg_quote( $to_sec, '/' ); } $ws = "(?:\s+[^>]+)?"; return "/(.*?)\n?]+\s+)?(?i:end)=" . "['\"]?\\1['\"]?" . "$ws\/?>/s"; } /** * Count headings in skipped text. * * Count skipped headings, so parser can skip them, to * prevent wrong heading links. * * @param string $text * @param int $limit * @return int */ private static function countHeadings( $text, $limit ) { $pat = '^(={1,6}).+\1\s*$()'; $count = 0; $offset = 0; $m = []; while ( preg_match( "/$pat/im", $text, $m, PREG_OFFSET_CAPTURE, $offset ) ) { if ( $m[2][1] > $limit ) { break; } $count++; $offset = $m[2][1]; } return $count; } public static function text( $parser, $page, &$title, &$text ) { $title = Title::newFromText( $page ); if ( $title === null ) { $text = ''; return true; } else { $text = $parser->fetchTemplateAndTitle( $title )[0]; } // if article doesn't exist, return a red link. if ( $text == false ) { $text = "[[" . $title->getPrefixedText() . "]]"; return false; } else { return true; } } // section inclusion - include all matching sections public static function includeSection( $parser, $page = '', $sec = '', $to = '', $recursionCheck = true, $trim = false, $skipPattern = [] ) { $output = []; if ( self::text( $parser, $page, $title, $text ) == false ) { $output[] = $text; return $output; } $any = false; $pat = self::createSectionPattern( $sec, $to, $any ); preg_match_all( $pat, $text, $m, PREG_PATTERN_ORDER ); foreach ( $m[2] as $nr => $piece ) { $piece = self::parse( $parser, $piece, "#lst:${page}|${sec}", 0, $recursionCheck, $trim, $skipPattern ); if ( $any ) { $output[] = $m[1][$nr] . '::' . $piece; } else { $output[] = $piece; } } return $output; } /** * Truncate a portion of wikitext so that .. * ... does not contain (open) html comments * ... it is not larger that $lim characters * ... it is balanced in terms of braces, brackets and tags * ... it is cut at a word boundary (white space) if possible * ... can be used as content of a wikitable field without spoiling the whole surrounding wikitext structure * * @param $text the wikitext to be truncated * @param $limit limit of character count for the result * @param $link an optional link which will be appended to the text if it was truncated * * @return string the truncated text; * note that the returned text may be longer than the limit if this is necessary * to return something at all. We do not want to return an empty string if the input is not empty * if the text is already shorter than the limit, the text * will be returned without any checks for balance of tags */ public static function limitTranscludedText( $text, $limit, $link = '' ) { // if text is smaller than limit return complete text if ( $limit >= strlen( $text ) ) { return $text; } // otherwise strip html comments and check again $text = preg_replace( '//s', '', $text ); if ( $limit >= strlen( $text ) ) { return $text; } // search latest position with balanced brackets/braces // store also the position of the last preceding space $brackets = 0; $cbrackets = 0; $n0 = -1; $nb = 0; for ( $i = 0; $i < $limit; $i++ ) { $c = $text[$i]; if ( $c == '[' ) { $brackets++; } if ( $c == ']' ) { $brackets--; } if ( $c == '{' ) { $cbrackets++; } if ( $c == '}' ) { $cbrackets--; } // we store the position if it is valid in terms of parentheses balancing if ( $brackets == 0 && $cbrackets == 0 ) { $n0 = $i; if ( $c == ' ' ) { $nb = $i; } } } // if there is a valid cut-off point we use it; it will be the largest one which is not above the limit if ( $n0 >= 0 ) { // we try to cut off at a word boundary, this may lead to a shortening of max. 15 chars // @phan-suppress-next-line PhanSuspiciousValueComparison if ( $nb > 0 && $nb + 15 > $n0 ) { $n0 = $nb; } $cut = substr( $text, 0, $n0 + 1 ); // an open html comment would be fatal, but this should not happen as we already have // eliminated html comments at the beginning // some tags are critical: ref, pre, nowiki // if these tags were not balanced they would spoil the result completely // we enforce balance by appending the necessary amount of corresponding closing tags // currently we ignore the nesting, i.e. all closing tags are appended at the end. // This simple approach may fail in some cases ... $matches = []; $noMatches = preg_match_all( '#<\s*(/?ref|/?pre|/?nowiki)(\s+[^>]*?)*>#im', $cut, $matches ); $tags = [ 'ref' => 0, 'pre' => 0, 'nowiki' => 0 ]; if ( $noMatches > 0 ) { // calculate tag count (ignoring nesting) foreach ( $matches[1] as $mm ) { if ( $mm[0] == '/' ) { $tags[substr( $mm, 1 )]--; } else { $tags[$mm]++; } } // append missing closing tags - should the tags be ordered by precedence ? foreach ( $tags as $tagName => $level ) { // @phan-suppress-next-line PhanPluginLoopVariableReuse while ( $level > 0 ) { // avoid empty ref tag if ( $tagName == 'ref' && substr( $cut, strlen( $cut ) - 5 ) == '' ) { $cut = substr( $cut, 0, strlen( $cut ) - 5 ); } else { $cut .= ''; } $level--; } } } return $cut . $link; } elseif ( $limit == 0 ) { return $link; } else { // otherwise we recurse and try again with twice the limit size; this will lead to bigger output but // it will at least produce some output at all; otherwise the reader might think that there // is no information at all return self::limitTranscludedText( $text, $limit * 2, $link ); } } public static function includeHeading( $parser, $page, $sec, $to, &$sectionHeading, $recursionCheck, $maxLength, $link, $trim, $skipPattern ) { $output = []; if ( self::text( $parser, $page, $title, $text ) == false ) { $output[0] = $text; return $output; } // throw away comments $text = preg_replace( '//s', '', $text ); return self::extractHeadingFromText( $parser, $page, $title, $text, $sec, $to, $sectionHeading, $recursionCheck, $maxLength, $link, $trim, $skipPattern ); } // section inclusion - include all matching sections (return array) public static function extractHeadingFromText( $parser, $page, $title, $text, $sec, $to, &$sectionHeading, $recursionCheck, $maxLength, $cLink, $trim, $skipPattern = [] ) { $continueSearch = true; $output = []; $n = 0; $output[$n] = ''; $nr = 0; // check if we are going to fetch the n-th section if ( preg_match( '/^%-?[1-9][0-9]*$/', $sec ) ) { $nr = substr( $sec, 1 ); } if ( preg_match( '/^%0$/', $sec ) ) { // transclude text before the first section $nr = -2; } // if the section name starts with a # or with a @ we use it as regexp, otherwise as plain string $isPlain = true; if ( $sec != '' && ( $sec[0] == '#' || $sec[0] == '@' ) ) { $sec = substr( $sec, 1 ); $isPlain = false; } do { // Generate a regex to match the === classical heading section(s) === we're //interested in. $headLine = ''; $begin_off = 0; if ( $sec == '' ) { $head_len = 6; } else { if ( $nr != 0 ) { $pat = '^(={1,6})\s*[^=\s\n][^\n=]*\s*\1\s*($)'; } elseif ( $isPlain ) { $pat = '^(={1,6})\s*' . preg_quote( $sec, '/' ) . '\s*\1\s*($)'; } else { $pat = '^(={1,6})\s*' . str_replace( '/', '\/', $sec ) . '\s*\1\s*($)'; } if ( preg_match( "/$pat/im", $text, $m, PREG_OFFSET_CAPTURE ) ) { $mata = []; $no_parenthesis = preg_match_all( '/\(/', $pat, $mata ); $begin_off = $m[$no_parenthesis][1]; $head_len = strlen( $m[1][0] ); $headLine = trim( $m[0][0], "\n =\t" ); } elseif ( $nr == -2 ) { // take whole article if no heading found $m[1][1] = strlen( $text ) + 1; } else { // match failed return $output; } } // create a link symbol (arrow, img, ...) in case we have to cut the text block to maxLength $link = $cLink; if ( $link == 'default' ) { $link = ' [[' . $page . '#' . $headLine . '|..→]]'; } elseif ( strstr( $link, 'img=' ) != false ) { $link = str_replace( 'img=', "page=" . $page . '#' . $headLine . "\nimg=Image:", $link ) . "\n"; } elseif ( strstr( $link, '%SECTION%' ) == false ) { $link = ' [[' . $page . '#' . $headLine . '|' . $link . ']]'; } else { $link = str_replace( '%SECTION%', $page . '#' . $headLine, $link ); } if ( $nr == -2 ) { // output text before first section and done $piece = substr( $text, 0, $m[1][1] - 1 ); $output[0] = self::parse( $parser, $piece, "#lsth:${page}|${sec}", 0, $recursionCheck, $maxLength, $link, $trim, $skipPattern ); return $output; } if ( isset( $end_off ) ) { unset( $end_off ); } if ( $to != '' ) { // if $to is supplied, try and match it. If we don't match, just ignore it. if ( $isPlain ) { $pat = '^(={1,6})\s*' . preg_quote( $to, '/' ) . '\s*\1\s*$'; } else { $pat = '^(={1,6})\s*' . str_replace( '/', '\/', $to ) . '\s*\1\s*$'; } if ( preg_match( "/$pat/im", $text, $mm, PREG_OFFSET_CAPTURE, $begin_off ) ) { $end_off = $mm[0][1] - 1; } } if ( !isset( $end_off ) ) { if ( $nr != 0 ) { $pat = '^(={1,6})\s*[^\s\n=][^\n=]*\s*\1\s*$'; } else { // @phan-suppress-next-line PhanPossiblyUndeclaredVariable $pat = '^(={1,' . $head_len . '})(?!=)\s*.*?\1\s*$'; } if ( preg_match( "/$pat/im", $text, $mm, PREG_OFFSET_CAPTURE, $begin_off ) ) { $end_off = $mm[0][1] - 1; } elseif ( $sec == '' ) { $end_off = -1; } } $nhead = self::countHeadings( $text, $begin_off ); wfDebug( "LSTH: head offset = $nhead" ); if ( isset( $end_off ) ) { if ( $end_off == -1 ) { return $output; } $piece = substr( $text, $begin_off, $end_off - $begin_off ); if ( $sec == '' ) { $continueSearch = false; } else { $text = substr( $text, $end_off ); } } else { $piece = substr( $text, $begin_off ); $continueSearch = false; } if ( $nr > 1 ) { // skip until we reach the n-th section $nr--; continue; } if ( isset( $m[0][0] ) ) { $sectionHeading[$n] = $headLine; } else { $sectionHeading[0] = $headLine; } if ( $nr == 1 ) { // output n-th section and done $output[0] = self::parse( $parser, $piece, "#lsth:${page}|${sec}", $nhead, $recursionCheck, $maxLength, $link, $trim, $skipPattern ); break; } if ( $nr == -1 ) { if ( !isset( $end_off ) ) { // output last section and done $output[0] = self::parse( $parser, $piece, "#lsth:${page}|${sec}", $nhead, $recursionCheck, $maxLength, $link, $trim, $skipPattern ); break; } } else { // output section by name and continue search for another section with the same name $output[$n++] = self::parse( $parser, $piece, "#lsth:${page}|${sec}", $nhead, $recursionCheck, $maxLength, $link, $trim, $skipPattern ); } } while ( $continueSearch ); return $output; } // template inclusion - find the place(s) where template1 is called, // replace its name by template2, then expand template2 and return the result // we return an array containing all occurences of the template call which match the condition "$mustMatch" // and do NOT match the condition "$mustNotMatch" (if specified) // we use a callback function to format retrieved parameters, accessible via $lister->formatTemplateArg() public static function includeTemplate( $parser, Lister $lister, $dplNr, $article, $template1, $template2, $defaultTemplate, $mustMatch, $mustNotMatch, $matchParsed, $catlist ) { $page = $article->mTitle->getPrefixedText(); $date = $article->myDate; $user = $article->mUserLink; $title = Title::newFromText( $page ); // get text and throw away html comments $text = preg_replace( '//s', '', $parser->fetchTemplateAndTitle( $title )[0] ); if ( $template1 != '' && $template1[0] == '#' ) { // --------------------------------------------- looking for a parser function call $template1 = substr( $template1, 1 ); $template2 = substr( $template2, 1 ); $defaultTemplate = substr( $defaultTemplate, 1 ); // when looking for parser function calls we accept regexp search patterns $text2 = preg_replace( "/\{\{\s*#(" . $template1 . ')(\s*[:}])/i', '°³²|%PFUNC%=\1\2|', $text ); $tCalls = preg_split( '/°³²/', ' ' . $text2 ); foreach ( $tCalls as $i => $tCall ) { $n = strpos( $tCall, ':' ); if ( $n !== false ) { $tCalls[$i][$n] = ' '; } } } elseif ( $template1 != '' && $template1[0] == '~' ) { // --------------------------------------------- looking for an xml-tag extension call $template1 = substr( $template1, 1 ); $template2 = substr( $template2, 1 ); $defaultTemplate = substr( $defaultTemplate, 1 ); // looking for tags $text2 = preg_replace( '/\<\s*(' . $template1 . ')\s*\>/i', '°³²|%TAG%=\1|%TAGBODY%=', $text ); $tCalls = preg_split( '/°³²/', ' ' . $text2 ); foreach ( $tCalls as $i => $tCall ) { $tCalls[$i] = preg_replace( '/\<\s*\/' . $template1 . '\s*\>.*/is', '}}', $tCall ); } } else { // --------------------------------------------- looking for template call // we accept plain text as a template name, space or underscore are the same // the localized name for "Template:" may preceed the template name // the name may start with a different namespace for the surrogate template, followed by :: $contLang = MediaWikiServices::getInstance()->getContentLanguage(); $nsNames = $contLang->getNamespaces(); $tCalls = preg_split( '/\{\{\s*(Template:|' . $nsNames[10] . ':)?' . self::spaceOrUnderscore( preg_quote( $template1, '/' ) ) . '\s*[|}]/i', ' ' . $text ); // We restore the first separator symbol (we had to include that symbol into the SPLIT, because we must make // sure that we only accept exact matches of the complete template name // (e.g. when looking for "foo" we must not accept "foo xyz") foreach ( $tCalls as $nr => $tCall ) { if ( $tCall[0] == '}' ) { $tCalls[$nr] = '}' . $tCall; } else { $tCalls[$nr] = '|' . $tCall; } } } $output = []; $extractParm = []; // check if we want to extract parameters directly from the call // in that case we won´t invoke template2 but will directly return the extracted parameters // as a sequence of table columns; if ( strlen( $template2 ) > strlen( $template1 ) && substr( $template2, 0, strlen( $template1 ) + 1 ) == ( $template1 . ':' ) ) { $extractParm = preg_split( '/:\s*/s', trim( substr( $template2, strlen( $template1 ) + 1 ) ) ); } if ( count( $tCalls ) <= 1 ) { // template was not called (note that count will be 1 if there is no template invocation) if ( count( $extractParm ) > 0 ) { // if parameters are required directly: return empty columns if ( count( $extractParm ) > 1 ) { $output[0] = $lister->formatTemplateArg( '', $dplNr, 0, true, -1, $article ); for ( $i = 1; $i < count( $extractParm ); $i++ ) { $output[0] .= "\n|" . $lister->formatTemplateArg( '', $dplNr, $i, true, -1, $article ); } } else { $output[0] = $lister->formatTemplateArg( '', $dplNr, 0, true, -1, $article ); } } else { // put a red link into the output $output[0] = $parser->preprocess( '{{' . $defaultTemplate . '|%PAGE%=' . $page . '|%TITLE%=' . $title->getText() . '|%DATE%=' . $date . '|%USER%=' . $user . '}}', $parser->getTitle(), $parser->getOptions() ); } unset( $title ); return $output; } $output[0] = ''; $n = -2; // loop for all template invocations $firstCall = true; foreach ( $tCalls as $iii => $tCall ) { if ( $n == -2 ) { $n++; continue; } $c = $tCall[0]; // normally we construct a call for template2 with the parameters of template1 if ( count( $extractParm ) == 0 ) { // find the end of the call: bracket level must be zero $cbrackets = 0; $templateCall = '{{' . $template2 . $tCall; $size = strlen( $templateCall ); for ( $i = 0; $i < $size; $i++ ) { $c = $templateCall[$i]; if ( $c == '{' ) { $cbrackets++; } if ( $c == '}' ) { $cbrackets--; } if ( $cbrackets == 0 ) { // if we must match a condition: test against it if ( ( $mustMatch == '' || preg_match( $mustMatch, substr( $templateCall, 0, $i - 1 ) ) ) && ( $mustNotMatch == '' || !preg_match( $mustNotMatch, substr( $templateCall, 0, $i - 1 ) ) ) ) { $invocation = substr( $templateCall, 0, $i - 1 ); $argChain = $invocation . '|%PAGE%=' . $page . '|%TITLE%=' . $title->getText(); if ( $catlist != '' ) { $argChain .= "|%CATLIST%=$catlist"; } $argChain .= '|%DATE%=' . $date . '|%USER%=' . $user . '|%ARGS%=' . str_replace( '|', '§', preg_replace( '/[}]+/', '}', preg_replace( '/[{]+/', '{', substr( $invocation, strlen( $template2 ) + 2 ) ) ) ) . '}}'; $output[++$n] = $parser->preprocess( $argChain, $parser->getTitle(), $parser->getOptions() ); } break; } } } else { // if the user wants parameters directly from the call line of template1 we return just those $cbrackets = 2; $templateCall = $tCall; $size = strlen( $templateCall ); $parms = []; $parm = ''; $hasParm = false; for ( $i = 0; $i < $size; $i++ ) { $c = $templateCall[$i]; if ( $c == '{' || $c == '[' ) { // we count both types of brackets $cbrackets++; } if ( $c == '}' || $c == ']' ) { $cbrackets--; } if ( $cbrackets == 2 && $c == '|' ) { $parms[] = trim( $parm ); $hasParm = true; $parm = ''; } else { $parm .= $c; } if ( $cbrackets == 0 ) { if ( $hasParm ) { $parms[] = trim( substr( $parm, 0, strlen( $parm ) - 2 ) ); } array_splice( $parms, 0, 1 ); // if we must match a condition: test against it $callText = substr( $templateCall, 0, $i - 1 ); if ( ( $mustMatch == '' || ( ( $matchParsed && preg_match( $mustMatch, $parser->recursiveTagParse( $callText ) ) ) || ( !$matchParsed && preg_match( $mustMatch, $callText ) ) ) ) && ( $mustNotMatch == '' || ( ( $matchParsed && !preg_match( $mustNotMatch, $parser->recursiveTagParse( $callText ) ) ) || ( !$matchParsed && !preg_match( $mustNotMatch, $callText ) ) ) ) ) { $output[++$n] = ''; $second = false; foreach ( $extractParm as $exParmKey => $exParm ) { $maxlen = -1; $limpos = strpos( $exParm, '[' ); if ( $limpos > 0 && $exParm[strlen( $exParm ) - 1] == ']' ) { $maxlen = intval( substr( $exParm, $limpos + 1, strlen( $exParm ) - $limpos - 2 ) ); $exParm = substr( $exParm, 0, $limpos ); } if ( $second ) { // @phan-suppress-next-line PhanTypeInvalidDimOffset if ( $output[$n] == '' || $output[$n][strlen( $output[$n] ) - 1] != "\n" ) { $output[$n] .= "\n"; } $output[$n] .= "|"; } $found = false; // % in parameter name if ( strpos( $exParm, '%' ) !== false ) { // %% is a short form for inclusion of %PAGE% and %TITLE% $found = true; $output[$n] .= $lister->formatTemplateArg( $exParm, $dplNr, $exParmKey, $firstCall, $maxlen, $article ); } if ( !$found ) { // named parameter $exParmQuote = str_replace( '/', '\/', $exParm ); foreach ( $parms as $parm ) { if ( !preg_match( "/^\s*$exParmQuote\s*=/", $parm ) ) { continue; } $found = true; $output[$n] .= $lister->formatTemplateArg( preg_replace( "/^$exParmQuote\s*=\s*/", "", $parm ), $dplNr, $exParmKey, $firstCall, $maxlen, $article ); break; } } if ( !$found && is_numeric( $exParm ) && intval( $exParm ) == $exParm ) { // numeric parameter $np = 0; foreach ( $parms as $parm ) { if ( strstr( $parm, '=' ) === false ) { ++$np; } if ( $np != $exParm ) { continue; } $found = true; $output[$n] .= $lister->formatTemplateArg( $parm, $dplNr, $exParmKey, $firstCall, $maxlen, $article ); break; } } if ( !$found ) { $output[$n] .= $lister->formatTemplateArg( '', $dplNr, $exParmKey, $firstCall, $maxlen, $article ); } $second = true; } } break; } } } $firstCall = false; } return $output; } public static function spaceOrUnderscore( $pattern ) { // returns a pettern that matches underscores as well as spaces return str_replace( ' ', '[ _]', $pattern ); } }