parser = $parser; } /** * @return PPFrame_HipHop */ function newFrame() { return new PPFrame_HipHop( $this ); } /** * @param $args array * @return PPCustomFrame_HipHop */ function newCustomFrame( $args ) { return new PPCustomFrame_HipHop( $this, $args ); } /** * @param $values array * @return PPNode_HipHop_Array */ function newPartNodeArray( $values ) { $list = array(); foreach ( $values as $k => $val ) { $partNode = new PPNode_HipHop_Tree( 'part' ); $nameNode = new PPNode_HipHop_Tree( 'name' ); if ( is_int( $k ) ) { $nameNode->addChild( new PPNode_HipHop_Attr( 'index', $k ) ); $partNode->addChild( $nameNode ); } else { $nameNode->addChild( new PPNode_HipHop_Text( $k ) ); $partNode->addChild( $nameNode ); $partNode->addChild( new PPNode_HipHop_Text( '=' ) ); } $valueNode = new PPNode_HipHop_Tree( 'value' ); $valueNode->addChild( new PPNode_HipHop_Text( $val ) ); $partNode->addChild( $valueNode ); $list[] = $partNode; } $node = new PPNode_HipHop_Array( $list ); return $node; } /** * Preprocess some wikitext and return the document tree. * This is the ghost of Parser::replace_variables(). * * @param $text String: the text to parse * @param $flags Integer: bitwise combination of: * Parser::PTD_FOR_INCLUSION Handle / as if the text is being * included. Default is to assume a direct page view. * * The generated DOM tree must depend only on the input text and the flags. * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899. * * Any flag added to the $flags parameter here, or any other parameter liable to cause a * change in the DOM tree for a given text, must be passed through the section identifier * in the section edit link and thus back to extractSections(). * * The output of this function is currently only cached in process memory, but a persistent * cache may be implemented at a later date which takes further advantage of these strict * dependency requirements. * * @throws MWException * @return PPNode_HipHop_Tree */ function preprocessToObj( $text, $flags = 0 ) { wfProfileIn( __METHOD__ ); // Check cache. global $wgMemc, $wgPreprocessorCacheThreshold; $lengthText = strlen( $text ); $cacheable = ($wgPreprocessorCacheThreshold !== false && $lengthText > $wgPreprocessorCacheThreshold); if ( $cacheable ) { wfProfileIn( __METHOD__.'-cacheable' ); $cacheKey = strval( wfMemcKey( 'preprocess-hash', md5($text), $flags ) ); $cacheValue = strval( $wgMemc->get( $cacheKey ) ); if ( $cacheValue !== '' ) { $version = substr( $cacheValue, 0, 8 ); if ( intval( $version ) == self::CACHE_VERSION ) { $hash = unserialize( substr( $cacheValue, 8 ) ); // From the cache wfDebugLog( "Preprocessor", "Loaded preprocessor hash from memcached (key $cacheKey)" ); wfProfileOut( __METHOD__.'-cacheable' ); wfProfileOut( __METHOD__ ); return $hash; } } wfProfileIn( __METHOD__.'-cache-miss' ); } $rules = array( '{' => array( 'end' => '}', 'names' => array( 2 => 'template', 3 => 'tplarg', ), 'min' => 2, 'max' => 3, ), '[' => array( 'end' => ']', 'names' => array( 2 => 'LITERAL' ), 'min' => 2, 'max' => 2, ) ); $forInclusion = (bool)( $flags & Parser::PTD_FOR_INCLUSION ); $xmlishElements = (array)$this->parser->getStripList(); $enableOnlyinclude = false; if ( $forInclusion ) { $ignoredTags = array( 'includeonly', '/includeonly' ); $ignoredElements = array( 'noinclude' ); $xmlishElements[] = 'noinclude'; if ( strpos( $text, '' ) !== false && strpos( $text, '' ) !== false ) { $enableOnlyinclude = true; } } else if ( $this->parser->ot['wiki'] ) { $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude', 'includeonly', '/includeonly' ); $ignoredElements = array(); } else { $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' ); $ignoredElements = array( 'includeonly' ); $xmlishElements[] = 'includeonly'; } $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) ); // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA"; $stack = new PPDStack_HipHop; $searchBase = "[{<\n"; $revText = strrev( $text ); // For fast reverse searches $i = 0; # Input pointer, starts out pointing to a pseudo-newline before the start $accum = $stack->getAccum(); # Current accumulator $headingIndex = 1; $stackFlags = array( 'findPipe' => false, # True to take notice of pipe characters 'findEquals' => false, # True to find equals signs in arguments 'inHeading' => false, # True if $i is inside a possible heading ); $noMoreGT = false; # True if there are no more greater-than (>) signs right of $i $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next $fakeLineStart = true; # Do a line-start run without outputting an LF character while ( true ) { //$this->memCheck(); if ( $findOnlyinclude ) { // Ignore all input up to the next $variantStartPos = strpos( $text, '', $i ); if ( $variantStartPos === false ) { // Ignored section runs to the end $accum->addNodeWithText( 'ignore', strval( substr( $text, $i ) ) ); break; } $startPos1 = intval( $variantStartPos ); $tagEndPos = $startPos1 + strlen( '' ); // past-the-end $accum->addNodeWithText( 'ignore', strval( substr( $text, $i, $tagEndPos - $i ) ) ); $i = $tagEndPos; $findOnlyinclude = false; } if ( $fakeLineStart ) { $found = 'line-start'; $curChar = ''; } else { # Find next opening brace, closing brace or pipe $search = $searchBase; if ( $stack->top === false ) { $currentClosing = ''; } else { $currentClosing = strval( $stack->getTop()->close ); $search .= $currentClosing; } if ( $stackFlags['findPipe'] ) { $search .= '|'; } if ( $stackFlags['findEquals'] ) { // First equals will be for the template $search .= '='; } $rule = null; # Output literal section, advance input counter $literalLength = intval( strcspn( $text, $search, $i ) ); if ( $literalLength > 0 ) { $accum->addLiteral( strval( substr( $text, $i, $literalLength ) ) ); $i += $literalLength; } if ( $i >= $lengthText ) { if ( $currentClosing === "\n" ) { // Do a past-the-end run to finish off the heading $curChar = ''; $found = 'line-end'; } else { # All done break; } } else { $curChar = $text[$i]; if ( $curChar === '|' ) { $found = 'pipe'; } elseif ( $curChar === '=' ) { $found = 'equals'; } elseif ( $curChar === '<' ) { $found = 'angle'; } elseif ( $curChar === "\n" ) { if ( $stackFlags['inHeading'] ) { $found = 'line-end'; } else { $found = 'line-start'; } } elseif ( $curChar === $currentClosing ) { $found = 'close'; } elseif ( isset( $rules[$curChar] ) ) { $found = 'open'; $rule = $rules[$curChar]; } else { # Some versions of PHP have a strcspn which stops on null characters # Ignore and continue ++$i; continue; } } } if ( $found === 'angle' ) { $matches = false; // Handle if ( $enableOnlyinclude && substr( $text, $i, strlen( '' ) ) === '' ) { $findOnlyinclude = true; continue; } // Determine element name if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) { // Element name missing or not listed $accum->addLiteral( '<' ); ++$i; continue; } // Handle comments if ( isset( $matches[2] ) && $matches[2] === '!--' ) { // To avoid leaving blank lines, when a comment is both preceded // and followed by a newline (ignoring spaces), trim leading and // trailing spaces and one of the newlines. // Find the end $variantEndPos = strpos( $text, '-->', $i + 4 ); if ( $variantEndPos === false ) { // Unclosed comment in input, runs to end $inner = strval( substr( $text, $i ) ); $accum->addNodeWithText( 'comment', $inner ); $i = $lengthText; } else { $endPos = intval( $variantEndPos ); // Search backwards for leading whitespace if ( $i ) { $wsStart = $i - intval( strspn( $revText, ' ', $lengthText - $i ) ); } else { $wsStart = 0; } // Search forwards for trailing whitespace // $wsEnd will be the position of the last space (or the '>' if there's none) $wsEnd = $endPos + 2 + intval( strspn( $text, ' ', $endPos + 3 ) ); // Eat the line if possible // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but // it's a possible beneficial b/c break. if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) === "\n" && substr( $text, $wsEnd + 1, 1 ) === "\n" ) { $startPos2 = $wsStart; $endPos = $wsEnd + 1; // Remove leading whitespace from the end of the accumulator // Sanity check first though $wsLength = $i - $wsStart; if ( $wsLength > 0 && $accum->lastNode instanceof PPNode_HipHop_Text && substr( $accum->lastNode->value, -$wsLength ) === str_repeat( ' ', $wsLength ) ) { $accum->lastNode->value = strval( substr( $accum->lastNode->value, 0, -$wsLength ) ); } // Do a line-start run next time to look for headings after the comment $fakeLineStart = true; } else { // No line to eat, just take the comment itself $startPos2 = $i; $endPos += 2; } if ( $stack->top ) { $part = $stack->getTop()->getCurrentPart(); if ( ! (isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 )) { $part->visualEnd = $wsStart; } // Else comments abutting, no change in visual end $part->commentEnd = $endPos; } $i = $endPos + 1; $inner = strval( substr( $text, $startPos2, $endPos - $startPos2 + 1 ) ); $accum->addNodeWithText( 'comment', $inner ); } continue; } $name = strval( $matches[1] ); $lowerName = strtolower( $name ); $attrStart = $i + strlen( $name ) + 1; // Find end of tag $variantTagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart ); if ( $variantTagEndPos === false ) { // Infinite backtrack // Disable tag search to prevent worst-case O(N^2) performance $noMoreGT = true; $accum->addLiteral( '<' ); ++$i; continue; } $tagEndPos = intval( $variantTagEndPos ); // Handle ignored tags if ( in_array( $lowerName, $ignoredTags ) ) { $accum->addNodeWithText( 'ignore', strval( substr( $text, $i, $tagEndPos - $i + 1 ) ) ); $i = $tagEndPos + 1; continue; } $tagStartPos = $i; $close = ''; if ( $text[$tagEndPos-1] === '/' ) { // Short end tag $attrEnd = $tagEndPos - 1; $shortEnd = true; $inner = ''; $i = $tagEndPos + 1; $haveClose = false; } else { $attrEnd = $tagEndPos; $shortEnd = false; // Find closing tag if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) { $inner = strval( substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 ) ); $i = intval( $matches[0][1] ) + strlen( $matches[0][0] ); $close = strval( $matches[0][0] ); $haveClose = true; } else { // No end tag -- let it run out to the end of the text. $inner = strval( substr( $text, $tagEndPos + 1 ) ); $i = $lengthText; $haveClose = false; } } // and just become tags if ( in_array( $lowerName, $ignoredElements ) ) { $accum->addNodeWithText( 'ignore', strval( substr( $text, $tagStartPos, $i - $tagStartPos ) ) ); continue; } if ( $attrEnd <= $attrStart ) { $attr = ''; } else { // Note that the attr element contains the whitespace between name and attribute, // this is necessary for precise reconstruction during pre-save transform. $attr = strval( substr( $text, $attrStart, $attrEnd - $attrStart ) ); } $extNode = new PPNode_HipHop_Tree( 'ext' ); $extNode->addChild( PPNode_HipHop_Tree::newWithText( 'name', $name ) ); $extNode->addChild( PPNode_HipHop_Tree::newWithText( 'attr', $attr ) ); if ( !$shortEnd ) { $extNode->addChild( PPNode_HipHop_Tree::newWithText( 'inner', $inner ) ); } if ( $haveClose ) { $extNode->addChild( PPNode_HipHop_Tree::newWithText( 'close', $close ) ); } $accum->addNode( $extNode ); } elseif ( $found === 'line-start' ) { // Is this the start of a heading? // Line break belongs before the heading element in any case if ( $fakeLineStart ) { $fakeLineStart = false; } else { $accum->addLiteral( $curChar ); $i++; } $count = intval( strspn( $text, '=', $i, 6 ) ); if ( $count == 1 && $stackFlags['findEquals'] ) { // DWIM: This looks kind of like a name/value separator // Let's let the equals handler have it and break the potential heading // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex. } elseif ( $count > 0 ) { $partData = array( 'open' => "\n", 'close' => "\n", 'parts' => array( new PPDPart_HipHop( str_repeat( '=', $count ) ) ), 'startPos' => $i, 'count' => $count ); $stack->push( $partData ); $accum = $stack->getAccum(); $stackFlags = $stack->getFlags(); $i += $count; } } elseif ( $found === 'line-end' ) { $piece = $stack->getTop(); // A heading must be open, otherwise \n wouldn't have been in the search list assert( $piece->open === "\n" ); // Passing the assert condition directly instead of string, as // HPHP /compiler/ chokes on strings when ASSERT_ACTIVE != 0. $part = $piece->getCurrentPart(); // Search back through the input to see if it has a proper close // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient $wsLength = intval( strspn( $revText, " \t", $lengthText - $i ) ); $searchStart = $i - $wsLength; if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) { // Comment found at line end // Search for equals signs before the comment $searchStart = intval( $part->visualEnd ); $searchStart -= intval( strspn( $revText, " \t", $lengthText - $searchStart ) ); } $count = intval( $piece->count ); $equalsLength = intval( strspn( $revText, '=', $lengthText - $searchStart ) ); $isTreeNode = false; $resultAccum = $accum; if ( $equalsLength > 0 ) { if ( $searchStart - $equalsLength == $piece->startPos ) { // This is just a single string of equals signs on its own line // Replicate the doHeadings behaviour /={count}(.+)={count}/ // First find out how many equals signs there really are (don't stop at 6) $count = $equalsLength; if ( $count < 3 ) { $count = 0; } else { $count = intval( ( $count - 1 ) / 2 ); if ( $count > 6 ) { $count = 6; } } } else { if ( $count > $equalsLength ) { $count = $equalsLength; } } if ( $count > 0 ) { // Normal match, output $tree = new PPNode_HipHop_Tree( 'possible-h' ); $tree->addChild( new PPNode_HipHop_Attr( 'level', $count ) ); $tree->addChild( new PPNode_HipHop_Attr( 'i', $headingIndex++ ) ); $tree->lastChild->nextSibling = $accum->firstNode; $tree->lastChild = $accum->lastNode; $isTreeNode = true; } else { // Single equals sign on its own line, count=0 // Output $resultAccum } } else { // No match, no , just pass down the inner text // Output $resultAccum } // Unwind the stack $stack->pop(); $accum = $stack->getAccum(); $stackFlags = $stack->getFlags(); // Append the result to the enclosing accumulator if ( $isTreeNode ) { $accum->addNode( $tree ); } else { $accum->addAccum( $resultAccum ); } // Note that we do NOT increment the input pointer. // This is because the closing linebreak could be the opening linebreak of // another heading. Infinite loops are avoided because the next iteration MUST // hit the heading open case above, which unconditionally increments the // input pointer. } elseif ( $found === 'open' ) { # count opening brace characters $count = intval( strspn( $text, $curChar, $i ) ); # we need to add to stack only if opening brace count is enough for one of the rules if ( $count >= $rule['min'] ) { # Add it to the stack $partData = array( 'open' => $curChar, 'close' => $rule['end'], 'count' => $count, 'lineStart' => ($i == 0 || $text[$i-1] === "\n"), ); $stack->push( $partData ); $accum = $stack->getAccum(); $stackFlags = $stack->getFlags(); } else { # Add literal brace(s) $accum->addLiteral( str_repeat( $curChar, $count ) ); } $i += $count; } elseif ( $found === 'close' ) { $piece = $stack->getTop(); # lets check if there are enough characters for closing brace $maxCount = intval( $piece->count ); $count = intval( strspn( $text, $curChar, $i, $maxCount ) ); # check for maximum matching characters (if there are 5 closing # characters, we will probably need only 3 - depending on the rules) $rule = $rules[$piece->open]; if ( $count > $rule['max'] ) { # The specified maximum exists in the callback array, unless the caller # has made an error $matchingCount = intval( $rule['max'] ); } else { # Count is less than the maximum # Skip any gaps in the callback array to find the true largest match # Need to use array_key_exists not isset because the callback can be null $matchingCount = $count; while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) { --$matchingCount; } } if ($matchingCount <= 0) { # No matching element found in callback array # Output a literal closing brace and continue $accum->addLiteral( str_repeat( $curChar, $count ) ); $i += $count; continue; } $name = strval( $rule['names'][$matchingCount] ); $isTreeNode = false; if ( $name === 'LITERAL' ) { // No element, just literal text $resultAccum = $piece->breakSyntax( $matchingCount ); $resultAccum->addLiteral( str_repeat( $rule['end'], $matchingCount ) ); } else { # Create XML element # Note: $parts is already XML, does not need to be encoded further $isTreeNode = true; $parts = $piece->parts; $titleAccum = PPDAccum_HipHop::cast( $parts[0]->out ); unset( $parts[0] ); $tree = new PPNode_HipHop_Tree( $name ); # The invocation is at the start of the line if lineStart is set in # the stack, and all opening brackets are used up. if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) { $tree->addChild( new PPNode_HipHop_Attr( 'lineStart', 1 ) ); } $titleNode = new PPNode_HipHop_Tree( 'title' ); $titleNode->firstChild = $titleAccum->firstNode; $titleNode->lastChild = $titleAccum->lastNode; $tree->addChild( $titleNode ); $argIndex = 1; foreach ( $parts as $variantPart ) { $part = PPDPart_HipHop::cast( $variantPart ); if ( isset( $part->eqpos ) ) { // Find equals $lastNode = false; for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) { if ( $node === $part->eqpos ) { break; } $lastNode = $node; } if ( !$node ) { throw new MWException( __METHOD__. ': eqpos not found' ); } if ( $node->name !== 'equals' ) { throw new MWException( __METHOD__ .': eqpos is not equals' ); } $equalsNode = $node; // Construct name node $nameNode = new PPNode_HipHop_Tree( 'name' ); if ( $lastNode !== false ) { $lastNode->nextSibling = false; $nameNode->firstChild = $part->out->firstNode; $nameNode->lastChild = $lastNode; } // Construct value node $valueNode = new PPNode_HipHop_Tree( 'value' ); if ( $equalsNode->nextSibling !== false ) { $valueNode->firstChild = $equalsNode->nextSibling; $valueNode->lastChild = $part->out->lastNode; } $partNode = new PPNode_HipHop_Tree( 'part' ); $partNode->addChild( $nameNode ); $partNode->addChild( $equalsNode->firstChild ); $partNode->addChild( $valueNode ); $tree->addChild( $partNode ); } else { $partNode = new PPNode_HipHop_Tree( 'part' ); $nameNode = new PPNode_HipHop_Tree( 'name' ); $nameNode->addChild( new PPNode_HipHop_Attr( 'index', $argIndex++ ) ); $valueNode = new PPNode_HipHop_Tree( 'value' ); $valueNode->firstChild = $part->out->firstNode; $valueNode->lastChild = $part->out->lastNode; $partNode->addChild( $nameNode ); $partNode->addChild( $valueNode ); $tree->addChild( $partNode ); } } } # Advance input pointer $i += $matchingCount; # Unwind the stack $stack->pop(); $accum = $stack->getAccum(); # Re-add the old stack element if it still has unmatched opening characters remaining if ($matchingCount < $piece->count) { $piece->parts = array( new PPDPart_HipHop ); $piece->count -= $matchingCount; # do we still qualify for any callback with remaining count? $names = $rules[$piece->open]['names']; $skippedBraces = 0; $enclosingAccum = $accum; while ( $piece->count ) { if ( array_key_exists( $piece->count, $names ) ) { $stack->push( $piece ); $accum = $stack->getAccum(); break; } --$piece->count; $skippedBraces ++; } $enclosingAccum->addLiteral( str_repeat( $piece->open, $skippedBraces ) ); } $stackFlags = $stack->getFlags(); # Add XML element to the enclosing accumulator if ( $isTreeNode ) { $accum->addNode( $tree ); } else { $accum->addAccum( $resultAccum ); } } elseif ( $found === 'pipe' ) { $stackFlags['findEquals'] = true; // shortcut for getFlags() $stack->addPart(); $accum = $stack->getAccum(); ++$i; } elseif ( $found === 'equals' ) { $stackFlags['findEquals'] = false; // shortcut for getFlags() $accum->addNodeWithText( 'equals', '=' ); $stack->getCurrentPart()->eqpos = $accum->lastNode; ++$i; } } # Output any remaining unclosed brackets foreach ( $stack->stack as $variantPiece ) { $piece = PPDStackElement_HipHop::cast( $variantPiece ); $stack->rootAccum->addAccum( $piece->breakSyntax() ); } # Enable top-level headings for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) { if ( isset( $node->name ) && $node->name === 'possible-h' ) { $node->name = 'h'; } } $rootNode = new PPNode_HipHop_Tree( 'root' ); $rootNode->firstChild = $stack->rootAccum->firstNode; $rootNode->lastChild = $stack->rootAccum->lastNode; // Cache if ($cacheable) { $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . serialize( $rootNode ); $wgMemc->set( $cacheKey, $cacheValue, 86400 ); wfProfileOut( __METHOD__.'-cache-miss' ); wfProfileOut( __METHOD__.'-cacheable' ); wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" ); } wfProfileOut( __METHOD__ ); return $rootNode; } } /** * Stack class to help Preprocessor::preprocessToObj() * @ingroup Parser */ class PPDStack_HipHop { var $stack, $rootAccum; /** * @var PPDStack */ var $top; var $out; static $false = false; function __construct() { $this->stack = array(); $this->top = false; $this->rootAccum = new PPDAccum_HipHop; $this->accum = $this->rootAccum; } /** * @return int */ function count() { return count( $this->stack ); } function getAccum() { return PPDAccum_HipHop::cast( $this->accum ); } function getCurrentPart() { return $this->getTop()->getCurrentPart(); } function getTop() { return PPDStackElement_HipHop::cast( $this->top ); } function push( $data ) { if ( $data instanceof PPDStackElement_HipHop ) { $this->stack[] = $data; } else { $this->stack[] = new PPDStackElement_HipHop( $data ); } $this->top = $this->stack[ count( $this->stack ) - 1 ]; $this->accum = $this->top->getAccum(); } function pop() { if ( !count( $this->stack ) ) { throw new MWException( __METHOD__.': no elements remaining' ); } $temp = array_pop( $this->stack ); if ( count( $this->stack ) ) { $this->top = $this->stack[ count( $this->stack ) - 1 ]; $this->accum = $this->top->getAccum(); } else { $this->top = self::$false; $this->accum = $this->rootAccum; } return $temp; } function addPart( $s = '' ) { $this->top->addPart( $s ); $this->accum = $this->top->getAccum(); } /** * @return array */ function getFlags() { if ( !count( $this->stack ) ) { return array( 'findEquals' => false, 'findPipe' => false, 'inHeading' => false, ); } else { return $this->top->getFlags(); } } } /** * @ingroup Parser */ class PPDStackElement_HipHop { var $open, // Opening character (\n for heading) $close, // Matching closing character $count, // Number of opening characters found (number of "=" for heading) $parts, // Array of PPDPart objects describing pipe-separated parts. $lineStart; // True if the open char appeared at the start of the input line. Not set for headings. /** * @param $obj PPDStackElement_HipHop * @return PPDStackElement_HipHop */ static function cast( PPDStackElement_HipHop $obj ) { return $obj; } /** * @param $data array */ function __construct( $data = array() ) { $this->parts = array( new PPDPart_HipHop ); foreach ( $data as $name => $value ) { $this->$name = $value; } } /** * @return PPDAccum_HipHop */ function getAccum() { return PPDAccum_HipHop::cast( $this->parts[count($this->parts) - 1]->out ); } /** * @param $s string */ function addPart( $s = '' ) { $this->parts[] = new PPDPart_HipHop( $s ); } /** * @return PPDPart_HipHop */ function getCurrentPart() { return PPDPart_HipHop::cast( $this->parts[count($this->parts) - 1] ); } /** * @return array */ function getFlags() { $partCount = count( $this->parts ); $findPipe = $this->open !== "\n" && $this->open !== '['; return array( 'findPipe' => $findPipe, 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts[$partCount - 1]->eqpos ), 'inHeading' => $this->open === "\n", ); } /** * Get the accumulator that would result if the close is not found. * * @param $openingCount bool * @return PPDAccum_HipHop */ function breakSyntax( $openingCount = false ) { if ( $this->open === "\n" ) { $accum = PPDAccum_HipHop::cast( $this->parts[0]->out ); } else { if ( $openingCount === false ) { $openingCount = $this->count; } $accum = new PPDAccum_HipHop; $accum->addLiteral( str_repeat( $this->open, $openingCount ) ); $first = true; foreach ( $this->parts as $part ) { if ( $first ) { $first = false; } else { $accum->addLiteral( '|' ); } $accum->addAccum( $part->out ); } } return $accum; } } /** * @ingroup Parser */ class PPDPart_HipHop { var $out; // Output accumulator object // Optional member variables: // eqpos Position of equals sign in output accumulator // commentEnd Past-the-end input pointer for the last comment encountered // visualEnd Past-the-end input pointer for the end of the accumulator minus comments function __construct( $out = '' ) { $this->out = new PPDAccum_HipHop; if ( $out !== '' ) { $this->out->addLiteral( $out ); } } static function cast( PPDPart_HipHop $obj ) { return $obj; } } /** * @ingroup Parser */ class PPDAccum_HipHop { var $firstNode, $lastNode; function __construct() { $this->firstNode = $this->lastNode = false; } static function cast( PPDAccum_HipHop $obj ) { return $obj; } /** * Append a string literal */ function addLiteral( string $s ) { if ( $this->lastNode === false ) { $this->firstNode = $this->lastNode = new PPNode_HipHop_Text( $s ); } elseif ( $this->lastNode instanceof PPNode_HipHop_Text ) { $this->lastNode->value .= $s; } else { $this->lastNode->nextSibling = new PPNode_HipHop_Text( $s ); $this->lastNode = $this->lastNode->nextSibling; } } /** * Append a PPNode */ function addNode( PPNode $node ) { if ( $this->lastNode === false ) { $this->firstNode = $this->lastNode = $node; } else { $this->lastNode->nextSibling = $node; $this->lastNode = $node; } } /** * Append a tree node with text contents */ function addNodeWithText( string $name, string $value ) { $node = PPNode_HipHop_Tree::newWithText( $name, $value ); $this->addNode( $node ); } /** * Append a PPDAccum_HipHop * Takes over ownership of the nodes in the source argument. These nodes may * subsequently be modified, especially nextSibling. */ function addAccum( PPDAccum_HipHop $accum ) { if ( $accum->lastNode === false ) { // nothing to add } elseif ( $this->lastNode === false ) { $this->firstNode = $accum->firstNode; $this->lastNode = $accum->lastNode; } else { $this->lastNode->nextSibling = $accum->firstNode; $this->lastNode = $accum->lastNode; } } } /** * An expansion frame, used as a context to expand the result of preprocessToObj() * @ingroup Parser */ class PPFrame_HipHop implements PPFrame { /** * @var Parser */ var $parser; /** * @var Preprocessor */ var $preprocessor; /** * @var Title */ var $title; var $titleCache; /** * Hashtable listing templates which are disallowed for expansion in this frame, * having been encountered previously in parent frames. */ var $loopCheckHash; /** * Recursion depth of this frame, top = 0 * Note that this is NOT the same as expansion depth in expand() */ var $depth; /** * Construct a new preprocessor frame. * @param $preprocessor Preprocessor: the parent preprocessor */ function __construct( $preprocessor ) { $this->preprocessor = $preprocessor; $this->parser = $preprocessor->parser; $this->title = $this->parser->mTitle; $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false ); $this->loopCheckHash = array(); $this->depth = 0; } /** * Create a new child frame * $args is optionally a multi-root PPNode or array containing the template arguments * * @param $args PPNode_HipHop_Array|array|bool * @param $title Title|bool * @param $indexOffset A number subtracted from the index attributes of the arguments * * @throws MWException * @return PPTemplateFrame_HipHop */ function newChild( $args = false, $title = false, $indexOffset = 0 ) { $namedArgs = array(); $numberedArgs = array(); if ( $title === false ) { $title = $this->title; } if ( $args !== false ) { if ( $args instanceof PPNode_HipHop_Array ) { $args = $args->value; } elseif ( !is_array( $args ) ) { throw new MWException( __METHOD__ . ': $args must be array or PPNode_HipHop_Array' ); } foreach ( $args as $arg ) { $bits = $arg->splitArg(); if ( $bits['index'] !== '' ) { // Numbered parameter $numberedArgs[$bits['index']] = $bits['value']; unset( $namedArgs[$bits['index']] ); } else { // Named parameter $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) ); $namedArgs[$name] = $bits['value']; unset( $numberedArgs[$name] ); } } } return new PPTemplateFrame_HipHop( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title ); } /** * @throws MWException * @param $root * @param $flags int * @return string */ function expand( $root, $flags = 0 ) { static $expansionDepth = 0; if ( is_string( $root ) ) { return $root; } if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) { $this->parser->limitationWarn( 'node-count-exceeded', $this->parser->mPPNodeCount, $this->parser->mOptions->getMaxPPNodeCount() ); return 'Node-count limit exceeded'; } if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) { $this->parser->limitationWarn( 'expansion-depth-exceeded', $expansionDepth, $this->parser->mOptions->getMaxPPExpandDepth() ); return 'Expansion depth limit exceeded'; } ++$expansionDepth; if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) { $this->parser->mHighestExpansionDepth = $expansionDepth; } $outStack = array( '', '' ); $iteratorStack = array( false, $root ); $indexStack = array( 0, 0 ); while ( count( $iteratorStack ) > 1 ) { $level = count( $outStack ) - 1; $iteratorNode =& $iteratorStack[ $level ]; $out =& $outStack[$level]; $index =& $indexStack[$level]; if ( is_array( $iteratorNode ) ) { if ( $index >= count( $iteratorNode ) ) { // All done with this iterator $iteratorStack[$level] = false; $contextNode = false; } else { $contextNode = $iteratorNode[$index]; $index++; } } elseif ( $iteratorNode instanceof PPNode_HipHop_Array ) { if ( $index >= $iteratorNode->getLength() ) { // All done with this iterator $iteratorStack[$level] = false; $contextNode = false; } else { $contextNode = $iteratorNode->item( $index ); $index++; } } else { // Copy to $contextNode and then delete from iterator stack, // because this is not an iterator but we do have to execute it once $contextNode = $iteratorStack[$level]; $iteratorStack[$level] = false; } $newIterator = false; if ( $contextNode === false ) { // nothing to do } elseif ( is_string( $contextNode ) ) { $out .= $contextNode; } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_HipHop_Array ) { $newIterator = $contextNode; } elseif ( $contextNode instanceof PPNode_HipHop_Attr ) { // No output } elseif ( $contextNode instanceof PPNode_HipHop_Text ) { $out .= $contextNode->value; } elseif ( $contextNode instanceof PPNode_HipHop_Tree ) { if ( $contextNode->name === 'template' ) { # Double-brace expansion $bits = $contextNode->splitTemplate(); if ( $flags & PPFrame::NO_TEMPLATES ) { $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $bits['title'], $bits['parts'] ); } else { $ret = $this->parser->braceSubstitution( $bits, $this ); if ( isset( $ret['object'] ) ) { $newIterator = $ret['object']; } else { $out .= $ret['text']; } } } elseif ( $contextNode->name === 'tplarg' ) { # Triple-brace expansion $bits = $contextNode->splitTemplate(); if ( $flags & PPFrame::NO_ARGS ) { $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $bits['title'], $bits['parts'] ); } else { $ret = $this->parser->argSubstitution( $bits, $this ); if ( isset( $ret['object'] ) ) { $newIterator = $ret['object']; } else { $out .= $ret['text']; } } } elseif ( $contextNode->name === 'comment' ) { # HTML-style comment # Remove it in HTML, pre+remove and STRIP_COMMENTS modes if ( $this->parser->ot['html'] || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() ) || ( $flags & PPFrame::STRIP_COMMENTS ) ) { $out .= ''; } # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result # Not in RECOVER_COMMENTS mode (extractSections) though elseif ( $this->parser->ot['wiki'] && ! ( $flags & PPFrame::RECOVER_COMMENTS ) ) { $out .= $this->parser->insertStripItem( $contextNode->firstChild->value ); } # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove else { $out .= $contextNode->firstChild->value; } } elseif ( $contextNode->name === 'ignore' ) { # Output suppression used by etc. # OT_WIKI will only respect in substed templates. # The other output types respect it unless NO_IGNORE is set. # extractSections() sets NO_IGNORE and so never respects it. if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & PPFrame::NO_IGNORE ) ) { $out .= $contextNode->firstChild->value; } else { //$out .= ''; } } elseif ( $contextNode->name === 'ext' ) { # Extension tag $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null ); $out .= $this->parser->extensionSubstitution( $bits, $this ); } elseif ( $contextNode->name === 'h' ) { # Heading if ( $this->parser->ot['html'] ) { # Expand immediately and insert heading index marker $s = ''; for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) { $s .= $this->expand( $node, $flags ); } $bits = $contextNode->splitHeading(); $titleText = $this->title->getPrefixedDBkey(); $this->parser->mHeadings[] = array( $titleText, $bits['i'] ); $serial = count( $this->parser->mHeadings ) - 1; $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX; $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] ); $this->parser->mStripState->addGeneral( $marker, '' ); $out .= $s; } else { # Expand in virtual stack $newIterator = $contextNode->getChildren(); } } else { # Generic recursive expansion $newIterator = $contextNode->getChildren(); } } else { throw new MWException( __METHOD__.': Invalid parameter type' ); } if ( $newIterator !== false ) { $outStack[] = ''; $iteratorStack[] = $newIterator; $indexStack[] = 0; } elseif ( $iteratorStack[$level] === false ) { // Return accumulated value to parent // With tail recursion while ( $iteratorStack[$level] === false && $level > 0 ) { $outStack[$level - 1] .= $out; array_pop( $outStack ); array_pop( $iteratorStack ); array_pop( $indexStack ); $level--; } } } --$expansionDepth; return $outStack[0]; } /** * @param $sep * @param $flags * @return string */ function implodeWithFlags( $sep, $flags /*, ... */ ) { $args = array_slice( func_get_args(), 2 ); $first = true; $s = ''; foreach ( $args as $root ) { if ( $root instanceof PPNode_HipHop_Array ) { $root = $root->value; } if ( !is_array( $root ) ) { $root = array( $root ); } foreach ( $root as $node ) { if ( $first ) { $first = false; } else { $s .= $sep; } $s .= $this->expand( $node, $flags ); } } return $s; } /** * Implode with no flags specified * This previously called implodeWithFlags but has now been inlined to reduce stack depth * @param $sep * @return string */ function implode( $sep /*, ... */ ) { $args = array_slice( func_get_args(), 1 ); $first = true; $s = ''; foreach ( $args as $root ) { if ( $root instanceof PPNode_HipHop_Array ) { $root = $root->value; } if ( !is_array( $root ) ) { $root = array( $root ); } foreach ( $root as $node ) { if ( $first ) { $first = false; } else { $s .= $sep; } $s .= $this->expand( $node ); } } return $s; } /** * Makes an object that, when expand()ed, will be the same as one obtained * with implode() * * @param $sep * @return PPNode_HipHop_Array */ function virtualImplode( $sep /*, ... */ ) { $args = array_slice( func_get_args(), 1 ); $out = array(); $first = true; foreach ( $args as $root ) { if ( $root instanceof PPNode_HipHop_Array ) { $root = $root->value; } if ( !is_array( $root ) ) { $root = array( $root ); } foreach ( $root as $node ) { if ( $first ) { $first = false; } else { $out[] = $sep; } $out[] = $node; } } return new PPNode_HipHop_Array( $out ); } /** * Virtual implode with brackets * * @param $start * @param $sep * @param $end * @return PPNode_HipHop_Array */ function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) { $args = array_slice( func_get_args(), 3 ); $out = array( $start ); $first = true; foreach ( $args as $root ) { if ( $root instanceof PPNode_HipHop_Array ) { $root = $root->value; } if ( !is_array( $root ) ) { $root = array( $root ); } foreach ( $root as $node ) { if ( $first ) { $first = false; } else { $out[] = $sep; } $out[] = $node; } } $out[] = $end; return new PPNode_HipHop_Array( $out ); } function __toString() { return 'frame{}'; } /** * @param $level bool * @return array|bool|String */ function getPDBK( $level = false ) { if ( $level === false ) { return $this->title->getPrefixedDBkey(); } else { return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false; } } /** * @return array */ function getArguments() { return array(); } /** * @return array */ function getNumberedArguments() { return array(); } /** * @return array */ function getNamedArguments() { return array(); } /** * Returns true if there are no arguments in this frame * * @return bool */ function isEmpty() { return true; } /** * @param $name * @return bool */ function getArgument( $name ) { return false; } /** * Returns true if the infinite loop check is OK, false if a loop is detected * * @param $title Title * * @return bool */ function loopCheck( $title ) { return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] ); } /** * Return true if the frame is a template frame * * @return bool */ function isTemplate() { return false; } /** * Get a title of frame * * @return Title */ function getTitle() { return $this->title; } } /** * Expansion frame with template arguments * @ingroup Parser */ class PPTemplateFrame_HipHop extends PPFrame_HipHop { var $numberedArgs, $namedArgs, $parent; var $numberedExpansionCache, $namedExpansionCache; /** * @param $preprocessor Preprocessor_HipHop * @param $parent bool * @param $numberedArgs array * @param $namedArgs array * @param $title Title|bool */ function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) { parent::__construct( $preprocessor ); $this->parent = $parent; $this->numberedArgs = $numberedArgs; $this->namedArgs = $namedArgs; $this->title = $title; $pdbk = $title ? $title->getPrefixedDBkey() : false; $this->titleCache = $parent->titleCache; $this->titleCache[] = $pdbk; $this->loopCheckHash = /*clone*/ $parent->loopCheckHash; if ( $pdbk !== false ) { $this->loopCheckHash[$pdbk] = true; } $this->depth = $parent->depth + 1; $this->numberedExpansionCache = $this->namedExpansionCache = array(); } function __toString() { $s = 'tplframe{'; $first = true; $args = $this->numberedArgs + $this->namedArgs; foreach ( $args as $name => $value ) { if ( $first ) { $first = false; } else { $s .= ', '; } $s .= "\"$name\":\"" . str_replace( '"', '\\"', $value->__toString() ) . '"'; } $s .= '}'; return $s; } /** * Returns true if there are no arguments in this frame * * @return bool */ function isEmpty() { return !count( $this->numberedArgs ) && !count( $this->namedArgs ); } /** * @return array */ function getArguments() { $arguments = array(); foreach ( array_merge( array_keys($this->numberedArgs), array_keys($this->namedArgs)) as $key ) { $arguments[$key] = $this->getArgument($key); } return $arguments; } /** * @return array */ function getNumberedArguments() { $arguments = array(); foreach ( array_keys($this->numberedArgs) as $key ) { $arguments[$key] = $this->getArgument($key); } return $arguments; } /** * @return array */ function getNamedArguments() { $arguments = array(); foreach ( array_keys($this->namedArgs) as $key ) { $arguments[$key] = $this->getArgument($key); } return $arguments; } /** * @param $index * @return array|bool */ function getNumberedArgument( $index ) { if ( !isset( $this->numberedArgs[$index] ) ) { return false; } if ( !isset( $this->numberedExpansionCache[$index] ) ) { # No trimming for unnamed arguments $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], PPFrame::STRIP_COMMENTS ); } return $this->numberedExpansionCache[$index]; } /** * @param $name * @return bool */ function getNamedArgument( $name ) { if ( !isset( $this->namedArgs[$name] ) ) { return false; } if ( !isset( $this->namedExpansionCache[$name] ) ) { # Trim named arguments post-expand, for backwards compatibility $this->namedExpansionCache[$name] = trim( $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) ); } return $this->namedExpansionCache[$name]; } /** * @param $name * @return array|bool */ function getArgument( $name ) { $text = $this->getNumberedArgument( $name ); if ( $text === false ) { $text = $this->getNamedArgument( $name ); } return $text; } /** * Return true if the frame is a template frame * * @return bool */ function isTemplate() { return true; } } /** * Expansion frame with custom arguments * @ingroup Parser */ class PPCustomFrame_HipHop extends PPFrame_HipHop { var $args; function __construct( $preprocessor, $args ) { parent::__construct( $preprocessor ); $this->args = $args; } function __toString() { $s = 'cstmframe{'; $first = true; foreach ( $this->args as $name => $value ) { if ( $first ) { $first = false; } else { $s .= ', '; } $s .= "\"$name\":\"" . str_replace( '"', '\\"', $value->__toString() ) . '"'; } $s .= '}'; return $s; } /** * @return bool */ function isEmpty() { return !count( $this->args ); } /** * @param $index * @return bool */ function getArgument( $index ) { if ( !isset( $this->args[$index] ) ) { return false; } return $this->args[$index]; } } /** * @ingroup Parser */ class PPNode_HipHop_Tree implements PPNode { var $name, $firstChild, $lastChild, $nextSibling; function __construct( $name ) { $this->name = $name; $this->firstChild = $this->lastChild = $this->nextSibling = false; } function __toString() { $inner = ''; $attribs = ''; for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) { if ( $node instanceof PPNode_HipHop_Attr ) { $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"'; } else { $inner .= $node->__toString(); } } if ( $inner === '' ) { return "<{$this->name}$attribs/>"; } else { return "<{$this->name}$attribs>$innername}>"; } } /** * @param $name * @param $text * @return PPNode_HipHop_Tree */ static function newWithText( $name, $text ) { $obj = new self( $name ); $obj->addChild( new PPNode_HipHop_Text( $text ) ); return $obj; } function addChild( $node ) { if ( $this->lastChild === false ) { $this->firstChild = $this->lastChild = $node; } else { $this->lastChild->nextSibling = $node; $this->lastChild = $node; } } /** * @return PPNode_HipHop_Array */ function getChildren() { $children = array(); for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) { $children[] = $child; } return new PPNode_HipHop_Array( $children ); } function getFirstChild() { return $this->firstChild; } function getNextSibling() { return $this->nextSibling; } /** * @param $name string * @return array */ function getChildrenOfType( $name ) { $children = array(); for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) { if ( isset( $child->name ) && $child->name === $name ) { $children[] = $child; } } return $children; } /** * @return bool */ function getLength() { return false; } /** * @param $i * @return bool */ function item( $i ) { return false; } /** * @return string */ function getName() { return $this->name; } /** * Split a node into an associative array containing: * name PPNode name * index String index * value PPNode value * * @throws MWException * @return array */ function splitArg() { $bits = array(); for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) { if ( !isset( $child->name ) ) { continue; } if ( $child->name === 'name' ) { $bits['name'] = $child; if ( $child->firstChild instanceof PPNode_HipHop_Attr && $child->firstChild->name === 'index' ) { $bits['index'] = $child->firstChild->value; } } elseif ( $child->name === 'value' ) { $bits['value'] = $child; } } if ( !isset( $bits['name'] ) ) { throw new MWException( 'Invalid brace node passed to ' . __METHOD__ ); } if ( !isset( $bits['index'] ) ) { $bits['index'] = ''; } return $bits; } /** * Split an node into an associative array containing name, attr, inner and close * All values in the resulting array are PPNodes. Inner and close are optional. * * @throws MWException * @return array */ function splitExt() { $bits = array(); for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) { if ( !isset( $child->name ) ) { continue; } if ( $child->name === 'name' ) { $bits['name'] = $child; } elseif ( $child->name === 'attr' ) { $bits['attr'] = $child; } elseif ( $child->name === 'inner' ) { $bits['inner'] = $child; } elseif ( $child->name === 'close' ) { $bits['close'] = $child; } } if ( !isset( $bits['name'] ) ) { throw new MWException( 'Invalid ext node passed to ' . __METHOD__ ); } return $bits; } /** * Split an node * * @throws MWException * @return array */ function splitHeading() { if ( $this->name !== 'h' ) { throw new MWException( 'Invalid h node passed to ' . __METHOD__ ); } $bits = array(); for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) { if ( !isset( $child->name ) ) { continue; } if ( $child->name === 'i' ) { $bits['i'] = $child->value; } elseif ( $child->name === 'level' ) { $bits['level'] = $child->value; } } if ( !isset( $bits['i'] ) ) { throw new MWException( 'Invalid h node passed to ' . __METHOD__ ); } return $bits; } /** * Split a