parser = $parser;
$mem = ini_get( 'memory_limit' );
$this->memoryLimit = false;
if ( strval( $mem ) !== '' && $mem != -1 ) {
if ( preg_match( '/^\d+$/', $mem ) ) {
$this->memoryLimit = $mem;
} elseif ( preg_match( '/^(\d+)M$/i', $mem, $m ) ) {
$this->memoryLimit = $m[1] * 1048576;
}
}
}
/**
* @return PPFrame_DOM
*/
function newFrame() {
return new PPFrame_DOM( $this );
}
/**
* @param $args array
* @return PPCustomFrame_DOM
*/
function newCustomFrame( $args ) {
return new PPCustomFrame_DOM( $this, $args );
}
/**
* @param $values
* @return PPNode_DOM
*/
function newPartNodeArray( $values ) {
//NOTE: DOM manipulation is slower than building & parsing XML! (or so Tim sais)
$xml = "";
foreach ( $values as $k => $val ) {
if ( is_int( $k ) ) {
$xml .= "" . htmlspecialchars( $val ) ."";
} else {
$xml .= "" . htmlspecialchars( $k ) . "=" . htmlspecialchars( $val ) . "";
}
}
$xml .= "
";
$dom = new DOMDocument();
$dom->loadXML( $xml );
$root = $dom->documentElement;
$node = new PPNode_DOM( $root->childNodes );
return $node;
}
/**
* @throws MWException
* @return bool
*/
function memCheck() {
if ( $this->memoryLimit === false ) {
return true;
}
$usage = memory_get_usage();
if ( $usage > $this->memoryLimit * 0.9 ) {
$limit = intval( $this->memoryLimit * 0.9 / 1048576 + 0.5 );
throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
}
return $usage <= $this->memoryLimit * 0.8;
}
/**
* 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 "" and "" 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.
*
* @return PPNode_DOM
*/
function preprocessToObj( $text, $flags = 0 ) {
wfProfileIn( __METHOD__ );
global $wgMemc, $wgPreprocessorCacheThreshold;
$xml = false;
$cacheable = ( $wgPreprocessorCacheThreshold !== false
&& strlen( $text ) > $wgPreprocessorCacheThreshold );
if ( $cacheable ) {
wfProfileIn( __METHOD__.'-cacheable' );
$cacheKey = wfMemcKey( 'preprocess-xml', md5($text), $flags );
$cacheValue = $wgMemc->get( $cacheKey );
if ( $cacheValue ) {
$version = substr( $cacheValue, 0, 8 );
if ( intval( $version ) == self::CACHE_VERSION ) {
$xml = substr( $cacheValue, 8 );
// From the cache
wfDebugLog( "Preprocessor", "Loaded preprocessor XML from memcached (key $cacheKey)" );
}
}
}
if ( $xml === false ) {
if ( $cacheable ) {
wfProfileIn( __METHOD__.'-cache-miss' );
$xml = $this->preprocessToXml( $text, $flags );
$cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . $xml;
$wgMemc->set( $cacheKey, $cacheValue, 86400 );
wfProfileOut( __METHOD__.'-cache-miss' );
wfDebugLog( "Preprocessor", "Saved preprocessor XML to memcached (key $cacheKey)" );
} else {
$xml = $this->preprocessToXml( $text, $flags );
}
}
// Fail if the number of elements exceeds acceptable limits
// Do not attempt to generate the DOM
$this->parser->mGeneratedPPNodeCount += substr_count( $xml, '<' );
$max = $this->parser->mOptions->getMaxGeneratedPPNodeCount();
if ( $this->parser->mGeneratedPPNodeCount > $max ) {
throw new MWException( __METHOD__.': generated node count limit exceeded' );
}
wfProfileIn( __METHOD__.'-loadXML' );
$dom = new DOMDocument;
wfSuppressWarnings();
$result = $dom->loadXML( $xml );
wfRestoreWarnings();
if ( !$result ) {
// Try running the XML through UtfNormal to get rid of invalid characters
$xml = UtfNormal::cleanUp( $xml );
// 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2 don't barf when the XML is >256 levels deep
$result = $dom->loadXML( $xml, 1 << 19 );
if ( !$result ) {
throw new MWException( __METHOD__.' generated invalid XML' );
}
}
$obj = new PPNode_DOM( $dom->documentElement );
wfProfileOut( __METHOD__.'-loadXML' );
if ( $cacheable ) {
wfProfileOut( __METHOD__.'-cacheable' );
}
wfProfileOut( __METHOD__ );
return $obj;
}
/**
* @param $text string
* @param $flags int
* @return string
*/
function preprocessToXml( $text, $flags = 0 ) {
wfProfileIn( __METHOD__ );
$rules = array(
'{' => array(
'end' => '}',
'names' => array(
2 => 'template',
3 => 'tplarg',
),
'min' => 2,
'max' => 3,
),
'[' => array(
'end' => ']',
'names' => array( 2 => null ),
'min' => 2,
'max' => 2,
)
);
$forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
$xmlishElements = $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 {
$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;
$searchBase = "[{<\n"; #}
$revText = strrev( $text ); // For fast reverse searches
$lengthText = strlen( $text );
$i = 0; # Input pointer, starts out pointing to a pseudo-newline before the start
$accum =& $stack->getAccum(); # Current accumulator
$accum = '';
$findEquals = false; # True to find equals signs in arguments
$findPipe = false; # True to take notice of pipe characters
$headingIndex = 1;
$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
$startPos = strpos( $text, '', $i );
if ( $startPos === false ) {
// Ignored section runs to the end
$accum .= '' . htmlspecialchars( substr( $text, $i ) ) . '';
break;
}
$tagEndPos = $startPos + strlen( '' ); // past-the-end
$accum .= '' . htmlspecialchars( 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 = $stack->top->close;
$search .= $currentClosing;
}
if ( $findPipe ) {
$search .= '|';
}
if ( $findEquals ) {
// First equals will be for the template
$search .= '=';
}
$rule = null;
# Output literal section, advance input counter
$literalLength = strcspn( $text, $search, $i );
if ( $literalLength > 0 ) {
$accum .= htmlspecialchars( 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 ( $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
$accum .= '<';
++$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
$endPos = strpos( $text, '-->', $i + 4 );
if ( $endPos === false ) {
// Unclosed comment in input, runs to end
$inner = substr( $text, $i );
$accum .= '' . htmlspecialchars( $inner ) . '';
$i = $lengthText;
} else {
// Search backwards for leading whitespace
$wsStart = $i ? ( $i - strspn( $revText, ' ', $lengthText - $i ) ) : 0;
// Search forwards for trailing whitespace
// $wsEnd will be the position of the last space (or the '>' if there's none)
$wsEnd = $endPos + 2 + 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" )
{
$startPos = $wsStart;
$endPos = $wsEnd + 1;
// Remove leading whitespace from the end of the accumulator
// Sanity check first though
$wsLength = $i - $wsStart;
if ( $wsLength > 0 && substr( $accum, -$wsLength ) === str_repeat( ' ', $wsLength ) ) {
$accum = substr( $accum, 0, -$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
$startPos = $i;
$endPos += 2;
}
if ( $stack->top ) {
$part = $stack->top->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 = substr( $text, $startPos, $endPos - $startPos + 1 );
$accum .= '' . htmlspecialchars( $inner ) . '';
}
continue;
}
$name = $matches[1];
$lowerName = strtolower( $name );
$attrStart = $i + strlen( $name ) + 1;
// Find end of tag
$tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
if ( $tagEndPos === false ) {
// Infinite backtrack
// Disable tag search to prevent worst-case O(N^2) performance
$noMoreGT = true;
$accum .= '<';
++$i;
continue;
}
// Handle ignored tags
if ( in_array( $lowerName, $ignoredTags ) ) {
$accum .= '' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '';
$i = $tagEndPos + 1;
continue;
}
$tagStartPos = $i;
if ( $text[$tagEndPos-1] == '/' ) {
$attrEnd = $tagEndPos - 1;
$inner = null;
$i = $tagEndPos + 1;
$close = '';
} else {
$attrEnd = $tagEndPos;
// Find closing tag
if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
$text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) )
{
$inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
$i = $matches[0][1] + strlen( $matches[0][0] );
$close = '' . htmlspecialchars( $matches[0][0] ) . '';
} else {
// No end tag -- let it run out to the end of the text.
$inner = substr( $text, $tagEndPos + 1 );
$i = $lengthText;
$close = '';
}
}
// and just become tags
if ( in_array( $lowerName, $ignoredElements ) ) {
$accum .= '' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
. '';
continue;
}
$accum .= '';
if ( $attrEnd <= $attrStart ) {
$attr = '';
} else {
$attr = substr( $text, $attrStart, $attrEnd - $attrStart );
}
$accum .= '' . htmlspecialchars( $name ) . '' .
// Note that the attr element contains the whitespace between name and attribute,
// this is necessary for precise reconstruction during pre-save transform.
'' . htmlspecialchars( $attr ) . '';
if ( $inner !== null ) {
$accum .= '' . htmlspecialchars( $inner ) . '';
}
$accum .= $close . '';
} 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 .= $curChar;
$i++;
}
$count = strspn( $text, '=', $i, 6 );
if ( $count == 1 && $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 ) {
$piece = array(
'open' => "\n",
'close' => "\n",
'parts' => array( new PPDPart( str_repeat( '=', $count ) ) ),
'startPos' => $i,
'count' => $count );
$stack->push( $piece );
$accum =& $stack->getAccum();
$flags = $stack->getFlags();
extract( $flags );
$i += $count;
}
} elseif ( $found == 'line-end' ) {
$piece = $stack->top;
// A heading must be open, otherwise \n wouldn't have been in the search list
assert( '$piece->open == "\n"' );
$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 = 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 = $part->visualEnd;
$searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
}
$count = $piece->count;
$equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
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 = min( 6, intval( ( $count - 1 ) / 2 ) );
}
} else {
$count = min( $equalsLength, $count );
}
if ( $count > 0 ) {
// Normal match, output
$element = "$accum";
$headingIndex++;
} else {
// Single equals sign on its own line, count=0
$element = $accum;
}
} else {
// No match, no , just pass down the inner text
$element = $accum;
}
// Unwind the stack
$stack->pop();
$accum =& $stack->getAccum();
$flags = $stack->getFlags();
extract( $flags );
// Append the result to the enclosing accumulator
$accum .= $element;
// 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 = 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
$piece = array(
'open' => $curChar,
'close' => $rule['end'],
'count' => $count,
'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
);
$stack->push( $piece );
$accum =& $stack->getAccum();
$flags = $stack->getFlags();
extract( $flags );
} else {
# Add literal brace(s)
$accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
}
$i += $count;
} elseif ( $found == 'close' ) {
$piece = $stack->top;
# lets check if there are enough characters for closing brace
$maxCount = $piece->count;
$count = 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 = $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 .= htmlspecialchars( str_repeat( $curChar, $count ) );
$i += $count;
continue;
}
$name = $rule['names'][$matchingCount];
if ( $name === null ) {
// No element, just literal text
$element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
} else {
# Create XML element
# Note: $parts is already XML, does not need to be encoded further
$parts = $piece->parts;
$title = $parts[0]->out;
unset( $parts[0] );
# 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 ) ) {
$attr = ' lineStart="1"';
} else {
$attr = '';
}
$element = "<$name$attr>";
$element .= "$title";
$argIndex = 1;
foreach ( $parts as $part ) {
if ( isset( $part->eqpos ) ) {
$argName = substr( $part->out, 0, $part->eqpos );
$argValue = substr( $part->out, $part->eqpos + 1 );
$element .= "$argName=$argValue";
} else {
$element .= "{$part->out}";
$argIndex++;
}
}
$element .= "$name>";
}
# 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 );
$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 .= str_repeat( $piece->open, $skippedBraces );
}
$flags = $stack->getFlags();
extract( $flags );
# Add XML element to the enclosing accumulator
$accum .= $element;
} elseif ( $found == 'pipe' ) {
$findEquals = true; // shortcut for getFlags()
$stack->addPart();
$accum =& $stack->getAccum();
++$i;
} elseif ( $found == 'equals' ) {
$findEquals = false; // shortcut for getFlags()
$stack->getCurrentPart()->eqpos = strlen( $accum );
$accum .= '=';
++$i;
}
}
# Output any remaining unclosed brackets
foreach ( $stack->stack as $piece ) {
$stack->rootAccum .= $piece->breakSyntax();
}
$stack->rootAccum .= '';
$xml = $stack->rootAccum;
wfProfileOut( __METHOD__ );
return $xml;
}
}
/**
* Stack class to help Preprocessor::preprocessToObj()
* @ingroup Parser
*/
class PPDStack {
var $stack, $rootAccum;
/**
* @var PPDStack
*/
var $top;
var $out;
var $elementClass = 'PPDStackElement';
static $false = false;
function __construct() {
$this->stack = array();
$this->top = false;
$this->rootAccum = '';
$this->accum =& $this->rootAccum;
}
/**
* @return int
*/
function count() {
return count( $this->stack );
}
function &getAccum() {
return $this->accum;
}
function getCurrentPart() {
if ( $this->top === false ) {
return false;
} else {
return $this->top->getCurrentPart();
}
}
function push( $data ) {
if ( $data instanceof $this->elementClass ) {
$this->stack[] = $data;
} else {
$class = $this->elementClass;
$this->stack[] = new $class( $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 {
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.
var $partClass = 'PPDPart';
function __construct( $data = array() ) {
$class = $this->partClass;
$this->parts = array( new $class );
foreach ( $data as $name => $value ) {
$this->$name = $value;
}
}
function &getAccum() {
return $this->parts[count($this->parts) - 1]->out;
}
function addPart( $s = '' ) {
$class = $this->partClass;
$this->parts[] = new $class( $s );
}
function getCurrentPart() {
return $this->parts[count($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 output string that would result if the close is not found.
*
* @return string
*/
function breakSyntax( $openingCount = false ) {
if ( $this->open == "\n" ) {
$s = $this->parts[0]->out;
} else {
if ( $openingCount === false ) {
$openingCount = $this->count;
}
$s = str_repeat( $this->open, $openingCount );
$first = true;
foreach ( $this->parts as $part ) {
if ( $first ) {
$first = false;
} else {
$s .= '|';
}
$s .= $part->out;
}
}
return $s;
}
}
/**
* @ingroup Parser
*/
class PPDPart {
var $out; // Output accumulator string
// 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 = $out;
}
}
/**
* An expansion frame, used as a context to expand the result of preprocessToObj()
* @ingroup Parser
*/
class PPFrame_DOM implements PPFrame {
/**
* @var Preprocessor
*/
var $preprocessor;
/**
* @var Parser
*/
var $parser;
/**
* @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
*
* @return PPTemplateFrame_DOM
*/
function newChild( $args = false, $title = false, $indexOffset = 0 ) {
$namedArgs = array();
$numberedArgs = array();
if ( $title === false ) {
$title = $this->title;
}
if ( $args !== false ) {
$xpath = false;
if ( $args instanceof PPNode ) {
$args = $args->node;
}
foreach ( $args as $arg ) {
if ( $arg instanceof PPNode ) {
$arg = $arg->node;
}
if ( !$xpath ) {
$xpath = new DOMXPath( $arg->ownerDocument );
}
$nameNodes = $xpath->query( 'name', $arg );
$value = $xpath->query( 'value', $arg );
if ( $nameNodes->item( 0 )->hasAttributes() ) {
// Numbered parameter
$index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
$index = $index - $indexOffset;
$numberedArgs[$index] = $value->item( 0 );
unset( $namedArgs[$index] );
} else {
// Named parameter
$name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
$namedArgs[$name] = $value->item( 0 );
unset( $numberedArgs[$name] );
}
}
}
return new PPTemplateFrame_DOM( $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,
$this->parser->mOptions->getMaxPPExpandDepth()
);
return 'Expansion depth limit exceeded';
}
wfProfileIn( __METHOD__ );
++$expansionDepth;
if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
$this->parser->mHighestExpansionDepth = $expansionDepth;
}
if ( $root instanceof PPNode_DOM ) {
$root = $root->node;
}
if ( $root instanceof DOMDocument ) {
$root = $root->documentElement;
}
$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 ( $iteratorNode instanceof PPNode_DOM ) $iteratorNode = $iteratorNode->node;
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 DOMNodeList ) {
if ( $index >= $iteratorNode->length ) {
// 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;
}
if ( $contextNode instanceof PPNode_DOM ) {
$contextNode = $contextNode->node;
}
$newIterator = false;
if ( $contextNode === false ) {
// nothing to do
} elseif ( is_string( $contextNode ) ) {
$out .= $contextNode;
} elseif ( is_array( $contextNode ) || $contextNode instanceof DOMNodeList ) {
$newIterator = $contextNode;
} elseif ( $contextNode instanceof DOMNode ) {
if ( $contextNode->nodeType == XML_TEXT_NODE ) {
$out .= $contextNode->nodeValue;
} elseif ( $contextNode->nodeName == 'template' ) {
# Double-brace expansion
$xpath = new DOMXPath( $contextNode->ownerDocument );
$titles = $xpath->query( 'title', $contextNode );
$title = $titles->item( 0 );
$parts = $xpath->query( 'part', $contextNode );
if ( $flags & PPFrame::NO_TEMPLATES ) {
$newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
} else {
$lineStart = $contextNode->getAttribute( 'lineStart' );
$params = array(
'title' => new PPNode_DOM( $title ),
'parts' => new PPNode_DOM( $parts ),
'lineStart' => $lineStart );
$ret = $this->parser->braceSubstitution( $params, $this );
if ( isset( $ret['object'] ) ) {
$newIterator = $ret['object'];
} else {
$out .= $ret['text'];
}
}
} elseif ( $contextNode->nodeName == 'tplarg' ) {
# Triple-brace expansion
$xpath = new DOMXPath( $contextNode->ownerDocument );
$titles = $xpath->query( 'title', $contextNode );
$title = $titles->item( 0 );
$parts = $xpath->query( 'part', $contextNode );
if ( $flags & PPFrame::NO_ARGS ) {
$newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
} else {
$params = array(
'title' => new PPNode_DOM( $title ),
'parts' => new PPNode_DOM( $parts ) );
$ret = $this->parser->argSubstitution( $params, $this );
if ( isset( $ret['object'] ) ) {
$newIterator = $ret['object'];
} else {
$out .= $ret['text'];
}
}
} elseif ( $contextNode->nodeName == '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->textContent );
}
# Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
else {
$out .= $contextNode->textContent;
}
} elseif ( $contextNode->nodeName == '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->textContent;
} else {
$out .= '';
}
} elseif ( $contextNode->nodeName == 'ext' ) {
# Extension tag
$xpath = new DOMXPath( $contextNode->ownerDocument );
$names = $xpath->query( 'name', $contextNode );
$attrs = $xpath->query( 'attr', $contextNode );
$inners = $xpath->query( 'inner', $contextNode );
$closes = $xpath->query( 'close', $contextNode );
$params = array(
'name' => new PPNode_DOM( $names->item( 0 ) ),
'attr' => $attrs->length > 0 ? new PPNode_DOM( $attrs->item( 0 ) ) : null,
'inner' => $inners->length > 0 ? new PPNode_DOM( $inners->item( 0 ) ) : null,
'close' => $closes->length > 0 ? new PPNode_DOM( $closes->item( 0 ) ) : null,
);
$out .= $this->parser->extensionSubstitution( $params, $this );
} elseif ( $contextNode->nodeName == 'h' ) {
# Heading
$s = $this->expand( $contextNode->childNodes, $flags );
# Insert a heading marker only for children of
# This is to stop extractSections from going over multiple tree levels
if ( $contextNode->parentNode->nodeName == 'root'
&& $this->parser->ot['html'] )
{
# Insert heading index marker
$headingIndex = $contextNode->getAttribute( 'i' );
$titleText = $this->title->getPrefixedDBkey();
$this->parser->mHeadings[] = array( $titleText, $headingIndex );
$serial = count( $this->parser->mHeadings ) - 1;
$marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
$count = $contextNode->getAttribute( 'level' );
$s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
$this->parser->mStripState->addGeneral( $marker, '' );
}
$out .= $s;
} else {
# Generic recursive expansion
$newIterator = $contextNode->childNodes;
}
} else {
wfProfileOut( __METHOD__ );
throw new MWException( __METHOD__.': Invalid parameter type' );
}
if ( $newIterator !== false ) {
if ( $newIterator instanceof PPNode_DOM ) {
$newIterator = $newIterator->node;
}
$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;
wfProfileOut( __METHOD__ );
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_DOM ) $root = $root->node;
if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
$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
*
* @return string
*/
function implode( $sep /*, ... */ ) {
$args = array_slice( func_get_args(), 1 );
$first = true;
$s = '';
foreach ( $args as $root ) {
if ( $root instanceof PPNode_DOM ) {
$root = $root->node;
}
if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
$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()
*
* @return array
*/
function virtualImplode( $sep /*, ... */ ) {
$args = array_slice( func_get_args(), 1 );
$out = array();
$first = true;
foreach ( $args as $root ) {
if ( $root instanceof PPNode_DOM ) {
$root = $root->node;
}
if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
$root = array( $root );
}
foreach ( $root as $node ) {
if ( $first ) {
$first = false;
} else {
$out[] = $sep;
}
$out[] = $node;
}
}
return $out;
}
/**
* Virtual implode with brackets
* @return 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_DOM ) {
$root = $root->node;
}
if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
$root = array( $root );
}
foreach ( $root as $node ) {
if ( $first ) {
$first = false;
} else {
$out[] = $sep;
}
$out[] = $node;
}
}
$out[] = $end;
return $out;
}
function __toString() {
return 'frame{}';
}
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;
}
function getArgument( $name ) {
return false;
}
/**
* Returns true if the infinite loop check is OK, false if a loop is detected
*
* @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_DOM extends PPFrame_DOM {
var $numberedArgs, $namedArgs;
/**
* @var PPFrame_DOM
*/
var $parent;
var $numberedExpansionCache, $namedExpansionCache;
/**
* @param $preprocessor
* @param $parent PPFrame_DOM
* @param $numberedArgs array
* @param $namedArgs array
* @param $title Title
*/
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->ownerDocument->saveXML( $value ) ) . '"';
}
$s .= '}';
return $s;
}
/**
* Returns true if there are no arguments in this frame
*
* @return bool
*/
function isEmpty() {
return !count( $this->numberedArgs ) && !count( $this->namedArgs );
}
function getArguments() {
$arguments = array();
foreach ( array_merge(
array_keys($this->numberedArgs),
array_keys($this->namedArgs)) as $key ) {
$arguments[$key] = $this->getArgument($key);
}
return $arguments;
}
function getNumberedArguments() {
$arguments = array();
foreach ( array_keys($this->numberedArgs) as $key ) {
$arguments[$key] = $this->getArgument($key);
}
return $arguments;
}
function getNamedArguments() {
$arguments = array();
foreach ( array_keys($this->namedArgs) as $key ) {
$arguments[$key] = $this->getArgument($key);
}
return $arguments;
}
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];
}
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];
}
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_DOM extends PPFrame_DOM {
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 );
}
function getArgument( $index ) {
if ( !isset( $this->args[$index] ) ) {
return false;
}
return $this->args[$index];
}
function getArguments() {
return $this->args;
}
}
/**
* @ingroup Parser
*/
class PPNode_DOM implements PPNode {
/**
* @var DOMElement
*/
var $node;
var $xpath;
function __construct( $node, $xpath = false ) {
$this->node = $node;
}
/**
* @return DOMXPath
*/
function getXPath() {
if ( $this->xpath === null ) {
$this->xpath = new DOMXPath( $this->node->ownerDocument );
}
return $this->xpath;
}
function __toString() {
if ( $this->node instanceof DOMNodeList ) {
$s = '';
foreach ( $this->node as $node ) {
$s .= $node->ownerDocument->saveXML( $node );
}
} else {
$s = $this->node->ownerDocument->saveXML( $this->node );
}
return $s;
}
/**
* @return bool|PPNode_DOM
*/
function getChildren() {
return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
}
/**
* @return bool|PPNode_DOM
*/
function getFirstChild() {
return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
}
/**
* @return bool|PPNode_DOM
*/
function getNextSibling() {
return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
}
/**
* @param $type
*
* @return bool|PPNode_DOM
*/
function getChildrenOfType( $type ) {
return new self( $this->getXPath()->query( $type, $this->node ) );
}
/**
* @return int
*/
function getLength() {
if ( $this->node instanceof DOMNodeList ) {
return $this->node->length;
} else {
return false;
}
}
/**
* @param $i
* @return bool|PPNode_DOM
*/
function item( $i ) {
$item = $this->node->item( $i );
return $item ? new self( $item ) : false;
}
/**
* @return string
*/
function getName() {
if ( $this->node instanceof DOMNodeList ) {
return '#nodelist';
} else {
return $this->node->nodeName;
}
}
/**
* Split a "" node into an associative array containing:
* - name PPNode name
* - index String index
* - value PPNode value
*
* @return array
*/
function splitArg() {
$xpath = $this->getXPath();
$names = $xpath->query( 'name', $this->node );
$values = $xpath->query( 'value', $this->node );
if ( !$names->length || !$values->length ) {
throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
}
$name = $names->item( 0 );
$index = $name->getAttribute( 'index' );
return array(
'name' => new self( $name ),
'index' => $index,
'value' => new self( $values->item( 0 ) ) );
}
/**
* 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.
*
* @return array
*/
function splitExt() {
$xpath = $this->getXPath();
$names = $xpath->query( 'name', $this->node );
$attrs = $xpath->query( 'attr', $this->node );
$inners = $xpath->query( 'inner', $this->node );
$closes = $xpath->query( 'close', $this->node );
if ( !$names->length || !$attrs->length ) {
throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
}
$parts = array(
'name' => new self( $names->item( 0 ) ),
'attr' => new self( $attrs->item( 0 ) ) );
if ( $inners->length ) {
$parts['inner'] = new self( $inners->item( 0 ) );
}
if ( $closes->length ) {
$parts['close'] = new self( $closes->item( 0 ) );
}
return $parts;
}
/**
* Split a "" node
* @return array
*/
function splitHeading() {
if ( $this->getName() !== 'h' ) {
throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
}
return array(
'i' => $this->node->getAttribute( 'i' ),
'level' => $this->node->getAttribute( 'level' ),
'contents' => $this->getChildren()
);
}
}