* Reduced stack depth by using an internal stack in expand(), and by having some common code paths (e.g. non-subst double-brace during PST) return objects which can be expanded in that internal stack instead of the PHP stack. This is friendly to xdebug but slightly slower than the original version. Also it probably helps robustness when you don't add 7 stack levels per pair of double braces.
* Profiling indicates that expand and PPD are now good targets for porting to C. Abstracted and refactored the relevant code to allow for a drop-in replacement. A factor of 2 reduction in average-case replaceVariables() time may be possible.
* Verified with preprocessorFuzzTest.php against r29950, updated to allow better PST tests.
* Made parserTests.php respect $wgParserConf
* LST and ParserFunctions need a simultaneous update with the core due to changed interfaces. DOM objects are now wrapped rather than directly exposed.
'ParserOutput' => 'includes/ParserOutput.php',
'ParserOptions' => 'includes/ParserOptions.php',
'PatrolLog' => 'includes/PatrolLog.php',
+ 'Preprocessor' => 'includes/Preprocessor.php',
+ 'PPFrame' => 'includes/Preprocessor.php',
+ 'PPNode' => 'includes/Preprocessor.php',
+ 'Preprocessor_DOM' => 'includes/Preprocessor_DOM.php',
+ 'PPFrame_DOM' => 'includes/Preprocessor_DOM.php',
+ 'PPTemplateFrame_DOM' => 'includes/Preprocessor_DOM.php',
+ 'PPDStack' => 'includes/Preprocessor_DOM.php',
+ 'PPDStackElement' => 'includes/Preprocessor_DOM.php',
+ 'PPNode_DOM' => 'includes/Preprocessor_DOM.php',
'ProfilerSimple' => 'includes/ProfilerSimple.php',
'ProfilerSimpleUDP' => 'includes/ProfilerSimpleUDP.php',
'Profiler' => 'includes/Profiler.php',
# Persistent:
var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
$mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
- $mExtLinkBracketedRegex;
+ $mExtLinkBracketedRegex, $mPreprocessor;
# Cleared with clearState():
var $mOutput, $mAutonumber, $mDTopen, $mStripState;
$this->mMarkerSuffix = "-QINU\x7f";
$this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
'[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
+ if ( isset( $conf['preprocessorClass'] ) ) {
+ $this->mPreprocessorClass = $conf['preprocessorClass'];
+ } else {
+ $this->mPreprocessorClass = 'Preprocessor_DOM';
+ }
$this->mFirstCall = true;
}
return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
}
+ /**
+ * Get a preprocessor object
+ */
+ function getPreprocessor() {
+ if ( !isset( $this->mPreprocessor ) ) {
+ $class = $this->mPreprocessorClass;
+ $this->mPreprocessor = new $class( $this );
+ }
+ return $this->mPreprocessor;
+ }
+
/**
* Replaces all occurrences of HTML-style comments and the given tags
* in the text with a random marker and returns teh next text. The output
* @private
*/
function preprocessToDom ( $text, $flags = 0 ) {
- wfProfileIn( __METHOD__ );
- wfProfileIn( __METHOD__.'-makexml' );
-
- $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 & self::PTD_FOR_INCLUSION;
-
- $xmlishElements = $this->getStripList();
- $enableOnlyinclude = false;
- if ( $forInclusion ) {
- $ignoredTags = array( 'includeonly', '/includeonly' );
- $ignoredElements = array( 'noinclude' );
- $xmlishElements[] = 'noinclude';
- if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== 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 = '[{<';
- $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 text accumulator
- $accum = '<root>';
- $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 <onlyinclude>
- $fakeLineStart = true; # Do a line-start run without outputting an LF character
-
- while ( true ) {
- if ( $findOnlyinclude ) {
- // Ignore all input up to the next <onlyinclude>
- $startPos = strpos( $text, '<onlyinclude>', $i );
- if ( $startPos === false ) {
- // Ignored section runs to the end
- $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
- break;
- }
- $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
- $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
- $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 .= '=';
- } else {
- // Look for headings
- // We can't look for headings when $findEquals is true, because the ambiguity
- // between template name/value separators and heading starts would be unresolved
- // until the closing double-brace is found. This would mean either infinite
- // backtrack, or creating and updating two separate tree structures until the
- // end of the ambiguity -- one tree structure assuming a heading, and the other
- // assuming a template argument.
- //
- // Easier to just break some section edit links.
- $search .= "\n";
- }
- $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 >= strlen( $text ) ) {
- 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 </onlyinclude>
- if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
- $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 .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
- $i = strlen( $text );
- } else {
- // Search backwards for leading whitespace
- $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
- // Search forwards for trailing whitespace
- // $wsEnd will be the position of the last space
- $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() does 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,
- // but only if stack->top===false, because headings don't exist at deeper levels.
- if ( $stack->top === false ) {
- $fakeLineStart = true;
- }
- } else {
- // No line to eat, just take the comment itself
- $startPos = $i;
- $endPos += 2;
- }
-
- $i = $endPos + 1;
- $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
- $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
- }
- continue;
- }
- $name = $matches[1];
- $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( $name, $ignoredTags ) ) {
- $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
- $i = $tagEndPos + 1;
- continue;
- }
-
- $tagStartPos = $i;
- if ( $text[$tagEndPos-1] == '/' ) {
- $attrEnd = $tagEndPos - 1;
- $inner = null;
- $i = $tagEndPos + 1;
- $close = '';
- } else {
- $attrEnd = $tagEndPos;
- // Find closing tag
- if ( preg_match( "/<\/$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 = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
- } else {
- // No end tag -- let it run out to the end of the text.
- $inner = substr( $text, $tagEndPos + 1 );
- $i = strlen( $text );
- $close = '';
- }
- }
- // <includeonly> and <noinclude> just become <ignore> tags
- if ( in_array( $name, $ignoredElements ) ) {
- $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
- . '</ignore>';
- continue;
- }
-
- $accum .= '<ext>';
- if ( $attrEnd <= $attrStart ) {
- $attr = '';
- } else {
- $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
- }
- $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
- // Note that the attr element contains the whitespace between name and attribute,
- // this is necessary for precise reconstruction during pre-save transform.
- '<attr>' . htmlspecialchars( $attr ) . '</attr>';
- if ( $inner !== null ) {
- $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
- }
- $accum .= $close . '</ext>';
- }
-
- 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 > 0 ) {
- $piece = array(
- 'open' => "\n",
- 'close' => "\n",
- 'parts' => array( str_repeat( '=', $count ) ),
- 'startPos' => $i,
- 'count' => $count );
- $stack->push( $piece );
- $accum =& $stack->getAccum();
- extract( $stack->getFlags() );
- $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" );
- // 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
- $m = false;
- $count = $piece->count;
- if ( preg_match( "/\s*(=+)/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
- if ( $i - strlen( $m[0] ) == $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 = strlen( $m[1] );
- if ( $count < 3 ) {
- $count = 0;
- } else {
- $count = min( 6, intval( ( $count - 1 ) / 2 ) );
- }
- } else {
- $count = min( strlen( $m[1] ), $count );
- }
- if ( $count > 0 ) {
- // Normal match, output <h>
- $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
- $headingIndex++;
- } else {
- // Single equals sign on its own line, count=0
- $element = $accum;
- }
- } else {
- // No match, no <h>, just pass down the inner text
- $element = $accum;
- }
- // Unwind the stack
- $stack->pop();
- $accum =& $stack->getAccum();
- extract( $stack->getFlags() );
-
- // 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' => $count,
- 'parts' => array( '' ),
- 'eqpos' => array(),
- 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
- );
-
- $stack->push( $piece );
- $accum =& $stack->getAccum();
- extract( $stack->getFlags() );
- } 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)
- $matchingCount = 0;
- $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 = str_repeat( $piece->open, $matchingCount ) .
- implode( '|', $piece->parts ) .
- 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];
- 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>";
- $element .= "<title>$title</title>";
- $argIndex = 1;
- foreach ( $parts as $partIndex => $part ) {
- if ( isset( $piece->eqpos[$partIndex] ) ) {
- $eqpos = $piece->eqpos[$partIndex];
- $argName = substr( $part, 0, $eqpos );
- $argValue = substr( $part, $eqpos + 1 );
- $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
- } else {
- $element .= "<part><name index=\"$argIndex\" /><value>$part</value></part>";
- $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( '' );
- $piece->count -= $matchingCount;
- $piece->eqpos = array();
- # 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 );
- }
-
- extract( $stack->getFlags() );
-
- # Add XML element to the enclosing accumulator
- $accum .= $element;
- }
-
- elseif ( $found == 'pipe' ) {
- $findEquals = true; // shortcut for getFlags()
- $stack->top->addPart();
- $accum =& $stack->getAccum();
- ++$i;
- }
-
- elseif ( $found == 'equals' ) {
- $findEquals = false; // shortcut for getFlags()
- $partsCount = count( $stack->top->parts );
- $stack->top->eqpos[$partsCount - 1] = strlen( $accum );
- $accum .= '=';
- ++$i;
- }
- }
-
- # Output any remaining unclosed brackets
- foreach ( $stack->stack as $piece ) {
- if ( $piece->open == "\n" ) {
- $stack->topAccum .= $piece->parts[0];
- } else {
- $stack->topAccum .= str_repeat( $piece->open, $piece->count ) . implode( '|', $piece->parts );
- }
- }
- $stack->topAccum .= '</root>';
- $xml = $stack->topAccum;
-
- wfProfileOut( __METHOD__.'-makexml' );
- 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 );
- $result = $dom->loadXML( $xml );
- if ( !$result ) {
- throw new MWException( __METHOD__.' generated invalid XML' );
- }
- }
- wfProfileOut( __METHOD__.'-loadXML' );
- wfProfileOut( __METHOD__ );
+ $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
return $dom;
}
wfProfileIn( $fname );
if ( $frame === false ) {
- $frame = new PPFrame( $this );
+ $frame = $this->getPreprocessor()->newFrame();
} elseif ( !( $frame instanceof PPFrame ) ) {
throw new MWException( __METHOD__ . ' called using the old argument format' );
}
* replacing any variables or templates within the template.
*
* @param array $piece The parts of the template
- * $piece['text']: matched text
* $piece['title']: the title, i.e. the part before the |
* $piece['parts']: the parameter array
+ * $piece['lineStart']: whether the brace was at the start of a line
* @param PPFrame The current frame, contains template arguments
* @return string the text of the template
* @private
$nowiki = false; # wiki markup in $text should be escaped
$isHTML = false; # $text is HTML, armour it against wikitext transformation
$forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
- $isDOM = false; # $text is a DOM node needing expansion
+ $isChildObj = false; # $text is a DOM node needing expansion in a child frame
+ $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
# Title object, where $text came from
$title = NULL;
# 1) Found SUBST but not in the PST phase
# 2) Didn't find SUBST and in the PST phase
# In either case, return without further processing
- $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
+ $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
+ $isLocalObj = true;
$found = true;
}
}
# Variables
- if ( !$found && $args->length == 0 ) {
+ if ( !$found && $args->getLength() == 0 ) {
$id = $this->mVariables->matchStartToEnd( $part1 );
if ( $id !== false ) {
$text = $this->getVariableValue( $id );
# Add a frame parameter, and pass the arguments as an array
$allArgs = $initialArgs;
$allArgs[] = $frame;
- foreach ( $args as $arg ) {
- $funcArgs[] = $arg;
+ for ( $i = 0; $i < $args->getLength(); $i++ ) {
+ $funcArgs[] = $args->item( $i );
}
$allArgs[] = $funcArgs;
} else {
# Convert arguments to plain text
- foreach ( $args as $arg ) {
- $funcArgs[] = trim( $frame->expand( $arg ) );
+ for ( $i = 0; $i < $args->getLength(); $i++ ) {
+ $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
}
$allArgs = array_merge( $initialArgs, $funcArgs );
}
list( $text, $title ) = $this->getTemplateDom( $title );
if ( $text !== false ) {
$found = true;
- $isDOM = true;
+ $isChildObj = true;
}
}
$text = $this->interwikiTransclude( $title, 'raw' );
// Preprocess it like a template
$text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
- $isDOM = true;
+ $isChildObj = true;
}
$found = true;
}
# If we haven't found text to substitute by now, we're done
# Recover the source wikitext and return it
if ( !$found ) {
- $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
+ $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
wfProfileOut( $fname );
- return $text;
+ return array( 'object' => $text );
}
# Expand DOM-style return values in a child frame
- if ( $isDOM ) {
+ if ( $isChildObj ) {
# Clean up argument array
$newFrame = $frame->newChild( $args, $title );
# Bug 529: if the template begins with a table or block-level
# element, it should be treated as beginning a new line.
# This behaviour is somewhat controversial.
- elseif ( !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
+ elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
$text = "\n" . $text;
}
- if ( !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
+ if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
# Error, oversize inclusion
$text = "[[$originalTitle]]" .
$this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
}
+ if ( $isLocalObj ) {
+ $ret = array( 'object' => $text );
+ } else {
+ $ret = array( 'text' => $text );
+ }
+
wfProfileOut( $fname );
- return $text;
+ return $ret;
}
/**
$parts = $piece['parts'];
$nameWithSpaces = $frame->expand( $piece['title'] );
$argName = trim( $nameWithSpaces );
-
+ $object = false;
$text = $frame->getArgument( $argName );
- if ( $text === false && ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
+ if ( $text === false && ( $this->ot['html'] || $this->ot['pre'] ) && $parts->getLength() > 0 ) {
# No match in frame, use the supplied default
- $text = $frame->expand( $parts->item( 0 ) );
+ $object = $parts->item( 0 )->getChildren();
}
if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
$error = '<!-- WARNING: argument omitted, expansion size too large -->';
}
- if ( $text === false ) {
+ if ( $text === false && $object === false ) {
# No match anywhere
- $text = '{{{' . $frame->implode( '|', $nameWithSpaces, $parts ) . '}}}';
+ $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
}
if ( $error !== false ) {
$text .= $error;
}
+ if ( $object !== false ) {
+ $ret = array( 'object' => $object );
+ } else {
+ $ret = array( 'text' => $text );
+ }
wfProfileOut( __METHOD__ );
- return $text;
+ return $ret;
}
/**
* This is the ghost of strip().
*
* @param array $params Associative array of parameters:
- * name DOMNode for the tag name
- * attr DOMNode for unparsed text where tag attributes are thought to be
+ * name PPNode for the tag name
+ * attr PPNode for unparsed text where tag attributes are thought to be
* attributes Optional associative array of parsed attributes
* inner Contents of extension element
* noClose Original text did not have a close tag
$text = preg_replace( $substRegex, $substText, $text );
$text = $this->cleanSigInSig( $text );
$dom = $this->preprocessToDom( $text );
- $frame = new PPFrame( $this );
- $text = $frame->expand( $dom->documentElement );
+ $frame = $this->getPreprocessor()->newFrame();
+ $text = $frame->expand( $dom );
if ( !$parsing ) {
$text = $this->mStripState->unstripBoth( $text );
$this->setOutputType( OT_WIKI );
$curIndex = 0;
$outText = '';
- $frame = new PPFrame( $this );
+ $frame = $this->getPreprocessor()->newFrame();
// Process section extraction flags
$flags = 0;
}
}
// Preprocess the text
- $dom = $this->preprocessToDom( $text, $flags );
- $root = $dom->documentElement;
+ $root = $this->preprocessToDom( $text, $flags );
// <h> nodes indicate section breaks
// They can only occur at the top level, so we can find them by iterating the root's children
- $node = $root->firstChild;
+ $node = $root->getFirstChild();
// Find the target section
if ( $sectionIndex == 0 ) {
$targetLevel = 1000;
} else {
while ( $node ) {
- if ( $node->nodeName == 'h' ) {
+ if ( $node->getName() == 'h' ) {
if ( $curIndex + 1 == $sectionIndex ) {
break;
}
if ( $mode == 'replace' ) {
$outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
}
- $node = $node->nextSibling;
+ $node = $node->getNextSibling();
}
if ( $node ) {
- $targetLevel = $node->getAttribute( 'level' );
+ $bits = $node->splitHeading();
+ $targetLevel = $bits['level'];
}
}
// Find the end of the section, including nested sections
do {
- if ( $node->nodeName == 'h' ) {
+ if ( $node->getName() == 'h' ) {
$curIndex++;
- $curLevel = $node->getAttribute( 'level' );
+ $bits = $node->splitHeading();
+ $curLevel = $bits['level'];
if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
break;
}
if ( $mode == 'get' ) {
$outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
}
- $node = $node->nextSibling;
+ $node = $node->getNextSibling();
} while ( $node );
// Write out the remainder (in replace mode only)
$outText .= $newText . "\n\n";
while ( $node ) {
$outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
- $node = $node->nextSibling;
+ $node = $node->getNextSibling();
}
}
return $text;
}
+ function srvus( $text ) {
+ return $this->testSrvus( $text, $this->mOutputType );
+ }
+
/**
* strip/replaceVariables/unstrip for preprocessor regression testing
*/
- function srvus( $text ) {
+ function testSrvus( $text, $title, $options, $outputType = OT_HTML ) {
+ $this->clearState();
+ $this->mTitle = $title;
+ $this->mOptions = $options;
+ $this->setOutputType( $outputType );
$text = $this->replaceVariables( $text );
$text = $this->mStripState->unstripBoth( $text );
$text = Sanitizer::removeHTMLtags( $text );
return $text;
}
+
+ function testPst( $text, $title, $options ) {
+ global $wgUser;
+ return $this->preSaveTransform( $text, $title, $wgUser, $options );
+ }
+
+ function testPreprocess( $text, $title, $options ) {
+ return $this->testSrvus( $text, $title, $options, OT_PREPROCESS );
+ }
}
/**
}
}
-/**
- * An expansion frame, used as a context to expand the result of preprocessToDom()
- */
-class PPFrame {
- var $parser, $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
- */
- var $depth;
-
- const NO_ARGS = 1;
- const NO_TEMPLATES = 2;
- const STRIP_COMMENTS = 4;
- const NO_IGNORE = 8;
-
- const RECOVER_ORIG = 11;
-
- /**
- * Construct a new preprocessor frame.
- * @param Parser $parser The parent parser
- * @param Title $title The context title, or false if there isn't one
- */
- function __construct( $parser ) {
- $this->parser = $parser;
- $this->title = $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 DOMNodeList containing the template arguments
- */
- function newChild( $args = false, $title = false ) {
- $namedArgs = array();
- $numberedArgs = array();
- if ( $title === false ) {
- $title = $this->title;
- }
- if ( $args !== false ) {
- $xpath = false;
- foreach ( $args as $arg ) {
- if ( !$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;
- $numberedArgs[$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( $this->parser, $this, $numberedArgs, $namedArgs, $title );
- }
-
- /**
- * Expand a DOMNode describing a preprocessed document into plain wikitext,
- * using the current context
- * @param $root the node
- */
- function expand( $root, $flags = 0 ) {
- if ( is_string( $root ) ) {
- return $root;
- }
-
- if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
- {
- return '<span class="error">Node-count limit exceeded</span>';
- }
-
- if ( is_array( $root ) || $root instanceof DOMNodeList ) {
- $s = '';
- foreach ( $root as $node ) {
- $s .= $this->expand( $node, $flags );
- }
- } elseif ( $root instanceof DOMNode ) {
- if ( $root->nodeType == XML_TEXT_NODE ) {
- $s = $root->nodeValue;
- } elseif ( $root->nodeName == 'template' ) {
- # Double-brace expansion
- $xpath = new DOMXPath( $root->ownerDocument );
- $titles = $xpath->query( 'title', $root );
- $title = $titles->item( 0 );
- $parts = $xpath->query( 'part', $root );
- if ( $flags & self::NO_TEMPLATES ) {
- $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
- } else {
- $lineStart = $root->getAttribute( 'lineStart' );
- $params = array(
- 'title' => $title,
- 'parts' => $parts,
- 'lineStart' => $lineStart,
- 'text' => 'FIXME' );
- $s = $this->parser->braceSubstitution( $params, $this );
- }
- } elseif ( $root->nodeName == 'tplarg' ) {
- # Triple-brace expansion
- $xpath = new DOMXPath( $root->ownerDocument );
- $titles = $xpath->query( 'title', $root );
- $title = $titles->item( 0 );
- $parts = $xpath->query( 'part', $root );
- if ( $flags & self::NO_ARGS ) {
- $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
- } else {
- $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
- $s = $this->parser->argSubstitution( $params, $this );
- }
- } elseif ( $root->nodeName == 'comment' ) {
- # HTML-style comment
- if ( $this->parser->ot['html']
- || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
- || ( $flags & self::STRIP_COMMENTS ) )
- {
- $s = '';
- } else {
- $s = $root->textContent;
- }
- } elseif ( $root->nodeName == 'ignore' ) {
- # Output suppression used by <includeonly> etc.
- # OT_WIKI will only respect <ignore> 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 & self::NO_IGNORE ) ) {
- $s = $root->textContent;
- } else {
- $s = '';
- }
- } elseif ( $root->nodeName == 'ext' ) {
- # Extension tag
- $xpath = new DOMXPath( $root->ownerDocument );
- $names = $xpath->query( 'name', $root );
- $attrs = $xpath->query( 'attr', $root );
- $inners = $xpath->query( 'inner', $root );
- $closes = $xpath->query( 'close', $root );
- $params = array(
- 'name' => $names->item( 0 ),
- 'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
- 'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
- 'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
- );
- $s = $this->parser->extensionSubstitution( $params, $this );
- } elseif ( $root->nodeName == 'h' ) {
- # Heading
- $s = $this->expand( $root->childNodes, $flags );
-
- if ( $this->parser->ot['html'] ) {
- # Insert heading index marker
- $headingIndex = $root->getAttribute( 'i' );
- $titleText = $this->title->getPrefixedDBkey();
- $this->parser->mHeadings[] = array( $titleText, $headingIndex );
- $serial = count( $this->parser->mHeadings ) - 1;
- $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
- $count = $root->getAttribute( 'level' );
- $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
- $this->parser->mStripState->general->setPair( $marker, '' );
- }
- } else {
- # Generic recursive expansion
- $s = '';
- for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
- if ( $node->nodeType == XML_TEXT_NODE ) {
- $s .= $node->nodeValue;
- } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
- $s .= $this->expand( $node, $flags );
- }
- }
- }
- } else {
- throw new MWException( __METHOD__.': Invalid parameter type' );
- }
- return $s;
- }
-
- function implodeWithFlags( $sep, $flags /*, ... */ ) {
- $args = array_slice( func_get_args(), 2 );
-
- $first = true;
- $s = '';
- foreach ( $args as $root ) {
- 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;
- }
-
- function implode( $sep /*, ... */ ) {
- $args = func_get_args();
- $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
- return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
- }
-
- /**
- * Split an <arg> or <template> node into a three-element array:
- * DOMNode name, string index and DOMNode value
- */
- function splitBraceNode( $node ) {
- $xpath = new DOMXPath( $node->ownerDocument );
- $names = $xpath->query( 'name', $node );
- $values = $xpath->query( 'value', $node );
- if ( !$names->length || !$values->length ) {
- throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
- }
- $name = $names->item( 0 );
- $index = $name->getAttribute( 'index' );
- return array( $name, $index, $values->item( 0 ) );
- }
-
- /**
- * Split an <ext> node into an associative array containing name, attr, inner and close
- * All values in the resulting array are DOMNodes. Inner and close are optional.
- */
- function splitExtNode( $node ) {
- $xpath = new DOMXPath( $node->ownerDocument );
- $names = $xpath->query( 'name', $node );
- $attrs = $xpath->query( 'attr', $node );
- $inners = $xpath->query( 'inner', $node );
- $closes = $xpath->query( 'close', $node );
- if ( !$names->length || !$attrs->length ) {
- throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
- }
- $parts = array(
- 'name' => $names->item( 0 ),
- 'attr' => $attrs->item( 0 ) );
- if ( $inners->length ) {
- $parts['inner'] = $inners->item( 0 );
- }
- if ( $closes->length ) {
- $parts['close'] = $closes->item( 0 );
- }
- return $parts;
- }
-
- 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;
- }
- }
-
- /**
- * Returns true if there are no arguments in this frame
- */
- function isEmpty() {
- return true;
- }
-
- function getArgument( $name ) {
- return false;
- }
-
- /**
- * Returns true if the infinite loop check is OK, false if a loop is detected
- */
- function loopCheck( $title ) {
- return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
- }
-}
-
-/**
- * Expansion frame with template arguments
- */
-class PPTemplateFrame extends PPFrame {
- var $numberedArgs, $namedArgs, $parent;
- var $numberedExpansionCache, $namedExpansionCache;
-
- function __construct( $parser, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
- $this->parser = $parser;
- $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
- */
- function isEmpty() {
- return !count( $this->numberedArgs ) && !count( $this->namedArgs );
- }
-
- 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], self::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], self::STRIP_COMMENTS ) );
- }
- return $this->namedExpansionCache[$name];
- }
-
- function getArgument( $name ) {
- $text = $this->getNumberedArgument( $name );
- if ( $text === false ) {
- $text = $this->getNamedArgument( $name );
- }
- return $text;
- }
-}
-
-/**
- * Stack class to help Parser::preprocessToDom()
- */
-class PPDStack {
- var $stack, $topAccum, $top;
-
- function __construct() {
- $this->stack = array();
- $this->topAccum = '';
- $this->top = false;
- }
-
- function &getAccum() {
- if ( count( $this->stack ) ) {
- return $this->top->getAccum();
- } else {
- return $this->topAccum;
- }
- }
-
- function push( $data ) {
- if ( $data instanceof PPDStackElement ) {
- $this->stack[] = $data;
- } else {
- $this->stack[] = new PPDStackElement( $data );
- }
- $this->top =& $this->stack[ count( $this->stack ) - 1 ];
- }
-
- 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 ];
- } else {
- $this->top = false;
- }
- }
-
- function getFlags() {
- if ( !count( $this->stack ) ) {
- return array(
- 'findEquals' => false,
- 'findPipe' => false,
- 'inHeading' => false,
- );
- } else {
- return $this->top->getFlags();
- }
- }
-}
-
-class PPDStackElement {
- var $open, $close, $count, $parts, $eqpos, $lineStart;
-
- function __construct( $data = array() ) {
- $this->parts = array( '' );
- $this->eqpos = array();
-
- foreach ( $data as $name => $value ) {
- $this->$name = $value;
- }
- }
-
- function &getAccum() {
- return $this->parts[count($this->parts) - 1];
- }
-
- function addPart( $s = '' ) {
- $this->parts[] = $s;
- }
-
- function getFlags() {
- $partCount = count( $this->parts );
- $findPipe = $this->open != "\n" && $this->open != '[';
- return array(
- 'findPipe' => $findPipe,
- 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->eqpos[$partCount - 1] ),
- 'inHeading' => $this->open == "\n",
- );
- }
-}
--- /dev/null
+<?php
+
+interface Preprocessor {
+ function __construct( $parser );
+ function newFrame();
+ function preprocessToObj( $text, $flags = 0 );
+}
+
+interface PPFrame {
+ const NO_ARGS = 1;
+ const NO_TEMPLATES = 2;
+ const STRIP_COMMENTS = 4;
+ const NO_IGNORE = 8;
+
+ const RECOVER_ORIG = 11;
+
+ /**
+ * Create a child frame
+ */
+ function newChild( $args = false, $title = false );
+
+ /**
+ * Expand a document tree node
+ */
+ function expand( $root, $flags = 0 );
+
+ /**
+ * Implode with flags for expand()
+ */
+ function implodeWithFlags( $sep, $flags /*, ... */ );
+
+ /**
+ * Implode with no flags specified
+ */
+ function implode( $sep /*, ... */ );
+
+ /**
+ * Makes an object that, when expand()ed, will be the same as one obtained
+ * with implode()
+ */
+ function virtualImplode( $sep /*, ... */ );
+
+ /**
+ * Virtual implode with brackets
+ */
+ function virtualBracketedImplode( $start, $sep, $end /*, ... */ );
+
+ /**
+ * Returns true if there are no arguments in this frame
+ */
+ function isEmpty();
+
+ function getArgument( $name );
+
+ /**
+ * Returns true if the infinite loop check is OK, false if a loop is detected
+ */
+ function loopCheck( $title );
+}
+
+interface PPNode {
+ function getChildren();
+ function getFirstChild();
+ function getNextSibling();
+ function getChildrenOfType( $type );
+ function getLength();
+ function item( $i );
+ function getName();
+
+ function splitArg();
+ function splitExt();
+ function splitHeading();
+}
+
--- /dev/null
+<?php
+
+class Preprocessor_DOM implements Preprocessor {
+ var $parser;
+
+ function __construct( $parser ) {
+ $this->parser = $parser;
+ }
+
+ function newFrame() {
+ return new PPFrame_DOM( $this );
+ }
+
+ /**
+ * Preprocess some wikitext and return the document tree.
+ * This is the ghost of Parser::replace_variables().
+ *
+ * @param string $text The text to parse
+ * @param integer flags Bitwise combination of:
+ * Parser::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> 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.
+ *
+ * @private
+ */
+ function preprocessToObj( $text, $flags = 0 ) {
+ wfProfileIn( __METHOD__ );
+ wfProfileIn( __METHOD__.'-makexml' );
+
+ $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, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== 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 = '[{<';
+ $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 text accumulator
+ $accum = '<root>';
+ $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 <onlyinclude>
+ $fakeLineStart = true; # Do a line-start run without outputting an LF character
+
+ while ( true ) {
+ if ( $findOnlyinclude ) {
+ // Ignore all input up to the next <onlyinclude>
+ $startPos = strpos( $text, '<onlyinclude>', $i );
+ if ( $startPos === false ) {
+ // Ignored section runs to the end
+ $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
+ break;
+ }
+ $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
+ $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
+ $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 .= '=';
+ } else {
+ // Look for headings
+ // We can't look for headings when $findEquals is true, because the ambiguity
+ // between template name/value separators and heading starts would be unresolved
+ // until the closing double-brace is found. This would mean either infinite
+ // backtrack, or creating and updating two separate tree structures until the
+ // end of the ambiguity -- one tree structure assuming a heading, and the other
+ // assuming a template argument.
+ //
+ // Easier to just break some section edit links.
+ $search .= "\n";
+ }
+ $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 >= strlen( $text ) ) {
+ 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 </onlyinclude>
+ if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
+ $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 .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
+ $i = strlen( $text );
+ } else {
+ // Search backwards for leading whitespace
+ $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
+ // Search forwards for trailing whitespace
+ // $wsEnd will be the position of the last space
+ $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() does 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,
+ // but only if stack->top===false, because headings don't exist at deeper levels.
+ if ( $stack->top === false ) {
+ $fakeLineStart = true;
+ }
+ } else {
+ // No line to eat, just take the comment itself
+ $startPos = $i;
+ $endPos += 2;
+ }
+
+ $i = $endPos + 1;
+ $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
+ $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
+ }
+ continue;
+ }
+ $name = $matches[1];
+ $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( $name, $ignoredTags ) ) {
+ $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
+ $i = $tagEndPos + 1;
+ continue;
+ }
+
+ $tagStartPos = $i;
+ if ( $text[$tagEndPos-1] == '/' ) {
+ $attrEnd = $tagEndPos - 1;
+ $inner = null;
+ $i = $tagEndPos + 1;
+ $close = '';
+ } else {
+ $attrEnd = $tagEndPos;
+ // Find closing tag
+ if ( preg_match( "/<\/$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 = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
+ } else {
+ // No end tag -- let it run out to the end of the text.
+ $inner = substr( $text, $tagEndPos + 1 );
+ $i = strlen( $text );
+ $close = '';
+ }
+ }
+ // <includeonly> and <noinclude> just become <ignore> tags
+ if ( in_array( $name, $ignoredElements ) ) {
+ $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
+ . '</ignore>';
+ continue;
+ }
+
+ $accum .= '<ext>';
+ if ( $attrEnd <= $attrStart ) {
+ $attr = '';
+ } else {
+ $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
+ }
+ $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
+ // Note that the attr element contains the whitespace between name and attribute,
+ // this is necessary for precise reconstruction during pre-save transform.
+ '<attr>' . htmlspecialchars( $attr ) . '</attr>';
+ if ( $inner !== null ) {
+ $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
+ }
+ $accum .= $close . '</ext>';
+ }
+
+ 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 > 0 ) {
+ $piece = array(
+ 'open' => "\n",
+ 'close' => "\n",
+ 'parts' => array( str_repeat( '=', $count ) ),
+ 'startPos' => $i,
+ 'count' => $count );
+ $stack->push( $piece );
+ $accum =& $stack->getAccum();
+ extract( $stack->getFlags() );
+ $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" );
+ // 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
+ $m = false;
+ $count = $piece->count;
+ if ( preg_match( "/\s*(=+)/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
+ if ( $i - strlen( $m[0] ) == $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 = strlen( $m[1] );
+ if ( $count < 3 ) {
+ $count = 0;
+ } else {
+ $count = min( 6, intval( ( $count - 1 ) / 2 ) );
+ }
+ } else {
+ $count = min( strlen( $m[1] ), $count );
+ }
+ if ( $count > 0 ) {
+ // Normal match, output <h>
+ $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
+ $headingIndex++;
+ } else {
+ // Single equals sign on its own line, count=0
+ $element = $accum;
+ }
+ } else {
+ // No match, no <h>, just pass down the inner text
+ $element = $accum;
+ }
+ // Unwind the stack
+ $stack->pop();
+ $accum =& $stack->getAccum();
+ extract( $stack->getFlags() );
+
+ // 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' => $count,
+ 'parts' => array( '' ),
+ 'eqpos' => array(),
+ 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
+ );
+
+ $stack->push( $piece );
+ $accum =& $stack->getAccum();
+ extract( $stack->getFlags() );
+ } 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)
+ $matchingCount = 0;
+ $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 = str_repeat( $piece->open, $matchingCount ) .
+ implode( '|', $piece->parts ) .
+ 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];
+ 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>";
+ $element .= "<title>$title</title>";
+ $argIndex = 1;
+ foreach ( $parts as $partIndex => $part ) {
+ if ( isset( $piece->eqpos[$partIndex] ) ) {
+ $eqpos = $piece->eqpos[$partIndex];
+ $argName = substr( $part, 0, $eqpos );
+ $argValue = substr( $part, $eqpos + 1 );
+ $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
+ } else {
+ $element .= "<part><name index=\"$argIndex\" /><value>$part</value></part>";
+ $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( '' );
+ $piece->count -= $matchingCount;
+ $piece->eqpos = array();
+ # 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 );
+ }
+
+ extract( $stack->getFlags() );
+
+ # Add XML element to the enclosing accumulator
+ $accum .= $element;
+ }
+
+ elseif ( $found == 'pipe' ) {
+ $findEquals = true; // shortcut for getFlags()
+ $stack->top->addPart();
+ $accum =& $stack->getAccum();
+ ++$i;
+ }
+
+ elseif ( $found == 'equals' ) {
+ $findEquals = false; // shortcut for getFlags()
+ $partsCount = count( $stack->top->parts );
+ $stack->top->eqpos[$partsCount - 1] = strlen( $accum );
+ $accum .= '=';
+ ++$i;
+ }
+ }
+
+ # Output any remaining unclosed brackets
+ foreach ( $stack->stack as $piece ) {
+ if ( $piece->open == "\n" ) {
+ $stack->topAccum .= $piece->parts[0];
+ } else {
+ $stack->topAccum .= str_repeat( $piece->open, $piece->count ) . implode( '|', $piece->parts );
+ }
+ }
+ $stack->topAccum .= '</root>';
+ $xml = $stack->topAccum;
+
+ wfProfileOut( __METHOD__.'-makexml' );
+ 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 );
+ $result = $dom->loadXML( $xml );
+ if ( !$result ) {
+ throw new MWException( __METHOD__.' generated invalid XML' );
+ }
+ }
+ $obj = new PPNode_DOM( $dom->documentElement );
+ wfProfileOut( __METHOD__.'-loadXML' );
+ wfProfileOut( __METHOD__ );
+ return $obj;
+ }
+}
+
+/**
+ * An expansion frame, used as a context to expand the result of preprocessToDom()
+ */
+class PPFrame_DOM implements PPFrame {
+ var $preprocessor, $parser, $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
+ */
+ 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
+ */
+ function newChild( $args = false, $title = false ) {
+ $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 $arg ) {
+ if ( !$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;
+ $numberedArgs[$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 );
+ }
+
+ function expand( $root, $flags = 0 ) {
+ if ( is_string( $root ) ) {
+ return $root;
+ }
+
+ if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
+ {
+ return '<span class="error">Node-count limit exceeded</span>';
+ }
+
+ 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 ) {
+ /*
+ print str_repeat( ' ', count( debug_backtrace() ) ) . $contextNode->nodeName;
+ if ( $contextNode->nodeName == 'title' ) {
+ print ' = ' . $contextNode->textContent;
+ }
+ print "<br/>\n";
+ */
+ 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 & self::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 & self::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
+ if ( $this->parser->ot['html']
+ || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
+ || ( $flags & self::STRIP_COMMENTS ) )
+ {
+ $out .= '';
+ } else {
+ $out .= $contextNode->textContent;
+ }
+ } elseif ( $contextNode->nodeName == 'ignore' ) {
+ # Output suppression used by <includeonly> etc.
+ # OT_WIKI will only respect <ignore> 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 & self::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 );
+
+ if ( $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-$serial-{$this->parser->mMarkerSuffix}";
+ $count = $contextNode->getAttribute( 'level' );
+ $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
+ $this->parser->mStripState->general->setPair( $marker, '' );
+ }
+ $out .= $s;
+ } else {
+ # Generic recursive expansion
+ $newIterator = $contextNode->childNodes;
+ }
+ } else {
+ 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--;
+ }
+ }
+ }
+ return $outStack[0];
+ }
+
+ 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
+ */
+ 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()
+ */
+ function virtualImplode( $sep /*, ... */ ) {
+ $args = array_slice( func_get_args(), 1 );
+ $out = array();
+ $first = true;
+ if ( $root instanceof PPNode_DOM ) $root = $root->node;
+
+ foreach ( $args as $root ) {
+ 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
+ */
+ 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;
+ }
+ }
+
+ /**
+ * Returns true if there are no arguments in this frame
+ */
+ function isEmpty() {
+ return true;
+ }
+
+ function getArgument( $name ) {
+ return false;
+ }
+
+ /**
+ * Returns true if the infinite loop check is OK, false if a loop is detected
+ */
+ function loopCheck( $title ) {
+ return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
+ }
+}
+
+/**
+ * Expansion frame with template arguments
+ */
+class PPTemplateFrame_DOM extends PPFrame_DOM {
+ var $numberedArgs, $namedArgs, $parent;
+ var $numberedExpansionCache, $namedExpansionCache;
+
+ function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
+ $this->preprocessor = $preprocessor;
+ $this->parser = $preprocessor->parser;
+ $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
+ */
+ function isEmpty() {
+ return !count( $this->numberedArgs ) && !count( $this->namedArgs );
+ }
+
+ 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], self::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], self::STRIP_COMMENTS ) );
+ }
+ return $this->namedExpansionCache[$name];
+ }
+
+ function getArgument( $name ) {
+ $text = $this->getNumberedArgument( $name );
+ if ( $text === false ) {
+ $text = $this->getNamedArgument( $name );
+ }
+ return $text;
+ }
+}
+
+/**
+ * Stack class to help Parser::preprocessToDom()
+ */
+class PPDStack {
+ var $stack, $topAccum, $top;
+
+ function __construct() {
+ $this->stack = array();
+ $this->topAccum = '';
+ $this->top = false;
+ }
+
+ function &getAccum() {
+ if ( count( $this->stack ) ) {
+ return $this->top->getAccum();
+ } else {
+ return $this->topAccum;
+ }
+ }
+
+ function push( $data ) {
+ if ( $data instanceof PPDStackElement ) {
+ $this->stack[] = $data;
+ } else {
+ $this->stack[] = new PPDStackElement( $data );
+ }
+ $this->top =& $this->stack[ count( $this->stack ) - 1 ];
+ }
+
+ 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 ];
+ } else {
+ $this->top = false;
+ }
+ }
+
+ function getFlags() {
+ if ( !count( $this->stack ) ) {
+ return array(
+ 'findEquals' => false,
+ 'findPipe' => false,
+ 'inHeading' => false,
+ );
+ } else {
+ return $this->top->getFlags();
+ }
+ }
+}
+
+class PPDStackElement {
+ var $open, $close, $count, $parts, $eqpos, $lineStart;
+
+ function __construct( $data = array() ) {
+ $this->parts = array( '' );
+ $this->eqpos = array();
+
+ foreach ( $data as $name => $value ) {
+ $this->$name = $value;
+ }
+ }
+
+ function &getAccum() {
+ return $this->parts[count($this->parts) - 1];
+ }
+
+ function addPart( $s = '' ) {
+ $this->parts[] = $s;
+ }
+
+ function getFlags() {
+ $partCount = count( $this->parts );
+ $findPipe = $this->open != "\n" && $this->open != '[';
+ return array(
+ 'findPipe' => $findPipe,
+ 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->eqpos[$partCount - 1] ),
+ 'inHeading' => $this->open == "\n",
+ );
+ }
+}
+
+class PPNode_DOM implements PPNode {
+ var $node;
+
+ function __construct( $node, $xpath = false ) {
+ $this->node = $node;
+ }
+
+ function __get( $name ) {
+ if ( $name == 'xpath' ) {
+ $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( $node );
+ }
+ return $s;
+ }
+
+ function getChildren() {
+ return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
+ }
+
+ function getFirstChild() {
+ return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
+ }
+
+ function getNextSibling() {
+ return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
+ }
+
+ function getChildrenOfType( $type ) {
+ return new self( $this->xpath->query( $type, $this->node ) );
+ }
+
+ function getLength() {
+ if ( $this->node instanceof DOMNodeList ) {
+ return $this->node->length;
+ } else {
+ return false;
+ }
+ }
+
+ function item( $i ) {
+ $item = $this->node->item( $i );
+ return $item ? new self( $item ) : false;
+ }
+
+ function getName() {
+ if ( $this->node instanceof DOMNodeList ) {
+ return '#nodelist';
+ } else {
+ return $this->node->nodeName;
+ }
+ }
+
+ /**
+ * Split an <arg> node into a three-element array:
+ * PPNode name, string index and PPNode value
+ */
+ function splitArg() {
+ $names = $this->xpath->query( 'name', $this->node );
+ $values = $this->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 <ext> node into an associative array containing name, attr, inner and close
+ * All values in the resulting array are PPNodes. Inner and close are optional.
+ */
+ function splitExt() {
+ $names = $this->xpath->query( 'name', $this->node );
+ $attrs = $this->xpath->query( 'attr', $this->node );
+ $inners = $this->xpath->query( 'inner', $this->node );
+ $closes = $this->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 $parts;
+ }
+
+ /**
+ * Split a <h> node
+ */
+ function splitHeading() {
+ if ( !$this->nodeName == '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()
+ );
+ }
+}
* @return bool
*/
private function runTest( $desc, $input, $result, $opts ) {
+ global $wgParserConf;
if( $this->showProgress ) {
$this->showTesting( $desc );
}
$noxml = (bool)preg_match( '~\\b noxml \\b~x', $opts );
- $parser = new Parser;
+ $class = $wgParserConf['class'];
+ $parser = new $class( $wgParserConf );
foreach( $this->hooks as $tag => $callback ) {
$parser->setHook( $tag, $callback );
}
var $hairs = array(
'[[', ']]', '{{', '}}', '{{{', '}}}',
'<', '>', '<nowiki', '<gallery', '</nowiki>', '</gallery>', '<nOwIkI>', '</NoWiKi>',
- //'<!--' , '-->',
- //'<ref>', '</ref>', '<references/>',
+ '<!--' , '-->',
"\n==", "==\n",
'|', '=', "\n", ' ', "\t", "\x7f",
+ '~~', '~~~', '~~~~', 'subst:',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',
'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't',
+
+ // extensions
+ //'<ref>', '</ref>', '<references/>',
);
var $minLength = 0;
var $maxLength = 20;
var $maxTemplates = 5;
- var $outputTypes = array( 'OT_HTML', 'OT_WIKI', 'OT_MSG', 'OT_PREPROCESS' );
+ //var $outputTypes = array( 'OT_HTML', 'OT_WIKI', 'OT_PREPROCESS' );
+ var $entryPoints = array( 'testSrvus', 'testPst', 'testPreprocess' );
static $currentTest = false;
function execute() {
return Title::newFromText( mt_rand( 0, 1000000 ), mt_rand( 0, 10 ) );
}
+ /*
function pickOutputType() {
$count = count( $this->outputTypes );
return $this->outputTypes[ mt_rand( 0, $count - 1 ) ];
+ }*/
+
+ function pickEntryPoint() {
+ $count = count( $this->entryPoints );
+ return $this->entryPoints[ mt_rand( 0, $count - 1 ) ];
}
}
class PPFuzzTest {
- var $templates, $mainText, $title;
+ var $templates, $mainText, $title, $entryPoint;
function __construct( $tester ) {
$this->parent = $tester;
$this->mainText = $tester->makeInputText();
$this->title = $tester->makeTitle();
- $this->outputType = $tester->pickOutputType();
+ //$this->outputType = $tester->pickOutputType();
+ $this->entryPoint = $tester->pickEntryPoint();
+ $this->nickname = $tester->makeInputText();
+ $this->fancySig = (bool)mt_rand( 0, 1 );
$this->templates = array();
}
function templateHook( $title ) {
$titleText = $title->getPrefixedDBkey();
-
+
if ( !isset( $this->templates[$titleText] ) ) {
$finalTitle = $title;
if ( count( $this->templates ) >= $this->parent->maxTemplates ) {
}
function execute() {
- global $wgParser;
+ global $wgParser, $wgUser;
+
+ $wgUser = new PPFuzzUser;
+ $wgUser->mName = 'Fuzz';
+ $wgUser->mFrom = 'name';
+ $wgUser->ppfz_test = $this;
+
$options = new ParserOptions;
$options->setTemplateCallback( array( $this, 'templateHook' ) );
- $wgParser->startExternalParse( $this->title, $options, constant( $this->outputType ) );
- return $wgParser->srvus( $this->mainText );
+ //$wgParser->startExternalParse( $this->title, $options, constant( $this->outputType ) );
+ return call_user_func( array( $wgParser, $this->entryPoint ), $this->mainText, $this->title, $options );
}
function getReport() {
$s = "Title: " . $this->title->getPrefixedDBkey() . "\n" .
- "Output type: {$this->outputType}\n" .
+// "Output type: {$this->outputType}\n" .
+ "Entry point: {$this->entryPoint}\n" .
+ "User: " . ( $this->fancySig ? 'fancy' : 'no-fancy' ) . ' ' . var_export( $this->nickname, true ) . "\n" .
"Main text: " . var_export( $this->mainText, true ) . "\n";
foreach ( $this->templates as $titleText => $template ) {
$finalTitle = $template['finalTitle'];
}
}
+class PPFuzzUser extends User {
+ var $ppfz_test;
+
+ function getOption( $option, $defaultOverride = '' ) {
+ if ( $option === 'fancysig' ) {
+ return $this->ppfz_test->fancySig;
+ } elseif ( $option === 'nickname' ) {
+ return $this->ppfz_test->nickname;
+ } else {
+ return parent::getOption( $option, $defaultOverride );
+ }
+ }
+}
+
ini_set( 'memory_limit', '50M' );
if ( isset( $args[0] ) ) {
$testText = file_get_contents( $args[0] );
dépôts / lhc / web / wiklou.git / commitdiff