3 * Preprocessor using PHP arrays
10 * Differences from DOM schema:
11 * * attribute nodes are children
12 * * <h> nodes that aren't at the top are replaced with <possible-h>
15 class Preprocessor_Hash
implements Preprocessor
{
21 const CACHE_VERSION
= 1;
23 function __construct( $parser ) {
24 $this->parser
= $parser;
28 return new PPFrame_Hash( $this );
31 function newCustomFrame( $args ) {
32 return new PPCustomFrame_Hash( $this, $args );
35 function newPartNodeArray( $values ) {
38 foreach ( $values as $k => $val ) {
39 $partNode = new PPNode_Hash_Tree( 'part' );
40 $nameNode = new PPNode_Hash_Tree( 'name' );
43 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $k ) );
44 $partNode->addChild( $nameNode );
46 $nameNode->addChild( new PPNode_Hash_Text( $k ) );
47 $partNode->addChild( $nameNode );
48 $partNode->addChild( new PPNode_Hash_Text( '=' ) );
51 $valueNode = new PPNode_Hash_Tree( 'value' );
52 $valueNode->addChild( new PPNode_Hash_Text( $val ) );
53 $partNode->addChild( $valueNode );
58 $node = new PPNode_Hash_Array( $list );
63 * Preprocess some wikitext and return the document tree.
64 * This is the ghost of Parser::replace_variables().
66 * @param $text String: the text to parse
67 * @param $flags Integer: bitwise combination of:
68 * Parser::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
69 * included. Default is to assume a direct page view.
71 * The generated DOM tree must depend only on the input text and the flags.
72 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
74 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
75 * change in the DOM tree for a given text, must be passed through the section identifier
76 * in the section edit link and thus back to extractSections().
78 * The output of this function is currently only cached in process memory, but a persistent
79 * cache may be implemented at a later date which takes further advantage of these strict
80 * dependency requirements.
84 function preprocessToObj( $text, $flags = 0 ) {
85 wfProfileIn( __METHOD__
);
88 global $wgMemc, $wgPreprocessorCacheThreshold;
90 $cacheable = $wgPreprocessorCacheThreshold !== false && strlen( $text ) > $wgPreprocessorCacheThreshold;
92 wfProfileIn( __METHOD__
.'-cacheable' );
94 $cacheKey = wfMemcKey( 'preprocess-hash', md5($text), $flags );
95 $cacheValue = $wgMemc->get( $cacheKey );
97 $version = substr( $cacheValue, 0, 8 );
98 if ( intval( $version ) == self
::CACHE_VERSION
) {
99 $hash = unserialize( substr( $cacheValue, 8 ) );
101 wfDebugLog( "Preprocessor",
102 "Loaded preprocessor hash from memcached (key $cacheKey)" );
103 wfProfileOut( __METHOD__
.'-cacheable' );
104 wfProfileOut( __METHOD__
);
108 wfProfileIn( __METHOD__
.'-cache-miss' );
123 'names' => array( 2 => null ),
129 $forInclusion = $flags & Parser
::PTD_FOR_INCLUSION
;
131 $xmlishElements = $this->parser
->getStripList();
132 $enableOnlyinclude = false;
133 if ( $forInclusion ) {
134 $ignoredTags = array( 'includeonly', '/includeonly' );
135 $ignoredElements = array( 'noinclude' );
136 $xmlishElements[] = 'noinclude';
137 if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
138 $enableOnlyinclude = true;
141 $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
142 $ignoredElements = array( 'includeonly' );
143 $xmlishElements[] = 'includeonly';
145 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
147 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
148 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
150 $stack = new PPDStack_Hash
;
152 $searchBase = "[{<\n";
153 $revText = strrev( $text ); // For fast reverse searches
155 $i = 0; # Input pointer, starts out pointing to a pseudo-newline before the start
156 $accum =& $stack->getAccum(); # Current accumulator
157 $findEquals = false; # True to find equals signs in arguments
158 $findPipe = false; # True to take notice of pipe characters
160 $inHeading = false; # True if $i is inside a possible heading
161 $noMoreGT = false; # True if there are no more greater-than (>) signs right of $i
162 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
163 $fakeLineStart = true; # Do a line-start run without outputting an LF character
168 if ( $findOnlyinclude ) {
169 // Ignore all input up to the next <onlyinclude>
170 $startPos = strpos( $text, '<onlyinclude>', $i );
171 if ( $startPos === false ) {
172 // Ignored section runs to the end
173 $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
176 $tagEndPos = $startPos +
strlen( '<onlyinclude>' ); // past-the-end
177 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
179 $findOnlyinclude = false;
182 if ( $fakeLineStart ) {
183 $found = 'line-start';
186 # Find next opening brace, closing brace or pipe
187 $search = $searchBase;
188 if ( $stack->top
=== false ) {
189 $currentClosing = '';
191 $currentClosing = $stack->top
->close
;
192 $search .= $currentClosing;
198 // First equals will be for the template
202 # Output literal section, advance input counter
203 $literalLength = strcspn( $text, $search, $i );
204 if ( $literalLength > 0 ) {
205 $accum->addLiteral( substr( $text, $i, $literalLength ) );
206 $i +
= $literalLength;
208 if ( $i >= strlen( $text ) ) {
209 if ( $currentClosing == "\n" ) {
210 // Do a past-the-end run to finish off the heading
218 $curChar = $text[$i];
219 if ( $curChar == '|' ) {
221 } elseif ( $curChar == '=' ) {
223 } elseif ( $curChar == '<' ) {
225 } elseif ( $curChar == "\n" ) {
229 $found = 'line-start';
231 } elseif ( $curChar == $currentClosing ) {
233 } elseif ( isset( $rules[$curChar] ) ) {
235 $rule = $rules[$curChar];
237 # Some versions of PHP have a strcspn which stops on null characters
238 # Ignore and continue
245 if ( $found == 'angle' ) {
247 // Handle </onlyinclude>
248 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
249 $findOnlyinclude = true;
253 // Determine element name
254 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i +
1 ) ) {
255 // Element name missing or not listed
256 $accum->addLiteral( '<' );
261 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
262 // To avoid leaving blank lines, when a comment is both preceded
263 // and followed by a newline (ignoring spaces), trim leading and
264 // trailing spaces and one of the newlines.
267 $endPos = strpos( $text, '-->', $i +
4 );
268 if ( $endPos === false ) {
269 // Unclosed comment in input, runs to end
270 $inner = substr( $text, $i );
271 $accum->addNodeWithText( 'comment', $inner );
272 $i = strlen( $text );
274 // Search backwards for leading whitespace
275 $wsStart = $i ?
( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
276 // Search forwards for trailing whitespace
277 // $wsEnd will be the position of the last space (or the '>' if there's none)
278 $wsEnd = $endPos +
2 +
strspn( $text, ' ', $endPos +
3 );
279 // Eat the line if possible
280 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
281 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
282 // it's a possible beneficial b/c break.
283 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
284 && substr( $text, $wsEnd +
1, 1 ) == "\n" )
286 $startPos = $wsStart;
287 $endPos = $wsEnd +
1;
288 // Remove leading whitespace from the end of the accumulator
289 // Sanity check first though
290 $wsLength = $i - $wsStart;
292 && $accum->lastNode
instanceof PPNode_Hash_Text
293 && substr( $accum->lastNode
->value
, -$wsLength ) === str_repeat( ' ', $wsLength ) )
295 $accum->lastNode
->value
= substr( $accum->lastNode
->value
, 0, -$wsLength );
297 // Do a line-start run next time to look for headings after the comment
298 $fakeLineStart = true;
300 // No line to eat, just take the comment itself
306 $part = $stack->top
->getCurrentPart();
307 if ( ! (isset( $part->commentEnd
) && $part->commentEnd
== $wsStart - 1 )) {
308 $part->visualEnd
= $wsStart;
310 // Else comments abutting, no change in visual end
311 $part->commentEnd
= $endPos;
314 $inner = substr( $text, $startPos, $endPos - $startPos +
1 );
315 $accum->addNodeWithText( 'comment', $inner );
320 $lowerName = strtolower( $name );
321 $attrStart = $i +
strlen( $name ) +
1;
324 $tagEndPos = $noMoreGT ?
false : strpos( $text, '>', $attrStart );
325 if ( $tagEndPos === false ) {
326 // Infinite backtrack
327 // Disable tag search to prevent worst-case O(N^2) performance
329 $accum->addLiteral( '<' );
334 // Handle ignored tags
335 if ( in_array( $lowerName, $ignoredTags ) ) {
336 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i +
1 ) );
342 if ( $text[$tagEndPos-1] == '/' ) {
344 $attrEnd = $tagEndPos - 1;
349 $attrEnd = $tagEndPos;
351 if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
352 $text, $matches, PREG_OFFSET_CAPTURE
, $tagEndPos +
1 ) )
354 $inner = substr( $text, $tagEndPos +
1, $matches[0][1] - $tagEndPos - 1 );
355 $i = $matches[0][1] +
strlen( $matches[0][0] );
356 $close = $matches[0][0];
358 // No end tag -- let it run out to the end of the text.
359 $inner = substr( $text, $tagEndPos +
1 );
360 $i = strlen( $text );
364 // <includeonly> and <noinclude> just become <ignore> tags
365 if ( in_array( $lowerName, $ignoredElements ) ) {
366 $accum->addNodeWithText( 'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
370 if ( $attrEnd <= $attrStart ) {
373 // Note that the attr element contains the whitespace between name and attribute,
374 // this is necessary for precise reconstruction during pre-save transform.
375 $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
378 $extNode = new PPNode_Hash_Tree( 'ext' );
379 $extNode->addChild( PPNode_Hash_Tree
::newWithText( 'name', $name ) );
380 $extNode->addChild( PPNode_Hash_Tree
::newWithText( 'attr', $attr ) );
381 if ( $inner !== null ) {
382 $extNode->addChild( PPNode_Hash_Tree
::newWithText( 'inner', $inner ) );
384 if ( $close !== null ) {
385 $extNode->addChild( PPNode_Hash_Tree
::newWithText( 'close', $close ) );
387 $accum->addNode( $extNode );
390 elseif ( $found == 'line-start' ) {
391 // Is this the start of a heading?
392 // Line break belongs before the heading element in any case
393 if ( $fakeLineStart ) {
394 $fakeLineStart = false;
396 $accum->addLiteral( $curChar );
400 $count = strspn( $text, '=', $i, 6 );
401 if ( $count == 1 && $findEquals ) {
402 // DWIM: This looks kind of like a name/value separator
403 // Let's let the equals handler have it and break the potential heading
404 // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex.
405 } elseif ( $count > 0 ) {
409 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
412 $stack->push( $piece );
413 $accum =& $stack->getAccum();
414 extract( $stack->getFlags() );
419 elseif ( $found == 'line-end' ) {
420 $piece = $stack->top
;
421 // A heading must be open, otherwise \n wouldn't have been in the search list
422 assert( $piece->open
== "\n" );
423 $part = $piece->getCurrentPart();
424 // Search back through the input to see if it has a proper close
425 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
426 $wsLength = strspn( $revText, " \t", strlen( $text ) - $i );
427 $searchStart = $i - $wsLength;
428 if ( isset( $part->commentEnd
) && $searchStart - 1 == $part->commentEnd
) {
429 // Comment found at line end
430 // Search for equals signs before the comment
431 $searchStart = $part->visualEnd
;
432 $searchStart -= strspn( $revText, " \t", strlen( $text ) - $searchStart );
434 $count = $piece->count
;
435 $equalsLength = strspn( $revText, '=', strlen( $text ) - $searchStart );
436 if ( $equalsLength > 0 ) {
437 if ( $searchStart - $equalsLength == $piece->startPos
) {
438 // This is just a single string of equals signs on its own line
439 // Replicate the doHeadings behaviour /={count}(.+)={count}/
440 // First find out how many equals signs there really are (don't stop at 6)
441 $count = $equalsLength;
445 $count = min( 6, intval( ( $count - 1 ) / 2 ) );
448 $count = min( $equalsLength, $count );
451 // Normal match, output <h>
452 $element = new PPNode_Hash_Tree( 'possible-h' );
453 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
454 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++
) );
455 $element->lastChild
->nextSibling
= $accum->firstNode
;
456 $element->lastChild
= $accum->lastNode
;
458 // Single equals sign on its own line, count=0
462 // No match, no <h>, just pass down the inner text
467 $accum =& $stack->getAccum();
468 extract( $stack->getFlags() );
470 // Append the result to the enclosing accumulator
471 if ( $element instanceof PPNode
) {
472 $accum->addNode( $element );
474 $accum->addAccum( $element );
476 // Note that we do NOT increment the input pointer.
477 // This is because the closing linebreak could be the opening linebreak of
478 // another heading. Infinite loops are avoided because the next iteration MUST
479 // hit the heading open case above, which unconditionally increments the
483 elseif ( $found == 'open' ) {
484 # count opening brace characters
485 $count = strspn( $text, $curChar, $i );
487 # we need to add to stack only if opening brace count is enough for one of the rules
488 if ( $count >= $rule['min'] ) {
489 # Add it to the stack
492 'close' => $rule['end'],
494 'lineStart' => ($i == 0 ||
$text[$i-1] == "\n"),
497 $stack->push( $piece );
498 $accum =& $stack->getAccum();
499 extract( $stack->getFlags() );
501 # Add literal brace(s)
502 $accum->addLiteral( str_repeat( $curChar, $count ) );
507 elseif ( $found == 'close' ) {
508 $piece = $stack->top
;
509 # lets check if there are enough characters for closing brace
510 $maxCount = $piece->count
;
511 $count = strspn( $text, $curChar, $i, $maxCount );
513 # check for maximum matching characters (if there are 5 closing
514 # characters, we will probably need only 3 - depending on the rules)
515 $rule = $rules[$piece->open
];
516 if ( $count > $rule['max'] ) {
517 # The specified maximum exists in the callback array, unless the caller
519 $matchingCount = $rule['max'];
521 # Count is less than the maximum
522 # Skip any gaps in the callback array to find the true largest match
523 # Need to use array_key_exists not isset because the callback can be null
524 $matchingCount = $count;
525 while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
530 if ($matchingCount <= 0) {
531 # No matching element found in callback array
532 # Output a literal closing brace and continue
533 $accum->addLiteral( str_repeat( $curChar, $count ) );
537 $name = $rule['names'][$matchingCount];
538 if ( $name === null ) {
539 // No element, just literal text
540 $element = $piece->breakSyntax( $matchingCount );
541 $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
544 # Note: $parts is already XML, does not need to be encoded further
545 $parts = $piece->parts
;
546 $titleAccum = $parts[0]->out
;
549 $element = new PPNode_Hash_Tree( $name );
551 # The invocation is at the start of the line if lineStart is set in
552 # the stack, and all opening brackets are used up.
553 if ( $maxCount == $matchingCount && !empty( $piece->lineStart
) ) {
554 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
556 $titleNode = new PPNode_Hash_Tree( 'title' );
557 $titleNode->firstChild
= $titleAccum->firstNode
;
558 $titleNode->lastChild
= $titleAccum->lastNode
;
559 $element->addChild( $titleNode );
561 foreach ( $parts as $part ) {
562 if ( isset( $part->eqpos
) ) {
565 for ( $node = $part->out
->firstNode
; $node; $node = $node->nextSibling
) {
566 if ( $node === $part->eqpos
) {
572 throw new MWException( __METHOD__
. ': eqpos not found' );
574 if ( $node->name
!== 'equals' ) {
575 throw new MWException( __METHOD__
.': eqpos is not equals' );
579 // Construct name node
580 $nameNode = new PPNode_Hash_Tree( 'name' );
581 if ( $lastNode !== false ) {
582 $lastNode->nextSibling
= false;
583 $nameNode->firstChild
= $part->out
->firstNode
;
584 $nameNode->lastChild
= $lastNode;
587 // Construct value node
588 $valueNode = new PPNode_Hash_Tree( 'value' );
589 if ( $equalsNode->nextSibling
!== false ) {
590 $valueNode->firstChild
= $equalsNode->nextSibling
;
591 $valueNode->lastChild
= $part->out
->lastNode
;
593 $partNode = new PPNode_Hash_Tree( 'part' );
594 $partNode->addChild( $nameNode );
595 $partNode->addChild( $equalsNode->firstChild
);
596 $partNode->addChild( $valueNode );
597 $element->addChild( $partNode );
599 $partNode = new PPNode_Hash_Tree( 'part' );
600 $nameNode = new PPNode_Hash_Tree( 'name' );
601 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++
) );
602 $valueNode = new PPNode_Hash_Tree( 'value' );
603 $valueNode->firstChild
= $part->out
->firstNode
;
604 $valueNode->lastChild
= $part->out
->lastNode
;
605 $partNode->addChild( $nameNode );
606 $partNode->addChild( $valueNode );
607 $element->addChild( $partNode );
612 # Advance input pointer
613 $i +
= $matchingCount;
617 $accum =& $stack->getAccum();
619 # Re-add the old stack element if it still has unmatched opening characters remaining
620 if ($matchingCount < $piece->count
) {
621 $piece->parts
= array( new PPDPart_Hash
);
622 $piece->count
-= $matchingCount;
623 # do we still qualify for any callback with remaining count?
624 $names = $rules[$piece->open
]['names'];
626 $enclosingAccum =& $accum;
627 while ( $piece->count
) {
628 if ( array_key_exists( $piece->count
, $names ) ) {
629 $stack->push( $piece );
630 $accum =& $stack->getAccum();
636 $enclosingAccum->addLiteral( str_repeat( $piece->open
, $skippedBraces ) );
639 extract( $stack->getFlags() );
641 # Add XML element to the enclosing accumulator
642 if ( $element instanceof PPNode
) {
643 $accum->addNode( $element );
645 $accum->addAccum( $element );
649 elseif ( $found == 'pipe' ) {
650 $findEquals = true; // shortcut for getFlags()
652 $accum =& $stack->getAccum();
656 elseif ( $found == 'equals' ) {
657 $findEquals = false; // shortcut for getFlags()
658 $accum->addNodeWithText( 'equals', '=' );
659 $stack->getCurrentPart()->eqpos
= $accum->lastNode
;
664 # Output any remaining unclosed brackets
665 foreach ( $stack->stack
as $piece ) {
666 $stack->rootAccum
->addAccum( $piece->breakSyntax() );
669 # Enable top-level headings
670 for ( $node = $stack->rootAccum
->firstNode
; $node; $node = $node->nextSibling
) {
671 if ( isset( $node->name
) && $node->name
=== 'possible-h' ) {
676 $rootNode = new PPNode_Hash_Tree( 'root' );
677 $rootNode->firstChild
= $stack->rootAccum
->firstNode
;
678 $rootNode->lastChild
= $stack->rootAccum
->lastNode
;
682 $cacheValue = sprintf( "%08d", self
::CACHE_VERSION
) . serialize( $rootNode );
683 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
684 wfProfileOut( __METHOD__
.'-cache-miss' );
685 wfProfileOut( __METHOD__
.'-cacheable' );
686 wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" );
689 wfProfileOut( __METHOD__
);
695 * Stack class to help Preprocessor::preprocessToObj()
698 class PPDStack_Hash
extends PPDStack
{
699 function __construct() {
700 $this->elementClass
= 'PPDStackElement_Hash';
701 parent
::__construct();
702 $this->rootAccum
= new PPDAccum_Hash
;
709 class PPDStackElement_Hash
extends PPDStackElement
{
710 function __construct( $data = array() ) {
711 $this->partClass
= 'PPDPart_Hash';
712 parent
::__construct( $data );
716 * Get the accumulator that would result if the close is not found.
718 function breakSyntax( $openingCount = false ) {
719 if ( $this->open
== "\n" ) {
720 $accum = $this->parts
[0]->out
;
722 if ( $openingCount === false ) {
723 $openingCount = $this->count
;
725 $accum = new PPDAccum_Hash
;
726 $accum->addLiteral( str_repeat( $this->open
, $openingCount ) );
728 foreach ( $this->parts
as $part ) {
732 $accum->addLiteral( '|' );
734 $accum->addAccum( $part->out
);
744 class PPDPart_Hash
extends PPDPart
{
745 function __construct( $out = '' ) {
746 $accum = new PPDAccum_Hash
;
748 $accum->addLiteral( $out );
750 parent
::__construct( $accum );
757 class PPDAccum_Hash
{
758 var $firstNode, $lastNode;
760 function __construct() {
761 $this->firstNode
= $this->lastNode
= false;
765 * Append a string literal
767 function addLiteral( $s ) {
768 if ( $this->lastNode
=== false ) {
769 $this->firstNode
= $this->lastNode
= new PPNode_Hash_Text( $s );
770 } elseif ( $this->lastNode
instanceof PPNode_Hash_Text
) {
771 $this->lastNode
->value
.= $s;
773 $this->lastNode
->nextSibling
= new PPNode_Hash_Text( $s );
774 $this->lastNode
= $this->lastNode
->nextSibling
;
781 function addNode( PPNode
$node ) {
782 if ( $this->lastNode
=== false ) {
783 $this->firstNode
= $this->lastNode
= $node;
785 $this->lastNode
->nextSibling
= $node;
786 $this->lastNode
= $node;
791 * Append a tree node with text contents
793 function addNodeWithText( $name, $value ) {
794 $node = PPNode_Hash_Tree
::newWithText( $name, $value );
795 $this->addNode( $node );
799 * Append a PPAccum_Hash
800 * Takes over ownership of the nodes in the source argument. These nodes may
801 * subsequently be modified, especially nextSibling.
803 function addAccum( $accum ) {
804 if ( $accum->lastNode
=== false ) {
806 } elseif ( $this->lastNode
=== false ) {
807 $this->firstNode
= $accum->firstNode
;
808 $this->lastNode
= $accum->lastNode
;
810 $this->lastNode
->nextSibling
= $accum->firstNode
;
811 $this->lastNode
= $accum->lastNode
;
817 * An expansion frame, used as a context to expand the result of preprocessToObj()
820 class PPFrame_Hash
implements PPFrame
{
839 * Hashtable listing templates which are disallowed for expansion in this frame,
840 * having been encountered previously in parent frames.
845 * Recursion depth of this frame, top = 0
846 * Note that this is NOT the same as expansion depth in expand()
852 * Construct a new preprocessor frame.
853 * @param $preprocessor Preprocessor: the parent preprocessor
855 function __construct( $preprocessor ) {
856 $this->preprocessor
= $preprocessor;
857 $this->parser
= $preprocessor->parser
;
858 $this->title
= $this->parser
->mTitle
;
859 $this->titleCache
= array( $this->title ?
$this->title
->getPrefixedDBkey() : false );
860 $this->loopCheckHash
= array();
865 * Create a new child frame
866 * $args is optionally a multi-root PPNode or array containing the template arguments
868 function newChild( $args = false, $title = false ) {
869 $namedArgs = array();
870 $numberedArgs = array();
871 if ( $title === false ) {
872 $title = $this->title
;
874 if ( $args !== false ) {
875 if ( $args instanceof PPNode_Hash_Array
) {
876 $args = $args->value
;
877 } elseif ( !is_array( $args ) ) {
878 throw new MWException( __METHOD__
. ': $args must be array or PPNode_Hash_Array' );
880 foreach ( $args as $arg ) {
881 $bits = $arg->splitArg();
882 if ( $bits['index'] !== '' ) {
883 // Numbered parameter
884 $numberedArgs[$bits['index']] = $bits['value'];
885 unset( $namedArgs[$bits['index']] );
888 $name = trim( $this->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
889 $namedArgs[$name] = $bits['value'];
890 unset( $numberedArgs[$name] );
894 return new PPTemplateFrame_Hash( $this->preprocessor
, $this, $numberedArgs, $namedArgs, $title );
897 function expand( $root, $flags = 0 ) {
898 static $expansionDepth = 0;
899 if ( is_string( $root ) ) {
903 if ( ++
$this->parser
->mPPNodeCount
> $this->parser
->mOptions
->getMaxPPNodeCount() ) {
904 return '<span class="error">Node-count limit exceeded</span>';
906 if ( $expansionDepth > $this->parser
->mOptions
->getMaxPPExpandDepth() ) {
907 return '<span class="error">Expansion depth limit exceeded</span>';
911 $outStack = array( '', '' );
912 $iteratorStack = array( false, $root );
913 $indexStack = array( 0, 0 );
915 while ( count( $iteratorStack ) > 1 ) {
916 $level = count( $outStack ) - 1;
917 $iteratorNode =& $iteratorStack[ $level ];
918 $out =& $outStack[$level];
919 $index =& $indexStack[$level];
921 if ( is_array( $iteratorNode ) ) {
922 if ( $index >= count( $iteratorNode ) ) {
923 // All done with this iterator
924 $iteratorStack[$level] = false;
925 $contextNode = false;
927 $contextNode = $iteratorNode[$index];
930 } elseif ( $iteratorNode instanceof PPNode_Hash_Array
) {
931 if ( $index >= $iteratorNode->getLength() ) {
932 // All done with this iterator
933 $iteratorStack[$level] = false;
934 $contextNode = false;
936 $contextNode = $iteratorNode->item( $index );
940 // Copy to $contextNode and then delete from iterator stack,
941 // because this is not an iterator but we do have to execute it once
942 $contextNode = $iteratorStack[$level];
943 $iteratorStack[$level] = false;
946 $newIterator = false;
948 if ( $contextNode === false ) {
950 } elseif ( is_string( $contextNode ) ) {
951 $out .= $contextNode;
952 } elseif ( is_array( $contextNode ) ||
$contextNode instanceof PPNode_Hash_Array
) {
953 $newIterator = $contextNode;
954 } elseif ( $contextNode instanceof PPNode_Hash_Attr
) {
956 } elseif ( $contextNode instanceof PPNode_Hash_Text
) {
957 $out .= $contextNode->value
;
958 } elseif ( $contextNode instanceof PPNode_Hash_Tree
) {
959 if ( $contextNode->name
== 'template' ) {
960 # Double-brace expansion
961 $bits = $contextNode->splitTemplate();
962 if ( $flags & PPFrame
::NO_TEMPLATES
) {
963 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $bits['title'], $bits['parts'] );
965 $ret = $this->parser
->braceSubstitution( $bits, $this );
966 if ( isset( $ret['object'] ) ) {
967 $newIterator = $ret['object'];
969 $out .= $ret['text'];
972 } elseif ( $contextNode->name
== 'tplarg' ) {
973 # Triple-brace expansion
974 $bits = $contextNode->splitTemplate();
975 if ( $flags & PPFrame
::NO_ARGS
) {
976 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $bits['title'], $bits['parts'] );
978 $ret = $this->parser
->argSubstitution( $bits, $this );
979 if ( isset( $ret['object'] ) ) {
980 $newIterator = $ret['object'];
982 $out .= $ret['text'];
985 } elseif ( $contextNode->name
== 'comment' ) {
987 # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
988 if ( $this->parser
->ot
['html']
989 ||
( $this->parser
->ot
['pre'] && $this->parser
->mOptions
->getRemoveComments() )
990 ||
( $flags & PPFrame
::STRIP_COMMENTS
) )
994 # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result
995 # Not in RECOVER_COMMENTS mode (extractSections) though
996 elseif ( $this->parser
->ot
['wiki'] && ! ( $flags & PPFrame
::RECOVER_COMMENTS
) ) {
997 $out .= $this->parser
->insertStripItem( $contextNode->firstChild
->value
);
999 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1001 $out .= $contextNode->firstChild
->value
;
1003 } elseif ( $contextNode->name
== 'ignore' ) {
1004 # Output suppression used by <includeonly> etc.
1005 # OT_WIKI will only respect <ignore> in substed templates.
1006 # The other output types respect it unless NO_IGNORE is set.
1007 # extractSections() sets NO_IGNORE and so never respects it.
1008 if ( ( !isset( $this->parent
) && $this->parser
->ot
['wiki'] ) ||
( $flags & PPFrame
::NO_IGNORE
) ) {
1009 $out .= $contextNode->firstChild
->value
;
1013 } elseif ( $contextNode->name
== 'ext' ) {
1015 $bits = $contextNode->splitExt() +
array( 'attr' => null, 'inner' => null, 'close' => null );
1016 $out .= $this->parser
->extensionSubstitution( $bits, $this );
1017 } elseif ( $contextNode->name
== 'h' ) {
1019 if ( $this->parser
->ot
['html'] ) {
1020 # Expand immediately and insert heading index marker
1022 for ( $node = $contextNode->firstChild
; $node; $node = $node->nextSibling
) {
1023 $s .= $this->expand( $node, $flags );
1026 $bits = $contextNode->splitHeading();
1027 $titleText = $this->title
->getPrefixedDBkey();
1028 $this->parser
->mHeadings
[] = array( $titleText, $bits['i'] );
1029 $serial = count( $this->parser
->mHeadings
) - 1;
1030 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser
::MARKER_SUFFIX
;
1031 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
1032 $this->parser
->mStripState
->addGeneral( $marker, '' );
1035 # Expand in virtual stack
1036 $newIterator = $contextNode->getChildren();
1039 # Generic recursive expansion
1040 $newIterator = $contextNode->getChildren();
1043 throw new MWException( __METHOD__
.': Invalid parameter type' );
1046 if ( $newIterator !== false ) {
1048 $iteratorStack[] = $newIterator;
1050 } elseif ( $iteratorStack[$level] === false ) {
1051 // Return accumulated value to parent
1052 // With tail recursion
1053 while ( $iteratorStack[$level] === false && $level > 0 ) {
1054 $outStack[$level - 1] .= $out;
1055 array_pop( $outStack );
1056 array_pop( $iteratorStack );
1057 array_pop( $indexStack );
1063 return $outStack[0];
1066 function implodeWithFlags( $sep, $flags /*, ... */ ) {
1067 $args = array_slice( func_get_args(), 2 );
1071 foreach ( $args as $root ) {
1072 if ( $root instanceof PPNode_Hash_Array
) {
1073 $root = $root->value
;
1075 if ( !is_array( $root ) ) {
1076 $root = array( $root );
1078 foreach ( $root as $node ) {
1084 $s .= $this->expand( $node, $flags );
1091 * Implode with no flags specified
1092 * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1094 function implode( $sep /*, ... */ ) {
1095 $args = array_slice( func_get_args(), 1 );
1099 foreach ( $args as $root ) {
1100 if ( $root instanceof PPNode_Hash_Array
) {
1101 $root = $root->value
;
1103 if ( !is_array( $root ) ) {
1104 $root = array( $root );
1106 foreach ( $root as $node ) {
1112 $s .= $this->expand( $node );
1119 * Makes an object that, when expand()ed, will be the same as one obtained
1122 function virtualImplode( $sep /*, ... */ ) {
1123 $args = array_slice( func_get_args(), 1 );
1127 foreach ( $args as $root ) {
1128 if ( $root instanceof PPNode_Hash_Array
) {
1129 $root = $root->value
;
1131 if ( !is_array( $root ) ) {
1132 $root = array( $root );
1134 foreach ( $root as $node ) {
1143 return new PPNode_Hash_Array( $out );
1147 * Virtual implode with brackets
1149 function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1150 $args = array_slice( func_get_args(), 3 );
1151 $out = array( $start );
1154 foreach ( $args as $root ) {
1155 if ( $root instanceof PPNode_Hash_Array
) {
1156 $root = $root->value
;
1158 if ( !is_array( $root ) ) {
1159 $root = array( $root );
1161 foreach ( $root as $node ) {
1171 return new PPNode_Hash_Array( $out );
1174 function __toString() {
1178 function getPDBK( $level = false ) {
1179 if ( $level === false ) {
1180 return $this->title
->getPrefixedDBkey();
1182 return isset( $this->titleCache
[$level] ) ?
$this->titleCache
[$level] : false;
1186 function getArguments() {
1190 function getNumberedArguments() {
1194 function getNamedArguments() {
1199 * Returns true if there are no arguments in this frame
1201 function isEmpty() {
1205 function getArgument( $name ) {
1210 * Returns true if the infinite loop check is OK, false if a loop is detected
1212 * @param $title Title
1214 function loopCheck( $title ) {
1215 return !isset( $this->loopCheckHash
[$title->getPrefixedDBkey()] );
1219 * Return true if the frame is a template frame
1221 function isTemplate() {
1227 * Expansion frame with template arguments
1230 class PPTemplateFrame_Hash
extends PPFrame_Hash
{
1231 var $numberedArgs, $namedArgs, $parent;
1232 var $numberedExpansionCache, $namedExpansionCache;
1234 function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
1235 parent
::__construct( $preprocessor );
1237 $this->parent
= $parent;
1238 $this->numberedArgs
= $numberedArgs;
1239 $this->namedArgs
= $namedArgs;
1240 $this->title
= $title;
1241 $pdbk = $title ?
$title->getPrefixedDBkey() : false;
1242 $this->titleCache
= $parent->titleCache
;
1243 $this->titleCache
[] = $pdbk;
1244 $this->loopCheckHash
= /*clone*/ $parent->loopCheckHash
;
1245 if ( $pdbk !== false ) {
1246 $this->loopCheckHash
[$pdbk] = true;
1248 $this->depth
= $parent->depth +
1;
1249 $this->numberedExpansionCache
= $this->namedExpansionCache
= array();
1252 function __toString() {
1255 $args = $this->numberedArgs +
$this->namedArgs
;
1256 foreach ( $args as $name => $value ) {
1262 $s .= "\"$name\":\"" .
1263 str_replace( '"', '\\"', $value->__toString() ) . '"';
1269 * Returns true if there are no arguments in this frame
1271 function isEmpty() {
1272 return !count( $this->numberedArgs
) && !count( $this->namedArgs
);
1275 function getArguments() {
1276 $arguments = array();
1277 foreach ( array_merge(
1278 array_keys($this->numberedArgs
),
1279 array_keys($this->namedArgs
)) as $key ) {
1280 $arguments[$key] = $this->getArgument($key);
1285 function getNumberedArguments() {
1286 $arguments = array();
1287 foreach ( array_keys($this->numberedArgs
) as $key ) {
1288 $arguments[$key] = $this->getArgument($key);
1293 function getNamedArguments() {
1294 $arguments = array();
1295 foreach ( array_keys($this->namedArgs
) as $key ) {
1296 $arguments[$key] = $this->getArgument($key);
1301 function getNumberedArgument( $index ) {
1302 if ( !isset( $this->numberedArgs
[$index] ) ) {
1305 if ( !isset( $this->numberedExpansionCache
[$index] ) ) {
1306 # No trimming for unnamed arguments
1307 $this->numberedExpansionCache
[$index] = $this->parent
->expand( $this->numberedArgs
[$index], PPFrame
::STRIP_COMMENTS
);
1309 return $this->numberedExpansionCache
[$index];
1312 function getNamedArgument( $name ) {
1313 if ( !isset( $this->namedArgs
[$name] ) ) {
1316 if ( !isset( $this->namedExpansionCache
[$name] ) ) {
1317 # Trim named arguments post-expand, for backwards compatibility
1318 $this->namedExpansionCache
[$name] = trim(
1319 $this->parent
->expand( $this->namedArgs
[$name], PPFrame
::STRIP_COMMENTS
) );
1321 return $this->namedExpansionCache
[$name];
1324 function getArgument( $name ) {
1325 $text = $this->getNumberedArgument( $name );
1326 if ( $text === false ) {
1327 $text = $this->getNamedArgument( $name );
1333 * Return true if the frame is a template frame
1335 function isTemplate() {
1341 * Expansion frame with custom arguments
1344 class PPCustomFrame_Hash
extends PPFrame_Hash
{
1347 function __construct( $preprocessor, $args ) {
1348 parent
::__construct( $preprocessor );
1349 $this->args
= $args;
1352 function __toString() {
1355 foreach ( $this->args
as $name => $value ) {
1361 $s .= "\"$name\":\"" .
1362 str_replace( '"', '\\"', $value->__toString() ) . '"';
1368 function isEmpty() {
1369 return !count( $this->args
);
1372 function getArgument( $index ) {
1373 if ( !isset( $this->args
[$index] ) ) {
1376 return $this->args
[$index];
1383 class PPNode_Hash_Tree
implements PPNode
{
1384 var $name, $firstChild, $lastChild, $nextSibling;
1386 function __construct( $name ) {
1387 $this->name
= $name;
1388 $this->firstChild
= $this->lastChild
= $this->nextSibling
= false;
1391 function __toString() {
1394 for ( $node = $this->firstChild
; $node; $node = $node->nextSibling
) {
1395 if ( $node instanceof PPNode_Hash_Attr
) {
1396 $attribs .= ' ' . $node->name
. '="' . htmlspecialchars( $node->value
) . '"';
1398 $inner .= $node->__toString();
1401 if ( $inner === '' ) {
1402 return "<{$this->name}$attribs/>";
1404 return "<{$this->name}$attribs>$inner</{$this->name}>";
1408 static function newWithText( $name, $text ) {
1409 $obj = new self( $name );
1410 $obj->addChild( new PPNode_Hash_Text( $text ) );
1414 function addChild( $node ) {
1415 if ( $this->lastChild
=== false ) {
1416 $this->firstChild
= $this->lastChild
= $node;
1418 $this->lastChild
->nextSibling
= $node;
1419 $this->lastChild
= $node;
1423 function getChildren() {
1424 $children = array();
1425 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1426 $children[] = $child;
1428 return new PPNode_Hash_Array( $children );
1431 function getFirstChild() {
1432 return $this->firstChild
;
1435 function getNextSibling() {
1436 return $this->nextSibling
;
1439 function getChildrenOfType( $name ) {
1440 $children = array();
1441 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1442 if ( isset( $child->name
) && $child->name
=== $name ) {
1443 $children[] = $name;
1449 function getLength() { return false; }
1450 function item( $i ) { return false; }
1452 function getName() {
1457 * Split a <part> node into an associative array containing:
1459 * index String index
1460 * value PPNode value
1462 function splitArg() {
1464 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1465 if ( !isset( $child->name
) ) {
1468 if ( $child->name
=== 'name' ) {
1469 $bits['name'] = $child;
1470 if ( $child->firstChild
instanceof PPNode_Hash_Attr
1471 && $child->firstChild
->name
=== 'index' )
1473 $bits['index'] = $child->firstChild
->value
;
1475 } elseif ( $child->name
=== 'value' ) {
1476 $bits['value'] = $child;
1480 if ( !isset( $bits['name'] ) ) {
1481 throw new MWException( 'Invalid brace node passed to ' . __METHOD__
);
1483 if ( !isset( $bits['index'] ) ) {
1484 $bits['index'] = '';
1490 * Split an <ext> node into an associative array containing name, attr, inner and close
1491 * All values in the resulting array are PPNodes. Inner and close are optional.
1493 function splitExt() {
1495 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1496 if ( !isset( $child->name
) ) {
1499 if ( $child->name
== 'name' ) {
1500 $bits['name'] = $child;
1501 } elseif ( $child->name
== 'attr' ) {
1502 $bits['attr'] = $child;
1503 } elseif ( $child->name
== 'inner' ) {
1504 $bits['inner'] = $child;
1505 } elseif ( $child->name
== 'close' ) {
1506 $bits['close'] = $child;
1509 if ( !isset( $bits['name'] ) ) {
1510 throw new MWException( 'Invalid ext node passed to ' . __METHOD__
);
1518 function splitHeading() {
1519 if ( $this->name
!== 'h' ) {
1520 throw new MWException( 'Invalid h node passed to ' . __METHOD__
);
1523 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1524 if ( !isset( $child->name
) ) {
1527 if ( $child->name
== 'i' ) {
1528 $bits['i'] = $child->value
;
1529 } elseif ( $child->name
== 'level' ) {
1530 $bits['level'] = $child->value
;
1533 if ( !isset( $bits['i'] ) ) {
1534 throw new MWException( 'Invalid h node passed to ' . __METHOD__
);
1540 * Split a <template> or <tplarg> node
1542 function splitTemplate() {
1544 $bits = array( 'lineStart' => '' );
1545 for ( $child = $this->firstChild
; $child; $child = $child->nextSibling
) {
1546 if ( !isset( $child->name
) ) {
1549 if ( $child->name
== 'title' ) {
1550 $bits['title'] = $child;
1552 if ( $child->name
== 'part' ) {
1555 if ( $child->name
== 'lineStart' ) {
1556 $bits['lineStart'] = '1';
1559 if ( !isset( $bits['title'] ) ) {
1560 throw new MWException( 'Invalid node passed to ' . __METHOD__
);
1562 $bits['parts'] = new PPNode_Hash_Array( $parts );
1570 class PPNode_Hash_Text
implements PPNode
{
1571 var $value, $nextSibling;
1573 function __construct( $value ) {
1574 if ( is_object( $value ) ) {
1575 throw new MWException( __CLASS__
. ' given object instead of string' );
1577 $this->value
= $value;
1580 function __toString() {
1581 return htmlspecialchars( $this->value
);
1584 function getNextSibling() {
1585 return $this->nextSibling
;
1588 function getChildren() { return false; }
1589 function getFirstChild() { return false; }
1590 function getChildrenOfType( $name ) { return false; }
1591 function getLength() { return false; }
1592 function item( $i ) { return false; }
1593 function getName() { return '#text'; }
1594 function splitArg() { throw new MWException( __METHOD__
. ': not supported' ); }
1595 function splitExt() { throw new MWException( __METHOD__
. ': not supported' ); }
1596 function splitHeading() { throw new MWException( __METHOD__
. ': not supported' ); }
1602 class PPNode_Hash_Array
implements PPNode
{
1603 var $value, $nextSibling;
1605 function __construct( $value ) {
1606 $this->value
= $value;
1609 function __toString() {
1610 return var_export( $this, true );
1613 function getLength() {
1614 return count( $this->value
);
1617 function item( $i ) {
1618 return $this->value
[$i];
1621 function getName() { return '#nodelist'; }
1623 function getNextSibling() {
1624 return $this->nextSibling
;
1627 function getChildren() { return false; }
1628 function getFirstChild() { return false; }
1629 function getChildrenOfType( $name ) { return false; }
1630 function splitArg() { throw new MWException( __METHOD__
. ': not supported' ); }
1631 function splitExt() { throw new MWException( __METHOD__
. ': not supported' ); }
1632 function splitHeading() { throw new MWException( __METHOD__
. ': not supported' ); }
1638 class PPNode_Hash_Attr
implements PPNode
{
1639 var $name, $value, $nextSibling;
1641 function __construct( $name, $value ) {
1642 $this->name
= $name;
1643 $this->value
= $value;
1646 function __toString() {
1647 return "<@{$this->name}>" . htmlspecialchars( $this->value
) . "</@{$this->name}>";
1650 function getName() {
1654 function getNextSibling() {
1655 return $this->nextSibling
;
1658 function getChildren() { return false; }
1659 function getFirstChild() { return false; }
1660 function getChildrenOfType( $name ) { return false; }
1661 function getLength() { return false; }
1662 function item( $i ) { return false; }
1663 function splitArg() { throw new MWException( __METHOD__
. ': not supported' ); }
1664 function splitExt() { throw new MWException( __METHOD__
. ': not supported' ); }
1665 function splitHeading() { throw new MWException( __METHOD__
. ': not supported' ); }