3 * Preprocessor using PHP's dom extension
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
27 // @codingStandardsIgnoreStart Squiz.Classes.ValidClassName.NotCamelCaps
28 class Preprocessor_DOM
extends Preprocessor
{
29 // @codingStandardsIgnoreEnd
38 const CACHE_PREFIX
= 'preprocess-xml';
40 public function __construct( $parser ) {
41 $this->parser
= $parser;
42 $mem = ini_get( 'memory_limit' );
43 $this->memoryLimit
= false;
44 if ( strval( $mem ) !== '' && $mem != -1 ) {
45 if ( preg_match( '/^\d+$/', $mem ) ) {
46 $this->memoryLimit
= $mem;
47 } elseif ( preg_match( '/^(\d+)M$/i', $mem, $m ) ) {
48 $this->memoryLimit
= $m[1] * 1048576;
56 public function newFrame() {
57 return new PPFrame_DOM( $this );
62 * @return PPCustomFrame_DOM
64 public function newCustomFrame( $args ) {
65 return new PPCustomFrame_DOM( $this, $args );
69 * @param array $values
73 public function newPartNodeArray( $values ) {
74 // NOTE: DOM manipulation is slower than building & parsing XML! (or so Tim sais)
77 foreach ( $values as $k => $val ) {
79 $xml .= "<part><name index=\"$k\"/><value>"
80 . htmlspecialchars( $val ) . "</value></part>";
82 $xml .= "<part><name>" . htmlspecialchars( $k )
83 . "</name>=<value>" . htmlspecialchars( $val ) . "</value></part>";
89 $dom = new DOMDocument();
90 MediaWiki\
suppressWarnings();
91 $result = $dom->loadXML( $xml );
92 MediaWiki\restoreWarnings
();
94 // Try running the XML through UtfNormal to get rid of invalid characters
95 $xml = UtfNormal\Validator
::cleanUp( $xml );
96 // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
97 // don't barf when the XML is >256 levels deep
98 $result = $dom->loadXML( $xml, 1 << 19 );
102 throw new MWException( 'Parameters passed to ' . __METHOD__
. ' result in invalid XML' );
105 $root = $dom->documentElement
;
106 $node = new PPNode_DOM( $root->childNodes
);
111 * @throws MWException
114 public function memCheck() {
115 if ( $this->memoryLimit
=== false ) {
118 $usage = memory_get_usage();
119 if ( $usage > $this->memoryLimit
* 0.9 ) {
120 $limit = intval( $this->memoryLimit
* 0.9 / 1048576 +
0.5 );
121 throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
123 return $usage <= $this->memoryLimit
* 0.8;
127 * Preprocess some wikitext and return the document tree.
128 * This is the ghost of Parser::replace_variables().
130 * @param string $text The text to parse
131 * @param int $flags Bitwise combination of:
132 * Parser::PTD_FOR_INCLUSION Handle "<noinclude>" and "<includeonly>"
133 * as if the text is being included. Default
134 * is to assume a direct page view.
136 * The generated DOM tree must depend only on the input text and the flags.
137 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
139 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
140 * change in the DOM tree for a given text, must be passed through the section identifier
141 * in the section edit link and thus back to extractSections().
143 * The output of this function is currently only cached in process memory, but a persistent
144 * cache may be implemented at a later date which takes further advantage of these strict
145 * dependency requirements.
147 * @throws MWException
150 public function preprocessToObj( $text, $flags = 0 ) {
152 $xml = $this->cacheGetTree( $text, $flags );
153 if ( $xml === false ) {
154 $xml = $this->preprocessToXml( $text, $flags );
155 $this->cacheSetTree( $text, $flags, $xml );
158 // Fail if the number of elements exceeds acceptable limits
159 // Do not attempt to generate the DOM
160 $this->parser
->mGeneratedPPNodeCount +
= substr_count( $xml, '<' );
161 $max = $this->parser
->mOptions
->getMaxGeneratedPPNodeCount();
162 if ( $this->parser
->mGeneratedPPNodeCount
> $max ) {
163 // if ( $cacheable ) { ... }
164 throw new MWException( __METHOD__
. ': generated node count limit exceeded' );
167 $dom = new DOMDocument
;
168 MediaWiki\
suppressWarnings();
169 $result = $dom->loadXML( $xml );
170 MediaWiki\restoreWarnings
();
172 // Try running the XML through UtfNormal to get rid of invalid characters
173 $xml = UtfNormal\Validator
::cleanUp( $xml );
174 // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
175 // don't barf when the XML is >256 levels deep.
176 $result = $dom->loadXML( $xml, 1 << 19 );
179 $obj = new PPNode_DOM( $dom->documentElement
);
182 // if ( $cacheable ) { ... }
185 throw new MWException( __METHOD__
. ' generated invalid XML' );
191 * @param string $text
195 public function preprocessToXml( $text, $flags = 0 ) {
196 global $wgDisableLangConversion;
198 $forInclusion = $flags & Parser
::PTD_FOR_INCLUSION
;
200 $xmlishElements = $this->parser
->getStripList();
201 $xmlishAllowMissingEndTag = [ 'includeonly', 'noinclude', 'onlyinclude' ];
202 $enableOnlyinclude = false;
203 if ( $forInclusion ) {
204 $ignoredTags = [ 'includeonly', '/includeonly' ];
205 $ignoredElements = [ 'noinclude' ];
206 $xmlishElements[] = 'noinclude';
207 if ( strpos( $text, '<onlyinclude>' ) !== false
208 && strpos( $text, '</onlyinclude>' ) !== false
210 $enableOnlyinclude = true;
213 $ignoredTags = [ 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' ];
214 $ignoredElements = [ 'includeonly' ];
215 $xmlishElements[] = 'includeonly';
217 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
219 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
220 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
222 $stack = new PPDStack
;
224 $searchBase = "[{<\n"; # }
225 if ( !$wgDisableLangConversion ) {
226 // FIXME: disabled due to T153761
227 // $searchBase .= '-';
230 // For fast reverse searches
231 $revText = strrev( $text );
232 $lengthText = strlen( $text );
234 // Input pointer, starts out pointing to a pseudo-newline before the start
236 // Current accumulator
237 $accum =& $stack->getAccum();
239 // True to find equals signs in arguments
241 // True to take notice of pipe characters
244 // True if $i is inside a possible heading
246 // True if there are no more greater-than (>) signs right of $i
248 // Map of tag name => true if there are no more closing tags of given type right of $i
249 $noMoreClosingTag = [];
250 // True to ignore all input up to the next <onlyinclude>
251 $findOnlyinclude = $enableOnlyinclude;
252 // Do a line-start run without outputting an LF character
253 $fakeLineStart = true;
256 // $this->memCheck();
258 if ( $findOnlyinclude ) {
259 // Ignore all input up to the next <onlyinclude>
260 $startPos = strpos( $text, '<onlyinclude>', $i );
261 if ( $startPos === false ) {
262 // Ignored section runs to the end
263 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
266 $tagEndPos = $startPos +
strlen( '<onlyinclude>' ); // past-the-end
267 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
269 $findOnlyinclude = false;
272 if ( $fakeLineStart ) {
273 $found = 'line-start';
276 # Find next opening brace, closing brace or pipe
277 $search = $searchBase;
278 if ( $stack->top
=== false ) {
279 $currentClosing = '';
281 $currentClosing = $stack->top
->close
;
282 $search .= $currentClosing;
288 // First equals will be for the template
292 # Output literal section, advance input counter
293 $literalLength = strcspn( $text, $search, $i );
294 if ( $literalLength > 0 ) {
295 $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
296 $i +
= $literalLength;
298 if ( $i >= $lengthText ) {
299 if ( $currentClosing == "\n" ) {
300 // Do a past-the-end run to finish off the heading
308 $curChar = $curTwoChar = $text[$i];
309 if ( ( $i +
1 ) < $lengthText ) {
310 $curTwoChar .= $text[$i +
1];
312 if ( $curChar == '|' ) {
314 } elseif ( $curChar == '=' ) {
316 } elseif ( $curChar == '<' ) {
318 } elseif ( $curChar == "\n" ) {
322 $found = 'line-start';
324 } elseif ( $curTwoChar == $currentClosing ) {
326 $curChar = $curTwoChar;
327 } elseif ( $curChar == $currentClosing ) {
329 } elseif ( isset( $this->rules
[$curTwoChar] ) ) {
330 $curChar = $curTwoChar;
332 $rule = $this->rules
[$curChar];
333 } elseif ( isset( $this->rules
[$curChar] ) ) {
335 $rule = $this->rules
[$curChar];
336 } elseif ( $curChar == '-' ) {
339 # Some versions of PHP have a strcspn which stops on null characters
340 # Ignore and continue
347 if ( $found == 'angle' ) {
349 // Handle </onlyinclude>
350 if ( $enableOnlyinclude
351 && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
353 $findOnlyinclude = true;
357 // Determine element name
358 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i +
1 ) ) {
359 // Element name missing or not listed
365 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
367 // To avoid leaving blank lines, when a sequence of
368 // space-separated comments is both preceded and followed by
369 // a newline (ignoring spaces), then
370 // trim leading and trailing spaces and the trailing newline.
373 $endPos = strpos( $text, '-->', $i +
4 );
374 if ( $endPos === false ) {
375 // Unclosed comment in input, runs to end
376 $inner = substr( $text, $i );
377 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
380 // Search backwards for leading whitespace
381 $wsStart = $i ?
( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
383 // Search forwards for trailing whitespace
384 // $wsEnd will be the position of the last space (or the '>' if there's none)
385 $wsEnd = $endPos +
2 +
strspn( $text, " \t", $endPos +
3 );
387 // Keep looking forward as long as we're finding more
389 $comments = [ [ $wsStart, $wsEnd ] ];
390 while ( substr( $text, $wsEnd +
1, 4 ) == '<!--' ) {
391 $c = strpos( $text, '-->', $wsEnd +
4 );
392 if ( $c === false ) {
395 $c = $c +
2 +
strspn( $text, " \t", $c +
3 );
396 $comments[] = [ $wsEnd +
1, $c ];
400 // Eat the line if possible
401 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
402 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
403 // it's a possible beneficial b/c break.
404 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
405 && substr( $text, $wsEnd +
1, 1 ) == "\n"
407 // Remove leading whitespace from the end of the accumulator
408 // Sanity check first though
409 $wsLength = $i - $wsStart;
411 && strspn( $accum, " \t", -$wsLength ) === $wsLength
413 $accum = substr( $accum, 0, -$wsLength );
416 // Dump all but the last comment to the accumulator
417 foreach ( $comments as $j => $com ) {
419 $endPos = $com[1] +
1;
420 if ( $j == ( count( $comments ) - 1 ) ) {
423 $inner = substr( $text, $startPos, $endPos - $startPos );
424 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
427 // Do a line-start run next time to look for headings after the comment
428 $fakeLineStart = true;
430 // No line to eat, just take the comment itself
436 $part = $stack->top
->getCurrentPart();
437 if ( !( isset( $part->commentEnd
) && $part->commentEnd
== $wsStart - 1 ) ) {
438 $part->visualEnd
= $wsStart;
440 // Else comments abutting, no change in visual end
441 $part->commentEnd
= $endPos;
444 $inner = substr( $text, $startPos, $endPos - $startPos +
1 );
445 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
450 $lowerName = strtolower( $name );
451 $attrStart = $i +
strlen( $name ) +
1;
454 $tagEndPos = $noMoreGT ?
false : strpos( $text, '>', $attrStart );
455 if ( $tagEndPos === false ) {
456 // Infinite backtrack
457 // Disable tag search to prevent worst-case O(N^2) performance
464 // Handle ignored tags
465 if ( in_array( $lowerName, $ignoredTags ) ) {
467 . htmlspecialchars( substr( $text, $i, $tagEndPos - $i +
1 ) )
474 if ( $text[$tagEndPos - 1] == '/' ) {
475 $attrEnd = $tagEndPos - 1;
480 $attrEnd = $tagEndPos;
483 !isset( $noMoreClosingTag[$name] ) &&
484 preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
485 $text, $matches, PREG_OFFSET_CAPTURE
, $tagEndPos +
1 )
487 $inner = substr( $text, $tagEndPos +
1, $matches[0][1] - $tagEndPos - 1 );
488 $i = $matches[0][1] +
strlen( $matches[0][0] );
489 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
492 if ( in_array( $name, $xmlishAllowMissingEndTag ) ) {
493 // Let it run out to the end of the text.
494 $inner = substr( $text, $tagEndPos +
1 );
498 // Don't match the tag, treat opening tag as literal and resume parsing.
500 $accum .= htmlspecialchars( substr( $text, $tagStartPos, $tagEndPos +
1 - $tagStartPos ) );
501 // Cache results, otherwise we have O(N^2) performance for input like <foo><foo><foo>...
502 $noMoreClosingTag[$name] = true;
507 // <includeonly> and <noinclude> just become <ignore> tags
508 if ( in_array( $lowerName, $ignoredElements ) ) {
509 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
515 if ( $attrEnd <= $attrStart ) {
518 $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
520 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
521 // Note that the attr element contains the whitespace between name and attribute,
522 // this is necessary for precise reconstruction during pre-save transform.
523 '<attr>' . htmlspecialchars( $attr ) . '</attr>';
524 if ( $inner !== null ) {
525 $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
527 $accum .= $close . '</ext>';
528 } elseif ( $found == 'line-start' ) {
529 // Is this the start of a heading?
530 // Line break belongs before the heading element in any case
531 if ( $fakeLineStart ) {
532 $fakeLineStart = false;
538 $count = strspn( $text, '=', $i, 6 );
539 if ( $count == 1 && $findEquals ) {
540 // DWIM: This looks kind of like a name/value separator.
541 // Let's let the equals handler have it and break the
542 // potential heading. This is heuristic, but AFAICT the
543 // methods for completely correct disambiguation are very
545 } elseif ( $count > 0 ) {
549 'parts' => [ new PPDPart( str_repeat( '=', $count ) ) ],
552 $stack->push( $piece );
553 $accum =& $stack->getAccum();
554 $flags = $stack->getFlags();
558 } elseif ( $found == 'line-end' ) {
559 $piece = $stack->top
;
560 // A heading must be open, otherwise \n wouldn't have been in the search list
561 assert( $piece->open
=== "\n" );
562 $part = $piece->getCurrentPart();
563 // Search back through the input to see if it has a proper close.
564 // Do this using the reversed string since the other solutions
565 // (end anchor, etc.) are inefficient.
566 $wsLength = strspn( $revText, " \t", $lengthText - $i );
567 $searchStart = $i - $wsLength;
568 if ( isset( $part->commentEnd
) && $searchStart - 1 == $part->commentEnd
) {
569 // Comment found at line end
570 // Search for equals signs before the comment
571 $searchStart = $part->visualEnd
;
572 $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
574 $count = $piece->count
;
575 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
576 if ( $equalsLength > 0 ) {
577 if ( $searchStart - $equalsLength == $piece->startPos
) {
578 // This is just a single string of equals signs on its own line
579 // Replicate the doHeadings behavior /={count}(.+)={count}/
580 // First find out how many equals signs there really are (don't stop at 6)
581 $count = $equalsLength;
585 $count = min( 6, intval( ( $count - 1 ) / 2 ) );
588 $count = min( $equalsLength, $count );
591 // Normal match, output <h>
592 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
595 // Single equals sign on its own line, count=0
599 // No match, no <h>, just pass down the inner text
604 $accum =& $stack->getAccum();
605 $flags = $stack->getFlags();
608 // Append the result to the enclosing accumulator
610 // Note that we do NOT increment the input pointer.
611 // This is because the closing linebreak could be the opening linebreak of
612 // another heading. Infinite loops are avoided because the next iteration MUST
613 // hit the heading open case above, which unconditionally increments the
615 } elseif ( $found == 'open' ) {
616 # count opening brace characters
617 $curLen = strlen( $curChar );
618 $count = ( $curLen > 1 ) ?
1 : strspn( $text, $curChar, $i );
620 # we need to add to stack only if opening brace count is enough for one of the rules
621 if ( $count >= $rule['min'] ) {
622 # Add it to the stack
625 'close' => $rule['end'],
627 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
630 $stack->push( $piece );
631 $accum =& $stack->getAccum();
632 $flags = $stack->getFlags();
635 # Add literal brace(s)
636 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
638 $i +
= $curLen * $count;
639 } elseif ( $found == 'close' ) {
640 $piece = $stack->top
;
641 # lets check if there are enough characters for closing brace
642 $maxCount = $piece->count
;
643 $curLen = strlen( $curChar );
644 $count = ( $curLen > 1 ) ?
1 : strspn( $text, $curChar, $i, $maxCount );
646 # check for maximum matching characters (if there are 5 closing
647 # characters, we will probably need only 3 - depending on the rules)
648 $rule = $this->rules
[$piece->open
];
649 if ( $count > $rule['max'] ) {
650 # The specified maximum exists in the callback array, unless the caller
652 $matchingCount = $rule['max'];
654 # Count is less than the maximum
655 # Skip any gaps in the callback array to find the true largest match
656 # Need to use array_key_exists not isset because the callback can be null
657 $matchingCount = $count;
658 while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
663 if ( $matchingCount <= 0 ) {
664 # No matching element found in callback array
665 # Output a literal closing brace and continue
666 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
667 $i +
= $curLen * $count;
670 $name = $rule['names'][$matchingCount];
671 if ( $name === null ) {
672 // No element, just literal text
673 $element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
676 # Note: $parts is already XML, does not need to be encoded further
677 $parts = $piece->parts
;
678 $title = $parts[0]->out
;
681 # The invocation is at the start of the line if lineStart is set in
682 # the stack, and all opening brackets are used up.
683 if ( $maxCount == $matchingCount && !empty( $piece->lineStart
) ) {
684 $attr = ' lineStart="1"';
689 $element = "<$name$attr>";
690 $element .= "<title>$title</title>";
692 foreach ( $parts as $part ) {
693 if ( isset( $part->eqpos
) ) {
694 $argName = substr( $part->out
, 0, $part->eqpos
);
695 $argValue = substr( $part->out
, $part->eqpos +
1 );
696 $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
698 $element .= "<part><name index=\"$argIndex\" /><value>{$part->out}</value></part>";
702 $element .= "</$name>";
705 # Advance input pointer
706 $i +
= $curLen * $matchingCount;
710 $accum =& $stack->getAccum();
712 # Re-add the old stack element if it still has unmatched opening characters remaining
713 if ( $matchingCount < $piece->count
) {
714 $piece->parts
= [ new PPDPart
];
715 $piece->count
-= $matchingCount;
716 # do we still qualify for any callback with remaining count?
717 $min = $this->rules
[$piece->open
]['min'];
718 if ( $piece->count
>= $min ) {
719 $stack->push( $piece );
720 $accum =& $stack->getAccum();
722 $accum .= str_repeat( $piece->open
, $piece->count
);
725 $flags = $stack->getFlags();
728 # Add XML element to the enclosing accumulator
730 } elseif ( $found == 'pipe' ) {
731 $findEquals = true; // shortcut for getFlags()
733 $accum =& $stack->getAccum();
735 } elseif ( $found == 'equals' ) {
736 $findEquals = false; // shortcut for getFlags()
737 $stack->getCurrentPart()->eqpos
= strlen( $accum );
740 } elseif ( $found == 'dash' ) {
746 # Output any remaining unclosed brackets
747 foreach ( $stack->stack
as $piece ) {
748 $stack->rootAccum
.= $piece->breakSyntax();
750 $stack->rootAccum
.= '</root>';
751 $xml = $stack->rootAccum
;
758 * Stack class to help Preprocessor::preprocessToObj()
762 public $stack, $rootAccum;
769 public $elementClass = 'PPDStackElement';
771 public static $false = false;
773 public function __construct() {
776 $this->rootAccum
= '';
777 $this->accum
=& $this->rootAccum
;
783 public function count() {
784 return count( $this->stack
);
787 public function &getAccum() {
791 public function getCurrentPart() {
792 if ( $this->top
=== false ) {
795 return $this->top
->getCurrentPart();
799 public function push( $data ) {
800 if ( $data instanceof $this->elementClass
) {
801 $this->stack
[] = $data;
803 $class = $this->elementClass
;
804 $this->stack
[] = new $class( $data );
806 $this->top
= $this->stack
[count( $this->stack
) - 1];
807 $this->accum
=& $this->top
->getAccum();
810 public function pop() {
811 if ( !count( $this->stack
) ) {
812 throw new MWException( __METHOD__
. ': no elements remaining' );
814 $temp = array_pop( $this->stack
);
816 if ( count( $this->stack
) ) {
817 $this->top
= $this->stack
[count( $this->stack
) - 1];
818 $this->accum
=& $this->top
->getAccum();
820 $this->top
= self
::$false;
821 $this->accum
=& $this->rootAccum
;
826 public function addPart( $s = '' ) {
827 $this->top
->addPart( $s );
828 $this->accum
=& $this->top
->getAccum();
834 public function getFlags() {
835 if ( !count( $this->stack
) ) {
837 'findEquals' => false,
839 'inHeading' => false,
842 return $this->top
->getFlags();
850 class PPDStackElement
{
852 * @var string Opening character (\n for heading)
857 * @var string Matching closing character
862 * @var int Number of opening characters found (number of "=" for heading)
867 * @var PPDPart[] Array of PPDPart objects describing pipe-separated parts.
872 * @var bool True if the open char appeared at the start of the input line.
873 * Not set for headings.
877 public $partClass = 'PPDPart';
879 public function __construct( $data = [] ) {
880 $class = $this->partClass
;
881 $this->parts
= [ new $class ];
883 foreach ( $data as $name => $value ) {
884 $this->$name = $value;
888 public function &getAccum() {
889 return $this->parts
[count( $this->parts
) - 1]->out
;
892 public function addPart( $s = '' ) {
893 $class = $this->partClass
;
894 $this->parts
[] = new $class( $s );
897 public function getCurrentPart() {
898 return $this->parts
[count( $this->parts
) - 1];
904 public function getFlags() {
905 $partCount = count( $this->parts
);
906 $findPipe = $this->open
!= "\n" && $this->open
!= '[';
908 'findPipe' => $findPipe,
909 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts
[$partCount - 1]->eqpos
),
910 'inHeading' => $this->open
== "\n",
915 * Get the output string that would result if the close is not found.
917 * @param bool|int $openingCount
920 public function breakSyntax( $openingCount = false ) {
921 if ( $this->open
== "\n" ) {
922 $s = $this->parts
[0]->out
;
924 if ( $openingCount === false ) {
925 $openingCount = $this->count
;
927 $s = str_repeat( $this->open
, $openingCount );
929 foreach ( $this->parts
as $part ) {
947 * @var string Output accumulator string
951 // Optional member variables:
952 // eqpos Position of equals sign in output accumulator
953 // commentEnd Past-the-end input pointer for the last comment encountered
954 // visualEnd Past-the-end input pointer for the end of the accumulator minus comments
956 public function __construct( $out = '' ) {
962 * An expansion frame, used as a context to expand the result of preprocessToObj()
965 // @codingStandardsIgnoreStart Squiz.Classes.ValidClassName.NotCamelCaps
966 class PPFrame_DOM
implements PPFrame
{
967 // @codingStandardsIgnoreEnd
972 public $preprocessor;
986 * Hashtable listing templates which are disallowed for expansion in this frame,
987 * having been encountered previously in parent frames.
989 public $loopCheckHash;
992 * Recursion depth of this frame, top = 0
993 * Note that this is NOT the same as expansion depth in expand()
997 private $volatile = false;
1003 protected $childExpansionCache;
1006 * Construct a new preprocessor frame.
1007 * @param Preprocessor $preprocessor The parent preprocessor
1009 public function __construct( $preprocessor ) {
1010 $this->preprocessor
= $preprocessor;
1011 $this->parser
= $preprocessor->parser
;
1012 $this->title
= $this->parser
->mTitle
;
1013 $this->titleCache
= [ $this->title ?
$this->title
->getPrefixedDBkey() : false ];
1014 $this->loopCheckHash
= [];
1016 $this->childExpansionCache
= [];
1020 * Create a new child frame
1021 * $args is optionally a multi-root PPNode or array containing the template arguments
1023 * @param bool|array $args
1024 * @param Title|bool $title
1025 * @param int $indexOffset
1026 * @return PPTemplateFrame_DOM
1028 public function newChild( $args = false, $title = false, $indexOffset = 0 ) {
1031 if ( $title === false ) {
1032 $title = $this->title
;
1034 if ( $args !== false ) {
1036 if ( $args instanceof PPNode
) {
1037 $args = $args->node
;
1039 foreach ( $args as $arg ) {
1040 if ( $arg instanceof PPNode
) {
1043 if ( !$xpath ||
$xpath->document
!== $arg->ownerDocument
) {
1044 $xpath = new DOMXPath( $arg->ownerDocument
);
1047 $nameNodes = $xpath->query( 'name', $arg );
1048 $value = $xpath->query( 'value', $arg );
1049 if ( $nameNodes->item( 0 )->hasAttributes() ) {
1050 // Numbered parameter
1051 $index = $nameNodes->item( 0 )->attributes
->getNamedItem( 'index' )->textContent
;
1052 $index = $index - $indexOffset;
1053 if ( isset( $namedArgs[$index] ) ||
isset( $numberedArgs[$index] ) ) {
1054 $this->parser
->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
1055 wfEscapeWikiText( $this->title
),
1056 wfEscapeWikiText( $title ),
1057 wfEscapeWikiText( $index ) )->text() );
1058 $this->parser
->addTrackingCategory( 'duplicate-args-category' );
1060 $numberedArgs[$index] = $value->item( 0 );
1061 unset( $namedArgs[$index] );
1064 $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame
::STRIP_COMMENTS
) );
1065 if ( isset( $namedArgs[$name] ) ||
isset( $numberedArgs[$name] ) ) {
1066 $this->parser
->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
1067 wfEscapeWikiText( $this->title
),
1068 wfEscapeWikiText( $title ),
1069 wfEscapeWikiText( $name ) )->text() );
1070 $this->parser
->addTrackingCategory( 'duplicate-args-category' );
1072 $namedArgs[$name] = $value->item( 0 );
1073 unset( $numberedArgs[$name] );
1077 return new PPTemplateFrame_DOM( $this->preprocessor
, $this, $numberedArgs, $namedArgs, $title );
1081 * @throws MWException
1082 * @param string|int $key
1083 * @param string|PPNode_DOM|DOMDocument $root
1087 public function cachedExpand( $key, $root, $flags = 0 ) {
1088 // we don't have a parent, so we don't have a cache
1089 return $this->expand( $root, $flags );
1093 * @throws MWException
1094 * @param string|PPNode_DOM|DOMDocument $root
1098 public function expand( $root, $flags = 0 ) {
1099 static $expansionDepth = 0;
1100 if ( is_string( $root ) ) {
1104 if ( ++
$this->parser
->mPPNodeCount
> $this->parser
->mOptions
->getMaxPPNodeCount() ) {
1105 $this->parser
->limitationWarn( 'node-count-exceeded',
1106 $this->parser
->mPPNodeCount
,
1107 $this->parser
->mOptions
->getMaxPPNodeCount()
1109 return '<span class="error">Node-count limit exceeded</span>';
1112 if ( $expansionDepth > $this->parser
->mOptions
->getMaxPPExpandDepth() ) {
1113 $this->parser
->limitationWarn( 'expansion-depth-exceeded',
1115 $this->parser
->mOptions
->getMaxPPExpandDepth()
1117 return '<span class="error">Expansion depth limit exceeded</span>';
1120 if ( $expansionDepth > $this->parser
->mHighestExpansionDepth
) {
1121 $this->parser
->mHighestExpansionDepth
= $expansionDepth;
1124 if ( $root instanceof PPNode_DOM
) {
1125 $root = $root->node
;
1127 if ( $root instanceof DOMDocument
) {
1128 $root = $root->documentElement
;
1131 $outStack = [ '', '' ];
1132 $iteratorStack = [ false, $root ];
1133 $indexStack = [ 0, 0 ];
1135 while ( count( $iteratorStack ) > 1 ) {
1136 $level = count( $outStack ) - 1;
1137 $iteratorNode =& $iteratorStack[$level];
1138 $out =& $outStack[$level];
1139 $index =& $indexStack[$level];
1141 if ( $iteratorNode instanceof PPNode_DOM
) {
1142 $iteratorNode = $iteratorNode->node
;
1145 if ( is_array( $iteratorNode ) ) {
1146 if ( $index >= count( $iteratorNode ) ) {
1147 // All done with this iterator
1148 $iteratorStack[$level] = false;
1149 $contextNode = false;
1151 $contextNode = $iteratorNode[$index];
1154 } elseif ( $iteratorNode instanceof DOMNodeList
) {
1155 if ( $index >= $iteratorNode->length
) {
1156 // All done with this iterator
1157 $iteratorStack[$level] = false;
1158 $contextNode = false;
1160 $contextNode = $iteratorNode->item( $index );
1164 // Copy to $contextNode and then delete from iterator stack,
1165 // because this is not an iterator but we do have to execute it once
1166 $contextNode = $iteratorStack[$level];
1167 $iteratorStack[$level] = false;
1170 if ( $contextNode instanceof PPNode_DOM
) {
1171 $contextNode = $contextNode->node
;
1174 $newIterator = false;
1176 if ( $contextNode === false ) {
1178 } elseif ( is_string( $contextNode ) ) {
1179 $out .= $contextNode;
1180 } elseif ( is_array( $contextNode ) ||
$contextNode instanceof DOMNodeList
) {
1181 $newIterator = $contextNode;
1182 } elseif ( $contextNode instanceof DOMNode
) {
1183 if ( $contextNode->nodeType
== XML_TEXT_NODE
) {
1184 $out .= $contextNode->nodeValue
;
1185 } elseif ( $contextNode->nodeName
== 'template' ) {
1186 # Double-brace expansion
1187 $xpath = new DOMXPath( $contextNode->ownerDocument
);
1188 $titles = $xpath->query( 'title', $contextNode );
1189 $title = $titles->item( 0 );
1190 $parts = $xpath->query( 'part', $contextNode );
1191 if ( $flags & PPFrame
::NO_TEMPLATES
) {
1192 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
1194 $lineStart = $contextNode->getAttribute( 'lineStart' );
1196 'title' => new PPNode_DOM( $title ),
1197 'parts' => new PPNode_DOM( $parts ),
1198 'lineStart' => $lineStart ];
1199 $ret = $this->parser
->braceSubstitution( $params, $this );
1200 if ( isset( $ret['object'] ) ) {
1201 $newIterator = $ret['object'];
1203 $out .= $ret['text'];
1206 } elseif ( $contextNode->nodeName
== 'tplarg' ) {
1207 # Triple-brace expansion
1208 $xpath = new DOMXPath( $contextNode->ownerDocument
);
1209 $titles = $xpath->query( 'title', $contextNode );
1210 $title = $titles->item( 0 );
1211 $parts = $xpath->query( 'part', $contextNode );
1212 if ( $flags & PPFrame
::NO_ARGS
) {
1213 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
1216 'title' => new PPNode_DOM( $title ),
1217 'parts' => new PPNode_DOM( $parts ) ];
1218 $ret = $this->parser
->argSubstitution( $params, $this );
1219 if ( isset( $ret['object'] ) ) {
1220 $newIterator = $ret['object'];
1222 $out .= $ret['text'];
1225 } elseif ( $contextNode->nodeName
== 'comment' ) {
1226 # HTML-style comment
1227 # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1228 # Not in RECOVER_COMMENTS mode (msgnw) though.
1229 if ( ( $this->parser
->ot
['html']
1230 ||
( $this->parser
->ot
['pre'] && $this->parser
->mOptions
->getRemoveComments() )
1231 ||
( $flags & PPFrame
::STRIP_COMMENTS
)
1232 ) && !( $flags & PPFrame
::RECOVER_COMMENTS
)
1235 } elseif ( $this->parser
->ot
['wiki'] && !( $flags & PPFrame
::RECOVER_COMMENTS
) ) {
1236 # Add a strip marker in PST mode so that pstPass2() can
1237 # run some old-fashioned regexes on the result.
1238 # Not in RECOVER_COMMENTS mode (extractSections) though.
1239 $out .= $this->parser
->insertStripItem( $contextNode->textContent
);
1241 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1242 $out .= $contextNode->textContent
;
1244 } elseif ( $contextNode->nodeName
== 'ignore' ) {
1245 # Output suppression used by <includeonly> etc.
1246 # OT_WIKI will only respect <ignore> in substed templates.
1247 # The other output types respect it unless NO_IGNORE is set.
1248 # extractSections() sets NO_IGNORE and so never respects it.
1249 if ( ( !isset( $this->parent
) && $this->parser
->ot
['wiki'] )
1250 ||
( $flags & PPFrame
::NO_IGNORE
)
1252 $out .= $contextNode->textContent
;
1256 } elseif ( $contextNode->nodeName
== 'ext' ) {
1258 $xpath = new DOMXPath( $contextNode->ownerDocument
);
1259 $names = $xpath->query( 'name', $contextNode );
1260 $attrs = $xpath->query( 'attr', $contextNode );
1261 $inners = $xpath->query( 'inner', $contextNode );
1262 $closes = $xpath->query( 'close', $contextNode );
1263 if ( $flags & PPFrame
::NO_TAGS
) {
1264 $s = '<' . $this->expand( $names->item( 0 ), $flags );
1265 if ( $attrs->length
> 0 ) {
1266 $s .= $this->expand( $attrs->item( 0 ), $flags );
1268 if ( $inners->length
> 0 ) {
1269 $s .= '>' . $this->expand( $inners->item( 0 ), $flags );
1270 if ( $closes->length
> 0 ) {
1271 $s .= $this->expand( $closes->item( 0 ), $flags );
1279 'name' => new PPNode_DOM( $names->item( 0 ) ),
1280 'attr' => $attrs->length
> 0 ?
new PPNode_DOM( $attrs->item( 0 ) ) : null,
1281 'inner' => $inners->length
> 0 ?
new PPNode_DOM( $inners->item( 0 ) ) : null,
1282 'close' => $closes->length
> 0 ?
new PPNode_DOM( $closes->item( 0 ) ) : null,
1284 $out .= $this->parser
->extensionSubstitution( $params, $this );
1286 } elseif ( $contextNode->nodeName
== 'h' ) {
1288 $s = $this->expand( $contextNode->childNodes
, $flags );
1290 # Insert a heading marker only for <h> children of <root>
1291 # This is to stop extractSections from going over multiple tree levels
1292 if ( $contextNode->parentNode
->nodeName
== 'root' && $this->parser
->ot
['html'] ) {
1293 # Insert heading index marker
1294 $headingIndex = $contextNode->getAttribute( 'i' );
1295 $titleText = $this->title
->getPrefixedDBkey();
1296 $this->parser
->mHeadings
[] = [ $titleText, $headingIndex ];
1297 $serial = count( $this->parser
->mHeadings
) - 1;
1298 $marker = Parser
::MARKER_PREFIX
. "-h-$serial-" . Parser
::MARKER_SUFFIX
;
1299 $count = $contextNode->getAttribute( 'level' );
1300 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
1301 $this->parser
->mStripState
->addGeneral( $marker, '' );
1305 # Generic recursive expansion
1306 $newIterator = $contextNode->childNodes
;
1309 throw new MWException( __METHOD__
. ': Invalid parameter type' );
1312 if ( $newIterator !== false ) {
1313 if ( $newIterator instanceof PPNode_DOM
) {
1314 $newIterator = $newIterator->node
;
1317 $iteratorStack[] = $newIterator;
1319 } elseif ( $iteratorStack[$level] === false ) {
1320 // Return accumulated value to parent
1321 // With tail recursion
1322 while ( $iteratorStack[$level] === false && $level > 0 ) {
1323 $outStack[$level - 1] .= $out;
1324 array_pop( $outStack );
1325 array_pop( $iteratorStack );
1326 array_pop( $indexStack );
1332 return $outStack[0];
1336 * @param string $sep
1338 * @param string|PPNode_DOM|DOMDocument $args,...
1341 public function implodeWithFlags( $sep, $flags /*, ... */ ) {
1342 $args = array_slice( func_get_args(), 2 );
1346 foreach ( $args as $root ) {
1347 if ( $root instanceof PPNode_DOM
) {
1348 $root = $root->node
;
1350 if ( !is_array( $root ) && !( $root instanceof DOMNodeList
) ) {
1353 foreach ( $root as $node ) {
1359 $s .= $this->expand( $node, $flags );
1366 * Implode with no flags specified
1367 * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1369 * @param string $sep
1370 * @param string|PPNode_DOM|DOMDocument $args,...
1373 public function implode( $sep /*, ... */ ) {
1374 $args = array_slice( func_get_args(), 1 );
1378 foreach ( $args as $root ) {
1379 if ( $root instanceof PPNode_DOM
) {
1380 $root = $root->node
;
1382 if ( !is_array( $root ) && !( $root instanceof DOMNodeList
) ) {
1385 foreach ( $root as $node ) {
1391 $s .= $this->expand( $node );
1398 * Makes an object that, when expand()ed, will be the same as one obtained
1401 * @param string $sep
1402 * @param string|PPNode_DOM|DOMDocument $args,...
1405 public function virtualImplode( $sep /*, ... */ ) {
1406 $args = array_slice( func_get_args(), 1 );
1410 foreach ( $args as $root ) {
1411 if ( $root instanceof PPNode_DOM
) {
1412 $root = $root->node
;
1414 if ( !is_array( $root ) && !( $root instanceof DOMNodeList
) ) {
1417 foreach ( $root as $node ) {
1430 * Virtual implode with brackets
1431 * @param string $start
1432 * @param string $sep
1433 * @param string $end
1434 * @param string|PPNode_DOM|DOMDocument $args,...
1437 public function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1438 $args = array_slice( func_get_args(), 3 );
1442 foreach ( $args as $root ) {
1443 if ( $root instanceof PPNode_DOM
) {
1444 $root = $root->node
;
1446 if ( !is_array( $root ) && !( $root instanceof DOMNodeList
) ) {
1449 foreach ( $root as $node ) {
1462 public function __toString() {
1466 public function getPDBK( $level = false ) {
1467 if ( $level === false ) {
1468 return $this->title
->getPrefixedDBkey();
1470 return isset( $this->titleCache
[$level] ) ?
$this->titleCache
[$level] : false;
1477 public function getArguments() {
1484 public function getNumberedArguments() {
1491 public function getNamedArguments() {
1496 * Returns true if there are no arguments in this frame
1500 public function isEmpty() {
1505 * @param int|string $name
1506 * @return bool Always false in this implementation.
1508 public function getArgument( $name ) {
1513 * Returns true if the infinite loop check is OK, false if a loop is detected
1515 * @param Title $title
1518 public function loopCheck( $title ) {
1519 return !isset( $this->loopCheckHash
[$title->getPrefixedDBkey()] );
1523 * Return true if the frame is a template frame
1527 public function isTemplate() {
1532 * Get a title of frame
1536 public function getTitle() {
1537 return $this->title
;
1541 * Set the volatile flag
1545 public function setVolatile( $flag = true ) {
1546 $this->volatile
= $flag;
1550 * Get the volatile flag
1554 public function isVolatile() {
1555 return $this->volatile
;
1563 public function setTTL( $ttl ) {
1564 if ( $ttl !== null && ( $this->ttl
=== null ||
$ttl < $this->ttl
) ) {
1574 public function getTTL() {
1580 * Expansion frame with template arguments
1583 // @codingStandardsIgnoreStart Squiz.Classes.ValidClassName.NotCamelCaps
1584 class PPTemplateFrame_DOM
extends PPFrame_DOM
{
1585 // @codingStandardsIgnoreEnd
1587 public $numberedArgs, $namedArgs;
1593 public $numberedExpansionCache, $namedExpansionCache;
1596 * @param Preprocessor $preprocessor
1597 * @param bool|PPFrame_DOM $parent
1598 * @param array $numberedArgs
1599 * @param array $namedArgs
1600 * @param bool|Title $title
1602 public function __construct( $preprocessor, $parent = false, $numberedArgs = [],
1603 $namedArgs = [], $title = false
1605 parent
::__construct( $preprocessor );
1607 $this->parent
= $parent;
1608 $this->numberedArgs
= $numberedArgs;
1609 $this->namedArgs
= $namedArgs;
1610 $this->title
= $title;
1611 $pdbk = $title ?
$title->getPrefixedDBkey() : false;
1612 $this->titleCache
= $parent->titleCache
;
1613 $this->titleCache
[] = $pdbk;
1614 $this->loopCheckHash
= /*clone*/ $parent->loopCheckHash
;
1615 if ( $pdbk !== false ) {
1616 $this->loopCheckHash
[$pdbk] = true;
1618 $this->depth
= $parent->depth +
1;
1619 $this->numberedExpansionCache
= $this->namedExpansionCache
= [];
1622 public function __toString() {
1625 $args = $this->numberedArgs +
$this->namedArgs
;
1626 foreach ( $args as $name => $value ) {
1632 $s .= "\"$name\":\"" .
1633 str_replace( '"', '\\"', $value->ownerDocument
->saveXML( $value ) ) . '"';
1640 * @throws MWException
1641 * @param string|int $key
1642 * @param string|PPNode_DOM|DOMDocument $root
1646 public function cachedExpand( $key, $root, $flags = 0 ) {
1647 if ( isset( $this->parent
->childExpansionCache
[$key] ) ) {
1648 return $this->parent
->childExpansionCache
[$key];
1650 $retval = $this->expand( $root, $flags );
1651 if ( !$this->isVolatile() ) {
1652 $this->parent
->childExpansionCache
[$key] = $retval;
1658 * Returns true if there are no arguments in this frame
1662 public function isEmpty() {
1663 return !count( $this->numberedArgs
) && !count( $this->namedArgs
);
1666 public function getArguments() {
1668 foreach ( array_merge(
1669 array_keys( $this->numberedArgs
),
1670 array_keys( $this->namedArgs
) ) as $key ) {
1671 $arguments[$key] = $this->getArgument( $key );
1676 public function getNumberedArguments() {
1678 foreach ( array_keys( $this->numberedArgs
) as $key ) {
1679 $arguments[$key] = $this->getArgument( $key );
1684 public function getNamedArguments() {
1686 foreach ( array_keys( $this->namedArgs
) as $key ) {
1687 $arguments[$key] = $this->getArgument( $key );
1694 * @return string|bool
1696 public function getNumberedArgument( $index ) {
1697 if ( !isset( $this->numberedArgs
[$index] ) ) {
1700 if ( !isset( $this->numberedExpansionCache
[$index] ) ) {
1701 # No trimming for unnamed arguments
1702 $this->numberedExpansionCache
[$index] = $this->parent
->expand(
1703 $this->numberedArgs
[$index],
1704 PPFrame
::STRIP_COMMENTS
1707 return $this->numberedExpansionCache
[$index];
1711 * @param string $name
1712 * @return string|bool
1714 public function getNamedArgument( $name ) {
1715 if ( !isset( $this->namedArgs
[$name] ) ) {
1718 if ( !isset( $this->namedExpansionCache
[$name] ) ) {
1719 # Trim named arguments post-expand, for backwards compatibility
1720 $this->namedExpansionCache
[$name] = trim(
1721 $this->parent
->expand( $this->namedArgs
[$name], PPFrame
::STRIP_COMMENTS
) );
1723 return $this->namedExpansionCache
[$name];
1727 * @param int|string $name
1728 * @return string|bool
1730 public function getArgument( $name ) {
1731 $text = $this->getNumberedArgument( $name );
1732 if ( $text === false ) {
1733 $text = $this->getNamedArgument( $name );
1739 * Return true if the frame is a template frame
1743 public function isTemplate() {
1747 public function setVolatile( $flag = true ) {
1748 parent
::setVolatile( $flag );
1749 $this->parent
->setVolatile( $flag );
1752 public function setTTL( $ttl ) {
1753 parent
::setTTL( $ttl );
1754 $this->parent
->setTTL( $ttl );
1759 * Expansion frame with custom arguments
1762 // @codingStandardsIgnoreStart Squiz.Classes.ValidClassName.NotCamelCaps
1763 class PPCustomFrame_DOM
extends PPFrame_DOM
{
1764 // @codingStandardsIgnoreEnd
1768 public function __construct( $preprocessor, $args ) {
1769 parent
::__construct( $preprocessor );
1770 $this->args
= $args;
1773 public function __toString() {
1776 foreach ( $this->args
as $name => $value ) {
1782 $s .= "\"$name\":\"" .
1783 str_replace( '"', '\\"', $value->__toString() ) . '"';
1792 public function isEmpty() {
1793 return !count( $this->args
);
1797 * @param int|string $index
1798 * @return string|bool
1800 public function getArgument( $index ) {
1801 if ( !isset( $this->args
[$index] ) ) {
1804 return $this->args
[$index];
1807 public function getArguments() {
1815 // @codingStandardsIgnoreStart Squiz.Classes.ValidClassName.NotCamelCaps
1816 class PPNode_DOM
implements PPNode
{
1817 // @codingStandardsIgnoreEnd
1825 public function __construct( $node, $xpath = false ) {
1826 $this->node
= $node;
1832 public function getXPath() {
1833 if ( $this->xpath
=== null ) {
1834 $this->xpath
= new DOMXPath( $this->node
->ownerDocument
);
1836 return $this->xpath
;
1839 public function __toString() {
1840 if ( $this->node
instanceof DOMNodeList
) {
1842 foreach ( $this->node
as $node ) {
1843 $s .= $node->ownerDocument
->saveXML( $node );
1846 $s = $this->node
->ownerDocument
->saveXML( $this->node
);
1852 * @return bool|PPNode_DOM
1854 public function getChildren() {
1855 return $this->node
->childNodes ?
new self( $this->node
->childNodes
) : false;
1859 * @return bool|PPNode_DOM
1861 public function getFirstChild() {
1862 return $this->node
->firstChild ?
new self( $this->node
->firstChild
) : false;
1866 * @return bool|PPNode_DOM
1868 public function getNextSibling() {
1869 return $this->node
->nextSibling ?
new self( $this->node
->nextSibling
) : false;
1873 * @param string $type
1875 * @return bool|PPNode_DOM
1877 public function getChildrenOfType( $type ) {
1878 return new self( $this->getXPath()->query( $type, $this->node
) );
1884 public function getLength() {
1885 if ( $this->node
instanceof DOMNodeList
) {
1886 return $this->node
->length
;
1894 * @return bool|PPNode_DOM
1896 public function item( $i ) {
1897 $item = $this->node
->item( $i );
1898 return $item ?
new self( $item ) : false;
1904 public function getName() {
1905 if ( $this->node
instanceof DOMNodeList
) {
1908 return $this->node
->nodeName
;
1913 * Split a "<part>" node into an associative array containing:
1914 * - name PPNode name
1915 * - index String index
1916 * - value PPNode value
1918 * @throws MWException
1921 public function splitArg() {
1922 $xpath = $this->getXPath();
1923 $names = $xpath->query( 'name', $this->node
);
1924 $values = $xpath->query( 'value', $this->node
);
1925 if ( !$names->length ||
!$values->length
) {
1926 throw new MWException( 'Invalid brace node passed to ' . __METHOD__
);
1928 $name = $names->item( 0 );
1929 $index = $name->getAttribute( 'index' );
1931 'name' => new self( $name ),
1933 'value' => new self( $values->item( 0 ) ) ];
1937 * Split an "<ext>" node into an associative array containing name, attr, inner and close
1938 * All values in the resulting array are PPNodes. Inner and close are optional.
1940 * @throws MWException
1943 public function splitExt() {
1944 $xpath = $this->getXPath();
1945 $names = $xpath->query( 'name', $this->node
);
1946 $attrs = $xpath->query( 'attr', $this->node
);
1947 $inners = $xpath->query( 'inner', $this->node
);
1948 $closes = $xpath->query( 'close', $this->node
);
1949 if ( !$names->length ||
!$attrs->length
) {
1950 throw new MWException( 'Invalid ext node passed to ' . __METHOD__
);
1953 'name' => new self( $names->item( 0 ) ),
1954 'attr' => new self( $attrs->item( 0 ) ) ];
1955 if ( $inners->length
) {
1956 $parts['inner'] = new self( $inners->item( 0 ) );
1958 if ( $closes->length
) {
1959 $parts['close'] = new self( $closes->item( 0 ) );
1965 * Split a "<h>" node
1966 * @throws MWException
1969 public function splitHeading() {
1970 if ( $this->getName() !== 'h' ) {
1971 throw new MWException( 'Invalid h node passed to ' . __METHOD__
);
1974 'i' => $this->node
->getAttribute( 'i' ),
1975 'level' => $this->node
->getAttribute( 'level' ),
1976 'contents' => $this->getChildren()