Merge "Introduce ContentGetParserOutput hook"
[lhc/web/wiklou.git] / includes / parser / Preprocessor_DOM.php
1 <?php
2 /**
3 * Preprocessor using PHP's dom extension
4 *
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.
9 *
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.
14 *
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
19 *
20 * @file
21 * @ingroup Parser
22 */
23
24 /**
25 * @ingroup Parser
26 */
27 class Preprocessor_DOM implements Preprocessor {
28
29 /**
30 * @var Parser
31 */
32 var $parser;
33
34 var $memoryLimit;
35
36 const CACHE_VERSION = 1;
37
38 function __construct( $parser ) {
39 $this->parser = $parser;
40 $mem = ini_get( 'memory_limit' );
41 $this->memoryLimit = false;
42 if ( strval( $mem ) !== '' && $mem != -1 ) {
43 if ( preg_match( '/^\d+$/', $mem ) ) {
44 $this->memoryLimit = $mem;
45 } elseif ( preg_match( '/^(\d+)M$/i', $mem, $m ) ) {
46 $this->memoryLimit = $m[1] * 1048576;
47 }
48 }
49 }
50
51 /**
52 * @return PPFrame_DOM
53 */
54 function newFrame() {
55 return new PPFrame_DOM( $this );
56 }
57
58 /**
59 * @param array $args
60 * @return PPCustomFrame_DOM
61 */
62 function newCustomFrame( $args ) {
63 return new PPCustomFrame_DOM( $this, $args );
64 }
65
66 /**
67 * @param array $values
68 * @return PPNode_DOM
69 */
70 function newPartNodeArray( $values ) {
71 //NOTE: DOM manipulation is slower than building & parsing XML! (or so Tim sais)
72 $xml = "<list>";
73
74 foreach ( $values as $k => $val ) {
75 if ( is_int( $k ) ) {
76 $xml .= "<part><name index=\"$k\"/><value>" . htmlspecialchars( $val ) . "</value></part>";
77 } else {
78 $xml .= "<part><name>" . htmlspecialchars( $k ) . "</name>=<value>" . htmlspecialchars( $val ) . "</value></part>";
79 }
80 }
81
82 $xml .= "</list>";
83
84 $dom = new DOMDocument();
85 $dom->loadXML( $xml );
86 $root = $dom->documentElement;
87
88 $node = new PPNode_DOM( $root->childNodes );
89 return $node;
90 }
91
92 /**
93 * @throws MWException
94 * @return bool
95 */
96 function memCheck() {
97 if ( $this->memoryLimit === false ) {
98 return true;
99 }
100 $usage = memory_get_usage();
101 if ( $usage > $this->memoryLimit * 0.9 ) {
102 $limit = intval( $this->memoryLimit * 0.9 / 1048576 + 0.5 );
103 throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
104 }
105 return $usage <= $this->memoryLimit * 0.8;
106 }
107
108 /**
109 * Preprocess some wikitext and return the document tree.
110 * This is the ghost of Parser::replace_variables().
111 *
112 * @param string $text The text to parse
113 * @param int $flags Bitwise combination of:
114 * Parser::PTD_FOR_INCLUSION Handle "<noinclude>" and "<includeonly>" as if the text is being
115 * included. Default is to assume a direct page view.
116 *
117 * The generated DOM tree must depend only on the input text and the flags.
118 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
119 *
120 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
121 * change in the DOM tree for a given text, must be passed through the section identifier
122 * in the section edit link and thus back to extractSections().
123 *
124 * The output of this function is currently only cached in process memory, but a persistent
125 * cache may be implemented at a later date which takes further advantage of these strict
126 * dependency requirements.
127 *
128 * @throws MWException
129 * @return PPNode_DOM
130 */
131 function preprocessToObj( $text, $flags = 0 ) {
132 wfProfileIn( __METHOD__ );
133 global $wgMemc, $wgPreprocessorCacheThreshold;
134
135 $xml = false;
136 $cacheable = ( $wgPreprocessorCacheThreshold !== false
137 && strlen( $text ) > $wgPreprocessorCacheThreshold );
138 if ( $cacheable ) {
139 wfProfileIn( __METHOD__ . '-cacheable' );
140
141 $cacheKey = wfMemcKey( 'preprocess-xml', md5( $text ), $flags );
142 $cacheValue = $wgMemc->get( $cacheKey );
143 if ( $cacheValue ) {
144 $version = substr( $cacheValue, 0, 8 );
145 if ( intval( $version ) == self::CACHE_VERSION ) {
146 $xml = substr( $cacheValue, 8 );
147 // From the cache
148 wfDebugLog( "Preprocessor", "Loaded preprocessor XML from memcached (key $cacheKey)" );
149 }
150 }
151 if ( $xml === false ) {
152 wfProfileIn( __METHOD__ . '-cache-miss' );
153 $xml = $this->preprocessToXml( $text, $flags );
154 $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . $xml;
155 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
156 wfProfileOut( __METHOD__ . '-cache-miss' );
157 wfDebugLog( "Preprocessor", "Saved preprocessor XML to memcached (key $cacheKey)" );
158 }
159 } else {
160 $xml = $this->preprocessToXml( $text, $flags );
161 }
162
163 // Fail if the number of elements exceeds acceptable limits
164 // Do not attempt to generate the DOM
165 $this->parser->mGeneratedPPNodeCount += substr_count( $xml, '<' );
166 $max = $this->parser->mOptions->getMaxGeneratedPPNodeCount();
167 if ( $this->parser->mGeneratedPPNodeCount > $max ) {
168 if ( $cacheable ) {
169 wfProfileOut( __METHOD__ . '-cacheable' );
170 }
171 wfProfileOut( __METHOD__ );
172 throw new MWException( __METHOD__ . ': generated node count limit exceeded' );
173 }
174
175 wfProfileIn( __METHOD__ . '-loadXML' );
176 $dom = new DOMDocument;
177 wfSuppressWarnings();
178 $result = $dom->loadXML( $xml );
179 wfRestoreWarnings();
180 if ( !$result ) {
181 // Try running the XML through UtfNormal to get rid of invalid characters
182 $xml = UtfNormal::cleanUp( $xml );
183 // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2 don't barf when the XML is >256 levels deep
184 $result = $dom->loadXML( $xml, 1 << 19 );
185 }
186 if ( $result ) {
187 $obj = new PPNode_DOM( $dom->documentElement );
188 }
189 wfProfileOut( __METHOD__ . '-loadXML' );
190
191 if ( $cacheable ) {
192 wfProfileOut( __METHOD__ . '-cacheable' );
193 }
194
195 wfProfileOut( __METHOD__ );
196
197 if ( !$result ) {
198 throw new MWException( __METHOD__ . ' generated invalid XML' );
199 }
200 return $obj;
201 }
202
203 /**
204 * @param string $text
205 * @param int $flags
206 * @return string
207 */
208 function preprocessToXml( $text, $flags = 0 ) {
209 wfProfileIn( __METHOD__ );
210 $rules = array(
211 '{' => array(
212 'end' => '}',
213 'names' => array(
214 2 => 'template',
215 3 => 'tplarg',
216 ),
217 'min' => 2,
218 'max' => 3,
219 ),
220 '[' => array(
221 'end' => ']',
222 'names' => array( 2 => null ),
223 'min' => 2,
224 'max' => 2,
225 )
226 );
227
228 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
229
230 $xmlishElements = $this->parser->getStripList();
231 $enableOnlyinclude = false;
232 if ( $forInclusion ) {
233 $ignoredTags = array( 'includeonly', '/includeonly' );
234 $ignoredElements = array( 'noinclude' );
235 $xmlishElements[] = 'noinclude';
236 if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
237 $enableOnlyinclude = true;
238 }
239 } else {
240 $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
241 $ignoredElements = array( 'includeonly' );
242 $xmlishElements[] = 'includeonly';
243 }
244 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
245
246 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
247 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
248
249 $stack = new PPDStack;
250
251 $searchBase = "[{<\n"; #}
252 $revText = strrev( $text ); // For fast reverse searches
253 $lengthText = strlen( $text );
254
255 $i = 0; # Input pointer, starts out pointing to a pseudo-newline before the start
256 $accum =& $stack->getAccum(); # Current accumulator
257 $accum = '<root>';
258 $findEquals = false; # True to find equals signs in arguments
259 $findPipe = false; # True to take notice of pipe characters
260 $headingIndex = 1;
261 $inHeading = false; # True if $i is inside a possible heading
262 $noMoreGT = false; # True if there are no more greater-than (>) signs right of $i
263 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
264 $fakeLineStart = true; # Do a line-start run without outputting an LF character
265
266 while ( true ) {
267 //$this->memCheck();
268
269 if ( $findOnlyinclude ) {
270 // Ignore all input up to the next <onlyinclude>
271 $startPos = strpos( $text, '<onlyinclude>', $i );
272 if ( $startPos === false ) {
273 // Ignored section runs to the end
274 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
275 break;
276 }
277 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
278 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
279 $i = $tagEndPos;
280 $findOnlyinclude = false;
281 }
282
283 if ( $fakeLineStart ) {
284 $found = 'line-start';
285 $curChar = '';
286 } else {
287 # Find next opening brace, closing brace or pipe
288 $search = $searchBase;
289 if ( $stack->top === false ) {
290 $currentClosing = '';
291 } else {
292 $currentClosing = $stack->top->close;
293 $search .= $currentClosing;
294 }
295 if ( $findPipe ) {
296 $search .= '|';
297 }
298 if ( $findEquals ) {
299 // First equals will be for the template
300 $search .= '=';
301 }
302 $rule = null;
303 # Output literal section, advance input counter
304 $literalLength = strcspn( $text, $search, $i );
305 if ( $literalLength > 0 ) {
306 $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
307 $i += $literalLength;
308 }
309 if ( $i >= $lengthText ) {
310 if ( $currentClosing == "\n" ) {
311 // Do a past-the-end run to finish off the heading
312 $curChar = '';
313 $found = 'line-end';
314 } else {
315 # All done
316 break;
317 }
318 } else {
319 $curChar = $text[$i];
320 if ( $curChar == '|' ) {
321 $found = 'pipe';
322 } elseif ( $curChar == '=' ) {
323 $found = 'equals';
324 } elseif ( $curChar == '<' ) {
325 $found = 'angle';
326 } elseif ( $curChar == "\n" ) {
327 if ( $inHeading ) {
328 $found = 'line-end';
329 } else {
330 $found = 'line-start';
331 }
332 } elseif ( $curChar == $currentClosing ) {
333 $found = 'close';
334 } elseif ( isset( $rules[$curChar] ) ) {
335 $found = 'open';
336 $rule = $rules[$curChar];
337 } else {
338 # Some versions of PHP have a strcspn which stops on null characters
339 # Ignore and continue
340 ++$i;
341 continue;
342 }
343 }
344 }
345
346 if ( $found == 'angle' ) {
347 $matches = false;
348 // Handle </onlyinclude>
349 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
350 $findOnlyinclude = true;
351 continue;
352 }
353
354 // Determine element name
355 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
356 // Element name missing or not listed
357 $accum .= '&lt;';
358 ++$i;
359 continue;
360 }
361 // Handle comments
362 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
363
364 // To avoid leaving blank lines, when a sequence of
365 // space-separated comments is both preceded and followed by
366 // a newline (ignoring spaces), then
367 // trim leading and trailing spaces and the trailing newline.
368
369 // Find the end
370 $endPos = strpos( $text, '-->', $i + 4 );
371 if ( $endPos === false ) {
372 // Unclosed comment in input, runs to end
373 $inner = substr( $text, $i );
374 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
375 $i = $lengthText;
376 } else {
377 // Search backwards for leading whitespace
378 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
379
380 // Search forwards for trailing whitespace
381 // $wsEnd will be the position of the last space (or the '>' if there's none)
382 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
383
384 // Keep looking forward as long as we're finding more
385 // comments.
386 $comments = array( array( $wsStart, $wsEnd ) );
387 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
388 $c = strpos( $text, '-->', $wsEnd + 4 );
389 if ( $c === false ) {
390 break;
391 }
392 $c = $c + 2 + strspn( $text, " \t", $c + 3 );
393 $comments[] = array( $wsEnd + 1, $c );
394 $wsEnd = $c;
395 }
396
397 // Eat the line if possible
398 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
399 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
400 // it's a possible beneficial b/c break.
401 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
402 && substr( $text, $wsEnd + 1, 1 ) == "\n"
403 ) {
404 // Remove leading whitespace from the end of the accumulator
405 // Sanity check first though
406 $wsLength = $i - $wsStart;
407 if ( $wsLength > 0
408 && strspn( $accum, " \t", -$wsLength ) === $wsLength
409 ) {
410 $accum = substr( $accum, 0, -$wsLength );
411 }
412
413 // Dump all but the last comment to the accumulator
414 foreach ( $comments as $j => $com ) {
415 $startPos = $com[0];
416 $endPos = $com[1] + 1;
417 if ( $j == ( count( $comments ) - 1 ) ) {
418 break;
419 }
420 $inner = substr( $text, $startPos, $endPos - $startPos );
421 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
422 }
423
424 // Do a line-start run next time to look for headings after the comment
425 $fakeLineStart = true;
426 } else {
427 // No line to eat, just take the comment itself
428 $startPos = $i;
429 $endPos += 2;
430 }
431
432 if ( $stack->top ) {
433 $part = $stack->top->getCurrentPart();
434 if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
435 $part->visualEnd = $wsStart;
436 }
437 // Else comments abutting, no change in visual end
438 $part->commentEnd = $endPos;
439 }
440 $i = $endPos + 1;
441 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
442 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
443 }
444 continue;
445 }
446 $name = $matches[1];
447 $lowerName = strtolower( $name );
448 $attrStart = $i + strlen( $name ) + 1;
449
450 // Find end of tag
451 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
452 if ( $tagEndPos === false ) {
453 // Infinite backtrack
454 // Disable tag search to prevent worst-case O(N^2) performance
455 $noMoreGT = true;
456 $accum .= '&lt;';
457 ++$i;
458 continue;
459 }
460
461 // Handle ignored tags
462 if ( in_array( $lowerName, $ignoredTags ) ) {
463 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
464 $i = $tagEndPos + 1;
465 continue;
466 }
467
468 $tagStartPos = $i;
469 if ( $text[$tagEndPos - 1] == '/' ) {
470 $attrEnd = $tagEndPos - 1;
471 $inner = null;
472 $i = $tagEndPos + 1;
473 $close = '';
474 } else {
475 $attrEnd = $tagEndPos;
476 // Find closing tag
477 if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
478 $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
479 ) {
480 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
481 $i = $matches[0][1] + strlen( $matches[0][0] );
482 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
483 } else {
484 // No end tag -- let it run out to the end of the text.
485 $inner = substr( $text, $tagEndPos + 1 );
486 $i = $lengthText;
487 $close = '';
488 }
489 }
490 // <includeonly> and <noinclude> just become <ignore> tags
491 if ( in_array( $lowerName, $ignoredElements ) ) {
492 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
493 . '</ignore>';
494 continue;
495 }
496
497 $accum .= '<ext>';
498 if ( $attrEnd <= $attrStart ) {
499 $attr = '';
500 } else {
501 $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
502 }
503 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
504 // Note that the attr element contains the whitespace between name and attribute,
505 // this is necessary for precise reconstruction during pre-save transform.
506 '<attr>' . htmlspecialchars( $attr ) . '</attr>';
507 if ( $inner !== null ) {
508 $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
509 }
510 $accum .= $close . '</ext>';
511 } elseif ( $found == 'line-start' ) {
512 // Is this the start of a heading?
513 // Line break belongs before the heading element in any case
514 if ( $fakeLineStart ) {
515 $fakeLineStart = false;
516 } else {
517 $accum .= $curChar;
518 $i++;
519 }
520
521 $count = strspn( $text, '=', $i, 6 );
522 if ( $count == 1 && $findEquals ) {
523 // DWIM: This looks kind of like a name/value separator
524 // Let's let the equals handler have it and break the potential heading
525 // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex.
526 } elseif ( $count > 0 ) {
527 $piece = array(
528 'open' => "\n",
529 'close' => "\n",
530 'parts' => array( new PPDPart( str_repeat( '=', $count ) ) ),
531 'startPos' => $i,
532 'count' => $count );
533 $stack->push( $piece );
534 $accum =& $stack->getAccum();
535 $flags = $stack->getFlags();
536 extract( $flags );
537 $i += $count;
538 }
539 } elseif ( $found == 'line-end' ) {
540 $piece = $stack->top;
541 // A heading must be open, otherwise \n wouldn't have been in the search list
542 assert( '$piece->open == "\n"' );
543 $part = $piece->getCurrentPart();
544 // Search back through the input to see if it has a proper close
545 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
546 $wsLength = strspn( $revText, " \t", $lengthText - $i );
547 $searchStart = $i - $wsLength;
548 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
549 // Comment found at line end
550 // Search for equals signs before the comment
551 $searchStart = $part->visualEnd;
552 $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
553 }
554 $count = $piece->count;
555 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
556 if ( $equalsLength > 0 ) {
557 if ( $searchStart - $equalsLength == $piece->startPos ) {
558 // This is just a single string of equals signs on its own line
559 // Replicate the doHeadings behavior /={count}(.+)={count}/
560 // First find out how many equals signs there really are (don't stop at 6)
561 $count = $equalsLength;
562 if ( $count < 3 ) {
563 $count = 0;
564 } else {
565 $count = min( 6, intval( ( $count - 1 ) / 2 ) );
566 }
567 } else {
568 $count = min( $equalsLength, $count );
569 }
570 if ( $count > 0 ) {
571 // Normal match, output <h>
572 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
573 $headingIndex++;
574 } else {
575 // Single equals sign on its own line, count=0
576 $element = $accum;
577 }
578 } else {
579 // No match, no <h>, just pass down the inner text
580 $element = $accum;
581 }
582 // Unwind the stack
583 $stack->pop();
584 $accum =& $stack->getAccum();
585 $flags = $stack->getFlags();
586 extract( $flags );
587
588 // Append the result to the enclosing accumulator
589 $accum .= $element;
590 // Note that we do NOT increment the input pointer.
591 // This is because the closing linebreak could be the opening linebreak of
592 // another heading. Infinite loops are avoided because the next iteration MUST
593 // hit the heading open case above, which unconditionally increments the
594 // input pointer.
595 } elseif ( $found == 'open' ) {
596 # count opening brace characters
597 $count = strspn( $text, $curChar, $i );
598
599 # we need to add to stack only if opening brace count is enough for one of the rules
600 if ( $count >= $rule['min'] ) {
601 # Add it to the stack
602 $piece = array(
603 'open' => $curChar,
604 'close' => $rule['end'],
605 'count' => $count,
606 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
607 );
608
609 $stack->push( $piece );
610 $accum =& $stack->getAccum();
611 $flags = $stack->getFlags();
612 extract( $flags );
613 } else {
614 # Add literal brace(s)
615 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
616 }
617 $i += $count;
618 } elseif ( $found == 'close' ) {
619 $piece = $stack->top;
620 # lets check if there are enough characters for closing brace
621 $maxCount = $piece->count;
622 $count = strspn( $text, $curChar, $i, $maxCount );
623
624 # check for maximum matching characters (if there are 5 closing
625 # characters, we will probably need only 3 - depending on the rules)
626 $rule = $rules[$piece->open];
627 if ( $count > $rule['max'] ) {
628 # The specified maximum exists in the callback array, unless the caller
629 # has made an error
630 $matchingCount = $rule['max'];
631 } else {
632 # Count is less than the maximum
633 # Skip any gaps in the callback array to find the true largest match
634 # Need to use array_key_exists not isset because the callback can be null
635 $matchingCount = $count;
636 while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
637 --$matchingCount;
638 }
639 }
640
641 if ( $matchingCount <= 0 ) {
642 # No matching element found in callback array
643 # Output a literal closing brace and continue
644 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
645 $i += $count;
646 continue;
647 }
648 $name = $rule['names'][$matchingCount];
649 if ( $name === null ) {
650 // No element, just literal text
651 $element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
652 } else {
653 # Create XML element
654 # Note: $parts is already XML, does not need to be encoded further
655 $parts = $piece->parts;
656 $title = $parts[0]->out;
657 unset( $parts[0] );
658
659 # The invocation is at the start of the line if lineStart is set in
660 # the stack, and all opening brackets are used up.
661 if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
662 $attr = ' lineStart="1"';
663 } else {
664 $attr = '';
665 }
666
667 $element = "<$name$attr>";
668 $element .= "<title>$title</title>";
669 $argIndex = 1;
670 foreach ( $parts as $part ) {
671 if ( isset( $part->eqpos ) ) {
672 $argName = substr( $part->out, 0, $part->eqpos );
673 $argValue = substr( $part->out, $part->eqpos + 1 );
674 $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
675 } else {
676 $element .= "<part><name index=\"$argIndex\" /><value>{$part->out}</value></part>";
677 $argIndex++;
678 }
679 }
680 $element .= "</$name>";
681 }
682
683 # Advance input pointer
684 $i += $matchingCount;
685
686 # Unwind the stack
687 $stack->pop();
688 $accum =& $stack->getAccum();
689
690 # Re-add the old stack element if it still has unmatched opening characters remaining
691 if ( $matchingCount < $piece->count ) {
692 $piece->parts = array( new PPDPart );
693 $piece->count -= $matchingCount;
694 # do we still qualify for any callback with remaining count?
695 $min = $rules[$piece->open]['min'];
696 if ( $piece->count >= $min ) {
697 $stack->push( $piece );
698 $accum =& $stack->getAccum();
699 } else {
700 $accum .= str_repeat( $piece->open, $piece->count );
701 }
702 }
703 $flags = $stack->getFlags();
704 extract( $flags );
705
706 # Add XML element to the enclosing accumulator
707 $accum .= $element;
708 } elseif ( $found == 'pipe' ) {
709 $findEquals = true; // shortcut for getFlags()
710 $stack->addPart();
711 $accum =& $stack->getAccum();
712 ++$i;
713 } elseif ( $found == 'equals' ) {
714 $findEquals = false; // shortcut for getFlags()
715 $stack->getCurrentPart()->eqpos = strlen( $accum );
716 $accum .= '=';
717 ++$i;
718 }
719 }
720
721 # Output any remaining unclosed brackets
722 foreach ( $stack->stack as $piece ) {
723 $stack->rootAccum .= $piece->breakSyntax();
724 }
725 $stack->rootAccum .= '</root>';
726 $xml = $stack->rootAccum;
727
728 wfProfileOut( __METHOD__ );
729
730 return $xml;
731 }
732 }
733
734 /**
735 * Stack class to help Preprocessor::preprocessToObj()
736 * @ingroup Parser
737 */
738 class PPDStack {
739 var $stack, $rootAccum;
740
741 /**
742 * @var PPDStack
743 */
744 var $top;
745 var $out;
746 var $elementClass = 'PPDStackElement';
747
748 static $false = false;
749
750 function __construct() {
751 $this->stack = array();
752 $this->top = false;
753 $this->rootAccum = '';
754 $this->accum =& $this->rootAccum;
755 }
756
757 /**
758 * @return int
759 */
760 function count() {
761 return count( $this->stack );
762 }
763
764 function &getAccum() {
765 return $this->accum;
766 }
767
768 function getCurrentPart() {
769 if ( $this->top === false ) {
770 return false;
771 } else {
772 return $this->top->getCurrentPart();
773 }
774 }
775
776 function push( $data ) {
777 if ( $data instanceof $this->elementClass ) {
778 $this->stack[] = $data;
779 } else {
780 $class = $this->elementClass;
781 $this->stack[] = new $class( $data );
782 }
783 $this->top = $this->stack[count( $this->stack ) - 1];
784 $this->accum =& $this->top->getAccum();
785 }
786
787 function pop() {
788 if ( !count( $this->stack ) ) {
789 throw new MWException( __METHOD__ . ': no elements remaining' );
790 }
791 $temp = array_pop( $this->stack );
792
793 if ( count( $this->stack ) ) {
794 $this->top = $this->stack[count( $this->stack ) - 1];
795 $this->accum =& $this->top->getAccum();
796 } else {
797 $this->top = self::$false;
798 $this->accum =& $this->rootAccum;
799 }
800 return $temp;
801 }
802
803 function addPart( $s = '' ) {
804 $this->top->addPart( $s );
805 $this->accum =& $this->top->getAccum();
806 }
807
808 /**
809 * @return array
810 */
811 function getFlags() {
812 if ( !count( $this->stack ) ) {
813 return array(
814 'findEquals' => false,
815 'findPipe' => false,
816 'inHeading' => false,
817 );
818 } else {
819 return $this->top->getFlags();
820 }
821 }
822 }
823
824 /**
825 * @ingroup Parser
826 */
827 class PPDStackElement {
828 var $open, // Opening character (\n for heading)
829 $close, // Matching closing character
830 $count, // Number of opening characters found (number of "=" for heading)
831 $parts, // Array of PPDPart objects describing pipe-separated parts.
832 $lineStart; // True if the open char appeared at the start of the input line. Not set for headings.
833
834 var $partClass = 'PPDPart';
835
836 function __construct( $data = array() ) {
837 $class = $this->partClass;
838 $this->parts = array( new $class );
839
840 foreach ( $data as $name => $value ) {
841 $this->$name = $value;
842 }
843 }
844
845 function &getAccum() {
846 return $this->parts[count( $this->parts ) - 1]->out;
847 }
848
849 function addPart( $s = '' ) {
850 $class = $this->partClass;
851 $this->parts[] = new $class( $s );
852 }
853
854 function getCurrentPart() {
855 return $this->parts[count( $this->parts ) - 1];
856 }
857
858 /**
859 * @return array
860 */
861 function getFlags() {
862 $partCount = count( $this->parts );
863 $findPipe = $this->open != "\n" && $this->open != '[';
864 return array(
865 'findPipe' => $findPipe,
866 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts[$partCount - 1]->eqpos ),
867 'inHeading' => $this->open == "\n",
868 );
869 }
870
871 /**
872 * Get the output string that would result if the close is not found.
873 *
874 * @return string
875 */
876 function breakSyntax( $openingCount = false ) {
877 if ( $this->open == "\n" ) {
878 $s = $this->parts[0]->out;
879 } else {
880 if ( $openingCount === false ) {
881 $openingCount = $this->count;
882 }
883 $s = str_repeat( $this->open, $openingCount );
884 $first = true;
885 foreach ( $this->parts as $part ) {
886 if ( $first ) {
887 $first = false;
888 } else {
889 $s .= '|';
890 }
891 $s .= $part->out;
892 }
893 }
894 return $s;
895 }
896 }
897
898 /**
899 * @ingroup Parser
900 */
901 class PPDPart {
902 var $out; // Output accumulator string
903
904 // Optional member variables:
905 // eqpos Position of equals sign in output accumulator
906 // commentEnd Past-the-end input pointer for the last comment encountered
907 // visualEnd Past-the-end input pointer for the end of the accumulator minus comments
908
909 function __construct( $out = '' ) {
910 $this->out = $out;
911 }
912 }
913
914 /**
915 * An expansion frame, used as a context to expand the result of preprocessToObj()
916 * @ingroup Parser
917 */
918 class PPFrame_DOM implements PPFrame {
919
920 /**
921 * @var Preprocessor
922 */
923 var $preprocessor;
924
925 /**
926 * @var Parser
927 */
928 var $parser;
929
930 /**
931 * @var Title
932 */
933 var $title;
934 var $titleCache;
935
936 /**
937 * Hashtable listing templates which are disallowed for expansion in this frame,
938 * having been encountered previously in parent frames.
939 */
940 var $loopCheckHash;
941
942 /**
943 * Recursion depth of this frame, top = 0
944 * Note that this is NOT the same as expansion depth in expand()
945 */
946 var $depth;
947
948 /**
949 * Construct a new preprocessor frame.
950 * @param Preprocessor $preprocessor The parent preprocessor
951 */
952 function __construct( $preprocessor ) {
953 $this->preprocessor = $preprocessor;
954 $this->parser = $preprocessor->parser;
955 $this->title = $this->parser->mTitle;
956 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
957 $this->loopCheckHash = array();
958 $this->depth = 0;
959 }
960
961 /**
962 * Create a new child frame
963 * $args is optionally a multi-root PPNode or array containing the template arguments
964 *
965 * @param bool|array $args
966 * @param Title|bool $title
967 * @param int $indexOffset
968 * @return PPTemplateFrame_DOM
969 */
970 function newChild( $args = false, $title = false, $indexOffset = 0 ) {
971 $namedArgs = array();
972 $numberedArgs = array();
973 if ( $title === false ) {
974 $title = $this->title;
975 }
976 if ( $args !== false ) {
977 $xpath = false;
978 if ( $args instanceof PPNode ) {
979 $args = $args->node;
980 }
981 foreach ( $args as $arg ) {
982 if ( $arg instanceof PPNode ) {
983 $arg = $arg->node;
984 }
985 if ( !$xpath ) {
986 $xpath = new DOMXPath( $arg->ownerDocument );
987 }
988
989 $nameNodes = $xpath->query( 'name', $arg );
990 $value = $xpath->query( 'value', $arg );
991 if ( $nameNodes->item( 0 )->hasAttributes() ) {
992 // Numbered parameter
993 $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
994 $index = $index - $indexOffset;
995 $numberedArgs[$index] = $value->item( 0 );
996 unset( $namedArgs[$index] );
997 } else {
998 // Named parameter
999 $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
1000 $namedArgs[$name] = $value->item( 0 );
1001 unset( $numberedArgs[$name] );
1002 }
1003 }
1004 }
1005 return new PPTemplateFrame_DOM( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
1006 }
1007
1008 /**
1009 * @throws MWException
1010 * @param string|PPNode_DOM|DOMDocument $root
1011 * @param int $flags
1012 * @return string
1013 */
1014 function expand( $root, $flags = 0 ) {
1015 static $expansionDepth = 0;
1016 if ( is_string( $root ) ) {
1017 return $root;
1018 }
1019
1020 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1021 $this->parser->limitationWarn( 'node-count-exceeded',
1022 $this->parser->mPPNodeCount,
1023 $this->parser->mOptions->getMaxPPNodeCount()
1024 );
1025 return '<span class="error">Node-count limit exceeded</span>';
1026 }
1027
1028 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1029 $this->parser->limitationWarn( 'expansion-depth-exceeded',
1030 $expansionDepth,
1031 $this->parser->mOptions->getMaxPPExpandDepth()
1032 );
1033 return '<span class="error">Expansion depth limit exceeded</span>';
1034 }
1035 wfProfileIn( __METHOD__ );
1036 ++$expansionDepth;
1037 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1038 $this->parser->mHighestExpansionDepth = $expansionDepth;
1039 }
1040
1041 if ( $root instanceof PPNode_DOM ) {
1042 $root = $root->node;
1043 }
1044 if ( $root instanceof DOMDocument ) {
1045 $root = $root->documentElement;
1046 }
1047
1048 $outStack = array( '', '' );
1049 $iteratorStack = array( false, $root );
1050 $indexStack = array( 0, 0 );
1051
1052 while ( count( $iteratorStack ) > 1 ) {
1053 $level = count( $outStack ) - 1;
1054 $iteratorNode =& $iteratorStack[$level];
1055 $out =& $outStack[$level];
1056 $index =& $indexStack[$level];
1057
1058 if ( $iteratorNode instanceof PPNode_DOM ) {
1059 $iteratorNode = $iteratorNode->node;
1060 }
1061
1062 if ( is_array( $iteratorNode ) ) {
1063 if ( $index >= count( $iteratorNode ) ) {
1064 // All done with this iterator
1065 $iteratorStack[$level] = false;
1066 $contextNode = false;
1067 } else {
1068 $contextNode = $iteratorNode[$index];
1069 $index++;
1070 }
1071 } elseif ( $iteratorNode instanceof DOMNodeList ) {
1072 if ( $index >= $iteratorNode->length ) {
1073 // All done with this iterator
1074 $iteratorStack[$level] = false;
1075 $contextNode = false;
1076 } else {
1077 $contextNode = $iteratorNode->item( $index );
1078 $index++;
1079 }
1080 } else {
1081 // Copy to $contextNode and then delete from iterator stack,
1082 // because this is not an iterator but we do have to execute it once
1083 $contextNode = $iteratorStack[$level];
1084 $iteratorStack[$level] = false;
1085 }
1086
1087 if ( $contextNode instanceof PPNode_DOM ) {
1088 $contextNode = $contextNode->node;
1089 }
1090
1091 $newIterator = false;
1092
1093 if ( $contextNode === false ) {
1094 // nothing to do
1095 } elseif ( is_string( $contextNode ) ) {
1096 $out .= $contextNode;
1097 } elseif ( is_array( $contextNode ) || $contextNode instanceof DOMNodeList ) {
1098 $newIterator = $contextNode;
1099 } elseif ( $contextNode instanceof DOMNode ) {
1100 if ( $contextNode->nodeType == XML_TEXT_NODE ) {
1101 $out .= $contextNode->nodeValue;
1102 } elseif ( $contextNode->nodeName == 'template' ) {
1103 # Double-brace expansion
1104 $xpath = new DOMXPath( $contextNode->ownerDocument );
1105 $titles = $xpath->query( 'title', $contextNode );
1106 $title = $titles->item( 0 );
1107 $parts = $xpath->query( 'part', $contextNode );
1108 if ( $flags & PPFrame::NO_TEMPLATES ) {
1109 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
1110 } else {
1111 $lineStart = $contextNode->getAttribute( 'lineStart' );
1112 $params = array(
1113 'title' => new PPNode_DOM( $title ),
1114 'parts' => new PPNode_DOM( $parts ),
1115 'lineStart' => $lineStart );
1116 $ret = $this->parser->braceSubstitution( $params, $this );
1117 if ( isset( $ret['object'] ) ) {
1118 $newIterator = $ret['object'];
1119 } else {
1120 $out .= $ret['text'];
1121 }
1122 }
1123 } elseif ( $contextNode->nodeName == 'tplarg' ) {
1124 # Triple-brace expansion
1125 $xpath = new DOMXPath( $contextNode->ownerDocument );
1126 $titles = $xpath->query( 'title', $contextNode );
1127 $title = $titles->item( 0 );
1128 $parts = $xpath->query( 'part', $contextNode );
1129 if ( $flags & PPFrame::NO_ARGS ) {
1130 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
1131 } else {
1132 $params = array(
1133 'title' => new PPNode_DOM( $title ),
1134 'parts' => new PPNode_DOM( $parts ) );
1135 $ret = $this->parser->argSubstitution( $params, $this );
1136 if ( isset( $ret['object'] ) ) {
1137 $newIterator = $ret['object'];
1138 } else {
1139 $out .= $ret['text'];
1140 }
1141 }
1142 } elseif ( $contextNode->nodeName == 'comment' ) {
1143 # HTML-style comment
1144 # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1145 if ( $this->parser->ot['html']
1146 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1147 || ( $flags & PPFrame::STRIP_COMMENTS )
1148 ) {
1149 $out .= '';
1150 } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1151 # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result
1152 # Not in RECOVER_COMMENTS mode (extractSections) though
1153 $out .= $this->parser->insertStripItem( $contextNode->textContent );
1154 } else {
1155 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1156 $out .= $contextNode->textContent;
1157 }
1158 } elseif ( $contextNode->nodeName == 'ignore' ) {
1159 # Output suppression used by <includeonly> etc.
1160 # OT_WIKI will only respect <ignore> in substed templates.
1161 # The other output types respect it unless NO_IGNORE is set.
1162 # extractSections() sets NO_IGNORE and so never respects it.
1163 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & PPFrame::NO_IGNORE ) ) {
1164 $out .= $contextNode->textContent;
1165 } else {
1166 $out .= '';
1167 }
1168 } elseif ( $contextNode->nodeName == 'ext' ) {
1169 # Extension tag
1170 $xpath = new DOMXPath( $contextNode->ownerDocument );
1171 $names = $xpath->query( 'name', $contextNode );
1172 $attrs = $xpath->query( 'attr', $contextNode );
1173 $inners = $xpath->query( 'inner', $contextNode );
1174 $closes = $xpath->query( 'close', $contextNode );
1175 $params = array(
1176 'name' => new PPNode_DOM( $names->item( 0 ) ),
1177 'attr' => $attrs->length > 0 ? new PPNode_DOM( $attrs->item( 0 ) ) : null,
1178 'inner' => $inners->length > 0 ? new PPNode_DOM( $inners->item( 0 ) ) : null,
1179 'close' => $closes->length > 0 ? new PPNode_DOM( $closes->item( 0 ) ) : null,
1180 );
1181 $out .= $this->parser->extensionSubstitution( $params, $this );
1182 } elseif ( $contextNode->nodeName == 'h' ) {
1183 # Heading
1184 $s = $this->expand( $contextNode->childNodes, $flags );
1185
1186 # Insert a heading marker only for <h> children of <root>
1187 # This is to stop extractSections from going over multiple tree levels
1188 if ( $contextNode->parentNode->nodeName == 'root' && $this->parser->ot['html'] ) {
1189 # Insert heading index marker
1190 $headingIndex = $contextNode->getAttribute( 'i' );
1191 $titleText = $this->title->getPrefixedDBkey();
1192 $this->parser->mHeadings[] = array( $titleText, $headingIndex );
1193 $serial = count( $this->parser->mHeadings ) - 1;
1194 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1195 $count = $contextNode->getAttribute( 'level' );
1196 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
1197 $this->parser->mStripState->addGeneral( $marker, '' );
1198 }
1199 $out .= $s;
1200 } else {
1201 # Generic recursive expansion
1202 $newIterator = $contextNode->childNodes;
1203 }
1204 } else {
1205 wfProfileOut( __METHOD__ );
1206 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1207 }
1208
1209 if ( $newIterator !== false ) {
1210 if ( $newIterator instanceof PPNode_DOM ) {
1211 $newIterator = $newIterator->node;
1212 }
1213 $outStack[] = '';
1214 $iteratorStack[] = $newIterator;
1215 $indexStack[] = 0;
1216 } elseif ( $iteratorStack[$level] === false ) {
1217 // Return accumulated value to parent
1218 // With tail recursion
1219 while ( $iteratorStack[$level] === false && $level > 0 ) {
1220 $outStack[$level - 1] .= $out;
1221 array_pop( $outStack );
1222 array_pop( $iteratorStack );
1223 array_pop( $indexStack );
1224 $level--;
1225 }
1226 }
1227 }
1228 --$expansionDepth;
1229 wfProfileOut( __METHOD__ );
1230 return $outStack[0];
1231 }
1232
1233 /**
1234 * @param string $sep
1235 * @param int $flags
1236 * @return string
1237 */
1238 function implodeWithFlags( $sep, $flags /*, ... */ ) {
1239 $args = array_slice( func_get_args(), 2 );
1240
1241 $first = true;
1242 $s = '';
1243 foreach ( $args as $root ) {
1244 if ( $root instanceof PPNode_DOM ) {
1245 $root = $root->node;
1246 }
1247 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1248 $root = array( $root );
1249 }
1250 foreach ( $root as $node ) {
1251 if ( $first ) {
1252 $first = false;
1253 } else {
1254 $s .= $sep;
1255 }
1256 $s .= $this->expand( $node, $flags );
1257 }
1258 }
1259 return $s;
1260 }
1261
1262 /**
1263 * Implode with no flags specified
1264 * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1265 *
1266 * @param string $sep
1267 * @return string
1268 */
1269 function implode( $sep /*, ... */ ) {
1270 $args = array_slice( func_get_args(), 1 );
1271
1272 $first = true;
1273 $s = '';
1274 foreach ( $args as $root ) {
1275 if ( $root instanceof PPNode_DOM ) {
1276 $root = $root->node;
1277 }
1278 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1279 $root = array( $root );
1280 }
1281 foreach ( $root as $node ) {
1282 if ( $first ) {
1283 $first = false;
1284 } else {
1285 $s .= $sep;
1286 }
1287 $s .= $this->expand( $node );
1288 }
1289 }
1290 return $s;
1291 }
1292
1293 /**
1294 * Makes an object that, when expand()ed, will be the same as one obtained
1295 * with implode()
1296 *
1297 * @param string $sep
1298 * @return array
1299 */
1300 function virtualImplode( $sep /*, ... */ ) {
1301 $args = array_slice( func_get_args(), 1 );
1302 $out = array();
1303 $first = true;
1304
1305 foreach ( $args as $root ) {
1306 if ( $root instanceof PPNode_DOM ) {
1307 $root = $root->node;
1308 }
1309 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1310 $root = array( $root );
1311 }
1312 foreach ( $root as $node ) {
1313 if ( $first ) {
1314 $first = false;
1315 } else {
1316 $out[] = $sep;
1317 }
1318 $out[] = $node;
1319 }
1320 }
1321 return $out;
1322 }
1323
1324 /**
1325 * Virtual implode with brackets
1326 * @param string $start
1327 * @param string $sep
1328 * @param string $end
1329 * @return array
1330 */
1331 function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1332 $args = array_slice( func_get_args(), 3 );
1333 $out = array( $start );
1334 $first = true;
1335
1336 foreach ( $args as $root ) {
1337 if ( $root instanceof PPNode_DOM ) {
1338 $root = $root->node;
1339 }
1340 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1341 $root = array( $root );
1342 }
1343 foreach ( $root as $node ) {
1344 if ( $first ) {
1345 $first = false;
1346 } else {
1347 $out[] = $sep;
1348 }
1349 $out[] = $node;
1350 }
1351 }
1352 $out[] = $end;
1353 return $out;
1354 }
1355
1356 function __toString() {
1357 return 'frame{}';
1358 }
1359
1360 function getPDBK( $level = false ) {
1361 if ( $level === false ) {
1362 return $this->title->getPrefixedDBkey();
1363 } else {
1364 return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1365 }
1366 }
1367
1368 /**
1369 * @return array
1370 */
1371 function getArguments() {
1372 return array();
1373 }
1374
1375 /**
1376 * @return array
1377 */
1378 function getNumberedArguments() {
1379 return array();
1380 }
1381
1382 /**
1383 * @return array
1384 */
1385 function getNamedArguments() {
1386 return array();
1387 }
1388
1389 /**
1390 * Returns true if there are no arguments in this frame
1391 *
1392 * @return bool
1393 */
1394 function isEmpty() {
1395 return true;
1396 }
1397
1398 function getArgument( $name ) {
1399 return false;
1400 }
1401
1402 /**
1403 * Returns true if the infinite loop check is OK, false if a loop is detected
1404 *
1405 * @param Title $title
1406 * @return bool
1407 */
1408 function loopCheck( $title ) {
1409 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1410 }
1411
1412 /**
1413 * Return true if the frame is a template frame
1414 *
1415 * @return bool
1416 */
1417 function isTemplate() {
1418 return false;
1419 }
1420
1421 /**
1422 * Get a title of frame
1423 *
1424 * @return Title
1425 */
1426 function getTitle() {
1427 return $this->title;
1428 }
1429 }
1430
1431 /**
1432 * Expansion frame with template arguments
1433 * @ingroup Parser
1434 */
1435 class PPTemplateFrame_DOM extends PPFrame_DOM {
1436 var $numberedArgs, $namedArgs;
1437
1438 /**
1439 * @var PPFrame_DOM
1440 */
1441 var $parent;
1442 var $numberedExpansionCache, $namedExpansionCache;
1443
1444 /**
1445 * @param Preprocessor $preprocessor
1446 * @param PPFrame_DOM $parent
1447 * @param array $numberedArgs
1448 * @param array $namedArgs
1449 * @param Title $title
1450 */
1451 function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
1452 parent::__construct( $preprocessor );
1453
1454 $this->parent = $parent;
1455 $this->numberedArgs = $numberedArgs;
1456 $this->namedArgs = $namedArgs;
1457 $this->title = $title;
1458 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1459 $this->titleCache = $parent->titleCache;
1460 $this->titleCache[] = $pdbk;
1461 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1462 if ( $pdbk !== false ) {
1463 $this->loopCheckHash[$pdbk] = true;
1464 }
1465 $this->depth = $parent->depth + 1;
1466 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1467 }
1468
1469 function __toString() {
1470 $s = 'tplframe{';
1471 $first = true;
1472 $args = $this->numberedArgs + $this->namedArgs;
1473 foreach ( $args as $name => $value ) {
1474 if ( $first ) {
1475 $first = false;
1476 } else {
1477 $s .= ', ';
1478 }
1479 $s .= "\"$name\":\"" .
1480 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
1481 }
1482 $s .= '}';
1483 return $s;
1484 }
1485
1486 /**
1487 * Returns true if there are no arguments in this frame
1488 *
1489 * @return bool
1490 */
1491 function isEmpty() {
1492 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1493 }
1494
1495 function getArguments() {
1496 $arguments = array();
1497 foreach ( array_merge(
1498 array_keys( $this->numberedArgs ),
1499 array_keys( $this->namedArgs ) ) as $key ) {
1500 $arguments[$key] = $this->getArgument( $key );
1501 }
1502 return $arguments;
1503 }
1504
1505 function getNumberedArguments() {
1506 $arguments = array();
1507 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1508 $arguments[$key] = $this->getArgument( $key );
1509 }
1510 return $arguments;
1511 }
1512
1513 function getNamedArguments() {
1514 $arguments = array();
1515 foreach ( array_keys( $this->namedArgs ) as $key ) {
1516 $arguments[$key] = $this->getArgument( $key );
1517 }
1518 return $arguments;
1519 }
1520
1521 function getNumberedArgument( $index ) {
1522 if ( !isset( $this->numberedArgs[$index] ) ) {
1523 return false;
1524 }
1525 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1526 # No trimming for unnamed arguments
1527 $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], PPFrame::STRIP_COMMENTS );
1528 }
1529 return $this->numberedExpansionCache[$index];
1530 }
1531
1532 function getNamedArgument( $name ) {
1533 if ( !isset( $this->namedArgs[$name] ) ) {
1534 return false;
1535 }
1536 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1537 # Trim named arguments post-expand, for backwards compatibility
1538 $this->namedExpansionCache[$name] = trim(
1539 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1540 }
1541 return $this->namedExpansionCache[$name];
1542 }
1543
1544 function getArgument( $name ) {
1545 $text = $this->getNumberedArgument( $name );
1546 if ( $text === false ) {
1547 $text = $this->getNamedArgument( $name );
1548 }
1549 return $text;
1550 }
1551
1552 /**
1553 * Return true if the frame is a template frame
1554 *
1555 * @return bool
1556 */
1557 function isTemplate() {
1558 return true;
1559 }
1560 }
1561
1562 /**
1563 * Expansion frame with custom arguments
1564 * @ingroup Parser
1565 */
1566 class PPCustomFrame_DOM extends PPFrame_DOM {
1567 var $args;
1568
1569 function __construct( $preprocessor, $args ) {
1570 parent::__construct( $preprocessor );
1571 $this->args = $args;
1572 }
1573
1574 function __toString() {
1575 $s = 'cstmframe{';
1576 $first = true;
1577 foreach ( $this->args as $name => $value ) {
1578 if ( $first ) {
1579 $first = false;
1580 } else {
1581 $s .= ', ';
1582 }
1583 $s .= "\"$name\":\"" .
1584 str_replace( '"', '\\"', $value->__toString() ) . '"';
1585 }
1586 $s .= '}';
1587 return $s;
1588 }
1589
1590 /**
1591 * @return bool
1592 */
1593 function isEmpty() {
1594 return !count( $this->args );
1595 }
1596
1597 function getArgument( $index ) {
1598 if ( !isset( $this->args[$index] ) ) {
1599 return false;
1600 }
1601 return $this->args[$index];
1602 }
1603
1604 function getArguments() {
1605 return $this->args;
1606 }
1607 }
1608
1609 /**
1610 * @ingroup Parser
1611 */
1612 class PPNode_DOM implements PPNode {
1613
1614 /**
1615 * @var DOMElement
1616 */
1617 var $node;
1618 var $xpath;
1619
1620 function __construct( $node, $xpath = false ) {
1621 $this->node = $node;
1622 }
1623
1624 /**
1625 * @return DOMXPath
1626 */
1627 function getXPath() {
1628 if ( $this->xpath === null ) {
1629 $this->xpath = new DOMXPath( $this->node->ownerDocument );
1630 }
1631 return $this->xpath;
1632 }
1633
1634 function __toString() {
1635 if ( $this->node instanceof DOMNodeList ) {
1636 $s = '';
1637 foreach ( $this->node as $node ) {
1638 $s .= $node->ownerDocument->saveXML( $node );
1639 }
1640 } else {
1641 $s = $this->node->ownerDocument->saveXML( $this->node );
1642 }
1643 return $s;
1644 }
1645
1646 /**
1647 * @return bool|PPNode_DOM
1648 */
1649 function getChildren() {
1650 return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
1651 }
1652
1653 /**
1654 * @return bool|PPNode_DOM
1655 */
1656 function getFirstChild() {
1657 return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
1658 }
1659
1660 /**
1661 * @return bool|PPNode_DOM
1662 */
1663 function getNextSibling() {
1664 return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
1665 }
1666
1667 /**
1668 * @param string $type
1669 *
1670 * @return bool|PPNode_DOM
1671 */
1672 function getChildrenOfType( $type ) {
1673 return new self( $this->getXPath()->query( $type, $this->node ) );
1674 }
1675
1676 /**
1677 * @return int
1678 */
1679 function getLength() {
1680 if ( $this->node instanceof DOMNodeList ) {
1681 return $this->node->length;
1682 } else {
1683 return false;
1684 }
1685 }
1686
1687 /**
1688 * @param int $i
1689 * @return bool|PPNode_DOM
1690 */
1691 function item( $i ) {
1692 $item = $this->node->item( $i );
1693 return $item ? new self( $item ) : false;
1694 }
1695
1696 /**
1697 * @return string
1698 */
1699 function getName() {
1700 if ( $this->node instanceof DOMNodeList ) {
1701 return '#nodelist';
1702 } else {
1703 return $this->node->nodeName;
1704 }
1705 }
1706
1707 /**
1708 * Split a "<part>" node into an associative array containing:
1709 * - name PPNode name
1710 * - index String index
1711 * - value PPNode value
1712 *
1713 * @throws MWException
1714 * @return array
1715 */
1716 function splitArg() {
1717 $xpath = $this->getXPath();
1718 $names = $xpath->query( 'name', $this->node );
1719 $values = $xpath->query( 'value', $this->node );
1720 if ( !$names->length || !$values->length ) {
1721 throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1722 }
1723 $name = $names->item( 0 );
1724 $index = $name->getAttribute( 'index' );
1725 return array(
1726 'name' => new self( $name ),
1727 'index' => $index,
1728 'value' => new self( $values->item( 0 ) ) );
1729 }
1730
1731 /**
1732 * Split an "<ext>" node into an associative array containing name, attr, inner and close
1733 * All values in the resulting array are PPNodes. Inner and close are optional.
1734 *
1735 * @throws MWException
1736 * @return array
1737 */
1738 function splitExt() {
1739 $xpath = $this->getXPath();
1740 $names = $xpath->query( 'name', $this->node );
1741 $attrs = $xpath->query( 'attr', $this->node );
1742 $inners = $xpath->query( 'inner', $this->node );
1743 $closes = $xpath->query( 'close', $this->node );
1744 if ( !$names->length || !$attrs->length ) {
1745 throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1746 }
1747 $parts = array(
1748 'name' => new self( $names->item( 0 ) ),
1749 'attr' => new self( $attrs->item( 0 ) ) );
1750 if ( $inners->length ) {
1751 $parts['inner'] = new self( $inners->item( 0 ) );
1752 }
1753 if ( $closes->length ) {
1754 $parts['close'] = new self( $closes->item( 0 ) );
1755 }
1756 return $parts;
1757 }
1758
1759 /**
1760 * Split a "<h>" node
1761 * @throws MWException
1762 * @return array
1763 */
1764 function splitHeading() {
1765 if ( $this->getName() !== 'h' ) {
1766 throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1767 }
1768 return array(
1769 'i' => $this->node->getAttribute( 'i' ),
1770 'level' => $this->node->getAttribute( 'level' ),
1771 'contents' => $this->getChildren()
1772 );
1773 }
1774 }