Merge "User::pingLimiter() profiles per action as well"
[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>"
77 . htmlspecialchars( $val ) . "</value></part>";
78 } else {
79 $xml .= "<part><name>" . htmlspecialchars( $k )
80 . "</name>=<value>" . htmlspecialchars( $val ) . "</value></part>";
81 }
82 }
83
84 $xml .= "</list>";
85
86 wfProfileIn( __METHOD__ . '-loadXML' );
87 $dom = new DOMDocument();
88 wfSuppressWarnings();
89 $result = $dom->loadXML( $xml );
90 wfRestoreWarnings();
91 if ( !$result ) {
92 // Try running the XML through UtfNormal to get rid of invalid characters
93 $xml = UtfNormal::cleanUp( $xml );
94 // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
95 // don't barf when the XML is >256 levels deep
96 $result = $dom->loadXML( $xml, 1 << 19 );
97 }
98 wfProfileOut( __METHOD__ . '-loadXML' );
99
100 if ( !$result ) {
101 throw new MWException( 'Parameters passed to ' . __METHOD__ . ' result in invalid XML' );
102 }
103
104 $root = $dom->documentElement;
105 $node = new PPNode_DOM( $root->childNodes );
106 return $node;
107 }
108
109 /**
110 * @throws MWException
111 * @return bool
112 */
113 function memCheck() {
114 if ( $this->memoryLimit === false ) {
115 return true;
116 }
117 $usage = memory_get_usage();
118 if ( $usage > $this->memoryLimit * 0.9 ) {
119 $limit = intval( $this->memoryLimit * 0.9 / 1048576 + 0.5 );
120 throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
121 }
122 return $usage <= $this->memoryLimit * 0.8;
123 }
124
125 /**
126 * Preprocess some wikitext and return the document tree.
127 * This is the ghost of Parser::replace_variables().
128 *
129 * @param string $text The text to parse
130 * @param int $flags Bitwise combination of:
131 * Parser::PTD_FOR_INCLUSION Handle "<noinclude>" and "<includeonly>"
132 * as if the text is being included. Default
133 * is to assume a direct page view.
134 *
135 * The generated DOM tree must depend only on the input text and the flags.
136 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
137 *
138 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
139 * change in the DOM tree for a given text, must be passed through the section identifier
140 * in the section edit link and thus back to extractSections().
141 *
142 * The output of this function is currently only cached in process memory, but a persistent
143 * cache may be implemented at a later date which takes further advantage of these strict
144 * dependency requirements.
145 *
146 * @throws MWException
147 * @return PPNode_DOM
148 */
149 function preprocessToObj( $text, $flags = 0 ) {
150 wfProfileIn( __METHOD__ );
151 global $wgMemc, $wgPreprocessorCacheThreshold;
152
153 $xml = false;
154 $cacheable = ( $wgPreprocessorCacheThreshold !== false
155 && strlen( $text ) > $wgPreprocessorCacheThreshold );
156 if ( $cacheable ) {
157 wfProfileIn( __METHOD__ . '-cacheable' );
158
159 $cacheKey = wfMemcKey( 'preprocess-xml', md5( $text ), $flags );
160 $cacheValue = $wgMemc->get( $cacheKey );
161 if ( $cacheValue ) {
162 $version = substr( $cacheValue, 0, 8 );
163 if ( intval( $version ) == self::CACHE_VERSION ) {
164 $xml = substr( $cacheValue, 8 );
165 // From the cache
166 wfDebugLog( "Preprocessor", "Loaded preprocessor XML from memcached (key $cacheKey)" );
167 }
168 }
169 if ( $xml === false ) {
170 wfProfileIn( __METHOD__ . '-cache-miss' );
171 $xml = $this->preprocessToXml( $text, $flags );
172 $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . $xml;
173 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
174 wfProfileOut( __METHOD__ . '-cache-miss' );
175 wfDebugLog( "Preprocessor", "Saved preprocessor XML to memcached (key $cacheKey)" );
176 }
177 } else {
178 $xml = $this->preprocessToXml( $text, $flags );
179 }
180
181 // Fail if the number of elements exceeds acceptable limits
182 // Do not attempt to generate the DOM
183 $this->parser->mGeneratedPPNodeCount += substr_count( $xml, '<' );
184 $max = $this->parser->mOptions->getMaxGeneratedPPNodeCount();
185 if ( $this->parser->mGeneratedPPNodeCount > $max ) {
186 if ( $cacheable ) {
187 wfProfileOut( __METHOD__ . '-cacheable' );
188 }
189 wfProfileOut( __METHOD__ );
190 throw new MWException( __METHOD__ . ': generated node count limit exceeded' );
191 }
192
193 wfProfileIn( __METHOD__ . '-loadXML' );
194 $dom = new DOMDocument;
195 wfSuppressWarnings();
196 $result = $dom->loadXML( $xml );
197 wfRestoreWarnings();
198 if ( !$result ) {
199 // Try running the XML through UtfNormal to get rid of invalid characters
200 $xml = UtfNormal::cleanUp( $xml );
201 // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
202 // don't barf when the XML is >256 levels deep.
203 $result = $dom->loadXML( $xml, 1 << 19 );
204 }
205 if ( $result ) {
206 $obj = new PPNode_DOM( $dom->documentElement );
207 }
208 wfProfileOut( __METHOD__ . '-loadXML' );
209
210 if ( $cacheable ) {
211 wfProfileOut( __METHOD__ . '-cacheable' );
212 }
213
214 wfProfileOut( __METHOD__ );
215
216 if ( !$result ) {
217 throw new MWException( __METHOD__ . ' generated invalid XML' );
218 }
219 return $obj;
220 }
221
222 /**
223 * @param string $text
224 * @param int $flags
225 * @return string
226 */
227 function preprocessToXml( $text, $flags = 0 ) {
228 wfProfileIn( __METHOD__ );
229 $rules = array(
230 '{' => array(
231 'end' => '}',
232 'names' => array(
233 2 => 'template',
234 3 => 'tplarg',
235 ),
236 'min' => 2,
237 'max' => 3,
238 ),
239 '[' => array(
240 'end' => ']',
241 'names' => array( 2 => null ),
242 'min' => 2,
243 'max' => 2,
244 )
245 );
246
247 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
248
249 $xmlishElements = $this->parser->getStripList();
250 $enableOnlyinclude = false;
251 if ( $forInclusion ) {
252 $ignoredTags = array( 'includeonly', '/includeonly' );
253 $ignoredElements = array( 'noinclude' );
254 $xmlishElements[] = 'noinclude';
255 if ( strpos( $text, '<onlyinclude>' ) !== false
256 && strpos( $text, '</onlyinclude>' ) !== false
257 ) {
258 $enableOnlyinclude = true;
259 }
260 } else {
261 $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
262 $ignoredElements = array( 'includeonly' );
263 $xmlishElements[] = 'includeonly';
264 }
265 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
266
267 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
268 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
269
270 $stack = new PPDStack;
271
272 $searchBase = "[{<\n"; #}
273 // For fast reverse searches
274 $revText = strrev( $text );
275 $lengthText = strlen( $text );
276
277 // Input pointer, starts out pointing to a pseudo-newline before the start
278 $i = 0;
279 // Current accumulator
280 $accum =& $stack->getAccum();
281 $accum = '<root>';
282 // True to find equals signs in arguments
283 $findEquals = false;
284 // True to take notice of pipe characters
285 $findPipe = false;
286 $headingIndex = 1;
287 // True if $i is inside a possible heading
288 $inHeading = false;
289 // True if there are no more greater-than (>) signs right of $i
290 $noMoreGT = false;
291 // True to ignore all input up to the next <onlyinclude>
292 $findOnlyinclude = $enableOnlyinclude;
293 // Do a line-start run without outputting an LF character
294 $fakeLineStart = true;
295
296 while ( true ) {
297 //$this->memCheck();
298
299 if ( $findOnlyinclude ) {
300 // Ignore all input up to the next <onlyinclude>
301 $startPos = strpos( $text, '<onlyinclude>', $i );
302 if ( $startPos === false ) {
303 // Ignored section runs to the end
304 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
305 break;
306 }
307 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
308 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
309 $i = $tagEndPos;
310 $findOnlyinclude = false;
311 }
312
313 if ( $fakeLineStart ) {
314 $found = 'line-start';
315 $curChar = '';
316 } else {
317 # Find next opening brace, closing brace or pipe
318 $search = $searchBase;
319 if ( $stack->top === false ) {
320 $currentClosing = '';
321 } else {
322 $currentClosing = $stack->top->close;
323 $search .= $currentClosing;
324 }
325 if ( $findPipe ) {
326 $search .= '|';
327 }
328 if ( $findEquals ) {
329 // First equals will be for the template
330 $search .= '=';
331 }
332 $rule = null;
333 # Output literal section, advance input counter
334 $literalLength = strcspn( $text, $search, $i );
335 if ( $literalLength > 0 ) {
336 $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
337 $i += $literalLength;
338 }
339 if ( $i >= $lengthText ) {
340 if ( $currentClosing == "\n" ) {
341 // Do a past-the-end run to finish off the heading
342 $curChar = '';
343 $found = 'line-end';
344 } else {
345 # All done
346 break;
347 }
348 } else {
349 $curChar = $text[$i];
350 if ( $curChar == '|' ) {
351 $found = 'pipe';
352 } elseif ( $curChar == '=' ) {
353 $found = 'equals';
354 } elseif ( $curChar == '<' ) {
355 $found = 'angle';
356 } elseif ( $curChar == "\n" ) {
357 if ( $inHeading ) {
358 $found = 'line-end';
359 } else {
360 $found = 'line-start';
361 }
362 } elseif ( $curChar == $currentClosing ) {
363 $found = 'close';
364 } elseif ( isset( $rules[$curChar] ) ) {
365 $found = 'open';
366 $rule = $rules[$curChar];
367 } else {
368 # Some versions of PHP have a strcspn which stops on null characters
369 # Ignore and continue
370 ++$i;
371 continue;
372 }
373 }
374 }
375
376 if ( $found == 'angle' ) {
377 $matches = false;
378 // Handle </onlyinclude>
379 if ( $enableOnlyinclude
380 && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
381 ) {
382 $findOnlyinclude = true;
383 continue;
384 }
385
386 // Determine element name
387 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
388 // Element name missing or not listed
389 $accum .= '&lt;';
390 ++$i;
391 continue;
392 }
393 // Handle comments
394 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
395
396 // To avoid leaving blank lines, when a sequence of
397 // space-separated comments is both preceded and followed by
398 // a newline (ignoring spaces), then
399 // trim leading and trailing spaces and the trailing newline.
400
401 // Find the end
402 $endPos = strpos( $text, '-->', $i + 4 );
403 if ( $endPos === false ) {
404 // Unclosed comment in input, runs to end
405 $inner = substr( $text, $i );
406 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
407 $i = $lengthText;
408 } else {
409 // Search backwards for leading whitespace
410 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
411
412 // Search forwards for trailing whitespace
413 // $wsEnd will be the position of the last space (or the '>' if there's none)
414 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
415
416 // Keep looking forward as long as we're finding more
417 // comments.
418 $comments = array( array( $wsStart, $wsEnd ) );
419 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
420 $c = strpos( $text, '-->', $wsEnd + 4 );
421 if ( $c === false ) {
422 break;
423 }
424 $c = $c + 2 + strspn( $text, " \t", $c + 3 );
425 $comments[] = array( $wsEnd + 1, $c );
426 $wsEnd = $c;
427 }
428
429 // Eat the line if possible
430 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
431 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
432 // it's a possible beneficial b/c break.
433 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
434 && substr( $text, $wsEnd + 1, 1 ) == "\n"
435 ) {
436 // Remove leading whitespace from the end of the accumulator
437 // Sanity check first though
438 $wsLength = $i - $wsStart;
439 if ( $wsLength > 0
440 && strspn( $accum, " \t", -$wsLength ) === $wsLength
441 ) {
442 $accum = substr( $accum, 0, -$wsLength );
443 }
444
445 // Dump all but the last comment to the accumulator
446 foreach ( $comments as $j => $com ) {
447 $startPos = $com[0];
448 $endPos = $com[1] + 1;
449 if ( $j == ( count( $comments ) - 1 ) ) {
450 break;
451 }
452 $inner = substr( $text, $startPos, $endPos - $startPos );
453 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
454 }
455
456 // Do a line-start run next time to look for headings after the comment
457 $fakeLineStart = true;
458 } else {
459 // No line to eat, just take the comment itself
460 $startPos = $i;
461 $endPos += 2;
462 }
463
464 if ( $stack->top ) {
465 $part = $stack->top->getCurrentPart();
466 if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
467 $part->visualEnd = $wsStart;
468 }
469 // Else comments abutting, no change in visual end
470 $part->commentEnd = $endPos;
471 }
472 $i = $endPos + 1;
473 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
474 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
475 }
476 continue;
477 }
478 $name = $matches[1];
479 $lowerName = strtolower( $name );
480 $attrStart = $i + strlen( $name ) + 1;
481
482 // Find end of tag
483 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
484 if ( $tagEndPos === false ) {
485 // Infinite backtrack
486 // Disable tag search to prevent worst-case O(N^2) performance
487 $noMoreGT = true;
488 $accum .= '&lt;';
489 ++$i;
490 continue;
491 }
492
493 // Handle ignored tags
494 if ( in_array( $lowerName, $ignoredTags ) ) {
495 $accum .= '<ignore>'
496 . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) )
497 . '</ignore>';
498 $i = $tagEndPos + 1;
499 continue;
500 }
501
502 $tagStartPos = $i;
503 if ( $text[$tagEndPos - 1] == '/' ) {
504 $attrEnd = $tagEndPos - 1;
505 $inner = null;
506 $i = $tagEndPos + 1;
507 $close = '';
508 } else {
509 $attrEnd = $tagEndPos;
510 // Find closing tag
511 if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
512 $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
513 ) {
514 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
515 $i = $matches[0][1] + strlen( $matches[0][0] );
516 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
517 } else {
518 // No end tag -- let it run out to the end of the text.
519 $inner = substr( $text, $tagEndPos + 1 );
520 $i = $lengthText;
521 $close = '';
522 }
523 }
524 // <includeonly> and <noinclude> just become <ignore> tags
525 if ( in_array( $lowerName, $ignoredElements ) ) {
526 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
527 . '</ignore>';
528 continue;
529 }
530
531 $accum .= '<ext>';
532 if ( $attrEnd <= $attrStart ) {
533 $attr = '';
534 } else {
535 $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
536 }
537 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
538 // Note that the attr element contains the whitespace between name and attribute,
539 // this is necessary for precise reconstruction during pre-save transform.
540 '<attr>' . htmlspecialchars( $attr ) . '</attr>';
541 if ( $inner !== null ) {
542 $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
543 }
544 $accum .= $close . '</ext>';
545 } elseif ( $found == 'line-start' ) {
546 // Is this the start of a heading?
547 // Line break belongs before the heading element in any case
548 if ( $fakeLineStart ) {
549 $fakeLineStart = false;
550 } else {
551 $accum .= $curChar;
552 $i++;
553 }
554
555 $count = strspn( $text, '=', $i, 6 );
556 if ( $count == 1 && $findEquals ) {
557 // DWIM: This looks kind of like a name/value separator.
558 // Let's let the equals handler have it and break the
559 // potential heading. This is heuristic, but AFAICT the
560 // methods for completely correct disambiguation are very
561 // complex.
562 } elseif ( $count > 0 ) {
563 $piece = array(
564 'open' => "\n",
565 'close' => "\n",
566 'parts' => array( new PPDPart( str_repeat( '=', $count ) ) ),
567 'startPos' => $i,
568 'count' => $count );
569 $stack->push( $piece );
570 $accum =& $stack->getAccum();
571 $flags = $stack->getFlags();
572 extract( $flags );
573 $i += $count;
574 }
575 } elseif ( $found == 'line-end' ) {
576 $piece = $stack->top;
577 // A heading must be open, otherwise \n wouldn't have been in the search list
578 assert( '$piece->open == "\n"' );
579 $part = $piece->getCurrentPart();
580 // Search back through the input to see if it has a proper close.
581 // Do this using the reversed string since the other solutions
582 // (end anchor, etc.) are inefficient.
583 $wsLength = strspn( $revText, " \t", $lengthText - $i );
584 $searchStart = $i - $wsLength;
585 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
586 // Comment found at line end
587 // Search for equals signs before the comment
588 $searchStart = $part->visualEnd;
589 $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
590 }
591 $count = $piece->count;
592 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
593 if ( $equalsLength > 0 ) {
594 if ( $searchStart - $equalsLength == $piece->startPos ) {
595 // This is just a single string of equals signs on its own line
596 // Replicate the doHeadings behavior /={count}(.+)={count}/
597 // First find out how many equals signs there really are (don't stop at 6)
598 $count = $equalsLength;
599 if ( $count < 3 ) {
600 $count = 0;
601 } else {
602 $count = min( 6, intval( ( $count - 1 ) / 2 ) );
603 }
604 } else {
605 $count = min( $equalsLength, $count );
606 }
607 if ( $count > 0 ) {
608 // Normal match, output <h>
609 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
610 $headingIndex++;
611 } else {
612 // Single equals sign on its own line, count=0
613 $element = $accum;
614 }
615 } else {
616 // No match, no <h>, just pass down the inner text
617 $element = $accum;
618 }
619 // Unwind the stack
620 $stack->pop();
621 $accum =& $stack->getAccum();
622 $flags = $stack->getFlags();
623 extract( $flags );
624
625 // Append the result to the enclosing accumulator
626 $accum .= $element;
627 // Note that we do NOT increment the input pointer.
628 // This is because the closing linebreak could be the opening linebreak of
629 // another heading. Infinite loops are avoided because the next iteration MUST
630 // hit the heading open case above, which unconditionally increments the
631 // input pointer.
632 } elseif ( $found == 'open' ) {
633 # count opening brace characters
634 $count = strspn( $text, $curChar, $i );
635
636 # we need to add to stack only if opening brace count is enough for one of the rules
637 if ( $count >= $rule['min'] ) {
638 # Add it to the stack
639 $piece = array(
640 'open' => $curChar,
641 'close' => $rule['end'],
642 'count' => $count,
643 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
644 );
645
646 $stack->push( $piece );
647 $accum =& $stack->getAccum();
648 $flags = $stack->getFlags();
649 extract( $flags );
650 } else {
651 # Add literal brace(s)
652 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
653 }
654 $i += $count;
655 } elseif ( $found == 'close' ) {
656 $piece = $stack->top;
657 # lets check if there are enough characters for closing brace
658 $maxCount = $piece->count;
659 $count = strspn( $text, $curChar, $i, $maxCount );
660
661 # check for maximum matching characters (if there are 5 closing
662 # characters, we will probably need only 3 - depending on the rules)
663 $rule = $rules[$piece->open];
664 if ( $count > $rule['max'] ) {
665 # The specified maximum exists in the callback array, unless the caller
666 # has made an error
667 $matchingCount = $rule['max'];
668 } else {
669 # Count is less than the maximum
670 # Skip any gaps in the callback array to find the true largest match
671 # Need to use array_key_exists not isset because the callback can be null
672 $matchingCount = $count;
673 while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
674 --$matchingCount;
675 }
676 }
677
678 if ( $matchingCount <= 0 ) {
679 # No matching element found in callback array
680 # Output a literal closing brace and continue
681 $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
682 $i += $count;
683 continue;
684 }
685 $name = $rule['names'][$matchingCount];
686 if ( $name === null ) {
687 // No element, just literal text
688 $element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
689 } else {
690 # Create XML element
691 # Note: $parts is already XML, does not need to be encoded further
692 $parts = $piece->parts;
693 $title = $parts[0]->out;
694 unset( $parts[0] );
695
696 # The invocation is at the start of the line if lineStart is set in
697 # the stack, and all opening brackets are used up.
698 if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
699 $attr = ' lineStart="1"';
700 } else {
701 $attr = '';
702 }
703
704 $element = "<$name$attr>";
705 $element .= "<title>$title</title>";
706 $argIndex = 1;
707 foreach ( $parts as $part ) {
708 if ( isset( $part->eqpos ) ) {
709 $argName = substr( $part->out, 0, $part->eqpos );
710 $argValue = substr( $part->out, $part->eqpos + 1 );
711 $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
712 } else {
713 $element .= "<part><name index=\"$argIndex\" /><value>{$part->out}</value></part>";
714 $argIndex++;
715 }
716 }
717 $element .= "</$name>";
718 }
719
720 # Advance input pointer
721 $i += $matchingCount;
722
723 # Unwind the stack
724 $stack->pop();
725 $accum =& $stack->getAccum();
726
727 # Re-add the old stack element if it still has unmatched opening characters remaining
728 if ( $matchingCount < $piece->count ) {
729 $piece->parts = array( new PPDPart );
730 $piece->count -= $matchingCount;
731 # do we still qualify for any callback with remaining count?
732 $min = $rules[$piece->open]['min'];
733 if ( $piece->count >= $min ) {
734 $stack->push( $piece );
735 $accum =& $stack->getAccum();
736 } else {
737 $accum .= str_repeat( $piece->open, $piece->count );
738 }
739 }
740 $flags = $stack->getFlags();
741 extract( $flags );
742
743 # Add XML element to the enclosing accumulator
744 $accum .= $element;
745 } elseif ( $found == 'pipe' ) {
746 $findEquals = true; // shortcut for getFlags()
747 $stack->addPart();
748 $accum =& $stack->getAccum();
749 ++$i;
750 } elseif ( $found == 'equals' ) {
751 $findEquals = false; // shortcut for getFlags()
752 $stack->getCurrentPart()->eqpos = strlen( $accum );
753 $accum .= '=';
754 ++$i;
755 }
756 }
757
758 # Output any remaining unclosed brackets
759 foreach ( $stack->stack as $piece ) {
760 $stack->rootAccum .= $piece->breakSyntax();
761 }
762 $stack->rootAccum .= '</root>';
763 $xml = $stack->rootAccum;
764
765 wfProfileOut( __METHOD__ );
766
767 return $xml;
768 }
769 }
770
771 /**
772 * Stack class to help Preprocessor::preprocessToObj()
773 * @ingroup Parser
774 */
775 class PPDStack {
776 var $stack, $rootAccum;
777
778 /**
779 * @var PPDStack
780 */
781 var $top;
782 var $out;
783 var $elementClass = 'PPDStackElement';
784
785 static $false = false;
786
787 function __construct() {
788 $this->stack = array();
789 $this->top = false;
790 $this->rootAccum = '';
791 $this->accum =& $this->rootAccum;
792 }
793
794 /**
795 * @return int
796 */
797 function count() {
798 return count( $this->stack );
799 }
800
801 function &getAccum() {
802 return $this->accum;
803 }
804
805 function getCurrentPart() {
806 if ( $this->top === false ) {
807 return false;
808 } else {
809 return $this->top->getCurrentPart();
810 }
811 }
812
813 function push( $data ) {
814 if ( $data instanceof $this->elementClass ) {
815 $this->stack[] = $data;
816 } else {
817 $class = $this->elementClass;
818 $this->stack[] = new $class( $data );
819 }
820 $this->top = $this->stack[count( $this->stack ) - 1];
821 $this->accum =& $this->top->getAccum();
822 }
823
824 function pop() {
825 if ( !count( $this->stack ) ) {
826 throw new MWException( __METHOD__ . ': no elements remaining' );
827 }
828 $temp = array_pop( $this->stack );
829
830 if ( count( $this->stack ) ) {
831 $this->top = $this->stack[count( $this->stack ) - 1];
832 $this->accum =& $this->top->getAccum();
833 } else {
834 $this->top = self::$false;
835 $this->accum =& $this->rootAccum;
836 }
837 return $temp;
838 }
839
840 function addPart( $s = '' ) {
841 $this->top->addPart( $s );
842 $this->accum =& $this->top->getAccum();
843 }
844
845 /**
846 * @return array
847 */
848 function getFlags() {
849 if ( !count( $this->stack ) ) {
850 return array(
851 'findEquals' => false,
852 'findPipe' => false,
853 'inHeading' => false,
854 );
855 } else {
856 return $this->top->getFlags();
857 }
858 }
859 }
860
861 /**
862 * @ingroup Parser
863 */
864 class PPDStackElement {
865 var $open, // Opening character (\n for heading)
866 $close, // Matching closing character
867 $count, // Number of opening characters found (number of "=" for heading)
868 $parts, // Array of PPDPart objects describing pipe-separated parts.
869 $lineStart; // True if the open char appeared at the start of the input line. Not set for headings.
870
871 var $partClass = 'PPDPart';
872
873 function __construct( $data = array() ) {
874 $class = $this->partClass;
875 $this->parts = array( new $class );
876
877 foreach ( $data as $name => $value ) {
878 $this->$name = $value;
879 }
880 }
881
882 function &getAccum() {
883 return $this->parts[count( $this->parts ) - 1]->out;
884 }
885
886 function addPart( $s = '' ) {
887 $class = $this->partClass;
888 $this->parts[] = new $class( $s );
889 }
890
891 function getCurrentPart() {
892 return $this->parts[count( $this->parts ) - 1];
893 }
894
895 /**
896 * @return array
897 */
898 function getFlags() {
899 $partCount = count( $this->parts );
900 $findPipe = $this->open != "\n" && $this->open != '[';
901 return array(
902 'findPipe' => $findPipe,
903 'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts[$partCount - 1]->eqpos ),
904 'inHeading' => $this->open == "\n",
905 );
906 }
907
908 /**
909 * Get the output string that would result if the close is not found.
910 *
911 * @param bool|int $openingCount
912 * @return string
913 */
914 function breakSyntax( $openingCount = false ) {
915 if ( $this->open == "\n" ) {
916 $s = $this->parts[0]->out;
917 } else {
918 if ( $openingCount === false ) {
919 $openingCount = $this->count;
920 }
921 $s = str_repeat( $this->open, $openingCount );
922 $first = true;
923 foreach ( $this->parts as $part ) {
924 if ( $first ) {
925 $first = false;
926 } else {
927 $s .= '|';
928 }
929 $s .= $part->out;
930 }
931 }
932 return $s;
933 }
934 }
935
936 /**
937 * @ingroup Parser
938 */
939 class PPDPart {
940 var $out; // Output accumulator string
941
942 // Optional member variables:
943 // eqpos Position of equals sign in output accumulator
944 // commentEnd Past-the-end input pointer for the last comment encountered
945 // visualEnd Past-the-end input pointer for the end of the accumulator minus comments
946
947 function __construct( $out = '' ) {
948 $this->out = $out;
949 }
950 }
951
952 /**
953 * An expansion frame, used as a context to expand the result of preprocessToObj()
954 * @ingroup Parser
955 */
956 class PPFrame_DOM implements PPFrame {
957
958 /**
959 * @var Preprocessor
960 */
961 var $preprocessor;
962
963 /**
964 * @var Parser
965 */
966 var $parser;
967
968 /**
969 * @var Title
970 */
971 var $title;
972 var $titleCache;
973
974 /**
975 * Hashtable listing templates which are disallowed for expansion in this frame,
976 * having been encountered previously in parent frames.
977 */
978 var $loopCheckHash;
979
980 /**
981 * Recursion depth of this frame, top = 0
982 * Note that this is NOT the same as expansion depth in expand()
983 */
984 var $depth;
985
986 /**
987 * Construct a new preprocessor frame.
988 * @param Preprocessor $preprocessor The parent preprocessor
989 */
990 function __construct( $preprocessor ) {
991 $this->preprocessor = $preprocessor;
992 $this->parser = $preprocessor->parser;
993 $this->title = $this->parser->mTitle;
994 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
995 $this->loopCheckHash = array();
996 $this->depth = 0;
997 }
998
999 /**
1000 * Create a new child frame
1001 * $args is optionally a multi-root PPNode or array containing the template arguments
1002 *
1003 * @param bool|array $args
1004 * @param Title|bool $title
1005 * @param int $indexOffset
1006 * @return PPTemplateFrame_DOM
1007 */
1008 function newChild( $args = false, $title = false, $indexOffset = 0 ) {
1009 $namedArgs = array();
1010 $numberedArgs = array();
1011 if ( $title === false ) {
1012 $title = $this->title;
1013 }
1014 if ( $args !== false ) {
1015 $xpath = false;
1016 if ( $args instanceof PPNode ) {
1017 $args = $args->node;
1018 }
1019 foreach ( $args as $arg ) {
1020 if ( $arg instanceof PPNode ) {
1021 $arg = $arg->node;
1022 }
1023 if ( !$xpath ) {
1024 $xpath = new DOMXPath( $arg->ownerDocument );
1025 }
1026
1027 $nameNodes = $xpath->query( 'name', $arg );
1028 $value = $xpath->query( 'value', $arg );
1029 if ( $nameNodes->item( 0 )->hasAttributes() ) {
1030 // Numbered parameter
1031 $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
1032 $index = $index - $indexOffset;
1033 $numberedArgs[$index] = $value->item( 0 );
1034 unset( $namedArgs[$index] );
1035 } else {
1036 // Named parameter
1037 $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
1038 $namedArgs[$name] = $value->item( 0 );
1039 unset( $numberedArgs[$name] );
1040 }
1041 }
1042 }
1043 return new PPTemplateFrame_DOM( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
1044 }
1045
1046 /**
1047 * @throws MWException
1048 * @param string|PPNode_DOM|DOMDocument $root
1049 * @param int $flags
1050 * @return string
1051 */
1052 function expand( $root, $flags = 0 ) {
1053 static $expansionDepth = 0;
1054 if ( is_string( $root ) ) {
1055 return $root;
1056 }
1057
1058 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1059 $this->parser->limitationWarn( 'node-count-exceeded',
1060 $this->parser->mPPNodeCount,
1061 $this->parser->mOptions->getMaxPPNodeCount()
1062 );
1063 return '<span class="error">Node-count limit exceeded</span>';
1064 }
1065
1066 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1067 $this->parser->limitationWarn( 'expansion-depth-exceeded',
1068 $expansionDepth,
1069 $this->parser->mOptions->getMaxPPExpandDepth()
1070 );
1071 return '<span class="error">Expansion depth limit exceeded</span>';
1072 }
1073 wfProfileIn( __METHOD__ );
1074 ++$expansionDepth;
1075 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1076 $this->parser->mHighestExpansionDepth = $expansionDepth;
1077 }
1078
1079 if ( $root instanceof PPNode_DOM ) {
1080 $root = $root->node;
1081 }
1082 if ( $root instanceof DOMDocument ) {
1083 $root = $root->documentElement;
1084 }
1085
1086 $outStack = array( '', '' );
1087 $iteratorStack = array( false, $root );
1088 $indexStack = array( 0, 0 );
1089
1090 while ( count( $iteratorStack ) > 1 ) {
1091 $level = count( $outStack ) - 1;
1092 $iteratorNode =& $iteratorStack[$level];
1093 $out =& $outStack[$level];
1094 $index =& $indexStack[$level];
1095
1096 if ( $iteratorNode instanceof PPNode_DOM ) {
1097 $iteratorNode = $iteratorNode->node;
1098 }
1099
1100 if ( is_array( $iteratorNode ) ) {
1101 if ( $index >= count( $iteratorNode ) ) {
1102 // All done with this iterator
1103 $iteratorStack[$level] = false;
1104 $contextNode = false;
1105 } else {
1106 $contextNode = $iteratorNode[$index];
1107 $index++;
1108 }
1109 } elseif ( $iteratorNode instanceof DOMNodeList ) {
1110 if ( $index >= $iteratorNode->length ) {
1111 // All done with this iterator
1112 $iteratorStack[$level] = false;
1113 $contextNode = false;
1114 } else {
1115 $contextNode = $iteratorNode->item( $index );
1116 $index++;
1117 }
1118 } else {
1119 // Copy to $contextNode and then delete from iterator stack,
1120 // because this is not an iterator but we do have to execute it once
1121 $contextNode = $iteratorStack[$level];
1122 $iteratorStack[$level] = false;
1123 }
1124
1125 if ( $contextNode instanceof PPNode_DOM ) {
1126 $contextNode = $contextNode->node;
1127 }
1128
1129 $newIterator = false;
1130
1131 if ( $contextNode === false ) {
1132 // nothing to do
1133 } elseif ( is_string( $contextNode ) ) {
1134 $out .= $contextNode;
1135 } elseif ( is_array( $contextNode ) || $contextNode instanceof DOMNodeList ) {
1136 $newIterator = $contextNode;
1137 } elseif ( $contextNode instanceof DOMNode ) {
1138 if ( $contextNode->nodeType == XML_TEXT_NODE ) {
1139 $out .= $contextNode->nodeValue;
1140 } elseif ( $contextNode->nodeName == 'template' ) {
1141 # Double-brace expansion
1142 $xpath = new DOMXPath( $contextNode->ownerDocument );
1143 $titles = $xpath->query( 'title', $contextNode );
1144 $title = $titles->item( 0 );
1145 $parts = $xpath->query( 'part', $contextNode );
1146 if ( $flags & PPFrame::NO_TEMPLATES ) {
1147 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
1148 } else {
1149 $lineStart = $contextNode->getAttribute( 'lineStart' );
1150 $params = array(
1151 'title' => new PPNode_DOM( $title ),
1152 'parts' => new PPNode_DOM( $parts ),
1153 'lineStart' => $lineStart );
1154 $ret = $this->parser->braceSubstitution( $params, $this );
1155 if ( isset( $ret['object'] ) ) {
1156 $newIterator = $ret['object'];
1157 } else {
1158 $out .= $ret['text'];
1159 }
1160 }
1161 } elseif ( $contextNode->nodeName == 'tplarg' ) {
1162 # Triple-brace expansion
1163 $xpath = new DOMXPath( $contextNode->ownerDocument );
1164 $titles = $xpath->query( 'title', $contextNode );
1165 $title = $titles->item( 0 );
1166 $parts = $xpath->query( 'part', $contextNode );
1167 if ( $flags & PPFrame::NO_ARGS ) {
1168 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
1169 } else {
1170 $params = array(
1171 'title' => new PPNode_DOM( $title ),
1172 'parts' => new PPNode_DOM( $parts ) );
1173 $ret = $this->parser->argSubstitution( $params, $this );
1174 if ( isset( $ret['object'] ) ) {
1175 $newIterator = $ret['object'];
1176 } else {
1177 $out .= $ret['text'];
1178 }
1179 }
1180 } elseif ( $contextNode->nodeName == 'comment' ) {
1181 # HTML-style comment
1182 # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1183 if ( $this->parser->ot['html']
1184 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1185 || ( $flags & PPFrame::STRIP_COMMENTS )
1186 ) {
1187 $out .= '';
1188 } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1189 # Add a strip marker in PST mode so that pstPass2() can
1190 # run some old-fashioned regexes on the result.
1191 # Not in RECOVER_COMMENTS mode (extractSections) though.
1192 $out .= $this->parser->insertStripItem( $contextNode->textContent );
1193 } else {
1194 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1195 $out .= $contextNode->textContent;
1196 }
1197 } elseif ( $contextNode->nodeName == 'ignore' ) {
1198 # Output suppression used by <includeonly> etc.
1199 # OT_WIKI will only respect <ignore> in substed templates.
1200 # The other output types respect it unless NO_IGNORE is set.
1201 # extractSections() sets NO_IGNORE and so never respects it.
1202 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] )
1203 || ( $flags & PPFrame::NO_IGNORE )
1204 ) {
1205 $out .= $contextNode->textContent;
1206 } else {
1207 $out .= '';
1208 }
1209 } elseif ( $contextNode->nodeName == 'ext' ) {
1210 # Extension tag
1211 $xpath = new DOMXPath( $contextNode->ownerDocument );
1212 $names = $xpath->query( 'name', $contextNode );
1213 $attrs = $xpath->query( 'attr', $contextNode );
1214 $inners = $xpath->query( 'inner', $contextNode );
1215 $closes = $xpath->query( 'close', $contextNode );
1216 $params = array(
1217 'name' => new PPNode_DOM( $names->item( 0 ) ),
1218 'attr' => $attrs->length > 0 ? new PPNode_DOM( $attrs->item( 0 ) ) : null,
1219 'inner' => $inners->length > 0 ? new PPNode_DOM( $inners->item( 0 ) ) : null,
1220 'close' => $closes->length > 0 ? new PPNode_DOM( $closes->item( 0 ) ) : null,
1221 );
1222 $out .= $this->parser->extensionSubstitution( $params, $this );
1223 } elseif ( $contextNode->nodeName == 'h' ) {
1224 # Heading
1225 $s = $this->expand( $contextNode->childNodes, $flags );
1226
1227 # Insert a heading marker only for <h> children of <root>
1228 # This is to stop extractSections from going over multiple tree levels
1229 if ( $contextNode->parentNode->nodeName == 'root' && $this->parser->ot['html'] ) {
1230 # Insert heading index marker
1231 $headingIndex = $contextNode->getAttribute( 'i' );
1232 $titleText = $this->title->getPrefixedDBkey();
1233 $this->parser->mHeadings[] = array( $titleText, $headingIndex );
1234 $serial = count( $this->parser->mHeadings ) - 1;
1235 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1236 $count = $contextNode->getAttribute( 'level' );
1237 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
1238 $this->parser->mStripState->addGeneral( $marker, '' );
1239 }
1240 $out .= $s;
1241 } else {
1242 # Generic recursive expansion
1243 $newIterator = $contextNode->childNodes;
1244 }
1245 } else {
1246 wfProfileOut( __METHOD__ );
1247 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1248 }
1249
1250 if ( $newIterator !== false ) {
1251 if ( $newIterator instanceof PPNode_DOM ) {
1252 $newIterator = $newIterator->node;
1253 }
1254 $outStack[] = '';
1255 $iteratorStack[] = $newIterator;
1256 $indexStack[] = 0;
1257 } elseif ( $iteratorStack[$level] === false ) {
1258 // Return accumulated value to parent
1259 // With tail recursion
1260 while ( $iteratorStack[$level] === false && $level > 0 ) {
1261 $outStack[$level - 1] .= $out;
1262 array_pop( $outStack );
1263 array_pop( $iteratorStack );
1264 array_pop( $indexStack );
1265 $level--;
1266 }
1267 }
1268 }
1269 --$expansionDepth;
1270 wfProfileOut( __METHOD__ );
1271 return $outStack[0];
1272 }
1273
1274 /**
1275 * @param string $sep
1276 * @param int $flags
1277 * @return string
1278 */
1279 function implodeWithFlags( $sep, $flags /*, ... */ ) {
1280 $args = array_slice( func_get_args(), 2 );
1281
1282 $first = true;
1283 $s = '';
1284 foreach ( $args as $root ) {
1285 if ( $root instanceof PPNode_DOM ) {
1286 $root = $root->node;
1287 }
1288 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1289 $root = array( $root );
1290 }
1291 foreach ( $root as $node ) {
1292 if ( $first ) {
1293 $first = false;
1294 } else {
1295 $s .= $sep;
1296 }
1297 $s .= $this->expand( $node, $flags );
1298 }
1299 }
1300 return $s;
1301 }
1302
1303 /**
1304 * Implode with no flags specified
1305 * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1306 *
1307 * @param string $sep
1308 * @return string
1309 */
1310 function implode( $sep /*, ... */ ) {
1311 $args = array_slice( func_get_args(), 1 );
1312
1313 $first = true;
1314 $s = '';
1315 foreach ( $args as $root ) {
1316 if ( $root instanceof PPNode_DOM ) {
1317 $root = $root->node;
1318 }
1319 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1320 $root = array( $root );
1321 }
1322 foreach ( $root as $node ) {
1323 if ( $first ) {
1324 $first = false;
1325 } else {
1326 $s .= $sep;
1327 }
1328 $s .= $this->expand( $node );
1329 }
1330 }
1331 return $s;
1332 }
1333
1334 /**
1335 * Makes an object that, when expand()ed, will be the same as one obtained
1336 * with implode()
1337 *
1338 * @param string $sep
1339 * @return array
1340 */
1341 function virtualImplode( $sep /*, ... */ ) {
1342 $args = array_slice( func_get_args(), 1 );
1343 $out = array();
1344 $first = true;
1345
1346 foreach ( $args as $root ) {
1347 if ( $root instanceof PPNode_DOM ) {
1348 $root = $root->node;
1349 }
1350 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1351 $root = array( $root );
1352 }
1353 foreach ( $root as $node ) {
1354 if ( $first ) {
1355 $first = false;
1356 } else {
1357 $out[] = $sep;
1358 }
1359 $out[] = $node;
1360 }
1361 }
1362 return $out;
1363 }
1364
1365 /**
1366 * Virtual implode with brackets
1367 * @param string $start
1368 * @param string $sep
1369 * @param string $end
1370 * @return array
1371 */
1372 function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1373 $args = array_slice( func_get_args(), 3 );
1374 $out = array( $start );
1375 $first = true;
1376
1377 foreach ( $args as $root ) {
1378 if ( $root instanceof PPNode_DOM ) {
1379 $root = $root->node;
1380 }
1381 if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1382 $root = array( $root );
1383 }
1384 foreach ( $root as $node ) {
1385 if ( $first ) {
1386 $first = false;
1387 } else {
1388 $out[] = $sep;
1389 }
1390 $out[] = $node;
1391 }
1392 }
1393 $out[] = $end;
1394 return $out;
1395 }
1396
1397 function __toString() {
1398 return 'frame{}';
1399 }
1400
1401 function getPDBK( $level = false ) {
1402 if ( $level === false ) {
1403 return $this->title->getPrefixedDBkey();
1404 } else {
1405 return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1406 }
1407 }
1408
1409 /**
1410 * @return array
1411 */
1412 function getArguments() {
1413 return array();
1414 }
1415
1416 /**
1417 * @return array
1418 */
1419 function getNumberedArguments() {
1420 return array();
1421 }
1422
1423 /**
1424 * @return array
1425 */
1426 function getNamedArguments() {
1427 return array();
1428 }
1429
1430 /**
1431 * Returns true if there are no arguments in this frame
1432 *
1433 * @return bool
1434 */
1435 function isEmpty() {
1436 return true;
1437 }
1438
1439 function getArgument( $name ) {
1440 return false;
1441 }
1442
1443 /**
1444 * Returns true if the infinite loop check is OK, false if a loop is detected
1445 *
1446 * @param Title $title
1447 * @return bool
1448 */
1449 function loopCheck( $title ) {
1450 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1451 }
1452
1453 /**
1454 * Return true if the frame is a template frame
1455 *
1456 * @return bool
1457 */
1458 function isTemplate() {
1459 return false;
1460 }
1461
1462 /**
1463 * Get a title of frame
1464 *
1465 * @return Title
1466 */
1467 function getTitle() {
1468 return $this->title;
1469 }
1470 }
1471
1472 /**
1473 * Expansion frame with template arguments
1474 * @ingroup Parser
1475 */
1476 class PPTemplateFrame_DOM extends PPFrame_DOM {
1477 var $numberedArgs, $namedArgs;
1478
1479 /**
1480 * @var PPFrame_DOM
1481 */
1482 var $parent;
1483 var $numberedExpansionCache, $namedExpansionCache;
1484
1485 /**
1486 * @param Preprocessor $preprocessor
1487 * @param bool|PPFrame_DOM $parent
1488 * @param array $numberedArgs
1489 * @param array $namedArgs
1490 * @param bool|Title $title
1491 */
1492 function __construct( $preprocessor, $parent = false, $numberedArgs = array(),
1493 $namedArgs = array(), $title = false
1494 ) {
1495 parent::__construct( $preprocessor );
1496
1497 $this->parent = $parent;
1498 $this->numberedArgs = $numberedArgs;
1499 $this->namedArgs = $namedArgs;
1500 $this->title = $title;
1501 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1502 $this->titleCache = $parent->titleCache;
1503 $this->titleCache[] = $pdbk;
1504 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1505 if ( $pdbk !== false ) {
1506 $this->loopCheckHash[$pdbk] = true;
1507 }
1508 $this->depth = $parent->depth + 1;
1509 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1510 }
1511
1512 function __toString() {
1513 $s = 'tplframe{';
1514 $first = true;
1515 $args = $this->numberedArgs + $this->namedArgs;
1516 foreach ( $args as $name => $value ) {
1517 if ( $first ) {
1518 $first = false;
1519 } else {
1520 $s .= ', ';
1521 }
1522 $s .= "\"$name\":\"" .
1523 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
1524 }
1525 $s .= '}';
1526 return $s;
1527 }
1528
1529 /**
1530 * Returns true if there are no arguments in this frame
1531 *
1532 * @return bool
1533 */
1534 function isEmpty() {
1535 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1536 }
1537
1538 function getArguments() {
1539 $arguments = array();
1540 foreach ( array_merge(
1541 array_keys( $this->numberedArgs ),
1542 array_keys( $this->namedArgs ) ) as $key ) {
1543 $arguments[$key] = $this->getArgument( $key );
1544 }
1545 return $arguments;
1546 }
1547
1548 function getNumberedArguments() {
1549 $arguments = array();
1550 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1551 $arguments[$key] = $this->getArgument( $key );
1552 }
1553 return $arguments;
1554 }
1555
1556 function getNamedArguments() {
1557 $arguments = array();
1558 foreach ( array_keys( $this->namedArgs ) as $key ) {
1559 $arguments[$key] = $this->getArgument( $key );
1560 }
1561 return $arguments;
1562 }
1563
1564 function getNumberedArgument( $index ) {
1565 if ( !isset( $this->numberedArgs[$index] ) ) {
1566 return false;
1567 }
1568 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1569 # No trimming for unnamed arguments
1570 $this->numberedExpansionCache[$index] = $this->parent->expand(
1571 $this->numberedArgs[$index],
1572 PPFrame::STRIP_COMMENTS
1573 );
1574 }
1575 return $this->numberedExpansionCache[$index];
1576 }
1577
1578 function getNamedArgument( $name ) {
1579 if ( !isset( $this->namedArgs[$name] ) ) {
1580 return false;
1581 }
1582 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1583 # Trim named arguments post-expand, for backwards compatibility
1584 $this->namedExpansionCache[$name] = trim(
1585 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1586 }
1587 return $this->namedExpansionCache[$name];
1588 }
1589
1590 function getArgument( $name ) {
1591 $text = $this->getNumberedArgument( $name );
1592 if ( $text === false ) {
1593 $text = $this->getNamedArgument( $name );
1594 }
1595 return $text;
1596 }
1597
1598 /**
1599 * Return true if the frame is a template frame
1600 *
1601 * @return bool
1602 */
1603 function isTemplate() {
1604 return true;
1605 }
1606 }
1607
1608 /**
1609 * Expansion frame with custom arguments
1610 * @ingroup Parser
1611 */
1612 class PPCustomFrame_DOM extends PPFrame_DOM {
1613 var $args;
1614
1615 function __construct( $preprocessor, $args ) {
1616 parent::__construct( $preprocessor );
1617 $this->args = $args;
1618 }
1619
1620 function __toString() {
1621 $s = 'cstmframe{';
1622 $first = true;
1623 foreach ( $this->args as $name => $value ) {
1624 if ( $first ) {
1625 $first = false;
1626 } else {
1627 $s .= ', ';
1628 }
1629 $s .= "\"$name\":\"" .
1630 str_replace( '"', '\\"', $value->__toString() ) . '"';
1631 }
1632 $s .= '}';
1633 return $s;
1634 }
1635
1636 /**
1637 * @return bool
1638 */
1639 function isEmpty() {
1640 return !count( $this->args );
1641 }
1642
1643 function getArgument( $index ) {
1644 if ( !isset( $this->args[$index] ) ) {
1645 return false;
1646 }
1647 return $this->args[$index];
1648 }
1649
1650 function getArguments() {
1651 return $this->args;
1652 }
1653 }
1654
1655 /**
1656 * @ingroup Parser
1657 */
1658 class PPNode_DOM implements PPNode {
1659
1660 /**
1661 * @var DOMElement
1662 */
1663 var $node;
1664 var $xpath;
1665
1666 function __construct( $node, $xpath = false ) {
1667 $this->node = $node;
1668 }
1669
1670 /**
1671 * @return DOMXPath
1672 */
1673 function getXPath() {
1674 if ( $this->xpath === null ) {
1675 $this->xpath = new DOMXPath( $this->node->ownerDocument );
1676 }
1677 return $this->xpath;
1678 }
1679
1680 function __toString() {
1681 if ( $this->node instanceof DOMNodeList ) {
1682 $s = '';
1683 foreach ( $this->node as $node ) {
1684 $s .= $node->ownerDocument->saveXML( $node );
1685 }
1686 } else {
1687 $s = $this->node->ownerDocument->saveXML( $this->node );
1688 }
1689 return $s;
1690 }
1691
1692 /**
1693 * @return bool|PPNode_DOM
1694 */
1695 function getChildren() {
1696 return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
1697 }
1698
1699 /**
1700 * @return bool|PPNode_DOM
1701 */
1702 function getFirstChild() {
1703 return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
1704 }
1705
1706 /**
1707 * @return bool|PPNode_DOM
1708 */
1709 function getNextSibling() {
1710 return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
1711 }
1712
1713 /**
1714 * @param string $type
1715 *
1716 * @return bool|PPNode_DOM
1717 */
1718 function getChildrenOfType( $type ) {
1719 return new self( $this->getXPath()->query( $type, $this->node ) );
1720 }
1721
1722 /**
1723 * @return int
1724 */
1725 function getLength() {
1726 if ( $this->node instanceof DOMNodeList ) {
1727 return $this->node->length;
1728 } else {
1729 return false;
1730 }
1731 }
1732
1733 /**
1734 * @param int $i
1735 * @return bool|PPNode_DOM
1736 */
1737 function item( $i ) {
1738 $item = $this->node->item( $i );
1739 return $item ? new self( $item ) : false;
1740 }
1741
1742 /**
1743 * @return string
1744 */
1745 function getName() {
1746 if ( $this->node instanceof DOMNodeList ) {
1747 return '#nodelist';
1748 } else {
1749 return $this->node->nodeName;
1750 }
1751 }
1752
1753 /**
1754 * Split a "<part>" node into an associative array containing:
1755 * - name PPNode name
1756 * - index String index
1757 * - value PPNode value
1758 *
1759 * @throws MWException
1760 * @return array
1761 */
1762 function splitArg() {
1763 $xpath = $this->getXPath();
1764 $names = $xpath->query( 'name', $this->node );
1765 $values = $xpath->query( 'value', $this->node );
1766 if ( !$names->length || !$values->length ) {
1767 throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1768 }
1769 $name = $names->item( 0 );
1770 $index = $name->getAttribute( 'index' );
1771 return array(
1772 'name' => new self( $name ),
1773 'index' => $index,
1774 'value' => new self( $values->item( 0 ) ) );
1775 }
1776
1777 /**
1778 * Split an "<ext>" node into an associative array containing name, attr, inner and close
1779 * All values in the resulting array are PPNodes. Inner and close are optional.
1780 *
1781 * @throws MWException
1782 * @return array
1783 */
1784 function splitExt() {
1785 $xpath = $this->getXPath();
1786 $names = $xpath->query( 'name', $this->node );
1787 $attrs = $xpath->query( 'attr', $this->node );
1788 $inners = $xpath->query( 'inner', $this->node );
1789 $closes = $xpath->query( 'close', $this->node );
1790 if ( !$names->length || !$attrs->length ) {
1791 throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1792 }
1793 $parts = array(
1794 'name' => new self( $names->item( 0 ) ),
1795 'attr' => new self( $attrs->item( 0 ) ) );
1796 if ( $inners->length ) {
1797 $parts['inner'] = new self( $inners->item( 0 ) );
1798 }
1799 if ( $closes->length ) {
1800 $parts['close'] = new self( $closes->item( 0 ) );
1801 }
1802 return $parts;
1803 }
1804
1805 /**
1806 * Split a "<h>" node
1807 * @throws MWException
1808 * @return array
1809 */
1810 function splitHeading() {
1811 if ( $this->getName() !== 'h' ) {
1812 throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1813 }
1814 return array(
1815 'i' => $this->node->getAttribute( 'i' ),
1816 'level' => $this->node->getAttribute( 'level' ),
1817 'contents' => $this->getChildren()
1818 );
1819 }
1820 }