Make PPFrame::RECOVER_COMMENTS actually work
[lhc/web/wiklou.git] / includes / parser / Preprocessor_Hash.php
1 <?php
2 /**
3 * Preprocessor using PHP arrays
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 * Differences from DOM schema:
26 * * attribute nodes are children
27 * * "<h>" nodes that aren't at the top are replaced with <possible-h>
28 * @ingroup Parser
29 * @codingStandardsIgnoreStart
30 */
31 class Preprocessor_Hash implements Preprocessor {
32 // @codingStandardsIgnoreEnd
33
34 /**
35 * @var Parser
36 */
37 public $parser;
38
39 const CACHE_VERSION = 1;
40
41 public function __construct( $parser ) {
42 $this->parser = $parser;
43 }
44
45 /**
46 * @return PPFrame_Hash
47 */
48 public function newFrame() {
49 return new PPFrame_Hash( $this );
50 }
51
52 /**
53 * @param array $args
54 * @return PPCustomFrame_Hash
55 */
56 public function newCustomFrame( $args ) {
57 return new PPCustomFrame_Hash( $this, $args );
58 }
59
60 /**
61 * @param array $values
62 * @return PPNode_Hash_Array
63 */
64 public function newPartNodeArray( $values ) {
65 $list = array();
66
67 foreach ( $values as $k => $val ) {
68 $partNode = new PPNode_Hash_Tree( 'part' );
69 $nameNode = new PPNode_Hash_Tree( 'name' );
70
71 if ( is_int( $k ) ) {
72 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $k ) );
73 $partNode->addChild( $nameNode );
74 } else {
75 $nameNode->addChild( new PPNode_Hash_Text( $k ) );
76 $partNode->addChild( $nameNode );
77 $partNode->addChild( new PPNode_Hash_Text( '=' ) );
78 }
79
80 $valueNode = new PPNode_Hash_Tree( 'value' );
81 $valueNode->addChild( new PPNode_Hash_Text( $val ) );
82 $partNode->addChild( $valueNode );
83
84 $list[] = $partNode;
85 }
86
87 $node = new PPNode_Hash_Array( $list );
88 return $node;
89 }
90
91 /**
92 * Preprocess some wikitext and return the document tree.
93 * This is the ghost of Parser::replace_variables().
94 *
95 * @param string $text The text to parse
96 * @param int $flags Bitwise combination of:
97 * Parser::PTD_FOR_INCLUSION Handle "<noinclude>" and "<includeonly>" as if the text is being
98 * included. Default is to assume a direct page view.
99 *
100 * The generated DOM tree must depend only on the input text and the flags.
101 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
102 *
103 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
104 * change in the DOM tree for a given text, must be passed through the section identifier
105 * in the section edit link and thus back to extractSections().
106 *
107 * The output of this function is currently only cached in process memory, but a persistent
108 * cache may be implemented at a later date which takes further advantage of these strict
109 * dependency requirements.
110 *
111 * @throws MWException
112 * @return PPNode_Hash_Tree
113 */
114 public function preprocessToObj( $text, $flags = 0 ) {
115 // Check cache.
116 global $wgMemc, $wgPreprocessorCacheThreshold;
117
118 $cacheable = $wgPreprocessorCacheThreshold !== false
119 && strlen( $text ) > $wgPreprocessorCacheThreshold;
120
121 if ( $cacheable ) {
122 $cacheKey = wfMemcKey( 'preprocess-hash', md5( $text ), $flags );
123 $cacheValue = $wgMemc->get( $cacheKey );
124 if ( $cacheValue ) {
125 $version = substr( $cacheValue, 0, 8 );
126 if ( intval( $version ) == self::CACHE_VERSION ) {
127 $hash = unserialize( substr( $cacheValue, 8 ) );
128 // From the cache
129 wfDebugLog( "Preprocessor",
130 "Loaded preprocessor hash from memcached (key $cacheKey)" );
131 return $hash;
132 }
133 }
134 }
135
136 $rules = array(
137 '{' => array(
138 'end' => '}',
139 'names' => array(
140 2 => 'template',
141 3 => 'tplarg',
142 ),
143 'min' => 2,
144 'max' => 3,
145 ),
146 '[' => array(
147 'end' => ']',
148 'names' => array( 2 => null ),
149 'min' => 2,
150 'max' => 2,
151 )
152 );
153
154 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
155
156 $xmlishElements = $this->parser->getStripList();
157 $enableOnlyinclude = false;
158 if ( $forInclusion ) {
159 $ignoredTags = array( 'includeonly', '/includeonly' );
160 $ignoredElements = array( 'noinclude' );
161 $xmlishElements[] = 'noinclude';
162 if ( strpos( $text, '<onlyinclude>' ) !== false
163 && strpos( $text, '</onlyinclude>' ) !== false
164 ) {
165 $enableOnlyinclude = true;
166 }
167 } else {
168 $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
169 $ignoredElements = array( 'includeonly' );
170 $xmlishElements[] = 'includeonly';
171 }
172 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
173
174 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
175 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
176
177 $stack = new PPDStack_Hash;
178
179 $searchBase = "[{<\n";
180 // For fast reverse searches
181 $revText = strrev( $text );
182 $lengthText = strlen( $text );
183
184 // Input pointer, starts out pointing to a pseudo-newline before the start
185 $i = 0;
186 // Current accumulator
187 $accum =& $stack->getAccum();
188 // True to find equals signs in arguments
189 $findEquals = false;
190 // True to take notice of pipe characters
191 $findPipe = false;
192 $headingIndex = 1;
193 // True if $i is inside a possible heading
194 $inHeading = false;
195 // True if there are no more greater-than (>) signs right of $i
196 $noMoreGT = false;
197 // True to ignore all input up to the next <onlyinclude>
198 $findOnlyinclude = $enableOnlyinclude;
199 // Do a line-start run without outputting an LF character
200 $fakeLineStart = true;
201
202 while ( true ) {
203 //$this->memCheck();
204
205 if ( $findOnlyinclude ) {
206 // Ignore all input up to the next <onlyinclude>
207 $startPos = strpos( $text, '<onlyinclude>', $i );
208 if ( $startPos === false ) {
209 // Ignored section runs to the end
210 $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
211 break;
212 }
213 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
214 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
215 $i = $tagEndPos;
216 $findOnlyinclude = false;
217 }
218
219 if ( $fakeLineStart ) {
220 $found = 'line-start';
221 $curChar = '';
222 } else {
223 # Find next opening brace, closing brace or pipe
224 $search = $searchBase;
225 if ( $stack->top === false ) {
226 $currentClosing = '';
227 } else {
228 $currentClosing = $stack->top->close;
229 $search .= $currentClosing;
230 }
231 if ( $findPipe ) {
232 $search .= '|';
233 }
234 if ( $findEquals ) {
235 // First equals will be for the template
236 $search .= '=';
237 }
238 $rule = null;
239 # Output literal section, advance input counter
240 $literalLength = strcspn( $text, $search, $i );
241 if ( $literalLength > 0 ) {
242 $accum->addLiteral( substr( $text, $i, $literalLength ) );
243 $i += $literalLength;
244 }
245 if ( $i >= $lengthText ) {
246 if ( $currentClosing == "\n" ) {
247 // Do a past-the-end run to finish off the heading
248 $curChar = '';
249 $found = 'line-end';
250 } else {
251 # All done
252 break;
253 }
254 } else {
255 $curChar = $text[$i];
256 if ( $curChar == '|' ) {
257 $found = 'pipe';
258 } elseif ( $curChar == '=' ) {
259 $found = 'equals';
260 } elseif ( $curChar == '<' ) {
261 $found = 'angle';
262 } elseif ( $curChar == "\n" ) {
263 if ( $inHeading ) {
264 $found = 'line-end';
265 } else {
266 $found = 'line-start';
267 }
268 } elseif ( $curChar == $currentClosing ) {
269 $found = 'close';
270 } elseif ( isset( $rules[$curChar] ) ) {
271 $found = 'open';
272 $rule = $rules[$curChar];
273 } else {
274 # Some versions of PHP have a strcspn which stops on null characters
275 # Ignore and continue
276 ++$i;
277 continue;
278 }
279 }
280 }
281
282 if ( $found == 'angle' ) {
283 $matches = false;
284 // Handle </onlyinclude>
285 if ( $enableOnlyinclude
286 && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
287 ) {
288 $findOnlyinclude = true;
289 continue;
290 }
291
292 // Determine element name
293 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
294 // Element name missing or not listed
295 $accum->addLiteral( '<' );
296 ++$i;
297 continue;
298 }
299 // Handle comments
300 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
301
302 // To avoid leaving blank lines, when a sequence of
303 // space-separated comments is both preceded and followed by
304 // a newline (ignoring spaces), then
305 // trim leading and trailing spaces and the trailing newline.
306
307 // Find the end
308 $endPos = strpos( $text, '-->', $i + 4 );
309 if ( $endPos === false ) {
310 // Unclosed comment in input, runs to end
311 $inner = substr( $text, $i );
312 $accum->addNodeWithText( 'comment', $inner );
313 $i = $lengthText;
314 } else {
315 // Search backwards for leading whitespace
316 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
317
318 // Search forwards for trailing whitespace
319 // $wsEnd will be the position of the last space (or the '>' if there's none)
320 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
321
322 // Keep looking forward as long as we're finding more
323 // comments.
324 $comments = array( array( $wsStart, $wsEnd ) );
325 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
326 $c = strpos( $text, '-->', $wsEnd + 4 );
327 if ( $c === false ) {
328 break;
329 }
330 $c = $c + 2 + strspn( $text, " \t", $c + 3 );
331 $comments[] = array( $wsEnd + 1, $c );
332 $wsEnd = $c;
333 }
334
335 // Eat the line if possible
336 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
337 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
338 // it's a possible beneficial b/c break.
339 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
340 && substr( $text, $wsEnd + 1, 1 ) == "\n"
341 ) {
342 // Remove leading whitespace from the end of the accumulator
343 // Sanity check first though
344 $wsLength = $i - $wsStart;
345 if ( $wsLength > 0
346 && $accum->lastNode instanceof PPNode_Hash_Text
347 && strspn( $accum->lastNode->value, " \t", -$wsLength ) === $wsLength
348 ) {
349 $accum->lastNode->value = substr( $accum->lastNode->value, 0, -$wsLength );
350 }
351
352 // Dump all but the last comment to the accumulator
353 foreach ( $comments as $j => $com ) {
354 $startPos = $com[0];
355 $endPos = $com[1] + 1;
356 if ( $j == ( count( $comments ) - 1 ) ) {
357 break;
358 }
359 $inner = substr( $text, $startPos, $endPos - $startPos );
360 $accum->addNodeWithText( 'comment', $inner );
361 }
362
363 // Do a line-start run next time to look for headings after the comment
364 $fakeLineStart = true;
365 } else {
366 // No line to eat, just take the comment itself
367 $startPos = $i;
368 $endPos += 2;
369 }
370
371 if ( $stack->top ) {
372 $part = $stack->top->getCurrentPart();
373 if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
374 $part->visualEnd = $wsStart;
375 }
376 // Else comments abutting, no change in visual end
377 $part->commentEnd = $endPos;
378 }
379 $i = $endPos + 1;
380 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
381 $accum->addNodeWithText( 'comment', $inner );
382 }
383 continue;
384 }
385 $name = $matches[1];
386 $lowerName = strtolower( $name );
387 $attrStart = $i + strlen( $name ) + 1;
388
389 // Find end of tag
390 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
391 if ( $tagEndPos === false ) {
392 // Infinite backtrack
393 // Disable tag search to prevent worst-case O(N^2) performance
394 $noMoreGT = true;
395 $accum->addLiteral( '<' );
396 ++$i;
397 continue;
398 }
399
400 // Handle ignored tags
401 if ( in_array( $lowerName, $ignoredTags ) ) {
402 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i + 1 ) );
403 $i = $tagEndPos + 1;
404 continue;
405 }
406
407 $tagStartPos = $i;
408 if ( $text[$tagEndPos - 1] == '/' ) {
409 // Short end tag
410 $attrEnd = $tagEndPos - 1;
411 $inner = null;
412 $i = $tagEndPos + 1;
413 $close = null;
414 } else {
415 $attrEnd = $tagEndPos;
416 // Find closing tag
417 if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
418 $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
419 ) {
420 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
421 $i = $matches[0][1] + strlen( $matches[0][0] );
422 $close = $matches[0][0];
423 } else {
424 // No end tag -- let it run out to the end of the text.
425 $inner = substr( $text, $tagEndPos + 1 );
426 $i = $lengthText;
427 $close = null;
428 }
429 }
430 // <includeonly> and <noinclude> just become <ignore> tags
431 if ( in_array( $lowerName, $ignoredElements ) ) {
432 $accum->addNodeWithText( 'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
433 continue;
434 }
435
436 if ( $attrEnd <= $attrStart ) {
437 $attr = '';
438 } else {
439 // Note that the attr element contains the whitespace between name and attribute,
440 // this is necessary for precise reconstruction during pre-save transform.
441 $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
442 }
443
444 $extNode = new PPNode_Hash_Tree( 'ext' );
445 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'name', $name ) );
446 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'attr', $attr ) );
447 if ( $inner !== null ) {
448 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'inner', $inner ) );
449 }
450 if ( $close !== null ) {
451 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'close', $close ) );
452 }
453 $accum->addNode( $extNode );
454 } elseif ( $found == 'line-start' ) {
455 // Is this the start of a heading?
456 // Line break belongs before the heading element in any case
457 if ( $fakeLineStart ) {
458 $fakeLineStart = false;
459 } else {
460 $accum->addLiteral( $curChar );
461 $i++;
462 }
463
464 $count = strspn( $text, '=', $i, 6 );
465 if ( $count == 1 && $findEquals ) {
466 // DWIM: This looks kind of like a name/value separator.
467 // Let's let the equals handler have it and break the potential
468 // heading. This is heuristic, but AFAICT the methods for
469 // completely correct disambiguation are very complex.
470 } elseif ( $count > 0 ) {
471 $piece = array(
472 'open' => "\n",
473 'close' => "\n",
474 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
475 'startPos' => $i,
476 'count' => $count );
477 $stack->push( $piece );
478 $accum =& $stack->getAccum();
479 extract( $stack->getFlags() );
480 $i += $count;
481 }
482 } elseif ( $found == 'line-end' ) {
483 $piece = $stack->top;
484 // A heading must be open, otherwise \n wouldn't have been in the search list
485 assert( '$piece->open == "\n"' );
486 $part = $piece->getCurrentPart();
487 // Search back through the input to see if it has a proper close.
488 // Do this using the reversed string since the other solutions
489 // (end anchor, etc.) are inefficient.
490 $wsLength = strspn( $revText, " \t", $lengthText - $i );
491 $searchStart = $i - $wsLength;
492 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
493 // Comment found at line end
494 // Search for equals signs before the comment
495 $searchStart = $part->visualEnd;
496 $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
497 }
498 $count = $piece->count;
499 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
500 if ( $equalsLength > 0 ) {
501 if ( $searchStart - $equalsLength == $piece->startPos ) {
502 // This is just a single string of equals signs on its own line
503 // Replicate the doHeadings behavior /={count}(.+)={count}/
504 // First find out how many equals signs there really are (don't stop at 6)
505 $count = $equalsLength;
506 if ( $count < 3 ) {
507 $count = 0;
508 } else {
509 $count = min( 6, intval( ( $count - 1 ) / 2 ) );
510 }
511 } else {
512 $count = min( $equalsLength, $count );
513 }
514 if ( $count > 0 ) {
515 // Normal match, output <h>
516 $element = new PPNode_Hash_Tree( 'possible-h' );
517 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
518 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++ ) );
519 $element->lastChild->nextSibling = $accum->firstNode;
520 $element->lastChild = $accum->lastNode;
521 } else {
522 // Single equals sign on its own line, count=0
523 $element = $accum;
524 }
525 } else {
526 // No match, no <h>, just pass down the inner text
527 $element = $accum;
528 }
529 // Unwind the stack
530 $stack->pop();
531 $accum =& $stack->getAccum();
532 extract( $stack->getFlags() );
533
534 // Append the result to the enclosing accumulator
535 if ( $element instanceof PPNode ) {
536 $accum->addNode( $element );
537 } else {
538 $accum->addAccum( $element );
539 }
540 // Note that we do NOT increment the input pointer.
541 // This is because the closing linebreak could be the opening linebreak of
542 // another heading. Infinite loops are avoided because the next iteration MUST
543 // hit the heading open case above, which unconditionally increments the
544 // input pointer.
545 } elseif ( $found == 'open' ) {
546 # count opening brace characters
547 $count = strspn( $text, $curChar, $i );
548
549 # we need to add to stack only if opening brace count is enough for one of the rules
550 if ( $count >= $rule['min'] ) {
551 # Add it to the stack
552 $piece = array(
553 'open' => $curChar,
554 'close' => $rule['end'],
555 'count' => $count,
556 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
557 );
558
559 $stack->push( $piece );
560 $accum =& $stack->getAccum();
561 extract( $stack->getFlags() );
562 } else {
563 # Add literal brace(s)
564 $accum->addLiteral( str_repeat( $curChar, $count ) );
565 }
566 $i += $count;
567 } elseif ( $found == 'close' ) {
568 $piece = $stack->top;
569 # lets check if there are enough characters for closing brace
570 $maxCount = $piece->count;
571 $count = strspn( $text, $curChar, $i, $maxCount );
572
573 # check for maximum matching characters (if there are 5 closing
574 # characters, we will probably need only 3 - depending on the rules)
575 $rule = $rules[$piece->open];
576 if ( $count > $rule['max'] ) {
577 # The specified maximum exists in the callback array, unless the caller
578 # has made an error
579 $matchingCount = $rule['max'];
580 } else {
581 # Count is less than the maximum
582 # Skip any gaps in the callback array to find the true largest match
583 # Need to use array_key_exists not isset because the callback can be null
584 $matchingCount = $count;
585 while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
586 --$matchingCount;
587 }
588 }
589
590 if ( $matchingCount <= 0 ) {
591 # No matching element found in callback array
592 # Output a literal closing brace and continue
593 $accum->addLiteral( str_repeat( $curChar, $count ) );
594 $i += $count;
595 continue;
596 }
597 $name = $rule['names'][$matchingCount];
598 if ( $name === null ) {
599 // No element, just literal text
600 $element = $piece->breakSyntax( $matchingCount );
601 $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
602 } else {
603 # Create XML element
604 # Note: $parts is already XML, does not need to be encoded further
605 $parts = $piece->parts;
606 $titleAccum = $parts[0]->out;
607 unset( $parts[0] );
608
609 $element = new PPNode_Hash_Tree( $name );
610
611 # The invocation is at the start of the line if lineStart is set in
612 # the stack, and all opening brackets are used up.
613 if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
614 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
615 }
616 $titleNode = new PPNode_Hash_Tree( 'title' );
617 $titleNode->firstChild = $titleAccum->firstNode;
618 $titleNode->lastChild = $titleAccum->lastNode;
619 $element->addChild( $titleNode );
620 $argIndex = 1;
621 foreach ( $parts as $part ) {
622 if ( isset( $part->eqpos ) ) {
623 // Find equals
624 $lastNode = false;
625 for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) {
626 if ( $node === $part->eqpos ) {
627 break;
628 }
629 $lastNode = $node;
630 }
631 if ( !$node ) {
632 if ( $cacheable ) {
633 }
634 throw new MWException( __METHOD__ . ': eqpos not found' );
635 }
636 if ( $node->name !== 'equals' ) {
637 if ( $cacheable ) {
638 }
639 throw new MWException( __METHOD__ . ': eqpos is not equals' );
640 }
641 $equalsNode = $node;
642
643 // Construct name node
644 $nameNode = new PPNode_Hash_Tree( 'name' );
645 if ( $lastNode !== false ) {
646 $lastNode->nextSibling = false;
647 $nameNode->firstChild = $part->out->firstNode;
648 $nameNode->lastChild = $lastNode;
649 }
650
651 // Construct value node
652 $valueNode = new PPNode_Hash_Tree( 'value' );
653 if ( $equalsNode->nextSibling !== false ) {
654 $valueNode->firstChild = $equalsNode->nextSibling;
655 $valueNode->lastChild = $part->out->lastNode;
656 }
657 $partNode = new PPNode_Hash_Tree( 'part' );
658 $partNode->addChild( $nameNode );
659 $partNode->addChild( $equalsNode->firstChild );
660 $partNode->addChild( $valueNode );
661 $element->addChild( $partNode );
662 } else {
663 $partNode = new PPNode_Hash_Tree( 'part' );
664 $nameNode = new PPNode_Hash_Tree( 'name' );
665 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++ ) );
666 $valueNode = new PPNode_Hash_Tree( 'value' );
667 $valueNode->firstChild = $part->out->firstNode;
668 $valueNode->lastChild = $part->out->lastNode;
669 $partNode->addChild( $nameNode );
670 $partNode->addChild( $valueNode );
671 $element->addChild( $partNode );
672 }
673 }
674 }
675
676 # Advance input pointer
677 $i += $matchingCount;
678
679 # Unwind the stack
680 $stack->pop();
681 $accum =& $stack->getAccum();
682
683 # Re-add the old stack element if it still has unmatched opening characters remaining
684 if ( $matchingCount < $piece->count ) {
685 $piece->parts = array( new PPDPart_Hash );
686 $piece->count -= $matchingCount;
687 # do we still qualify for any callback with remaining count?
688 $min = $rules[$piece->open]['min'];
689 if ( $piece->count >= $min ) {
690 $stack->push( $piece );
691 $accum =& $stack->getAccum();
692 } else {
693 $accum->addLiteral( str_repeat( $piece->open, $piece->count ) );
694 }
695 }
696
697 extract( $stack->getFlags() );
698
699 # Add XML element to the enclosing accumulator
700 if ( $element instanceof PPNode ) {
701 $accum->addNode( $element );
702 } else {
703 $accum->addAccum( $element );
704 }
705 } elseif ( $found == 'pipe' ) {
706 $findEquals = true; // shortcut for getFlags()
707 $stack->addPart();
708 $accum =& $stack->getAccum();
709 ++$i;
710 } elseif ( $found == 'equals' ) {
711 $findEquals = false; // shortcut for getFlags()
712 $accum->addNodeWithText( 'equals', '=' );
713 $stack->getCurrentPart()->eqpos = $accum->lastNode;
714 ++$i;
715 }
716 }
717
718 # Output any remaining unclosed brackets
719 foreach ( $stack->stack as $piece ) {
720 $stack->rootAccum->addAccum( $piece->breakSyntax() );
721 }
722
723 # Enable top-level headings
724 for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) {
725 if ( isset( $node->name ) && $node->name === 'possible-h' ) {
726 $node->name = 'h';
727 }
728 }
729
730 $rootNode = new PPNode_Hash_Tree( 'root' );
731 $rootNode->firstChild = $stack->rootAccum->firstNode;
732 $rootNode->lastChild = $stack->rootAccum->lastNode;
733
734 // Cache
735 if ( $cacheable ) {
736 $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . serialize( $rootNode );
737 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
738 wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" );
739 }
740
741 return $rootNode;
742 }
743 }
744
745 /**
746 * Stack class to help Preprocessor::preprocessToObj()
747 * @ingroup Parser
748 * @codingStandardsIgnoreStart
749 */
750 class PPDStack_Hash extends PPDStack {
751 // @codingStandardsIgnoreEnd
752
753 public function __construct() {
754 $this->elementClass = 'PPDStackElement_Hash';
755 parent::__construct();
756 $this->rootAccum = new PPDAccum_Hash;
757 }
758 }
759
760 /**
761 * @ingroup Parser
762 * @codingStandardsIgnoreStart
763 */
764 class PPDStackElement_Hash extends PPDStackElement {
765 // @codingStandardsIgnoreENd
766
767 public function __construct( $data = array() ) {
768 $this->partClass = 'PPDPart_Hash';
769 parent::__construct( $data );
770 }
771
772 /**
773 * Get the accumulator that would result if the close is not found.
774 *
775 * @param int|bool $openingCount
776 * @return PPDAccum_Hash
777 */
778 public function breakSyntax( $openingCount = false ) {
779 if ( $this->open == "\n" ) {
780 $accum = $this->parts[0]->out;
781 } else {
782 if ( $openingCount === false ) {
783 $openingCount = $this->count;
784 }
785 $accum = new PPDAccum_Hash;
786 $accum->addLiteral( str_repeat( $this->open, $openingCount ) );
787 $first = true;
788 foreach ( $this->parts as $part ) {
789 if ( $first ) {
790 $first = false;
791 } else {
792 $accum->addLiteral( '|' );
793 }
794 $accum->addAccum( $part->out );
795 }
796 }
797 return $accum;
798 }
799 }
800
801 /**
802 * @ingroup Parser
803 * @codingStandardsIgnoreStart
804 */
805 class PPDPart_Hash extends PPDPart {
806 // @codingStandardsIgnoreEnd
807
808 public function __construct( $out = '' ) {
809 $accum = new PPDAccum_Hash;
810 if ( $out !== '' ) {
811 $accum->addLiteral( $out );
812 }
813 parent::__construct( $accum );
814 }
815 }
816
817 /**
818 * @ingroup Parser
819 * @codingStandardsIgnoreStart
820 */
821 class PPDAccum_Hash {
822 // @codingStandardsIgnoreEnd
823
824 public $firstNode, $lastNode;
825
826 public function __construct() {
827 $this->firstNode = $this->lastNode = false;
828 }
829
830 /**
831 * Append a string literal
832 * @param string $s
833 */
834 public function addLiteral( $s ) {
835 if ( $this->lastNode === false ) {
836 $this->firstNode = $this->lastNode = new PPNode_Hash_Text( $s );
837 } elseif ( $this->lastNode instanceof PPNode_Hash_Text ) {
838 $this->lastNode->value .= $s;
839 } else {
840 $this->lastNode->nextSibling = new PPNode_Hash_Text( $s );
841 $this->lastNode = $this->lastNode->nextSibling;
842 }
843 }
844
845 /**
846 * Append a PPNode
847 * @param PPNode $node
848 */
849 public function addNode( PPNode $node ) {
850 if ( $this->lastNode === false ) {
851 $this->firstNode = $this->lastNode = $node;
852 } else {
853 $this->lastNode->nextSibling = $node;
854 $this->lastNode = $node;
855 }
856 }
857
858 /**
859 * Append a tree node with text contents
860 * @param string $name
861 * @param string $value
862 */
863 public function addNodeWithText( $name, $value ) {
864 $node = PPNode_Hash_Tree::newWithText( $name, $value );
865 $this->addNode( $node );
866 }
867
868 /**
869 * Append a PPDAccum_Hash
870 * Takes over ownership of the nodes in the source argument. These nodes may
871 * subsequently be modified, especially nextSibling.
872 * @param PPDAccum_Hash $accum
873 */
874 public function addAccum( $accum ) {
875 if ( $accum->lastNode === false ) {
876 // nothing to add
877 } elseif ( $this->lastNode === false ) {
878 $this->firstNode = $accum->firstNode;
879 $this->lastNode = $accum->lastNode;
880 } else {
881 $this->lastNode->nextSibling = $accum->firstNode;
882 $this->lastNode = $accum->lastNode;
883 }
884 }
885 }
886
887 /**
888 * An expansion frame, used as a context to expand the result of preprocessToObj()
889 * @ingroup Parser
890 * @codingStandardsIgnoreStart
891 */
892 class PPFrame_Hash implements PPFrame {
893 // @codingStandardsIgnoreEnd
894
895 /**
896 * @var Parser
897 */
898 public $parser;
899
900 /**
901 * @var Preprocessor
902 */
903 public $preprocessor;
904
905 /**
906 * @var Title
907 */
908 public $title;
909 public $titleCache;
910
911 /**
912 * Hashtable listing templates which are disallowed for expansion in this frame,
913 * having been encountered previously in parent frames.
914 */
915 public $loopCheckHash;
916
917 /**
918 * Recursion depth of this frame, top = 0
919 * Note that this is NOT the same as expansion depth in expand()
920 */
921 public $depth;
922
923 private $volatile = false;
924 private $ttl = null;
925
926 /**
927 * @var array
928 */
929 protected $childExpansionCache;
930
931 /**
932 * Construct a new preprocessor frame.
933 * @param Preprocessor $preprocessor The parent preprocessor
934 */
935 public function __construct( $preprocessor ) {
936 $this->preprocessor = $preprocessor;
937 $this->parser = $preprocessor->parser;
938 $this->title = $this->parser->mTitle;
939 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
940 $this->loopCheckHash = array();
941 $this->depth = 0;
942 $this->childExpansionCache = array();
943 }
944
945 /**
946 * Create a new child frame
947 * $args is optionally a multi-root PPNode or array containing the template arguments
948 *
949 * @param array|bool|PPNode_Hash_Array $args
950 * @param Title|bool $title
951 * @param int $indexOffset
952 * @throws MWException
953 * @return PPTemplateFrame_Hash
954 */
955 public function newChild( $args = false, $title = false, $indexOffset = 0 ) {
956 $namedArgs = array();
957 $numberedArgs = array();
958 if ( $title === false ) {
959 $title = $this->title;
960 }
961 if ( $args !== false ) {
962 if ( $args instanceof PPNode_Hash_Array ) {
963 $args = $args->value;
964 } elseif ( !is_array( $args ) ) {
965 throw new MWException( __METHOD__ . ': $args must be array or PPNode_Hash_Array' );
966 }
967 foreach ( $args as $arg ) {
968 $bits = $arg->splitArg();
969 if ( $bits['index'] !== '' ) {
970 // Numbered parameter
971 $index = $bits['index'] - $indexOffset;
972 if ( isset( $namedArgs[$index] ) || isset( $numberedArgs[$index] ) ) {
973 $this->parser->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
974 wfEscapeWikiText( $this->title ),
975 wfEscapeWikiText( $title ),
976 wfEscapeWikiText( $index ) )->text() );
977 $this->parser->addTrackingCategory( 'duplicate-args-category' );
978 }
979 $numberedArgs[$index] = $bits['value'];
980 unset( $namedArgs[$index] );
981 } else {
982 // Named parameter
983 $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
984 if ( isset( $namedArgs[$name] ) || isset( $numberedArgs[$name] ) ) {
985 $this->parser->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
986 wfEscapeWikiText( $this->title ),
987 wfEscapeWikiText( $title ),
988 wfEscapeWikiText( $name ) )->text() );
989 $this->parser->addTrackingCategory( 'duplicate-args-category' );
990 }
991 $namedArgs[$name] = $bits['value'];
992 unset( $numberedArgs[$name] );
993 }
994 }
995 }
996 return new PPTemplateFrame_Hash( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
997 }
998
999 /**
1000 * @throws MWException
1001 * @param string|int $key
1002 * @param string|PPNode $root
1003 * @param int $flags
1004 * @return string
1005 */
1006 public function cachedExpand( $key, $root, $flags = 0 ) {
1007 // we don't have a parent, so we don't have a cache
1008 return $this->expand( $root, $flags );
1009 }
1010
1011 /**
1012 * @throws MWException
1013 * @param string|PPNode $root
1014 * @param int $flags
1015 * @return string
1016 */
1017 public function expand( $root, $flags = 0 ) {
1018 static $expansionDepth = 0;
1019 if ( is_string( $root ) ) {
1020 return $root;
1021 }
1022
1023 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1024 $this->parser->limitationWarn( 'node-count-exceeded',
1025 $this->parser->mPPNodeCount,
1026 $this->parser->mOptions->getMaxPPNodeCount()
1027 );
1028 return '<span class="error">Node-count limit exceeded</span>';
1029 }
1030 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1031 $this->parser->limitationWarn( 'expansion-depth-exceeded',
1032 $expansionDepth,
1033 $this->parser->mOptions->getMaxPPExpandDepth()
1034 );
1035 return '<span class="error">Expansion depth limit exceeded</span>';
1036 }
1037 ++$expansionDepth;
1038 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1039 $this->parser->mHighestExpansionDepth = $expansionDepth;
1040 }
1041
1042 $outStack = array( '', '' );
1043 $iteratorStack = array( false, $root );
1044 $indexStack = array( 0, 0 );
1045
1046 while ( count( $iteratorStack ) > 1 ) {
1047 $level = count( $outStack ) - 1;
1048 $iteratorNode =& $iteratorStack[$level];
1049 $out =& $outStack[$level];
1050 $index =& $indexStack[$level];
1051
1052 if ( is_array( $iteratorNode ) ) {
1053 if ( $index >= count( $iteratorNode ) ) {
1054 // All done with this iterator
1055 $iteratorStack[$level] = false;
1056 $contextNode = false;
1057 } else {
1058 $contextNode = $iteratorNode[$index];
1059 $index++;
1060 }
1061 } elseif ( $iteratorNode instanceof PPNode_Hash_Array ) {
1062 if ( $index >= $iteratorNode->getLength() ) {
1063 // All done with this iterator
1064 $iteratorStack[$level] = false;
1065 $contextNode = false;
1066 } else {
1067 $contextNode = $iteratorNode->item( $index );
1068 $index++;
1069 }
1070 } else {
1071 // Copy to $contextNode and then delete from iterator stack,
1072 // because this is not an iterator but we do have to execute it once
1073 $contextNode = $iteratorStack[$level];
1074 $iteratorStack[$level] = false;
1075 }
1076
1077 $newIterator = false;
1078
1079 if ( $contextNode === false ) {
1080 // nothing to do
1081 } elseif ( is_string( $contextNode ) ) {
1082 $out .= $contextNode;
1083 } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_Hash_Array ) {
1084 $newIterator = $contextNode;
1085 } elseif ( $contextNode instanceof PPNode_Hash_Attr ) {
1086 // No output
1087 } elseif ( $contextNode instanceof PPNode_Hash_Text ) {
1088 $out .= $contextNode->value;
1089 } elseif ( $contextNode instanceof PPNode_Hash_Tree ) {
1090 if ( $contextNode->name == 'template' ) {
1091 # Double-brace expansion
1092 $bits = $contextNode->splitTemplate();
1093 if ( $flags & PPFrame::NO_TEMPLATES ) {
1094 $newIterator = $this->virtualBracketedImplode(
1095 '{{', '|', '}}',
1096 $bits['title'],
1097 $bits['parts']
1098 );
1099 } else {
1100 $ret = $this->parser->braceSubstitution( $bits, $this );
1101 if ( isset( $ret['object'] ) ) {
1102 $newIterator = $ret['object'];
1103 } else {
1104 $out .= $ret['text'];
1105 }
1106 }
1107 } elseif ( $contextNode->name == 'tplarg' ) {
1108 # Triple-brace expansion
1109 $bits = $contextNode->splitTemplate();
1110 if ( $flags & PPFrame::NO_ARGS ) {
1111 $newIterator = $this->virtualBracketedImplode(
1112 '{{{', '|', '}}}',
1113 $bits['title'],
1114 $bits['parts']
1115 );
1116 } else {
1117 $ret = $this->parser->argSubstitution( $bits, $this );
1118 if ( isset( $ret['object'] ) ) {
1119 $newIterator = $ret['object'];
1120 } else {
1121 $out .= $ret['text'];
1122 }
1123 }
1124 } elseif ( $contextNode->name == 'comment' ) {
1125 # HTML-style comment
1126 # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1127 # Not in RECOVER_COMMENTS mode (msgnw) though.
1128 if ( ( $this->parser->ot['html']
1129 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1130 || ( $flags & PPFrame::STRIP_COMMENTS )
1131 ) && !( $flags & PPFrame::RECOVER_COMMENTS )
1132 ) {
1133 $out .= '';
1134 } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1135 # Add a strip marker in PST mode so that pstPass2() can
1136 # run some old-fashioned regexes on the result.
1137 # Not in RECOVER_COMMENTS mode (extractSections) though.
1138 $out .= $this->parser->insertStripItem( $contextNode->firstChild->value );
1139 } else {
1140 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1141 $out .= $contextNode->firstChild->value;
1142 }
1143 } elseif ( $contextNode->name == 'ignore' ) {
1144 # Output suppression used by <includeonly> etc.
1145 # OT_WIKI will only respect <ignore> in substed templates.
1146 # The other output types respect it unless NO_IGNORE is set.
1147 # extractSections() sets NO_IGNORE and so never respects it.
1148 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] )
1149 || ( $flags & PPFrame::NO_IGNORE )
1150 ) {
1151 $out .= $contextNode->firstChild->value;
1152 } else {
1153 //$out .= '';
1154 }
1155 } elseif ( $contextNode->name == 'ext' ) {
1156 # Extension tag
1157 $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null );
1158 if ( $flags & PPFrame::NO_TAGS ) {
1159 $s = '<' . $bits['name']->firstChild->value;
1160 if ( $bits['attr'] ) {
1161 $s .= $bits['attr']->firstChild->value;
1162 }
1163 if ( $bits['inner'] ) {
1164 $s .= '>' . $bits['inner']->firstChild->value;
1165 if ( $bits['close'] ) {
1166 $s .= $bits['close']->firstChild->value;
1167 }
1168 } else {
1169 $s .= '/>';
1170 }
1171 $out .= $s;
1172 } else {
1173 $out .= $this->parser->extensionSubstitution( $bits, $this );
1174 }
1175 } elseif ( $contextNode->name == 'h' ) {
1176 # Heading
1177 if ( $this->parser->ot['html'] ) {
1178 # Expand immediately and insert heading index marker
1179 $s = '';
1180 for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) {
1181 $s .= $this->expand( $node, $flags );
1182 }
1183
1184 $bits = $contextNode->splitHeading();
1185 $titleText = $this->title->getPrefixedDBkey();
1186 $this->parser->mHeadings[] = array( $titleText, $bits['i'] );
1187 $serial = count( $this->parser->mHeadings ) - 1;
1188 $marker = Parser::MARKER_PREFIX . "-h-$serial-" . Parser::MARKER_SUFFIX;
1189 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
1190 $this->parser->mStripState->addGeneral( $marker, '' );
1191 $out .= $s;
1192 } else {
1193 # Expand in virtual stack
1194 $newIterator = $contextNode->getChildren();
1195 }
1196 } else {
1197 # Generic recursive expansion
1198 $newIterator = $contextNode->getChildren();
1199 }
1200 } else {
1201 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1202 }
1203
1204 if ( $newIterator !== false ) {
1205 $outStack[] = '';
1206 $iteratorStack[] = $newIterator;
1207 $indexStack[] = 0;
1208 } elseif ( $iteratorStack[$level] === false ) {
1209 // Return accumulated value to parent
1210 // With tail recursion
1211 while ( $iteratorStack[$level] === false && $level > 0 ) {
1212 $outStack[$level - 1] .= $out;
1213 array_pop( $outStack );
1214 array_pop( $iteratorStack );
1215 array_pop( $indexStack );
1216 $level--;
1217 }
1218 }
1219 }
1220 --$expansionDepth;
1221 return $outStack[0];
1222 }
1223
1224 /**
1225 * @param string $sep
1226 * @param int $flags
1227 * @param string|PPNode $args,...
1228 * @return string
1229 */
1230 public function implodeWithFlags( $sep, $flags /*, ... */ ) {
1231 $args = array_slice( func_get_args(), 2 );
1232
1233 $first = true;
1234 $s = '';
1235 foreach ( $args as $root ) {
1236 if ( $root instanceof PPNode_Hash_Array ) {
1237 $root = $root->value;
1238 }
1239 if ( !is_array( $root ) ) {
1240 $root = array( $root );
1241 }
1242 foreach ( $root as $node ) {
1243 if ( $first ) {
1244 $first = false;
1245 } else {
1246 $s .= $sep;
1247 }
1248 $s .= $this->expand( $node, $flags );
1249 }
1250 }
1251 return $s;
1252 }
1253
1254 /**
1255 * Implode with no flags specified
1256 * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1257 * @param string $sep
1258 * @param string|PPNode $args,...
1259 * @return string
1260 */
1261 public function implode( $sep /*, ... */ ) {
1262 $args = array_slice( func_get_args(), 1 );
1263
1264 $first = true;
1265 $s = '';
1266 foreach ( $args as $root ) {
1267 if ( $root instanceof PPNode_Hash_Array ) {
1268 $root = $root->value;
1269 }
1270 if ( !is_array( $root ) ) {
1271 $root = array( $root );
1272 }
1273 foreach ( $root as $node ) {
1274 if ( $first ) {
1275 $first = false;
1276 } else {
1277 $s .= $sep;
1278 }
1279 $s .= $this->expand( $node );
1280 }
1281 }
1282 return $s;
1283 }
1284
1285 /**
1286 * Makes an object that, when expand()ed, will be the same as one obtained
1287 * with implode()
1288 *
1289 * @param string $sep
1290 * @param string|PPNode $args,...
1291 * @return PPNode_Hash_Array
1292 */
1293 public function virtualImplode( $sep /*, ... */ ) {
1294 $args = array_slice( func_get_args(), 1 );
1295 $out = array();
1296 $first = true;
1297
1298 foreach ( $args as $root ) {
1299 if ( $root instanceof PPNode_Hash_Array ) {
1300 $root = $root->value;
1301 }
1302 if ( !is_array( $root ) ) {
1303 $root = array( $root );
1304 }
1305 foreach ( $root as $node ) {
1306 if ( $first ) {
1307 $first = false;
1308 } else {
1309 $out[] = $sep;
1310 }
1311 $out[] = $node;
1312 }
1313 }
1314 return new PPNode_Hash_Array( $out );
1315 }
1316
1317 /**
1318 * Virtual implode with brackets
1319 *
1320 * @param string $start
1321 * @param string $sep
1322 * @param string $end
1323 * @param string|PPNode $args,...
1324 * @return PPNode_Hash_Array
1325 */
1326 public function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1327 $args = array_slice( func_get_args(), 3 );
1328 $out = array( $start );
1329 $first = true;
1330
1331 foreach ( $args as $root ) {
1332 if ( $root instanceof PPNode_Hash_Array ) {
1333 $root = $root->value;
1334 }
1335 if ( !is_array( $root ) ) {
1336 $root = array( $root );
1337 }
1338 foreach ( $root as $node ) {
1339 if ( $first ) {
1340 $first = false;
1341 } else {
1342 $out[] = $sep;
1343 }
1344 $out[] = $node;
1345 }
1346 }
1347 $out[] = $end;
1348 return new PPNode_Hash_Array( $out );
1349 }
1350
1351 public function __toString() {
1352 return 'frame{}';
1353 }
1354
1355 /**
1356 * @param bool $level
1357 * @return array|bool|string
1358 */
1359 public function getPDBK( $level = false ) {
1360 if ( $level === false ) {
1361 return $this->title->getPrefixedDBkey();
1362 } else {
1363 return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1364 }
1365 }
1366
1367 /**
1368 * @return array
1369 */
1370 public function getArguments() {
1371 return array();
1372 }
1373
1374 /**
1375 * @return array
1376 */
1377 public function getNumberedArguments() {
1378 return array();
1379 }
1380
1381 /**
1382 * @return array
1383 */
1384 public function getNamedArguments() {
1385 return array();
1386 }
1387
1388 /**
1389 * Returns true if there are no arguments in this frame
1390 *
1391 * @return bool
1392 */
1393 public function isEmpty() {
1394 return true;
1395 }
1396
1397 /**
1398 * @param string $name
1399 * @return bool
1400 */
1401 public function getArgument( $name ) {
1402 return false;
1403 }
1404
1405 /**
1406 * Returns true if the infinite loop check is OK, false if a loop is detected
1407 *
1408 * @param Title $title
1409 *
1410 * @return bool
1411 */
1412 public function loopCheck( $title ) {
1413 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1414 }
1415
1416 /**
1417 * Return true if the frame is a template frame
1418 *
1419 * @return bool
1420 */
1421 public function isTemplate() {
1422 return false;
1423 }
1424
1425 /**
1426 * Get a title of frame
1427 *
1428 * @return Title
1429 */
1430 public function getTitle() {
1431 return $this->title;
1432 }
1433
1434 /**
1435 * Set the volatile flag
1436 *
1437 * @param bool $flag
1438 */
1439 public function setVolatile( $flag = true ) {
1440 $this->volatile = $flag;
1441 }
1442
1443 /**
1444 * Get the volatile flag
1445 *
1446 * @return bool
1447 */
1448 public function isVolatile() {
1449 return $this->volatile;
1450 }
1451
1452 /**
1453 * Set the TTL
1454 *
1455 * @param int $ttl
1456 */
1457 public function setTTL( $ttl ) {
1458 if ( $ttl !== null && ( $this->ttl === null || $ttl < $this->ttl ) ) {
1459 $this->ttl = $ttl;
1460 }
1461 }
1462
1463 /**
1464 * Get the TTL
1465 *
1466 * @return int|null
1467 */
1468 public function getTTL() {
1469 return $this->ttl;
1470 }
1471 }
1472
1473 /**
1474 * Expansion frame with template arguments
1475 * @ingroup Parser
1476 * @codingStandardsIgnoreStart
1477 */
1478 class PPTemplateFrame_Hash extends PPFrame_Hash {
1479 // @codingStandardsIgnoreEnd
1480
1481 public $numberedArgs, $namedArgs, $parent;
1482 public $numberedExpansionCache, $namedExpansionCache;
1483
1484 /**
1485 * @param Preprocessor $preprocessor
1486 * @param bool|PPFrame $parent
1487 * @param array $numberedArgs
1488 * @param array $namedArgs
1489 * @param bool|Title $title
1490 */
1491 public function __construct( $preprocessor, $parent = false, $numberedArgs = array(),
1492 $namedArgs = array(), $title = false
1493 ) {
1494 parent::__construct( $preprocessor );
1495
1496 $this->parent = $parent;
1497 $this->numberedArgs = $numberedArgs;
1498 $this->namedArgs = $namedArgs;
1499 $this->title = $title;
1500 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1501 $this->titleCache = $parent->titleCache;
1502 $this->titleCache[] = $pdbk;
1503 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1504 if ( $pdbk !== false ) {
1505 $this->loopCheckHash[$pdbk] = true;
1506 }
1507 $this->depth = $parent->depth + 1;
1508 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1509 }
1510
1511 public function __toString() {
1512 $s = 'tplframe{';
1513 $first = true;
1514 $args = $this->numberedArgs + $this->namedArgs;
1515 foreach ( $args as $name => $value ) {
1516 if ( $first ) {
1517 $first = false;
1518 } else {
1519 $s .= ', ';
1520 }
1521 $s .= "\"$name\":\"" .
1522 str_replace( '"', '\\"', $value->__toString() ) . '"';
1523 }
1524 $s .= '}';
1525 return $s;
1526 }
1527
1528 /**
1529 * @throws MWException
1530 * @param string|int $key
1531 * @param string|PPNode $root
1532 * @param int $flags
1533 * @return string
1534 */
1535 public function cachedExpand( $key, $root, $flags = 0 ) {
1536 if ( isset( $this->parent->childExpansionCache[$key] ) ) {
1537 return $this->parent->childExpansionCache[$key];
1538 }
1539 $retval = $this->expand( $root, $flags );
1540 if ( !$this->isVolatile() ) {
1541 $this->parent->childExpansionCache[$key] = $retval;
1542 }
1543 return $retval;
1544 }
1545
1546 /**
1547 * Returns true if there are no arguments in this frame
1548 *
1549 * @return bool
1550 */
1551 public function isEmpty() {
1552 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1553 }
1554
1555 /**
1556 * @return array
1557 */
1558 public function getArguments() {
1559 $arguments = array();
1560 foreach ( array_merge(
1561 array_keys( $this->numberedArgs ),
1562 array_keys( $this->namedArgs ) ) as $key ) {
1563 $arguments[$key] = $this->getArgument( $key );
1564 }
1565 return $arguments;
1566 }
1567
1568 /**
1569 * @return array
1570 */
1571 public function getNumberedArguments() {
1572 $arguments = array();
1573 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1574 $arguments[$key] = $this->getArgument( $key );
1575 }
1576 return $arguments;
1577 }
1578
1579 /**
1580 * @return array
1581 */
1582 public function getNamedArguments() {
1583 $arguments = array();
1584 foreach ( array_keys( $this->namedArgs ) as $key ) {
1585 $arguments[$key] = $this->getArgument( $key );
1586 }
1587 return $arguments;
1588 }
1589
1590 /**
1591 * @param int $index
1592 * @return array|bool
1593 */
1594 public function getNumberedArgument( $index ) {
1595 if ( !isset( $this->numberedArgs[$index] ) ) {
1596 return false;
1597 }
1598 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1599 # No trimming for unnamed arguments
1600 $this->numberedExpansionCache[$index] = $this->parent->expand(
1601 $this->numberedArgs[$index],
1602 PPFrame::STRIP_COMMENTS
1603 );
1604 }
1605 return $this->numberedExpansionCache[$index];
1606 }
1607
1608 /**
1609 * @param string $name
1610 * @return bool
1611 */
1612 public function getNamedArgument( $name ) {
1613 if ( !isset( $this->namedArgs[$name] ) ) {
1614 return false;
1615 }
1616 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1617 # Trim named arguments post-expand, for backwards compatibility
1618 $this->namedExpansionCache[$name] = trim(
1619 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1620 }
1621 return $this->namedExpansionCache[$name];
1622 }
1623
1624 /**
1625 * @param string $name
1626 * @return array|bool
1627 */
1628 public function getArgument( $name ) {
1629 $text = $this->getNumberedArgument( $name );
1630 if ( $text === false ) {
1631 $text = $this->getNamedArgument( $name );
1632 }
1633 return $text;
1634 }
1635
1636 /**
1637 * Return true if the frame is a template frame
1638 *
1639 * @return bool
1640 */
1641 public function isTemplate() {
1642 return true;
1643 }
1644
1645 public function setVolatile( $flag = true ) {
1646 parent::setVolatile( $flag );
1647 $this->parent->setVolatile( $flag );
1648 }
1649
1650 public function setTTL( $ttl ) {
1651 parent::setTTL( $ttl );
1652 $this->parent->setTTL( $ttl );
1653 }
1654 }
1655
1656 /**
1657 * Expansion frame with custom arguments
1658 * @ingroup Parser
1659 * @codingStandardsIgnoreStart
1660 */
1661 class PPCustomFrame_Hash extends PPFrame_Hash {
1662 // @codingStandardsIgnoreEnd
1663
1664 public $args;
1665
1666 public function __construct( $preprocessor, $args ) {
1667 parent::__construct( $preprocessor );
1668 $this->args = $args;
1669 }
1670
1671 public function __toString() {
1672 $s = 'cstmframe{';
1673 $first = true;
1674 foreach ( $this->args as $name => $value ) {
1675 if ( $first ) {
1676 $first = false;
1677 } else {
1678 $s .= ', ';
1679 }
1680 $s .= "\"$name\":\"" .
1681 str_replace( '"', '\\"', $value->__toString() ) . '"';
1682 }
1683 $s .= '}';
1684 return $s;
1685 }
1686
1687 /**
1688 * @return bool
1689 */
1690 public function isEmpty() {
1691 return !count( $this->args );
1692 }
1693
1694 /**
1695 * @param int $index
1696 * @return bool
1697 */
1698 public function getArgument( $index ) {
1699 if ( !isset( $this->args[$index] ) ) {
1700 return false;
1701 }
1702 return $this->args[$index];
1703 }
1704
1705 public function getArguments() {
1706 return $this->args;
1707 }
1708 }
1709
1710 /**
1711 * @ingroup Parser
1712 * @codingStandardsIgnoreStart
1713 */
1714 class PPNode_Hash_Tree implements PPNode {
1715 // @codingStandardsIgnoreEnd
1716
1717 public $name, $firstChild, $lastChild, $nextSibling;
1718
1719 public function __construct( $name ) {
1720 $this->name = $name;
1721 $this->firstChild = $this->lastChild = $this->nextSibling = false;
1722 }
1723
1724 public function __toString() {
1725 $inner = '';
1726 $attribs = '';
1727 for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) {
1728 if ( $node instanceof PPNode_Hash_Attr ) {
1729 $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"';
1730 } else {
1731 $inner .= $node->__toString();
1732 }
1733 }
1734 if ( $inner === '' ) {
1735 return "<{$this->name}$attribs/>";
1736 } else {
1737 return "<{$this->name}$attribs>$inner</{$this->name}>";
1738 }
1739 }
1740
1741 /**
1742 * @param string $name
1743 * @param string $text
1744 * @return PPNode_Hash_Tree
1745 */
1746 public static function newWithText( $name, $text ) {
1747 $obj = new self( $name );
1748 $obj->addChild( new PPNode_Hash_Text( $text ) );
1749 return $obj;
1750 }
1751
1752 public function addChild( $node ) {
1753 if ( $this->lastChild === false ) {
1754 $this->firstChild = $this->lastChild = $node;
1755 } else {
1756 $this->lastChild->nextSibling = $node;
1757 $this->lastChild = $node;
1758 }
1759 }
1760
1761 /**
1762 * @return PPNode_Hash_Array
1763 */
1764 public function getChildren() {
1765 $children = array();
1766 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1767 $children[] = $child;
1768 }
1769 return new PPNode_Hash_Array( $children );
1770 }
1771
1772 public function getFirstChild() {
1773 return $this->firstChild;
1774 }
1775
1776 public function getNextSibling() {
1777 return $this->nextSibling;
1778 }
1779
1780 public function getChildrenOfType( $name ) {
1781 $children = array();
1782 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1783 if ( isset( $child->name ) && $child->name === $name ) {
1784 $children[] = $child;
1785 }
1786 }
1787 return $children;
1788 }
1789
1790 /**
1791 * @return bool
1792 */
1793 public function getLength() {
1794 return false;
1795 }
1796
1797 /**
1798 * @param int $i
1799 * @return bool
1800 */
1801 public function item( $i ) {
1802 return false;
1803 }
1804
1805 /**
1806 * @return string
1807 */
1808 public function getName() {
1809 return $this->name;
1810 }
1811
1812 /**
1813 * Split a "<part>" node into an associative array containing:
1814 * - name PPNode name
1815 * - index String index
1816 * - value PPNode value
1817 *
1818 * @throws MWException
1819 * @return array
1820 */
1821 public function splitArg() {
1822 $bits = array();
1823 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1824 if ( !isset( $child->name ) ) {
1825 continue;
1826 }
1827 if ( $child->name === 'name' ) {
1828 $bits['name'] = $child;
1829 if ( $child->firstChild instanceof PPNode_Hash_Attr
1830 && $child->firstChild->name === 'index'
1831 ) {
1832 $bits['index'] = $child->firstChild->value;
1833 }
1834 } elseif ( $child->name === 'value' ) {
1835 $bits['value'] = $child;
1836 }
1837 }
1838
1839 if ( !isset( $bits['name'] ) ) {
1840 throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1841 }
1842 if ( !isset( $bits['index'] ) ) {
1843 $bits['index'] = '';
1844 }
1845 return $bits;
1846 }
1847
1848 /**
1849 * Split an "<ext>" node into an associative array containing name, attr, inner and close
1850 * All values in the resulting array are PPNodes. Inner and close are optional.
1851 *
1852 * @throws MWException
1853 * @return array
1854 */
1855 public function splitExt() {
1856 $bits = array();
1857 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1858 if ( !isset( $child->name ) ) {
1859 continue;
1860 }
1861 if ( $child->name == 'name' ) {
1862 $bits['name'] = $child;
1863 } elseif ( $child->name == 'attr' ) {
1864 $bits['attr'] = $child;
1865 } elseif ( $child->name == 'inner' ) {
1866 $bits['inner'] = $child;
1867 } elseif ( $child->name == 'close' ) {
1868 $bits['close'] = $child;
1869 }
1870 }
1871 if ( !isset( $bits['name'] ) ) {
1872 throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1873 }
1874 return $bits;
1875 }
1876
1877 /**
1878 * Split an "<h>" node
1879 *
1880 * @throws MWException
1881 * @return array
1882 */
1883 public function splitHeading() {
1884 if ( $this->name !== 'h' ) {
1885 throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1886 }
1887 $bits = array();
1888 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1889 if ( !isset( $child->name ) ) {
1890 continue;
1891 }
1892 if ( $child->name == 'i' ) {
1893 $bits['i'] = $child->value;
1894 } elseif ( $child->name == 'level' ) {
1895 $bits['level'] = $child->value;
1896 }
1897 }
1898 if ( !isset( $bits['i'] ) ) {
1899 throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1900 }
1901 return $bits;
1902 }
1903
1904 /**
1905 * Split a "<template>" or "<tplarg>" node
1906 *
1907 * @throws MWException
1908 * @return array
1909 */
1910 public function splitTemplate() {
1911 $parts = array();
1912 $bits = array( 'lineStart' => '' );
1913 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1914 if ( !isset( $child->name ) ) {
1915 continue;
1916 }
1917 if ( $child->name == 'title' ) {
1918 $bits['title'] = $child;
1919 }
1920 if ( $child->name == 'part' ) {
1921 $parts[] = $child;
1922 }
1923 if ( $child->name == 'lineStart' ) {
1924 $bits['lineStart'] = '1';
1925 }
1926 }
1927 if ( !isset( $bits['title'] ) ) {
1928 throw new MWException( 'Invalid node passed to ' . __METHOD__ );
1929 }
1930 $bits['parts'] = new PPNode_Hash_Array( $parts );
1931 return $bits;
1932 }
1933 }
1934
1935 /**
1936 * @ingroup Parser
1937 * @codingStandardsIgnoreStart
1938 */
1939 class PPNode_Hash_Text implements PPNode {
1940 // @codingStandardsIgnoreEnd
1941
1942 public $value, $nextSibling;
1943
1944 public function __construct( $value ) {
1945 if ( is_object( $value ) ) {
1946 throw new MWException( __CLASS__ . ' given object instead of string' );
1947 }
1948 $this->value = $value;
1949 }
1950
1951 public function __toString() {
1952 return htmlspecialchars( $this->value );
1953 }
1954
1955 public function getNextSibling() {
1956 return $this->nextSibling;
1957 }
1958
1959 public function getChildren() {
1960 return false;
1961 }
1962
1963 public function getFirstChild() {
1964 return false;
1965 }
1966
1967 public function getChildrenOfType( $name ) {
1968 return false;
1969 }
1970
1971 public function getLength() {
1972 return false;
1973 }
1974
1975 public function item( $i ) {
1976 return false;
1977 }
1978
1979 public function getName() {
1980 return '#text';
1981 }
1982
1983 public function splitArg() {
1984 throw new MWException( __METHOD__ . ': not supported' );
1985 }
1986
1987 public function splitExt() {
1988 throw new MWException( __METHOD__ . ': not supported' );
1989 }
1990
1991 public function splitHeading() {
1992 throw new MWException( __METHOD__ . ': not supported' );
1993 }
1994 }
1995
1996 /**
1997 * @ingroup Parser
1998 * @codingStandardsIgnoreStart
1999 */
2000 class PPNode_Hash_Array implements PPNode {
2001 // @codingStandardsIgnoreEnd
2002
2003 public $value, $nextSibling;
2004
2005 public function __construct( $value ) {
2006 $this->value = $value;
2007 }
2008
2009 public function __toString() {
2010 return var_export( $this, true );
2011 }
2012
2013 public function getLength() {
2014 return count( $this->value );
2015 }
2016
2017 public function item( $i ) {
2018 return $this->value[$i];
2019 }
2020
2021 public function getName() {
2022 return '#nodelist';
2023 }
2024
2025 public function getNextSibling() {
2026 return $this->nextSibling;
2027 }
2028
2029 public function getChildren() {
2030 return false;
2031 }
2032
2033 public function getFirstChild() {
2034 return false;
2035 }
2036
2037 public function getChildrenOfType( $name ) {
2038 return false;
2039 }
2040
2041 public function splitArg() {
2042 throw new MWException( __METHOD__ . ': not supported' );
2043 }
2044
2045 public function splitExt() {
2046 throw new MWException( __METHOD__ . ': not supported' );
2047 }
2048
2049 public function splitHeading() {
2050 throw new MWException( __METHOD__ . ': not supported' );
2051 }
2052 }
2053
2054 /**
2055 * @ingroup Parser
2056 * @codingStandardsIgnoreStart
2057 */
2058 class PPNode_Hash_Attr implements PPNode {
2059 // @codingStandardsIgnoreEnd
2060
2061 public $name, $value, $nextSibling;
2062
2063 public function __construct( $name, $value ) {
2064 $this->name = $name;
2065 $this->value = $value;
2066 }
2067
2068 public function __toString() {
2069 return "<@{$this->name}>" . htmlspecialchars( $this->value ) . "</@{$this->name}>";
2070 }
2071
2072 public function getName() {
2073 return $this->name;
2074 }
2075
2076 public function getNextSibling() {
2077 return $this->nextSibling;
2078 }
2079
2080 public function getChildren() {
2081 return false;
2082 }
2083
2084 public function getFirstChild() {
2085 return false;
2086 }
2087
2088 public function getChildrenOfType( $name ) {
2089 return false;
2090 }
2091
2092 public function getLength() {
2093 return false;
2094 }
2095
2096 public function item( $i ) {
2097 return false;
2098 }
2099
2100 public function splitArg() {
2101 throw new MWException( __METHOD__ . ': not supported' );
2102 }
2103
2104 public function splitExt() {
2105 throw new MWException( __METHOD__ . ': not supported' );
2106 }
2107
2108 public function splitHeading() {
2109 throw new MWException( __METHOD__ . ': not supported' );
2110 }
2111 }