Merge "Improve docs for Title::getInternalURL/getCanonicalURL"
[lhc/web/wiklou.git] / includes / parser / BlockLevelPass.php
1 <?php
2
3 /**
4 * This is the part of the wikitext parser which handles automatic paragraphs
5 * and conversion of start-of-line prefixes to HTML lists.
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License along
18 * with this program; if not, write to the Free Software Foundation, Inc.,
19 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
20 * http://www.gnu.org/copyleft/gpl.html
21 *
22 * @file
23 * @ingroup Parser
24 */
25 class BlockLevelPass {
26 private $DTopen = false;
27 private $inPre = false;
28 private $lastParagraph = '';
29 private $lineStart;
30 private $text;
31
32 # State constants for the definition list colon extraction
33 const COLON_STATE_TEXT = 0;
34 const COLON_STATE_TAG = 1;
35 const COLON_STATE_TAGSTART = 2;
36 const COLON_STATE_CLOSETAG = 3;
37 const COLON_STATE_TAGSLASH = 4;
38 const COLON_STATE_COMMENT = 5;
39 const COLON_STATE_COMMENTDASH = 6;
40 const COLON_STATE_COMMENTDASHDASH = 7;
41 const COLON_STATE_LC = 8;
42
43 /**
44 * Make lists from lines starting with ':', '*', '#', etc.
45 *
46 * @param string $text
47 * @param bool $lineStart Whether or not this is at the start of a line.
48 * @return string The lists rendered as HTML
49 */
50 public static function doBlockLevels( $text, $lineStart ) {
51 $pass = new self( $text, $lineStart );
52 return $pass->execute();
53 }
54
55 /**
56 * @param string $text
57 * @param bool $lineStart
58 */
59 private function __construct( $text, $lineStart ) {
60 $this->text = $text;
61 $this->lineStart = $lineStart;
62 }
63
64 /**
65 * @return bool
66 */
67 private function hasOpenParagraph() {
68 return $this->lastParagraph !== '';
69 }
70
71 /**
72 * If a pre or p is open, return the corresponding close tag and update
73 * the state. If no tag is open, return an empty string.
74 * @param bool $atTheEnd Omit trailing newline if we've reached the end.
75 * @return string
76 */
77 private function closeParagraph( $atTheEnd = false ) {
78 $result = '';
79 if ( $this->hasOpenParagraph() ) {
80 $result = '</' . $this->lastParagraph . '>';
81 if ( !$atTheEnd ) {
82 $result .= "\n";
83 }
84 }
85 $this->inPre = false;
86 $this->lastParagraph = '';
87 return $result;
88 }
89
90 /**
91 * getCommon() returns the length of the longest common substring
92 * of both arguments, starting at the beginning of both.
93 *
94 * @param string $st1
95 * @param string $st2
96 *
97 * @return int
98 */
99 private function getCommon( $st1, $st2 ) {
100 $shorter = min( strlen( $st1 ), strlen( $st2 ) );
101
102 for ( $i = 0; $i < $shorter; ++$i ) {
103 if ( $st1[$i] !== $st2[$i] ) {
104 break;
105 }
106 }
107 return $i;
108 }
109
110 /**
111 * Open the list item element identified by the prefix character.
112 *
113 * @param string $char
114 *
115 * @return string
116 */
117 private function openList( $char ) {
118 $result = $this->closeParagraph();
119
120 if ( $char === '*' ) {
121 $result .= "<ul><li>";
122 } elseif ( $char === '#' ) {
123 $result .= "<ol><li>";
124 } elseif ( $char === ':' ) {
125 $result .= "<dl><dd>";
126 } elseif ( $char === ';' ) {
127 $result .= "<dl><dt>";
128 $this->DTopen = true;
129 } else {
130 $result = '<!-- ERR 1 -->';
131 }
132
133 return $result;
134 }
135
136 /**
137 * Close the current list item and open the next one.
138 * @param string $char
139 *
140 * @return string
141 */
142 private function nextItem( $char ) {
143 if ( $char === '*' || $char === '#' ) {
144 return "</li>\n<li>";
145 } elseif ( $char === ':' || $char === ';' ) {
146 $close = "</dd>\n";
147 if ( $this->DTopen ) {
148 $close = "</dt>\n";
149 }
150 if ( $char === ';' ) {
151 $this->DTopen = true;
152 return $close . '<dt>';
153 } else {
154 $this->DTopen = false;
155 return $close . '<dd>';
156 }
157 }
158 return '<!-- ERR 2 -->';
159 }
160
161 /**
162 * Close the current list item identified by the prefix character.
163 * @param string $char
164 *
165 * @return string
166 */
167 private function closeList( $char ) {
168 if ( $char === '*' ) {
169 $text = "</li></ul>";
170 } elseif ( $char === '#' ) {
171 $text = "</li></ol>";
172 } elseif ( $char === ':' ) {
173 if ( $this->DTopen ) {
174 $this->DTopen = false;
175 $text = "</dt></dl>";
176 } else {
177 $text = "</dd></dl>";
178 }
179 } else {
180 return '<!-- ERR 3 -->';
181 }
182 return $text;
183 }
184
185 /**
186 * Execute the pass.
187 * @return string
188 */
189 private function execute() {
190 $text = $this->text;
191 # Parsing through the text line by line. The main thing
192 # happening here is handling of block-level elements p, pre,
193 # and making lists from lines starting with * # : etc.
194 $textLines = StringUtils::explode( "\n", $text );
195
196 $lastPrefix = $output = '';
197 $this->DTopen = $inBlockElem = false;
198 $prefixLength = 0;
199 $pendingPTag = false;
200 $inBlockquote = false;
201
202 for ( $textLines->rewind(); $textLines->valid(); ) {
203 $inputLine = $textLines->current();
204 $textLines->next();
205 $notLastLine = $textLines->valid();
206
207 # Fix up $lineStart
208 if ( !$this->lineStart ) {
209 $output .= $inputLine;
210 $this->lineStart = true;
211 continue;
212 }
213 # * = ul
214 # # = ol
215 # ; = dt
216 # : = dd
217
218 $lastPrefixLength = strlen( $lastPrefix );
219 $preCloseMatch = preg_match( '/<\\/pre/i', $inputLine );
220 $preOpenMatch = preg_match( '/<pre/i', $inputLine );
221 # If not in a <pre> element, scan for and figure out what prefixes are there.
222 if ( !$this->inPre ) {
223 # Multiple prefixes may abut each other for nested lists.
224 $prefixLength = strspn( $inputLine, '*#:;' );
225 $prefix = substr( $inputLine, 0, $prefixLength );
226
227 # eh?
228 # ; and : are both from definition-lists, so they're equivalent
229 # for the purposes of determining whether or not we need to open/close
230 # elements.
231 $prefix2 = str_replace( ';', ':', $prefix );
232 $t = substr( $inputLine, $prefixLength );
233 $this->inPre = (bool)$preOpenMatch;
234 } else {
235 # Don't interpret any other prefixes in preformatted text
236 $prefixLength = 0;
237 $prefix = $prefix2 = '';
238 $t = $inputLine;
239 }
240
241 # List generation
242 if ( $prefixLength && $lastPrefix === $prefix2 ) {
243 # Same as the last item, so no need to deal with nesting or opening stuff
244 $output .= $this->nextItem( substr( $prefix, -1 ) );
245 $pendingPTag = false;
246
247 if ( substr( $prefix, -1 ) === ';' ) {
248 # The one nasty exception: definition lists work like this:
249 # ; title : definition text
250 # So we check for : in the remainder text to split up the
251 # title and definition, without b0rking links.
252 $term = $t2 = '';
253 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
254 $t = $t2;
255 // Trim whitespace in list items
256 $output .= trim( $term ) . $this->nextItem( ':' );
257 }
258 }
259 } elseif ( $prefixLength || $lastPrefixLength ) {
260 # We need to open or close prefixes, or both.
261
262 # Either open or close a level...
263 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
264 $pendingPTag = false;
265
266 # Close all the prefixes which aren't shared.
267 while ( $commonPrefixLength < $lastPrefixLength ) {
268 $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
269 --$lastPrefixLength;
270 }
271
272 # Continue the current prefix if appropriate.
273 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
274 $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
275 }
276
277 # Close an open <dt> if we have a <dd> (":") starting on this line
278 if ( $this->DTopen && $commonPrefixLength > 0 && $prefix[$commonPrefixLength - 1] === ':' ) {
279 $output .= $this->nextItem( ':' );
280 }
281
282 # Open prefixes where appropriate.
283 if ( $lastPrefix && $prefixLength > $commonPrefixLength ) {
284 $output .= "\n";
285 }
286 while ( $prefixLength > $commonPrefixLength ) {
287 $char = $prefix[$commonPrefixLength];
288 $output .= $this->openList( $char );
289
290 if ( $char === ';' ) {
291 # @todo FIXME: This is dupe of code above
292 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
293 $t = $t2;
294 // Trim whitespace in list items
295 $output .= trim( $term ) . $this->nextItem( ':' );
296 }
297 }
298 ++$commonPrefixLength;
299 }
300 if ( !$prefixLength && $lastPrefix ) {
301 $output .= "\n";
302 }
303 $lastPrefix = $prefix2;
304 }
305
306 # If we have no prefixes, go to paragraph mode.
307 if ( $prefixLength == 0 ) {
308 # No prefix (not in list)--go to paragraph mode
309 # @todo consider using a stack for nestable elements like span, table and div
310
311 // P-wrapping and indent-pre are suppressed inside, not outside
312 $blockElems = 'table|h1|h2|h3|h4|h5|h6|pre|p|ul|ol|dl';
313 // P-wrapping and indent-pre are suppressed outside, not inside
314 $antiBlockElems = 'td|th';
315
316 $openMatch = preg_match(
317 '/<('
318 . "({$blockElems})|\\/({$antiBlockElems})|"
319 // Always suppresses
320 . '\\/?(tr|dt|dd|li)'
321 . ')\\b/iS',
322 $t
323 );
324 $closeMatch = preg_match(
325 '/<('
326 . "\\/({$blockElems})|({$antiBlockElems})|"
327 // Never suppresses
328 . '\\/?(center|blockquote|div|hr|mw:)'
329 . ')\\b/iS',
330 $t
331 );
332
333 // Any match closes the paragraph, but only when `!$closeMatch`
334 // do we enter block mode. The oddities with table rows and
335 // cells are to avoid paragraph wrapping in interstitial spaces
336 // leading to fostered content.
337
338 if ( $openMatch || $closeMatch ) {
339 $pendingPTag = false;
340 // Only close the paragraph if we're not inside a <pre> tag, or if
341 // that <pre> tag has just been opened
342 if ( !$this->inPre || $preOpenMatch ) {
343 // @todo T7718: paragraph closed
344 $output .= $this->closeParagraph();
345 }
346 if ( $preOpenMatch && !$preCloseMatch ) {
347 $this->inPre = true;
348 }
349 $bqOffset = 0;
350 while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t,
351 $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset )
352 ) {
353 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
354 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
355 }
356 $inBlockElem = !$closeMatch;
357 } elseif ( !$inBlockElem && !$this->inPre ) {
358 if ( substr( $t, 0, 1 ) == ' '
359 && ( $this->lastParagraph === 'pre' || trim( $t ) != '' )
360 && !$inBlockquote
361 ) {
362 # pre
363 if ( $this->lastParagraph !== 'pre' ) {
364 $pendingPTag = false;
365 $output .= $this->closeParagraph() . '<pre>';
366 $this->lastParagraph = 'pre';
367 }
368 $t = substr( $t, 1 );
369 } elseif ( preg_match( '/^(?:<style\\b[^>]*>.*?<\\/style>\s*|<link\\b[^>]*>\s*)+$/iS', $t ) ) {
370 # T186965: <style> or <link> by itself on a line shouldn't open or close paragraphs.
371 # But it should clear $pendingPTag.
372 if ( $pendingPTag ) {
373 $output .= $this->closeParagraph();
374 $pendingPTag = false;
375 }
376 } else {
377 # paragraph
378 if ( trim( $t ) === '' ) {
379 if ( $pendingPTag ) {
380 $output .= $pendingPTag . '<br />';
381 $pendingPTag = false;
382 $this->lastParagraph = 'p';
383 } elseif ( $this->lastParagraph !== 'p' ) {
384 $output .= $this->closeParagraph();
385 $pendingPTag = '<p>';
386 } else {
387 $pendingPTag = '</p><p>';
388 }
389 } elseif ( $pendingPTag ) {
390 $output .= $pendingPTag;
391 $pendingPTag = false;
392 $this->lastParagraph = 'p';
393 } elseif ( $this->lastParagraph !== 'p' ) {
394 $output .= $this->closeParagraph() . '<p>';
395 $this->lastParagraph = 'p';
396 }
397 }
398 }
399 }
400 # somewhere above we forget to get out of pre block (T2785)
401 if ( $preCloseMatch && $this->inPre ) {
402 $this->inPre = false;
403 }
404 if ( $pendingPTag === false ) {
405 if ( $prefixLength === 0 ) {
406 $output .= $t;
407 // Add a newline if there's an open paragraph
408 // or we've yet to reach the last line.
409 if ( $notLastLine || $this->hasOpenParagraph() ) {
410 $output .= "\n";
411 }
412 } else {
413 // Trim whitespace in list items
414 $output .= trim( $t );
415 }
416 }
417 }
418 while ( $prefixLength ) {
419 $output .= $this->closeList( $prefix2[$prefixLength - 1] );
420 --$prefixLength;
421 // Note that a paragraph is only ever opened when `prefixLength`
422 // is zero, but we'll choose to be overly cautious.
423 if ( !$prefixLength && $this->hasOpenParagraph() ) {
424 $output .= "\n";
425 }
426 }
427 $output .= $this->closeParagraph( true );
428 return $output;
429 }
430
431 /**
432 * Split up a string on ':', ignoring any occurrences inside tags
433 * to prevent illegal overlapping.
434 *
435 * @param string $str The string to split
436 * @param string &$before Set to everything before the ':'
437 * @param string &$after Set to everything after the ':'
438 * @throws MWException
439 * @return string The position of the ':', or false if none found
440 */
441 private function findColonNoLinks( $str, &$before, &$after ) {
442 if ( !preg_match( '/:|<|-\{/', $str, $m, PREG_OFFSET_CAPTURE ) ) {
443 # Nothing to find!
444 return false;
445 }
446
447 if ( $m[0][0] === ':' ) {
448 # Easy; no tag nesting to worry about
449 $colonPos = $m[0][1];
450 $before = substr( $str, 0, $colonPos );
451 $after = substr( $str, $colonPos + 1 );
452 return $colonPos;
453 }
454
455 # Ugly state machine to walk through avoiding tags.
456 $state = self::COLON_STATE_TEXT;
457 $ltLevel = 0;
458 $lcLevel = 0;
459 $len = strlen( $str );
460 for ( $i = $m[0][1]; $i < $len; $i++ ) {
461 $c = $str[$i];
462
463 switch ( $state ) {
464 case self::COLON_STATE_TEXT:
465 switch ( $c ) {
466 case "<":
467 # Could be either a <start> tag or an </end> tag
468 $state = self::COLON_STATE_TAGSTART;
469 break;
470 case ":":
471 if ( $ltLevel === 0 ) {
472 # We found it!
473 $before = substr( $str, 0, $i );
474 $after = substr( $str, $i + 1 );
475 return $i;
476 }
477 # Embedded in a tag; don't break it.
478 break;
479 default:
480 # Skip ahead looking for something interesting
481 if ( !preg_match( '/:|<|-\{/', $str, $m, PREG_OFFSET_CAPTURE, $i ) ) {
482 # Nothing else interesting
483 return false;
484 }
485 if ( $m[0][0] === '-{' ) {
486 $state = self::COLON_STATE_LC;
487 $lcLevel++;
488 $i = $m[0][1] + 1;
489 } else {
490 # Skip ahead to next interesting character.
491 $i = $m[0][1] - 1;
492 }
493 break;
494 }
495 break;
496 case self::COLON_STATE_LC:
497 # In language converter markup -{ ... }-
498 if ( !preg_match( '/-\{|\}-/', $str, $m, PREG_OFFSET_CAPTURE, $i ) ) {
499 # Nothing else interesting to find; abort!
500 # We're nested in language converter markup, but there
501 # are no close tags left. Abort!
502 break 2;
503 } elseif ( $m[0][0] === '-{' ) {
504 $i = $m[0][1] + 1;
505 $lcLevel++;
506 } elseif ( $m[0][0] === '}-' ) {
507 $i = $m[0][1] + 1;
508 $lcLevel--;
509 if ( $lcLevel === 0 ) {
510 $state = self::COLON_STATE_TEXT;
511 }
512 }
513 break;
514 case self::COLON_STATE_TAG:
515 # In a <tag>
516 switch ( $c ) {
517 case ">":
518 $ltLevel++;
519 $state = self::COLON_STATE_TEXT;
520 break;
521 case "/":
522 # Slash may be followed by >?
523 $state = self::COLON_STATE_TAGSLASH;
524 break;
525 default:
526 # ignore
527 }
528 break;
529 case self::COLON_STATE_TAGSTART:
530 switch ( $c ) {
531 case "/":
532 $state = self::COLON_STATE_CLOSETAG;
533 break;
534 case "!":
535 $state = self::COLON_STATE_COMMENT;
536 break;
537 case ">":
538 # Illegal early close? This shouldn't happen D:
539 $state = self::COLON_STATE_TEXT;
540 break;
541 default:
542 $state = self::COLON_STATE_TAG;
543 }
544 break;
545 case self::COLON_STATE_CLOSETAG:
546 # In a </tag>
547 if ( $c === ">" ) {
548 if ( $ltLevel > 0 ) {
549 $ltLevel--;
550 } else {
551 # ignore the excess close tag, but keep looking for
552 # colons. (This matches Parsoid behavior.)
553 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
554 }
555 $state = self::COLON_STATE_TEXT;
556 }
557 break;
558 case self::COLON_STATE_TAGSLASH:
559 if ( $c === ">" ) {
560 # Yes, a self-closed tag <blah/>
561 $state = self::COLON_STATE_TEXT;
562 } else {
563 # Probably we're jumping the gun, and this is an attribute
564 $state = self::COLON_STATE_TAG;
565 }
566 break;
567 case self::COLON_STATE_COMMENT:
568 if ( $c === "-" ) {
569 $state = self::COLON_STATE_COMMENTDASH;
570 }
571 break;
572 case self::COLON_STATE_COMMENTDASH:
573 if ( $c === "-" ) {
574 $state = self::COLON_STATE_COMMENTDASHDASH;
575 } else {
576 $state = self::COLON_STATE_COMMENT;
577 }
578 break;
579 case self::COLON_STATE_COMMENTDASHDASH:
580 if ( $c === ">" ) {
581 $state = self::COLON_STATE_TEXT;
582 } else {
583 $state = self::COLON_STATE_COMMENT;
584 }
585 break;
586 default:
587 throw new MWException( "State machine error in " . __METHOD__ );
588 }
589 }
590 if ( $ltLevel > 0 || $lcLevel > 0 ) {
591 wfDebug(
592 __METHOD__ . ": Invalid input; not enough close tags " .
593 "(level $ltLevel/$lcLevel, state $state)\n"
594 );
595 return false;
596 }
597 return false;
598 }
599 }