dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Improve/rename Parser::replaceUnusualEscapes
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
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 * @defgroup Parser Parser
26 */
27
28 /**
29 * PHP Parser - Processes wiki markup (which uses a more user-friendly
30 * syntax, such as "[[link]]" for making links), and provides a one-way
31 * transformation of that wiki markup it into (X)HTML output / markup
32 * (which in turn the browser understands, and can display).
33 *
34 * There are seven main entry points into the Parser class:
35 *
36 * - Parser::parse()
37 * produces HTML output
38 * - Parser::preSaveTransform().
39 * produces altered wiki markup.
40 * - Parser::preprocess()
41 * removes HTML comments and expands templates
42 * - Parser::cleanSig() and Parser::cleanSigInSig()
43 * Cleans a signature before saving it to preferences
44 * - Parser::getSection()
45 * Return the content of a section from an article for section editing
46 * - Parser::replaceSection()
47 * Replaces a section by number inside an article
48 * - Parser::getPreloadText()
49 * Removes <noinclude> sections, and <includeonly> tags.
50 *
51 * Globals used:
52 * object: $wgContLang
53 *
54 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
55 *
56 * @par Settings:
57 * $wgNamespacesWithSubpages
58 *
59 * @par Settings only within ParserOptions:
60 * $wgAllowExternalImages
61 * $wgAllowSpecialInclusion
62 * $wgInterwikiMagic
63 * $wgMaxArticleSize
64 *
65 * @ingroup Parser
66 */
67 class Parser {
68 /**
69 * Update this version number when the ParserOutput format
70 * changes in an incompatible way, so the parser cache
71 * can automatically discard old data.
72 */
73 const VERSION = '1.6.4';
74
75 /**
76 * Update this version number when the output of serialiseHalfParsedText()
77 * changes in an incompatible way
78 */
79 const HALF_PARSED_VERSION = 2;
80
81 # Flags for Parser::setFunctionHook
82 # Also available as global constants from Defines.php
83 const SFH_NO_HASH = 1;
84 const SFH_OBJECT_ARGS = 2;
85
86 # Constants needed for external link processing
87 # Everything except bracket, space, or control characters
88 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
89 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
90 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
91 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
92 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
93
94 # State constants for the definition list colon extraction
95 const COLON_STATE_TEXT = 0;
96 const COLON_STATE_TAG = 1;
97 const COLON_STATE_TAGSTART = 2;
98 const COLON_STATE_CLOSETAG = 3;
99 const COLON_STATE_TAGSLASH = 4;
100 const COLON_STATE_COMMENT = 5;
101 const COLON_STATE_COMMENTDASH = 6;
102 const COLON_STATE_COMMENTDASHDASH = 7;
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 # Marker Suffix needs to be accessible staticly.
116 const MARKER_SUFFIX = "-QINU\x7f";
117
118 # Markers used for wrapping the table of contents
119 const TOC_START = '<mw:toc>';
120 const TOC_END = '</mw:toc>';
121
122 # Persistent:
123 public $mTagHooks = array();
124 public $mTransparentTagHooks = array();
125 public $mFunctionHooks = array();
126 public $mFunctionSynonyms = array( 0 => array(), 1 => array() );
127 public $mFunctionTagHooks = array();
128 public $mStripList = array();
129 public $mDefaultStripList = array();
130 public $mVarCache = array();
131 public $mImageParams = array();
132 public $mImageParamsMagicArray = array();
133 public $mMarkerIndex = 0;
134 public $mFirstCall = true;
135
136 # Initialised by initialiseVariables()
137
138 /**
139 * @var MagicWordArray
140 */
141 public $mVariables;
142
143 /**
144 * @var MagicWordArray
145 */
146 public $mSubstWords;
147 public $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
148
149 # Cleared with clearState():
150 /**
151 * @var ParserOutput
152 */
153 public $mOutput;
154 public $mAutonumber, $mDTopen;
155
156 /**
157 * @var StripState
158 */
159 public $mStripState;
160
161 public $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
162 /**
163 * @var LinkHolderArray
164 */
165 public $mLinkHolders;
166
167 public $mLinkID;
168 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
169 public $mDefaultSort;
170 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
171 public $mExpensiveFunctionCount; # number of expensive parser function calls
172 public $mShowToc, $mForceTocPosition;
173
174 /**
175 * @var User
176 */
177 public $mUser; # User object; only used when doing pre-save transform
178
179 # Temporary
180 # These are variables reset at least once per parse regardless of $clearState
181
182 /**
183 * @var ParserOptions
184 */
185 public $mOptions;
186
187 /**
188 * @var Title
189 */
190 public $mTitle; # Title context, used for self-link rendering and similar things
191 public $mOutputType; # Output type, one of the OT_xxx constants
192 public $ot; # Shortcut alias, see setOutputType()
193 public $mRevisionObject; # The revision object of the specified revision ID
194 public $mRevisionId; # ID to display in {{REVISIONID}} tags
195 public $mRevisionTimestamp; # The timestamp of the specified revision ID
196 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
197 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
198 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
199 public $mInputSize = false; # For {{PAGESIZE}} on current page.
200
201 /**
202 * @var string
203 */
204 public $mUniqPrefix;
205
206 /**
207 * @var array Array with the language name of each language link (i.e. the
208 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
209 * duplicate language links to the ParserOutput.
210 */
211 public $mLangLinkLanguages;
212
213 /**
214 * @var bool Recursive call protection.
215 * This variable should be treated as if it were private.
216 */
217 public $mInParse = false;
218
219 /**
220 * @param array $conf
221 */
222 public function __construct( $conf = array() ) {
223 $this->mConf = $conf;
224 $this->mUrlProtocols = wfUrlProtocols();
225 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
226 self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
227 if ( isset( $conf['preprocessorClass'] ) ) {
228 $this->mPreprocessorClass = $conf['preprocessorClass'];
229 } elseif ( defined( 'HPHP_VERSION' ) ) {
230 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
231 $this->mPreprocessorClass = 'Preprocessor_Hash';
232 } elseif ( extension_loaded( 'domxml' ) ) {
233 # PECL extension that conflicts with the core DOM extension (bug 13770)
234 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
235 $this->mPreprocessorClass = 'Preprocessor_Hash';
236 } elseif ( extension_loaded( 'dom' ) ) {
237 $this->mPreprocessorClass = 'Preprocessor_DOM';
238 } else {
239 $this->mPreprocessorClass = 'Preprocessor_Hash';
240 }
241 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
242 }
243
244 /**
245 * Reduce memory usage to reduce the impact of circular references
246 */
247 public function __destruct() {
248 if ( isset( $this->mLinkHolders ) ) {
249 unset( $this->mLinkHolders );
250 }
251 foreach ( $this as $name => $value ) {
252 unset( $this->$name );
253 }
254 }
255
256 /**
257 * Allow extensions to clean up when the parser is cloned
258 */
259 public function __clone() {
260 $this->mInParse = false;
261 wfRunHooks( 'ParserCloned', array( $this ) );
262 }
263
264 /**
265 * Do various kinds of initialisation on the first call of the parser
266 */
267 public function firstCallInit() {
268 if ( !$this->mFirstCall ) {
269 return;
270 }
271 $this->mFirstCall = false;
272
273 wfProfileIn( __METHOD__ );
274
275 CoreParserFunctions::register( $this );
276 CoreTagHooks::register( $this );
277 $this->initialiseVariables();
278
279 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
280 wfProfileOut( __METHOD__ );
281 }
282
283 /**
284 * Clear Parser state
285 *
286 * @private
287 */
288 public function clearState() {
289 wfProfileIn( __METHOD__ );
290 if ( $this->mFirstCall ) {
291 $this->firstCallInit();
292 }
293 $this->mOutput = new ParserOutput;
294 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
295 $this->mAutonumber = 0;
296 $this->mLastSection = '';
297 $this->mDTopen = false;
298 $this->mIncludeCount = array();
299 $this->mArgStack = false;
300 $this->mInPre = false;
301 $this->mLinkHolders = new LinkHolderArray( $this );
302 $this->mLinkID = 0;
303 $this->mRevisionObject = $this->mRevisionTimestamp =
304 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
305 $this->mVarCache = array();
306 $this->mUser = null;
307 $this->mLangLinkLanguages = array();
308
309 /**
310 * Prefix for temporary replacement strings for the multipass parser.
311 * \x07 should never appear in input as it's disallowed in XML.
312 * Using it at the front also gives us a little extra robustness
313 * since it shouldn't match when butted up against identifier-like
314 * string constructs.
315 *
316 * Must not consist of all title characters, or else it will change
317 * the behavior of <nowiki> in a link.
318 */
319 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
320 $this->mStripState = new StripState( $this->mUniqPrefix );
321
322 # Clear these on every parse, bug 4549
323 $this->mTplRedirCache = $this->mTplDomCache = array();
324
325 $this->mShowToc = true;
326 $this->mForceTocPosition = false;
327 $this->mIncludeSizes = array(
328 'post-expand' => 0,
329 'arg' => 0,
330 );
331 $this->mPPNodeCount = 0;
332 $this->mGeneratedPPNodeCount = 0;
333 $this->mHighestExpansionDepth = 0;
334 $this->mDefaultSort = false;
335 $this->mHeadings = array();
336 $this->mDoubleUnderscores = array();
337 $this->mExpensiveFunctionCount = 0;
338
339 # Fix cloning
340 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
341 $this->mPreprocessor = null;
342 }
343
344 wfRunHooks( 'ParserClearState', array( &$this ) );
345 wfProfileOut( __METHOD__ );
346 }
347
348 /**
349 * Convert wikitext to HTML
350 * Do not call this function recursively.
351 *
352 * @param string $text Text we want to parse
353 * @param Title $title
354 * @param ParserOptions $options
355 * @param bool $linestart
356 * @param bool $clearState
357 * @param int $revid Number to pass in {{REVISIONID}}
358 * @return ParserOutput A ParserOutput
359 */
360 public function parse( $text, Title $title, ParserOptions $options,
361 $linestart = true, $clearState = true, $revid = null
362 ) {
363 /**
364 * First pass--just handle <nowiki> sections, pass the rest off
365 * to internalParse() which does all the real work.
366 */
367
368 global $wgUseTidy, $wgAlwaysUseTidy, $wgShowHostnames;
369 $fname = __METHOD__ . '-' . wfGetCaller();
370 wfProfileIn( __METHOD__ );
371 wfProfileIn( $fname );
372
373 if ( $clearState ) {
374 $magicScopeVariable = $this->lock();
375 }
376
377 $this->startParse( $title, $options, self::OT_HTML, $clearState );
378
379 $this->mInputSize = strlen( $text );
380 if ( $this->mOptions->getEnableLimitReport() ) {
381 $this->mOutput->resetParseStartTime();
382 }
383
384 # Remove the strip marker tag prefix from the input, if present.
385 if ( $clearState ) {
386 $text = str_replace( $this->mUniqPrefix, '', $text );
387 }
388
389 $oldRevisionId = $this->mRevisionId;
390 $oldRevisionObject = $this->mRevisionObject;
391 $oldRevisionTimestamp = $this->mRevisionTimestamp;
392 $oldRevisionUser = $this->mRevisionUser;
393 $oldRevisionSize = $this->mRevisionSize;
394 if ( $revid !== null ) {
395 $this->mRevisionId = $revid;
396 $this->mRevisionObject = null;
397 $this->mRevisionTimestamp = null;
398 $this->mRevisionUser = null;
399 $this->mRevisionSize = null;
400 }
401
402 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
403 # No more strip!
404 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
405 $text = $this->internalParse( $text );
406 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
407
408 $text = $this->mStripState->unstripGeneral( $text );
409
410 # Clean up special characters, only run once, next-to-last before doBlockLevels
411 $fixtags = array(
412 # french spaces, last one Guillemet-left
413 # only if there is something before the space
414 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
415 # french spaces, Guillemet-right
416 '/(\\302\\253) /' => '\\1&#160;',
417 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
418 );
419 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
420
421 $text = $this->doBlockLevels( $text, $linestart );
422
423 $this->replaceLinkHolders( $text );
424
425 /**
426 * The input doesn't get language converted if
427 * a) It's disabled
428 * b) Content isn't converted
429 * c) It's a conversion table
430 * d) it is an interface message (which is in the user language)
431 */
432 if ( !( $options->getDisableContentConversion()
433 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
434 ) {
435 if ( !$this->mOptions->getInterfaceMessage() ) {
436 # The position of the convert() call should not be changed. it
437 # assumes that the links are all replaced and the only thing left
438 # is the <nowiki> mark.
439 $text = $this->getConverterLanguage()->convert( $text );
440 }
441 }
442
443 /**
444 * A converted title will be provided in the output object if title and
445 * content conversion are enabled, the article text does not contain
446 * a conversion-suppressing double-underscore tag, and no
447 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
448 * automatic link conversion.
449 */
450 if ( !( $options->getDisableTitleConversion()
451 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
452 || isset( $this->mDoubleUnderscores['notitleconvert'] )
453 || $this->mOutput->getDisplayTitle() !== false )
454 ) {
455 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
456 if ( $convruletitle ) {
457 $this->mOutput->setTitleText( $convruletitle );
458 } else {
459 $titleText = $this->getConverterLanguage()->convertTitle( $title );
460 $this->mOutput->setTitleText( $titleText );
461 }
462 }
463
464 $text = $this->mStripState->unstripNoWiki( $text );
465
466 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
467
468 $text = $this->replaceTransparentTags( $text );
469 $text = $this->mStripState->unstripGeneral( $text );
470
471 $text = Sanitizer::normalizeCharReferences( $text );
472
473 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
474 $text = MWTidy::tidy( $text );
475 } else {
476 # attempt to sanitize at least some nesting problems
477 # (bug #2702 and quite a few others)
478 $tidyregs = array(
479 # ''Something [http://www.cool.com cool''] -->
480 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
481 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
482 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
483 # fix up an anchor inside another anchor, only
484 # at least for a single single nested link (bug 3695)
485 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
486 '\\1\\2</a>\\3</a>\\1\\4</a>',
487 # fix div inside inline elements- doBlockLevels won't wrap a line which
488 # contains a div, so fix it up here; replace
489 # div with escaped text
490 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
491 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
492 # remove empty italic or bold tag pairs, some
493 # introduced by rules above
494 '/<([bi])><\/\\1>/' => '',
495 );
496
497 $text = preg_replace(
498 array_keys( $tidyregs ),
499 array_values( $tidyregs ),
500 $text );
501 }
502
503 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
504 $this->limitationWarn( 'expensive-parserfunction',
505 $this->mExpensiveFunctionCount,
506 $this->mOptions->getExpensiveParserFunctionLimit()
507 );
508 }
509
510 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
511
512 # Information on include size limits, for the benefit of users who try to skirt them
513 if ( $this->mOptions->getEnableLimitReport() ) {
514 $max = $this->mOptions->getMaxIncludeSize();
515
516 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
517 if ( $cpuTime !== null ) {
518 $this->mOutput->setLimitReportData( 'limitreport-cputime',
519 sprintf( "%.3f", $cpuTime )
520 );
521 }
522
523 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
524 $this->mOutput->setLimitReportData( 'limitreport-walltime',
525 sprintf( "%.3f", $wallTime )
526 );
527
528 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
529 array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
530 );
531 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
532 array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
533 );
534 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
535 array( $this->mIncludeSizes['post-expand'], $max )
536 );
537 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
538 array( $this->mIncludeSizes['arg'], $max )
539 );
540 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
541 array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
542 );
543 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
544 array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
545 );
546 wfRunHooks( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );
547
548 $limitReport = "NewPP limit report\n";
549 if ( $wgShowHostnames ) {
550 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
551 }
552 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
553 if ( wfRunHooks( 'ParserLimitReportFormat',
554 array( $key, &$value, &$limitReport, false, false )
555 ) ) {
556 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
557 $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
558 ->inLanguage( 'en' )->useDatabase( false );
559 if ( !$valueMsg->exists() ) {
560 $valueMsg = new RawMessage( '$1' );
561 }
562 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
563 $valueMsg->params( $value );
564 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
565 }
566 }
567 }
568 // Since we're not really outputting HTML, decode the entities and
569 // then re-encode the things that need hiding inside HTML comments.
570 $limitReport = htmlspecialchars_decode( $limitReport );
571 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
572
573 // Sanitize for comment. Note '‐' in the replacement is U+2010,
574 // which looks much like the problematic '-'.
575 $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
576 $text .= "\n<!-- \n$limitReport-->\n";
577
578 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
579 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
580 $this->mTitle->getPrefixedDBkey() );
581 }
582 }
583 $this->mOutput->setText( $text );
584
585 $this->mRevisionId = $oldRevisionId;
586 $this->mRevisionObject = $oldRevisionObject;
587 $this->mRevisionTimestamp = $oldRevisionTimestamp;
588 $this->mRevisionUser = $oldRevisionUser;
589 $this->mRevisionSize = $oldRevisionSize;
590 $this->mInputSize = false;
591 wfProfileOut( $fname );
592 wfProfileOut( __METHOD__ );
593
594 return $this->mOutput;
595 }
596
597 /**
598 * Recursive parser entry point that can be called from an extension tag
599 * hook.
600 *
601 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
602 *
603 * @param string $text Text extension wants to have parsed
604 * @param bool|PPFrame $frame The frame to use for expanding any template variables
605 *
606 * @return string
607 */
608 public function recursiveTagParse( $text, $frame = false ) {
609 wfProfileIn( __METHOD__ );
610 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
611 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
612 $text = $this->internalParse( $text, false, $frame );
613 wfProfileOut( __METHOD__ );
614 return $text;
615 }
616
617 /**
618 * Expand templates and variables in the text, producing valid, static wikitext.
619 * Also removes comments.
620 * Do not call this function recursively.
621 * @param string $text
622 * @param Title $title
623 * @param ParserOptions $options
624 * @param int|null $revid
625 * @param bool|PPFrame $frame
626 * @return mixed|string
627 */
628 public function preprocess( $text, Title $title = null, ParserOptions $options, $revid = null, $frame = false ) {
629 wfProfileIn( __METHOD__ );
630 $magicScopeVariable = $this->lock();
631 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
632 if ( $revid !== null ) {
633 $this->mRevisionId = $revid;
634 }
635 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
636 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
637 $text = $this->replaceVariables( $text, $frame );
638 $text = $this->mStripState->unstripBoth( $text );
639 wfProfileOut( __METHOD__ );
640 return $text;
641 }
642
643 /**
644 * Recursive parser entry point that can be called from an extension tag
645 * hook.
646 *
647 * @param string $text Text to be expanded
648 * @param bool|PPFrame $frame The frame to use for expanding any template variables
649 * @return string
650 * @since 1.19
651 */
652 public function recursivePreprocess( $text, $frame = false ) {
653 wfProfileIn( __METHOD__ );
654 $text = $this->replaceVariables( $text, $frame );
655 $text = $this->mStripState->unstripBoth( $text );
656 wfProfileOut( __METHOD__ );
657 return $text;
658 }
659
660 /**
661 * Process the wikitext for the "?preload=" feature. (bug 5210)
662 *
663 * "<noinclude>", "<includeonly>" etc. are parsed as for template
664 * transclusion, comments, templates, arguments, tags hooks and parser
665 * functions are untouched.
666 *
667 * @param string $text
668 * @param Title $title
669 * @param ParserOptions $options
670 * @param array $params
671 * @return string
672 */
673 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = array() ) {
674 $msg = new RawMessage( $text );
675 $text = $msg->params( $params )->plain();
676
677 # Parser (re)initialisation
678 $magicScopeVariable = $this->lock();
679 $this->startParse( $title, $options, self::OT_PLAIN, true );
680
681 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
682 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
683 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
684 $text = $this->mStripState->unstripBoth( $text );
685 return $text;
686 }
687
688 /**
689 * Get a random string
690 *
691 * @return string
692 */
693 public static function getRandomString() {
694 return wfRandomString( 16 );
695 }
696
697 /**
698 * Set the current user.
699 * Should only be used when doing pre-save transform.
700 *
701 * @param User|null $user User object or null (to reset)
702 */
703 public function setUser( $user ) {
704 $this->mUser = $user;
705 }
706
707 /**
708 * Accessor for mUniqPrefix.
709 *
710 * @return string
711 */
712 public function uniqPrefix() {
713 if ( !isset( $this->mUniqPrefix ) ) {
714 # @todo FIXME: This is probably *horribly wrong*
715 # LanguageConverter seems to want $wgParser's uniqPrefix, however
716 # if this is called for a parser cache hit, the parser may not
717 # have ever been initialized in the first place.
718 # Not really sure what the heck is supposed to be going on here.
719 return '';
720 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
721 }
722 return $this->mUniqPrefix;
723 }
724
725 /**
726 * Set the context title
727 *
728 * @param Title $t
729 */
730 public function setTitle( $t ) {
731 if ( !$t ) {
732 $t = Title::newFromText( 'NO TITLE' );
733 }
734
735 if ( $t->hasFragment() ) {
736 # Strip the fragment to avoid various odd effects
737 $this->mTitle = clone $t;
738 $this->mTitle->setFragment( '' );
739 } else {
740 $this->mTitle = $t;
741 }
742 }
743
744 /**
745 * Accessor for the Title object
746 *
747 * @return Title
748 */
749 public function getTitle() {
750 return $this->mTitle;
751 }
752
753 /**
754 * Accessor/mutator for the Title object
755 *
756 * @param Title $x Title object or null to just get the current one
757 * @return Title
758 */
759 public function Title( $x = null ) {
760 return wfSetVar( $this->mTitle, $x );
761 }
762
763 /**
764 * Set the output type
765 *
766 * @param int $ot New value
767 */
768 public function setOutputType( $ot ) {
769 $this->mOutputType = $ot;
770 # Shortcut alias
771 $this->ot = array(
772 'html' => $ot == self::OT_HTML,
773 'wiki' => $ot == self::OT_WIKI,
774 'pre' => $ot == self::OT_PREPROCESS,
775 'plain' => $ot == self::OT_PLAIN,
776 );
777 }
778
779 /**
780 * Accessor/mutator for the output type
781 *
782 * @param int|null $x New value or null to just get the current one
783 * @return int
784 */
785 public function OutputType( $x = null ) {
786 return wfSetVar( $this->mOutputType, $x );
787 }
788
789 /**
790 * Get the ParserOutput object
791 *
792 * @return ParserOutput
793 */
794 public function getOutput() {
795 return $this->mOutput;
796 }
797
798 /**
799 * Get the ParserOptions object
800 *
801 * @return ParserOptions
802 */
803 public function getOptions() {
804 return $this->mOptions;
805 }
806
807 /**
808 * Accessor/mutator for the ParserOptions object
809 *
810 * @param ParserOptions $x New value or null to just get the current one
811 * @return ParserOptions Current ParserOptions object
812 */
813 public function Options( $x = null ) {
814 return wfSetVar( $this->mOptions, $x );
815 }
816
817 /**
818 * @return int
819 */
820 public function nextLinkID() {
821 return $this->mLinkID++;
822 }
823
824 /**
825 * @param int $id
826 */
827 public function setLinkID( $id ) {
828 $this->mLinkID = $id;
829 }
830
831 /**
832 * Get a language object for use in parser functions such as {{FORMATNUM:}}
833 * @return Language
834 */
835 public function getFunctionLang() {
836 return $this->getTargetLanguage();
837 }
838
839 /**
840 * Get the target language for the content being parsed. This is usually the
841 * language that the content is in.
842 *
843 * @since 1.19
844 *
845 * @throws MWException
846 * @return Language
847 */
848 public function getTargetLanguage() {
849 $target = $this->mOptions->getTargetLanguage();
850
851 if ( $target !== null ) {
852 return $target;
853 } elseif ( $this->mOptions->getInterfaceMessage() ) {
854 return $this->mOptions->getUserLangObj();
855 } elseif ( is_null( $this->mTitle ) ) {
856 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
857 }
858
859 return $this->mTitle->getPageLanguage();
860 }
861
862 /**
863 * Get the language object for language conversion
864 * @return Language|null
865 */
866 public function getConverterLanguage() {
867 return $this->getTargetLanguage();
868 }
869
870 /**
871 * Get a User object either from $this->mUser, if set, or from the
872 * ParserOptions object otherwise
873 *
874 * @return User
875 */
876 public function getUser() {
877 if ( !is_null( $this->mUser ) ) {
878 return $this->mUser;
879 }
880 return $this->mOptions->getUser();
881 }
882
883 /**
884 * Get a preprocessor object
885 *
886 * @return Preprocessor
887 */
888 public function getPreprocessor() {
889 if ( !isset( $this->mPreprocessor ) ) {
890 $class = $this->mPreprocessorClass;
891 $this->mPreprocessor = new $class( $this );
892 }
893 return $this->mPreprocessor;
894 }
895
896 /**
897 * Replaces all occurrences of HTML-style comments and the given tags
898 * in the text with a random marker and returns the next text. The output
899 * parameter $matches will be an associative array filled with data in
900 * the form:
901 *
902 * @code
903 * 'UNIQ-xxxxx' => array(
904 * 'element',
905 * 'tag content',
906 * array( 'param' => 'x' ),
907 * '<element param="x">tag content</element>' ) )
908 * @endcode
909 *
910 * @param array $elements List of element names. Comments are always extracted.
911 * @param string $text Source text string.
912 * @param array $matches Out parameter, Array: extracted tags
913 * @param string $uniq_prefix
914 * @return string Stripped text
915 */
916 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
917 static $n = 1;
918 $stripped = '';
919 $matches = array();
920
921 $taglist = implode( '|', $elements );
922 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
923
924 while ( $text != '' ) {
925 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
926 $stripped .= $p[0];
927 if ( count( $p ) < 5 ) {
928 break;
929 }
930 if ( count( $p ) > 5 ) {
931 # comment
932 $element = $p[4];
933 $attributes = '';
934 $close = '';
935 $inside = $p[5];
936 } else {
937 # tag
938 $element = $p[1];
939 $attributes = $p[2];
940 $close = $p[3];
941 $inside = $p[4];
942 }
943
944 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
945 $stripped .= $marker;
946
947 if ( $close === '/>' ) {
948 # Empty element tag, <tag />
949 $content = null;
950 $text = $inside;
951 $tail = null;
952 } else {
953 if ( $element === '!--' ) {
954 $end = '/(-->)/';
955 } else {
956 $end = "/(<\\/$element\\s*>)/i";
957 }
958 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
959 $content = $q[0];
960 if ( count( $q ) < 3 ) {
961 # No end tag -- let it run out to the end of the text.
962 $tail = '';
963 $text = '';
964 } else {
965 $tail = $q[1];
966 $text = $q[2];
967 }
968 }
969
970 $matches[$marker] = array( $element,
971 $content,
972 Sanitizer::decodeTagAttributes( $attributes ),
973 "<$element$attributes$close$content$tail" );
974 }
975 return $stripped;
976 }
977
978 /**
979 * Get a list of strippable XML-like elements
980 *
981 * @return array
982 */
983 public function getStripList() {
984 return $this->mStripList;
985 }
986
987 /**
988 * Add an item to the strip state
989 * Returns the unique tag which must be inserted into the stripped text
990 * The tag will be replaced with the original text in unstrip()
991 *
992 * @param string $text
993 *
994 * @return string
995 */
996 public function insertStripItem( $text ) {
997 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
998 $this->mMarkerIndex++;
999 $this->mStripState->addGeneral( $rnd, $text );
1000 return $rnd;
1001 }
1002
1003 /**
1004 * parse the wiki syntax used to render tables
1005 *
1006 * @private
1007 * @param string $text
1008 * @return string
1009 */
1010 public function doTableStuff( $text ) {
1011 wfProfileIn( __METHOD__ );
1012
1013 $lines = StringUtils::explode( "\n", $text );
1014 $out = '';
1015 $td_history = array(); # Is currently a td tag open?
1016 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
1017 $tr_history = array(); # Is currently a tr tag open?
1018 $tr_attributes = array(); # history of tr attributes
1019 $has_opened_tr = array(); # Did this table open a <tr> element?
1020 $indent_level = 0; # indent level of the table
1021
1022 foreach ( $lines as $outLine ) {
1023 $line = trim( $outLine );
1024
1025 if ( $line === '' ) { # empty line, go to next line
1026 $out .= $outLine . "\n";
1027 continue;
1028 }
1029
1030 $first_character = $line[0];
1031 $matches = array();
1032
1033 if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
1034 # First check if we are starting a new table
1035 $indent_level = strlen( $matches[1] );
1036
1037 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1038 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1039
1040 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1041 array_push( $td_history, false );
1042 array_push( $last_tag_history, '' );
1043 array_push( $tr_history, false );
1044 array_push( $tr_attributes, '' );
1045 array_push( $has_opened_tr, false );
1046 } elseif ( count( $td_history ) == 0 ) {
1047 # Don't do any of the following
1048 $out .= $outLine . "\n";
1049 continue;
1050 } elseif ( substr( $line, 0, 2 ) === '|}' ) {
1051 # We are ending a table
1052 $line = '</table>' . substr( $line, 2 );
1053 $last_tag = array_pop( $last_tag_history );
1054
1055 if ( !array_pop( $has_opened_tr ) ) {
1056 $line = "<tr><td></td></tr>{$line}";
1057 }
1058
1059 if ( array_pop( $tr_history ) ) {
1060 $line = "</tr>{$line}";
1061 }
1062
1063 if ( array_pop( $td_history ) ) {
1064 $line = "</{$last_tag}>{$line}";
1065 }
1066 array_pop( $tr_attributes );
1067 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1068 } elseif ( substr( $line, 0, 2 ) === '|-' ) {
1069 # Now we have a table row
1070 $line = preg_replace( '#^\|-+#', '', $line );
1071
1072 # Whats after the tag is now only attributes
1073 $attributes = $this->mStripState->unstripBoth( $line );
1074 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1075 array_pop( $tr_attributes );
1076 array_push( $tr_attributes, $attributes );
1077
1078 $line = '';
1079 $last_tag = array_pop( $last_tag_history );
1080 array_pop( $has_opened_tr );
1081 array_push( $has_opened_tr, true );
1082
1083 if ( array_pop( $tr_history ) ) {
1084 $line = '</tr>';
1085 }
1086
1087 if ( array_pop( $td_history ) ) {
1088 $line = "</{$last_tag}>{$line}";
1089 }
1090
1091 $outLine = $line;
1092 array_push( $tr_history, false );
1093 array_push( $td_history, false );
1094 array_push( $last_tag_history, '' );
1095 } elseif ( $first_character === '|'
1096 || $first_character === '!'
1097 || substr( $line, 0, 2 ) === '|+'
1098 ) {
1099 # This might be cell elements, td, th or captions
1100 if ( substr( $line, 0, 2 ) === '|+' ) {
1101 $first_character = '+';
1102 $line = substr( $line, 1 );
1103 }
1104
1105 $line = substr( $line, 1 );
1106
1107 if ( $first_character === '!' ) {
1108 $line = str_replace( '!!', '||', $line );
1109 }
1110
1111 # Split up multiple cells on the same line.
1112 # FIXME : This can result in improper nesting of tags processed
1113 # by earlier parser steps, but should avoid splitting up eg
1114 # attribute values containing literal "||".
1115 $cells = StringUtils::explodeMarkup( '||', $line );
1116
1117 $outLine = '';
1118
1119 # Loop through each table cell
1120 foreach ( $cells as $cell ) {
1121 $previous = '';
1122 if ( $first_character !== '+' ) {
1123 $tr_after = array_pop( $tr_attributes );
1124 if ( !array_pop( $tr_history ) ) {
1125 $previous = "<tr{$tr_after}>\n";
1126 }
1127 array_push( $tr_history, true );
1128 array_push( $tr_attributes, '' );
1129 array_pop( $has_opened_tr );
1130 array_push( $has_opened_tr, true );
1131 }
1132
1133 $last_tag = array_pop( $last_tag_history );
1134
1135 if ( array_pop( $td_history ) ) {
1136 $previous = "</{$last_tag}>\n{$previous}";
1137 }
1138
1139 if ( $first_character === '|' ) {
1140 $last_tag = 'td';
1141 } elseif ( $first_character === '!' ) {
1142 $last_tag = 'th';
1143 } elseif ( $first_character === '+' ) {
1144 $last_tag = 'caption';
1145 } else {
1146 $last_tag = '';
1147 }
1148
1149 array_push( $last_tag_history, $last_tag );
1150
1151 # A cell could contain both parameters and data
1152 $cell_data = explode( '|', $cell, 2 );
1153
1154 # Bug 553: Note that a '|' inside an invalid link should not
1155 # be mistaken as delimiting cell parameters
1156 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1157 $cell = "{$previous}<{$last_tag}>{$cell}";
1158 } elseif ( count( $cell_data ) == 1 ) {
1159 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1160 } else {
1161 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1162 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1163 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1164 }
1165
1166 $outLine .= $cell;
1167 array_push( $td_history, true );
1168 }
1169 }
1170 $out .= $outLine . "\n";
1171 }
1172
1173 # Closing open td, tr && table
1174 while ( count( $td_history ) > 0 ) {
1175 if ( array_pop( $td_history ) ) {
1176 $out .= "</td>\n";
1177 }
1178 if ( array_pop( $tr_history ) ) {
1179 $out .= "</tr>\n";
1180 }
1181 if ( !array_pop( $has_opened_tr ) ) {
1182 $out .= "<tr><td></td></tr>\n";
1183 }
1184
1185 $out .= "</table>\n";
1186 }
1187
1188 # Remove trailing line-ending (b/c)
1189 if ( substr( $out, -1 ) === "\n" ) {
1190 $out = substr( $out, 0, -1 );
1191 }
1192
1193 # special case: don't return empty table
1194 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1195 $out = '';
1196 }
1197
1198 wfProfileOut( __METHOD__ );
1199
1200 return $out;
1201 }
1202
1203 /**
1204 * Helper function for parse() that transforms wiki markup into
1205 * HTML. Only called for $mOutputType == self::OT_HTML.
1206 *
1207 * @private
1208 *
1209 * @param string $text
1210 * @param bool $isMain
1211 * @param bool $frame
1212 *
1213 * @return string
1214 */
1215 public function internalParse( $text, $isMain = true, $frame = false ) {
1216 wfProfileIn( __METHOD__ );
1217
1218 $origText = $text;
1219
1220 # Hook to suspend the parser in this state
1221 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1222 wfProfileOut( __METHOD__ );
1223 return $text;
1224 }
1225
1226 # if $frame is provided, then use $frame for replacing any variables
1227 if ( $frame ) {
1228 # use frame depth to infer how include/noinclude tags should be handled
1229 # depth=0 means this is the top-level document; otherwise it's an included document
1230 if ( !$frame->depth ) {
1231 $flag = 0;
1232 } else {
1233 $flag = Parser::PTD_FOR_INCLUSION;
1234 }
1235 $dom = $this->preprocessToDom( $text, $flag );
1236 $text = $frame->expand( $dom );
1237 } else {
1238 # if $frame is not provided, then use old-style replaceVariables
1239 $text = $this->replaceVariables( $text );
1240 }
1241
1242 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1243 $text = Sanitizer::removeHTMLtags(
1244 $text,
1245 array( &$this, 'attributeStripCallback' ),
1246 false,
1247 array_keys( $this->mTransparentTagHooks )
1248 );
1249 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1250
1251 # Tables need to come after variable replacement for things to work
1252 # properly; putting them before other transformations should keep
1253 # exciting things like link expansions from showing up in surprising
1254 # places.
1255 $text = $this->doTableStuff( $text );
1256
1257 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1258
1259 $text = $this->doDoubleUnderscore( $text );
1260
1261 $text = $this->doHeadings( $text );
1262 $text = $this->replaceInternalLinks( $text );
1263 $text = $this->doAllQuotes( $text );
1264 $text = $this->replaceExternalLinks( $text );
1265
1266 # replaceInternalLinks may sometimes leave behind
1267 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1268 $text = str_replace( $this->mUniqPrefix . 'NOPARSE', '', $text );
1269
1270 $text = $this->doMagicLinks( $text );
1271 $text = $this->formatHeadings( $text, $origText, $isMain );
1272
1273 wfProfileOut( __METHOD__ );
1274 return $text;
1275 }
1276
1277 /**
1278 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1279 * magic external links.
1280 *
1281 * DML
1282 * @private
1283 *
1284 * @param string $text
1285 *
1286 * @return string
1287 */
1288 public function doMagicLinks( $text ) {
1289 wfProfileIn( __METHOD__ );
1290 $prots = wfUrlProtocolsWithoutProtRel();
1291 $urlChar = self::EXT_LINK_URL_CLASS;
1292 $text = preg_replace_callback(
1293 '!(?: # Start cases
1294 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1295 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1296 (\\b(?i:$prots)$urlChar+) | # m[3]: Free external links" . '
1297 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1298 ISBN\s+(\b # m[5]: ISBN, capture number
1299 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1300 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1301 [0-9Xx] # check digit
1302 \b)
1303 )!xu', array( &$this, 'magicLinkCallback' ), $text );
1304 wfProfileOut( __METHOD__ );
1305 return $text;
1306 }
1307
1308 /**
1309 * @throws MWException
1310 * @param array $m
1311 * @return HTML|string
1312 */
1313 public function magicLinkCallback( $m ) {
1314 if ( isset( $m[1] ) && $m[1] !== '' ) {
1315 # Skip anchor
1316 return $m[0];
1317 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1318 # Skip HTML element
1319 return $m[0];
1320 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1321 # Free external link
1322 return $this->makeFreeExternalLink( $m[0] );
1323 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1324 # RFC or PMID
1325 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1326 $keyword = 'RFC';
1327 $urlmsg = 'rfcurl';
1328 $cssClass = 'mw-magiclink-rfc';
1329 $id = $m[4];
1330 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1331 $keyword = 'PMID';
1332 $urlmsg = 'pubmedurl';
1333 $cssClass = 'mw-magiclink-pmid';
1334 $id = $m[4];
1335 } else {
1336 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1337 substr( $m[0], 0, 20 ) . '"' );
1338 }
1339 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1340 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
1341 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1342 # ISBN
1343 $isbn = $m[5];
1344 $num = strtr( $isbn, array(
1345 '-' => '',
1346 ' ' => '',
1347 'x' => 'X',
1348 ));
1349 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1350 return '<a href="' .
1351 htmlspecialchars( $titleObj->getLocalURL() ) .
1352 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1353 } else {
1354 return $m[0];
1355 }
1356 }
1357
1358 /**
1359 * Make a free external link, given a user-supplied URL
1360 *
1361 * @param string $url
1362 *
1363 * @return string HTML
1364 * @private
1365 */
1366 public function makeFreeExternalLink( $url ) {
1367 wfProfileIn( __METHOD__ );
1368
1369 $trail = '';
1370
1371 # The characters '<' and '>' (which were escaped by
1372 # removeHTMLtags()) should not be included in
1373 # URLs, per RFC 2396.
1374 $m2 = array();
1375 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1376 $trail = substr( $url, $m2[0][1] ) . $trail;
1377 $url = substr( $url, 0, $m2[0][1] );
1378 }
1379
1380 # Move trailing punctuation to $trail
1381 $sep = ',;\.:!?';
1382 # If there is no left bracket, then consider right brackets fair game too
1383 if ( strpos( $url, '(' ) === false ) {
1384 $sep .= ')';
1385 }
1386
1387 $numSepChars = strspn( strrev( $url ), $sep );
1388 if ( $numSepChars ) {
1389 $trail = substr( $url, -$numSepChars ) . $trail;
1390 $url = substr( $url, 0, -$numSepChars );
1391 }
1392
1393 $url = Sanitizer::cleanUrl( $url );
1394
1395 # Is this an external image?
1396 $text = $this->maybeMakeExternalImage( $url );
1397 if ( $text === false ) {
1398 # Not an image, make a link
1399 $text = Linker::makeExternalLink( $url,
1400 $this->getConverterLanguage()->markNoConversion( $url, true ),
1401 true, 'free',
1402 $this->getExternalLinkAttribs( $url ) );
1403 # Register it in the output object...
1404 # Replace unnecessary URL escape codes with their equivalent characters
1405 $pasteurized = self::normalizeLinkUrl( $url );
1406 $this->mOutput->addExternalLink( $pasteurized );
1407 }
1408 wfProfileOut( __METHOD__ );
1409 return $text . $trail;
1410 }
1411
1412 /**
1413 * Parse headers and return html
1414 *
1415 * @private
1416 *
1417 * @param string $text
1418 *
1419 * @return string
1420 */
1421 public function doHeadings( $text ) {
1422 wfProfileIn( __METHOD__ );
1423 for ( $i = 6; $i >= 1; --$i ) {
1424 $h = str_repeat( '=', $i );
1425 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1426 }
1427 wfProfileOut( __METHOD__ );
1428 return $text;
1429 }
1430
1431 /**
1432 * Replace single quotes with HTML markup
1433 * @private
1434 *
1435 * @param string $text
1436 *
1437 * @return string The altered text
1438 */
1439 public function doAllQuotes( $text ) {
1440 wfProfileIn( __METHOD__ );
1441 $outtext = '';
1442 $lines = StringUtils::explode( "\n", $text );
1443 foreach ( $lines as $line ) {
1444 $outtext .= $this->doQuotes( $line ) . "\n";
1445 }
1446 $outtext = substr( $outtext, 0, -1 );
1447 wfProfileOut( __METHOD__ );
1448 return $outtext;
1449 }
1450
1451 /**
1452 * Helper function for doAllQuotes()
1453 *
1454 * @param string $text
1455 *
1456 * @return string
1457 */
1458 public function doQuotes( $text ) {
1459 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1460 $countarr = count( $arr );
1461 if ( $countarr == 1 ) {
1462 return $text;
1463 }
1464
1465 // First, do some preliminary work. This may shift some apostrophes from
1466 // being mark-up to being text. It also counts the number of occurrences
1467 // of bold and italics mark-ups.
1468 $numbold = 0;
1469 $numitalics = 0;
1470 for ( $i = 1; $i < $countarr; $i += 2 ) {
1471 $thislen = strlen( $arr[$i] );
1472 // If there are ever four apostrophes, assume the first is supposed to
1473 // be text, and the remaining three constitute mark-up for bold text.
1474 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1475 if ( $thislen == 4 ) {
1476 $arr[$i - 1] .= "'";
1477 $arr[$i] = "'''";
1478 $thislen = 3;
1479 } elseif ( $thislen > 5 ) {
1480 // If there are more than 5 apostrophes in a row, assume they're all
1481 // text except for the last 5.
1482 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1483 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1484 $arr[$i] = "'''''";
1485 $thislen = 5;
1486 }
1487 // Count the number of occurrences of bold and italics mark-ups.
1488 if ( $thislen == 2 ) {
1489 $numitalics++;
1490 } elseif ( $thislen == 3 ) {
1491 $numbold++;
1492 } elseif ( $thislen == 5 ) {
1493 $numitalics++;
1494 $numbold++;
1495 }
1496 }
1497
1498 // If there is an odd number of both bold and italics, it is likely
1499 // that one of the bold ones was meant to be an apostrophe followed
1500 // by italics. Which one we cannot know for certain, but it is more
1501 // likely to be one that has a single-letter word before it.
1502 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1503 $firstsingleletterword = -1;
1504 $firstmultiletterword = -1;
1505 $firstspace = -1;
1506 for ( $i = 1; $i < $countarr; $i += 2 ) {
1507 if ( strlen( $arr[$i] ) == 3 ) {
1508 $x1 = substr( $arr[$i - 1], -1 );
1509 $x2 = substr( $arr[$i - 1], -2, 1 );
1510 if ( $x1 === ' ' ) {
1511 if ( $firstspace == -1 ) {
1512 $firstspace = $i;
1513 }
1514 } elseif ( $x2 === ' ' ) {
1515 if ( $firstsingleletterword == -1 ) {
1516 $firstsingleletterword = $i;
1517 // if $firstsingleletterword is set, we don't
1518 // look at the other options, so we can bail early.
1519 break;
1520 }
1521 } else {
1522 if ( $firstmultiletterword == -1 ) {
1523 $firstmultiletterword = $i;
1524 }
1525 }
1526 }
1527 }
1528
1529 // If there is a single-letter word, use it!
1530 if ( $firstsingleletterword > -1 ) {
1531 $arr[$firstsingleletterword] = "''";
1532 $arr[$firstsingleletterword - 1] .= "'";
1533 } elseif ( $firstmultiletterword > -1 ) {
1534 // If not, but there's a multi-letter word, use that one.
1535 $arr[$firstmultiletterword] = "''";
1536 $arr[$firstmultiletterword - 1] .= "'";
1537 } elseif ( $firstspace > -1 ) {
1538 // ... otherwise use the first one that has neither.
1539 // (notice that it is possible for all three to be -1 if, for example,
1540 // there is only one pentuple-apostrophe in the line)
1541 $arr[$firstspace] = "''";
1542 $arr[$firstspace - 1] .= "'";
1543 }
1544 }
1545
1546 // Now let's actually convert our apostrophic mush to HTML!
1547 $output = '';
1548 $buffer = '';
1549 $state = '';
1550 $i = 0;
1551 foreach ( $arr as $r ) {
1552 if ( ( $i % 2 ) == 0 ) {
1553 if ( $state === 'both' ) {
1554 $buffer .= $r;
1555 } else {
1556 $output .= $r;
1557 }
1558 } else {
1559 $thislen = strlen( $r );
1560 if ( $thislen == 2 ) {
1561 if ( $state === 'i' ) {
1562 $output .= '</i>';
1563 $state = '';
1564 } elseif ( $state === 'bi' ) {
1565 $output .= '</i>';
1566 $state = 'b';
1567 } elseif ( $state === 'ib' ) {
1568 $output .= '</b></i><b>';
1569 $state = 'b';
1570 } elseif ( $state === 'both' ) {
1571 $output .= '<b><i>' . $buffer . '</i>';
1572 $state = 'b';
1573 } else { // $state can be 'b' or ''
1574 $output .= '<i>';
1575 $state .= 'i';
1576 }
1577 } elseif ( $thislen == 3 ) {
1578 if ( $state === 'b' ) {
1579 $output .= '</b>';
1580 $state = '';
1581 } elseif ( $state === 'bi' ) {
1582 $output .= '</i></b><i>';
1583 $state = 'i';
1584 } elseif ( $state === 'ib' ) {
1585 $output .= '</b>';
1586 $state = 'i';
1587 } elseif ( $state === 'both' ) {
1588 $output .= '<i><b>' . $buffer . '</b>';
1589 $state = 'i';
1590 } else { // $state can be 'i' or ''
1591 $output .= '<b>';
1592 $state .= 'b';
1593 }
1594 } elseif ( $thislen == 5 ) {
1595 if ( $state === 'b' ) {
1596 $output .= '</b><i>';
1597 $state = 'i';
1598 } elseif ( $state === 'i' ) {
1599 $output .= '</i><b>';
1600 $state = 'b';
1601 } elseif ( $state === 'bi' ) {
1602 $output .= '</i></b>';
1603 $state = '';
1604 } elseif ( $state === 'ib' ) {
1605 $output .= '</b></i>';
1606 $state = '';
1607 } elseif ( $state === 'both' ) {
1608 $output .= '<i><b>' . $buffer . '</b></i>';
1609 $state = '';
1610 } else { // ($state == '')
1611 $buffer = '';
1612 $state = 'both';
1613 }
1614 }
1615 }
1616 $i++;
1617 }
1618 // Now close all remaining tags. Notice that the order is important.
1619 if ( $state === 'b' || $state === 'ib' ) {
1620 $output .= '</b>';
1621 }
1622 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1623 $output .= '</i>';
1624 }
1625 if ( $state === 'bi' ) {
1626 $output .= '</b>';
1627 }
1628 // There might be lonely ''''', so make sure we have a buffer
1629 if ( $state === 'both' && $buffer ) {
1630 $output .= '<b><i>' . $buffer . '</i></b>';
1631 }
1632 return $output;
1633 }
1634
1635 /**
1636 * Replace external links (REL)
1637 *
1638 * Note: this is all very hackish and the order of execution matters a lot.
1639 * Make sure to run tests/parserTests.php if you change this code.
1640 *
1641 * @private
1642 *
1643 * @param string $text
1644 *
1645 * @throws MWException
1646 * @return string
1647 */
1648 public function replaceExternalLinks( $text ) {
1649 wfProfileIn( __METHOD__ );
1650
1651 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1652 if ( $bits === false ) {
1653 wfProfileOut( __METHOD__ );
1654 throw new MWException( "PCRE needs to be compiled with "
1655 . "--enable-unicode-properties in order for MediaWiki to function" );
1656 }
1657 $s = array_shift( $bits );
1658
1659 $i = 0;
1660 while ( $i < count( $bits ) ) {
1661 $url = $bits[$i++];
1662 $i++; // protocol
1663 $text = $bits[$i++];
1664 $trail = $bits[$i++];
1665
1666 # The characters '<' and '>' (which were escaped by
1667 # removeHTMLtags()) should not be included in
1668 # URLs, per RFC 2396.
1669 $m2 = array();
1670 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1671 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1672 $url = substr( $url, 0, $m2[0][1] );
1673 }
1674
1675 # If the link text is an image URL, replace it with an <img> tag
1676 # This happened by accident in the original parser, but some people used it extensively
1677 $img = $this->maybeMakeExternalImage( $text );
1678 if ( $img !== false ) {
1679 $text = $img;
1680 }
1681
1682 $dtrail = '';
1683
1684 # Set linktype for CSS - if URL==text, link is essentially free
1685 $linktype = ( $text === $url ) ? 'free' : 'text';
1686
1687 # No link text, e.g. [http://domain.tld/some.link]
1688 if ( $text == '' ) {
1689 # Autonumber
1690 $langObj = $this->getTargetLanguage();
1691 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1692 $linktype = 'autonumber';
1693 } else {
1694 # Have link text, e.g. [http://domain.tld/some.link text]s
1695 # Check for trail
1696 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1697 }
1698
1699 $text = $this->getConverterLanguage()->markNoConversion( $text );
1700
1701 $url = Sanitizer::cleanUrl( $url );
1702
1703 # Use the encoded URL
1704 # This means that users can paste URLs directly into the text
1705 # Funny characters like ö aren't valid in URLs anyway
1706 # This was changed in August 2004
1707 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1708 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1709
1710 # Register link in the output object.
1711 # Replace unnecessary URL escape codes with the referenced character
1712 # This prevents spammers from hiding links from the filters
1713 $pasteurized = self::normalizeLinkUrl( $url );
1714 $this->mOutput->addExternalLink( $pasteurized );
1715 }
1716
1717 wfProfileOut( __METHOD__ );
1718 return $s;
1719 }
1720
1721 /**
1722 * Get the rel attribute for a particular external link.
1723 *
1724 * @since 1.21
1725 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1726 * nofollow if appropriate
1727 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1728 * @return string|null Rel attribute for $url
1729 */
1730 public static function getExternalLinkRel( $url = false, $title = null ) {
1731 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1732 $ns = $title ? $title->getNamespace() : false;
1733 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1734 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1735 ) {
1736 return 'nofollow';
1737 }
1738 return null;
1739 }
1740
1741 /**
1742 * Get an associative array of additional HTML attributes appropriate for a
1743 * particular external link. This currently may include rel => nofollow
1744 * (depending on configuration, namespace, and the URL's domain) and/or a
1745 * target attribute (depending on configuration).
1746 *
1747 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1748 * nofollow if appropriate
1749 * @return array Associative array of HTML attributes
1750 */
1751 public function getExternalLinkAttribs( $url = false ) {
1752 $attribs = array();
1753 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1754
1755 if ( $this->mOptions->getExternalLinkTarget() ) {
1756 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1757 }
1758 return $attribs;
1759 }
1760
1761 /**
1762 * Replace unusual escape codes in a URL with their equivalent characters
1763 *
1764 * @deprecated since 1.24, use normalizeLinkUrl
1765 * @param string $url
1766 * @return string
1767 */
1768 public static function replaceUnusualEscapes( $url ) {
1769 wfDeprecated( __METHOD__, '1.24' );
1770 return self::normalizeLinkUrl( $url );
1771 }
1772
1773 /**
1774 * Replace unusual escape codes in a URL with their equivalent characters
1775 *
1776 * This generally follows the syntax defined in RFC 3986, with special
1777 * consideration for HTTP query strings.
1778 *
1779 * @param string $url
1780 * @return string
1781 */
1782 public static function normalizeLinkUrl( $url ) {
1783 # First, make sure unsafe characters are encoded
1784 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1785 function ( $m ) {
1786 return rawurlencode( $m[0] );
1787 },
1788 $url
1789 );
1790
1791 $ret = '';
1792 $end = strlen( $url );
1793
1794 # Fragment part - 'fragment'
1795 $start = strpos( $url, '#' );
1796 if ( $start !== false && $start < $end ) {
1797 $ret = self::normalizeUrlComponent(
1798 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1799 $end = $start;
1800 }
1801
1802 # Query part - 'query' minus &=+;
1803 $start = strpos( $url, '?' );
1804 if ( $start !== false && $start < $end ) {
1805 $ret = self::normalizeUrlComponent(
1806 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1807 $end = $start;
1808 }
1809
1810 # Scheme and path part - 'pchar'
1811 # (we assume no userinfo or encoded colons in the host)
1812 $ret = self::normalizeUrlComponent(
1813 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1814
1815 return $ret;
1816 }
1817
1818 private static function normalizeUrlComponent( $component, $unsafe ) {
1819 $callback = function ( $matches ) use ( $unsafe ) {
1820 $char = urldecode( $matches[0] );
1821 $ord = ord( $char );
1822 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1823 # Unescape it
1824 return $char;
1825 } else {
1826 # Leave it escaped, but use uppercase for a-f
1827 return strtoupper( $matches[0] );
1828 }
1829 };
1830 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1831 }
1832
1833 /**
1834 * make an image if it's allowed, either through the global
1835 * option, through the exception, or through the on-wiki whitelist
1836 *
1837 * @param string $url
1838 *
1839 * @return string
1840 */
1841 private function maybeMakeExternalImage( $url ) {
1842 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1843 $imagesexception = !empty( $imagesfrom );
1844 $text = false;
1845 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1846 if ( $imagesexception && is_array( $imagesfrom ) ) {
1847 $imagematch = false;
1848 foreach ( $imagesfrom as $match ) {
1849 if ( strpos( $url, $match ) === 0 ) {
1850 $imagematch = true;
1851 break;
1852 }
1853 }
1854 } elseif ( $imagesexception ) {
1855 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1856 } else {
1857 $imagematch = false;
1858 }
1859
1860 if ( $this->mOptions->getAllowExternalImages()
1861 || ( $imagesexception && $imagematch )
1862 ) {
1863 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1864 # Image found
1865 $text = Linker::makeExternalImage( $url );
1866 }
1867 }
1868 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1869 && preg_match( self::EXT_IMAGE_REGEX, $url )
1870 ) {
1871 $whitelist = explode(
1872 "\n",
1873 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
1874 );
1875
1876 foreach ( $whitelist as $entry ) {
1877 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1878 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1879 continue;
1880 }
1881 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1882 # Image matches a whitelist entry
1883 $text = Linker::makeExternalImage( $url );
1884 break;
1885 }
1886 }
1887 }
1888 return $text;
1889 }
1890
1891 /**
1892 * Process [[ ]] wikilinks
1893 *
1894 * @param string $s
1895 *
1896 * @return string Processed text
1897 *
1898 * @private
1899 */
1900 public function replaceInternalLinks( $s ) {
1901 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1902 return $s;
1903 }
1904
1905 /**
1906 * Process [[ ]] wikilinks (RIL)
1907 * @param string $s
1908 * @throws MWException
1909 * @return LinkHolderArray
1910 *
1911 * @private
1912 */
1913 public function replaceInternalLinks2( &$s ) {
1914 global $wgExtraInterlanguageLinkPrefixes;
1915 wfProfileIn( __METHOD__ );
1916
1917 wfProfileIn( __METHOD__ . '-setup' );
1918 static $tc = false, $e1, $e1_img;
1919 # the % is needed to support urlencoded titles as well
1920 if ( !$tc ) {
1921 $tc = Title::legalChars() . '#%';
1922 # Match a link having the form [[namespace:link|alternate]]trail
1923 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1924 # Match cases where there is no "]]", which might still be images
1925 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1926 }
1927
1928 $holders = new LinkHolderArray( $this );
1929
1930 # split the entire text string on occurrences of [[
1931 $a = StringUtils::explode( '[[', ' ' . $s );
1932 # get the first element (all text up to first [[), and remove the space we added
1933 $s = $a->current();
1934 $a->next();
1935 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1936 $s = substr( $s, 1 );
1937
1938 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1939 $e2 = null;
1940 if ( $useLinkPrefixExtension ) {
1941 # Match the end of a line for a word that's not followed by whitespace,
1942 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1943 global $wgContLang;
1944 $charset = $wgContLang->linkPrefixCharset();
1945 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
1946 }
1947
1948 if ( is_null( $this->mTitle ) ) {
1949 wfProfileOut( __METHOD__ . '-setup' );
1950 wfProfileOut( __METHOD__ );
1951 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
1952 }
1953 $nottalk = !$this->mTitle->isTalkPage();
1954
1955 if ( $useLinkPrefixExtension ) {
1956 $m = array();
1957 if ( preg_match( $e2, $s, $m ) ) {
1958 $first_prefix = $m[2];
1959 } else {
1960 $first_prefix = false;
1961 }
1962 } else {
1963 $prefix = '';
1964 }
1965
1966 $useSubpages = $this->areSubpagesAllowed();
1967 wfProfileOut( __METHOD__ . '-setup' );
1968
1969 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
1970 # Loop for each link
1971 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
1972 // @codingStandardsIgnoreStart
1973
1974 # Check for excessive memory usage
1975 if ( $holders->isBig() ) {
1976 # Too big
1977 # Do the existence check, replace the link holders and clear the array
1978 $holders->replace( $s );
1979 $holders->clear();
1980 }
1981
1982 if ( $useLinkPrefixExtension ) {
1983 wfProfileIn( __METHOD__ . '-prefixhandling' );
1984 if ( preg_match( $e2, $s, $m ) ) {
1985 $prefix = $m[2];
1986 $s = $m[1];
1987 } else {
1988 $prefix = '';
1989 }
1990 # first link
1991 if ( $first_prefix ) {
1992 $prefix = $first_prefix;
1993 $first_prefix = false;
1994 }
1995 wfProfileOut( __METHOD__ . '-prefixhandling' );
1996 }
1997
1998 $might_be_img = false;
1999
2000 wfProfileIn( __METHOD__ . "-e1" );
2001 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2002 $text = $m[2];
2003 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2004 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2005 # the real problem is with the $e1 regex
2006 # See bug 1300.
2007 #
2008 # Still some problems for cases where the ] is meant to be outside punctuation,
2009 # and no image is in sight. See bug 2095.
2010 #
2011 if ( $text !== ''
2012 && substr( $m[3], 0, 1 ) === ']'
2013 && strpos( $text, '[' ) !== false
2014 ) {
2015 $text .= ']'; # so that replaceExternalLinks($text) works later
2016 $m[3] = substr( $m[3], 1 );
2017 }
2018 # fix up urlencoded title texts
2019 if ( strpos( $m[1], '%' ) !== false ) {
2020 # Should anchors '#' also be rejected?
2021 $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
2022 }
2023 $trail = $m[3];
2024 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2025 # Invalid, but might be an image with a link in its caption
2026 $might_be_img = true;
2027 $text = $m[2];
2028 if ( strpos( $m[1], '%' ) !== false ) {
2029 $m[1] = rawurldecode( $m[1] );
2030 }
2031 $trail = "";
2032 } else { # Invalid form; output directly
2033 $s .= $prefix . '[[' . $line;
2034 wfProfileOut( __METHOD__ . "-e1" );
2035 continue;
2036 }
2037 wfProfileOut( __METHOD__ . "-e1" );
2038 wfProfileIn( __METHOD__ . "-misc" );
2039
2040 $origLink = $m[1];
2041
2042 # Don't allow internal links to pages containing
2043 # PROTO: where PROTO is a valid URL protocol; these
2044 # should be external links.
2045 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2046 $s .= $prefix . '[[' . $line;
2047 wfProfileOut( __METHOD__ . "-misc" );
2048 continue;
2049 }
2050
2051 # Make subpage if necessary
2052 if ( $useSubpages ) {
2053 $link = $this->maybeDoSubpageLink( $origLink, $text );
2054 } else {
2055 $link = $origLink;
2056 }
2057
2058 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2059 if ( !$noforce ) {
2060 # Strip off leading ':'
2061 $link = substr( $link, 1 );
2062 }
2063
2064 wfProfileOut( __METHOD__ . "-misc" );
2065 wfProfileIn( __METHOD__ . "-title" );
2066 $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
2067 if ( $nt === null ) {
2068 $s .= $prefix . '[[' . $line;
2069 wfProfileOut( __METHOD__ . "-title" );
2070 continue;
2071 }
2072
2073 $ns = $nt->getNamespace();
2074 $iw = $nt->getInterwiki();
2075 wfProfileOut( __METHOD__ . "-title" );
2076
2077 if ( $might_be_img ) { # if this is actually an invalid link
2078 wfProfileIn( __METHOD__ . "-might_be_img" );
2079 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2080 $found = false;
2081 while ( true ) {
2082 # look at the next 'line' to see if we can close it there
2083 $a->next();
2084 $next_line = $a->current();
2085 if ( $next_line === false || $next_line === null ) {
2086 break;
2087 }
2088 $m = explode( ']]', $next_line, 3 );
2089 if ( count( $m ) == 3 ) {
2090 # the first ]] closes the inner link, the second the image
2091 $found = true;
2092 $text .= "[[{$m[0]}]]{$m[1]}";
2093 $trail = $m[2];
2094 break;
2095 } elseif ( count( $m ) == 2 ) {
2096 # if there's exactly one ]] that's fine, we'll keep looking
2097 $text .= "[[{$m[0]}]]{$m[1]}";
2098 } else {
2099 # if $next_line is invalid too, we need look no further
2100 $text .= '[[' . $next_line;
2101 break;
2102 }
2103 }
2104 if ( !$found ) {
2105 # we couldn't find the end of this imageLink, so output it raw
2106 # but don't ignore what might be perfectly normal links in the text we've examined
2107 $holders->merge( $this->replaceInternalLinks2( $text ) );
2108 $s .= "{$prefix}[[$link|$text";
2109 # note: no $trail, because without an end, there *is* no trail
2110 wfProfileOut( __METHOD__ . "-might_be_img" );
2111 continue;
2112 }
2113 } else { # it's not an image, so output it raw
2114 $s .= "{$prefix}[[$link|$text";
2115 # note: no $trail, because without an end, there *is* no trail
2116 wfProfileOut( __METHOD__ . "-might_be_img" );
2117 continue;
2118 }
2119 wfProfileOut( __METHOD__ . "-might_be_img" );
2120 }
2121
2122 $wasblank = ( $text == '' );
2123 if ( $wasblank ) {
2124 $text = $link;
2125 } else {
2126 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2127 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2128 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2129 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2130 $text = $this->doQuotes( $text );
2131 }
2132
2133 # Link not escaped by : , create the various objects
2134 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2135 # Interwikis
2136 wfProfileIn( __METHOD__ . "-interwiki" );
2137 if (
2138 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2139 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2140 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2141 )
2142 ) {
2143 # Bug 24502: filter duplicates
2144 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2145 $this->mLangLinkLanguages[$iw] = true;
2146 $this->mOutput->addLanguageLink( $nt->getFullText() );
2147 }
2148
2149 $s = rtrim( $s . $prefix );
2150 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2151 wfProfileOut( __METHOD__ . "-interwiki" );
2152 continue;
2153 }
2154 wfProfileOut( __METHOD__ . "-interwiki" );
2155
2156 if ( $ns == NS_FILE ) {
2157 wfProfileIn( __METHOD__ . "-image" );
2158 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2159 if ( $wasblank ) {
2160 # if no parameters were passed, $text
2161 # becomes something like "File:Foo.png",
2162 # which we don't want to pass on to the
2163 # image generator
2164 $text = '';
2165 } else {
2166 # recursively parse links inside the image caption
2167 # actually, this will parse them in any other parameters, too,
2168 # but it might be hard to fix that, and it doesn't matter ATM
2169 $text = $this->replaceExternalLinks( $text );
2170 $holders->merge( $this->replaceInternalLinks2( $text ) );
2171 }
2172 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2173 $s .= $prefix . $this->armorLinks(
2174 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2175 } else {
2176 $s .= $prefix . $trail;
2177 }
2178 wfProfileOut( __METHOD__ . "-image" );
2179 continue;
2180 }
2181
2182 if ( $ns == NS_CATEGORY ) {
2183 wfProfileIn( __METHOD__ . "-category" );
2184 $s = rtrim( $s . "\n" ); # bug 87
2185
2186 if ( $wasblank ) {
2187 $sortkey = $this->getDefaultSort();
2188 } else {
2189 $sortkey = $text;
2190 }
2191 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2192 $sortkey = str_replace( "\n", '', $sortkey );
2193 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2194 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2195
2196 /**
2197 * Strip the whitespace Category links produce, see bug 87
2198 */
2199 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2200
2201 wfProfileOut( __METHOD__ . "-category" );
2202 continue;
2203 }
2204 }
2205
2206 # Self-link checking. For some languages, variants of the title are checked in
2207 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2208 # for linking to a different variant.
2209 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2210 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2211 continue;
2212 }
2213
2214 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2215 # @todo FIXME: Should do batch file existence checks, see comment below
2216 if ( $ns == NS_MEDIA ) {
2217 wfProfileIn( __METHOD__ . "-media" );
2218 # Give extensions a chance to select the file revision for us
2219 $options = array();
2220 $descQuery = false;
2221 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2222 array( $this, $nt, &$options, &$descQuery ) );
2223 # Fetch and register the file (file title may be different via hooks)
2224 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2225 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2226 $s .= $prefix . $this->armorLinks(
2227 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2228 wfProfileOut( __METHOD__ . "-media" );
2229 continue;
2230 }
2231
2232 wfProfileIn( __METHOD__ . "-always_known" );
2233 # Some titles, such as valid special pages or files in foreign repos, should
2234 # be shown as bluelinks even though they're not included in the page table
2235 #
2236 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2237 # batch file existence checks for NS_FILE and NS_MEDIA
2238 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2239 $this->mOutput->addLink( $nt );
2240 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2241 } else {
2242 # Links will be added to the output link list after checking
2243 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2244 }
2245 wfProfileOut( __METHOD__ . "-always_known" );
2246 }
2247 wfProfileOut( __METHOD__ );
2248 return $holders;
2249 }
2250
2251 /**
2252 * Render a forced-blue link inline; protect against double expansion of
2253 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2254 * Since this little disaster has to split off the trail text to avoid
2255 * breaking URLs in the following text without breaking trails on the
2256 * wiki links, it's been made into a horrible function.
2257 *
2258 * @param Title $nt
2259 * @param string $text
2260 * @param array|string $query
2261 * @param string $trail
2262 * @param string $prefix
2263 * @return string HTML-wikitext mix oh yuck
2264 */
2265 public function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2266 list( $inside, $trail ) = Linker::splitTrail( $trail );
2267
2268 if ( is_string( $query ) ) {
2269 $query = wfCgiToArray( $query );
2270 }
2271 if ( $text == '' ) {
2272 $text = htmlspecialchars( $nt->getPrefixedText() );
2273 }
2274
2275 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2276
2277 return $this->armorLinks( $link ) . $trail;
2278 }
2279
2280 /**
2281 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2282 * going to go through further parsing steps before inline URL expansion.
2283 *
2284 * Not needed quite as much as it used to be since free links are a bit
2285 * more sensible these days. But bracketed links are still an issue.
2286 *
2287 * @param string $text More-or-less HTML
2288 * @return string Less-or-more HTML with NOPARSE bits
2289 */
2290 public function armorLinks( $text ) {
2291 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2292 "{$this->mUniqPrefix}NOPARSE$1", $text );
2293 }
2294
2295 /**
2296 * Return true if subpage links should be expanded on this page.
2297 * @return bool
2298 */
2299 public function areSubpagesAllowed() {
2300 # Some namespaces don't allow subpages
2301 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2302 }
2303
2304 /**
2305 * Handle link to subpage if necessary
2306 *
2307 * @param string $target The source of the link
2308 * @param string &$text The link text, modified as necessary
2309 * @return string The full name of the link
2310 * @private
2311 */
2312 public function maybeDoSubpageLink( $target, &$text ) {
2313 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2314 }
2315
2316 /**#@+
2317 * Used by doBlockLevels()
2318 * @private
2319 *
2320 * @return string
2321 */
2322 public function closeParagraph() {
2323 $result = '';
2324 if ( $this->mLastSection != '' ) {
2325 $result = '</' . $this->mLastSection . ">\n";
2326 }
2327 $this->mInPre = false;
2328 $this->mLastSection = '';
2329 return $result;
2330 }
2331
2332 /**
2333 * getCommon() returns the length of the longest common substring
2334 * of both arguments, starting at the beginning of both.
2335 * @private
2336 *
2337 * @param string $st1
2338 * @param string $st2
2339 *
2340 * @return int
2341 */
2342 public function getCommon( $st1, $st2 ) {
2343 $fl = strlen( $st1 );
2344 $shorter = strlen( $st2 );
2345 if ( $fl < $shorter ) {
2346 $shorter = $fl;
2347 }
2348
2349 for ( $i = 0; $i < $shorter; ++$i ) {
2350 if ( $st1[$i] != $st2[$i] ) {
2351 break;
2352 }
2353 }
2354 return $i;
2355 }
2356
2357 /**
2358 * These next three functions open, continue, and close the list
2359 * element appropriate to the prefix character passed into them.
2360 * @private
2361 *
2362 * @param string $char
2363 *
2364 * @return string
2365 */
2366 public function openList( $char ) {
2367 $result = $this->closeParagraph();
2368
2369 if ( '*' === $char ) {
2370 $result .= "<ul><li>";
2371 } elseif ( '#' === $char ) {
2372 $result .= "<ol><li>";
2373 } elseif ( ':' === $char ) {
2374 $result .= "<dl><dd>";
2375 } elseif ( ';' === $char ) {
2376 $result .= "<dl><dt>";
2377 $this->mDTopen = true;
2378 } else {
2379 $result = '<!-- ERR 1 -->';
2380 }
2381
2382 return $result;
2383 }
2384
2385 /**
2386 * TODO: document
2387 * @param string $char
2388 * @private
2389 *
2390 * @return string
2391 */
2392 public function nextItem( $char ) {
2393 if ( '*' === $char || '#' === $char ) {
2394 return "</li>\n<li>";
2395 } elseif ( ':' === $char || ';' === $char ) {
2396 $close = "</dd>\n";
2397 if ( $this->mDTopen ) {
2398 $close = "</dt>\n";
2399 }
2400 if ( ';' === $char ) {
2401 $this->mDTopen = true;
2402 return $close . '<dt>';
2403 } else {
2404 $this->mDTopen = false;
2405 return $close . '<dd>';
2406 }
2407 }
2408 return '<!-- ERR 2 -->';
2409 }
2410
2411 /**
2412 * @todo Document
2413 * @param string $char
2414 * @private
2415 *
2416 * @return string
2417 */
2418 public function closeList( $char ) {
2419 if ( '*' === $char ) {
2420 $text = "</li></ul>";
2421 } elseif ( '#' === $char ) {
2422 $text = "</li></ol>";
2423 } elseif ( ':' === $char ) {
2424 if ( $this->mDTopen ) {
2425 $this->mDTopen = false;
2426 $text = "</dt></dl>";
2427 } else {
2428 $text = "</dd></dl>";
2429 }
2430 } else {
2431 return '<!-- ERR 3 -->';
2432 }
2433 return $text;
2434 }
2435 /**#@-*/
2436
2437 /**
2438 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2439 *
2440 * @param string $text
2441 * @param bool $linestart Whether or not this is at the start of a line.
2442 * @private
2443 * @return string The lists rendered as HTML
2444 */
2445 public function doBlockLevels( $text, $linestart ) {
2446 wfProfileIn( __METHOD__ );
2447
2448 # Parsing through the text line by line. The main thing
2449 # happening here is handling of block-level elements p, pre,
2450 # and making lists from lines starting with * # : etc.
2451 #
2452 $textLines = StringUtils::explode( "\n", $text );
2453
2454 $lastPrefix = $output = '';
2455 $this->mDTopen = $inBlockElem = false;
2456 $prefixLength = 0;
2457 $paragraphStack = false;
2458 $inBlockquote = false;
2459
2460 foreach ( $textLines as $oLine ) {
2461 # Fix up $linestart
2462 if ( !$linestart ) {
2463 $output .= $oLine;
2464 $linestart = true;
2465 continue;
2466 }
2467 # * = ul
2468 # # = ol
2469 # ; = dt
2470 # : = dd
2471
2472 $lastPrefixLength = strlen( $lastPrefix );
2473 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2474 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2475 # If not in a <pre> element, scan for and figure out what prefixes are there.
2476 if ( !$this->mInPre ) {
2477 # Multiple prefixes may abut each other for nested lists.
2478 $prefixLength = strspn( $oLine, '*#:;' );
2479 $prefix = substr( $oLine, 0, $prefixLength );
2480
2481 # eh?
2482 # ; and : are both from definition-lists, so they're equivalent
2483 # for the purposes of determining whether or not we need to open/close
2484 # elements.
2485 $prefix2 = str_replace( ';', ':', $prefix );
2486 $t = substr( $oLine, $prefixLength );
2487 $this->mInPre = (bool)$preOpenMatch;
2488 } else {
2489 # Don't interpret any other prefixes in preformatted text
2490 $prefixLength = 0;
2491 $prefix = $prefix2 = '';
2492 $t = $oLine;
2493 }
2494
2495 # List generation
2496 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2497 # Same as the last item, so no need to deal with nesting or opening stuff
2498 $output .= $this->nextItem( substr( $prefix, -1 ) );
2499 $paragraphStack = false;
2500
2501 if ( substr( $prefix, -1 ) === ';' ) {
2502 # The one nasty exception: definition lists work like this:
2503 # ; title : definition text
2504 # So we check for : in the remainder text to split up the
2505 # title and definition, without b0rking links.
2506 $term = $t2 = '';
2507 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2508 $t = $t2;
2509 $output .= $term . $this->nextItem( ':' );
2510 }
2511 }
2512 } elseif ( $prefixLength || $lastPrefixLength ) {
2513 # We need to open or close prefixes, or both.
2514
2515 # Either open or close a level...
2516 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2517 $paragraphStack = false;
2518
2519 # Close all the prefixes which aren't shared.
2520 while ( $commonPrefixLength < $lastPrefixLength ) {
2521 $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2522 --$lastPrefixLength;
2523 }
2524
2525 # Continue the current prefix if appropriate.
2526 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2527 $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2528 }
2529
2530 # Open prefixes where appropriate.
2531 if ( $lastPrefix && $prefixLength > $commonPrefixLength ) {
2532 $output .= "\n";
2533 }
2534 while ( $prefixLength > $commonPrefixLength ) {
2535 $char = substr( $prefix, $commonPrefixLength, 1 );
2536 $output .= $this->openList( $char );
2537
2538 if ( ';' === $char ) {
2539 # @todo FIXME: This is dupe of code above
2540 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2541 $t = $t2;
2542 $output .= $term . $this->nextItem( ':' );
2543 }
2544 }
2545 ++$commonPrefixLength;
2546 }
2547 if ( !$prefixLength && $lastPrefix ) {
2548 $output .= "\n";
2549 }
2550 $lastPrefix = $prefix2;
2551 }
2552
2553 # If we have no prefixes, go to paragraph mode.
2554 if ( 0 == $prefixLength ) {
2555 wfProfileIn( __METHOD__ . "-paragraph" );
2556 # No prefix (not in list)--go to paragraph mode
2557 # XXX: use a stack for nestable elements like span, table and div
2558 $openmatch = preg_match(
2559 '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|'
2560 . '<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS',
2561 $t
2562 );
2563 $closematch = preg_match(
2564 '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'
2565 . '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|<\\/mw:|'
2566 . $this->mUniqPrefix
2567 . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS',
2568 $t
2569 );
2570
2571 if ( $openmatch or $closematch ) {
2572 $paragraphStack = false;
2573 # @todo bug 5718: paragraph closed
2574 $output .= $this->closeParagraph();
2575 if ( $preOpenMatch and !$preCloseMatch ) {
2576 $this->mInPre = true;
2577 }
2578 $bqOffset = 0;
2579 while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
2580 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2581 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2582 }
2583 $inBlockElem = !$closematch;
2584 } elseif ( !$inBlockElem && !$this->mInPre ) {
2585 if ( ' ' == substr( $t, 0, 1 )
2586 && ( $this->mLastSection === 'pre' || trim( $t ) != '' )
2587 && !$inBlockquote
2588 ) {
2589 # pre
2590 if ( $this->mLastSection !== 'pre' ) {
2591 $paragraphStack = false;
2592 $output .= $this->closeParagraph() . '<pre>';
2593 $this->mLastSection = 'pre';
2594 }
2595 $t = substr( $t, 1 );
2596 } else {
2597 # paragraph
2598 if ( trim( $t ) === '' ) {
2599 if ( $paragraphStack ) {
2600 $output .= $paragraphStack . '<br />';
2601 $paragraphStack = false;
2602 $this->mLastSection = 'p';
2603 } else {
2604 if ( $this->mLastSection !== 'p' ) {
2605 $output .= $this->closeParagraph();
2606 $this->mLastSection = '';
2607 $paragraphStack = '<p>';
2608 } else {
2609 $paragraphStack = '</p><p>';
2610 }
2611 }
2612 } else {
2613 if ( $paragraphStack ) {
2614 $output .= $paragraphStack;
2615 $paragraphStack = false;
2616 $this->mLastSection = 'p';
2617 } elseif ( $this->mLastSection !== 'p' ) {
2618 $output .= $this->closeParagraph() . '<p>';
2619 $this->mLastSection = 'p';
2620 }
2621 }
2622 }
2623 }
2624 wfProfileOut( __METHOD__ . "-paragraph" );
2625 }
2626 # somewhere above we forget to get out of pre block (bug 785)
2627 if ( $preCloseMatch && $this->mInPre ) {
2628 $this->mInPre = false;
2629 }
2630 if ( $paragraphStack === false ) {
2631 $output .= $t;
2632 if ( $prefixLength === 0 ) {
2633 $output .= "\n";
2634 }
2635 }
2636 }
2637 while ( $prefixLength ) {
2638 $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2639 --$prefixLength;
2640 if ( !$prefixLength ) {
2641 $output .= "\n";
2642 }
2643 }
2644 if ( $this->mLastSection != '' ) {
2645 $output .= '</' . $this->mLastSection . '>';
2646 $this->mLastSection = '';
2647 }
2648
2649 wfProfileOut( __METHOD__ );
2650 return $output;
2651 }
2652
2653 /**
2654 * Split up a string on ':', ignoring any occurrences inside tags
2655 * to prevent illegal overlapping.
2656 *
2657 * @param string $str The string to split
2658 * @param string &$before Set to everything before the ':'
2659 * @param string &$after Set to everything after the ':'
2660 * @throws MWException
2661 * @return string The position of the ':', or false if none found
2662 */
2663 public function findColonNoLinks( $str, &$before, &$after ) {
2664 wfProfileIn( __METHOD__ );
2665
2666 $pos = strpos( $str, ':' );
2667 if ( $pos === false ) {
2668 # Nothing to find!
2669 wfProfileOut( __METHOD__ );
2670 return false;
2671 }
2672
2673 $lt = strpos( $str, '<' );
2674 if ( $lt === false || $lt > $pos ) {
2675 # Easy; no tag nesting to worry about
2676 $before = substr( $str, 0, $pos );
2677 $after = substr( $str, $pos + 1 );
2678 wfProfileOut( __METHOD__ );
2679 return $pos;
2680 }
2681
2682 # Ugly state machine to walk through avoiding tags.
2683 $state = self::COLON_STATE_TEXT;
2684 $stack = 0;
2685 $len = strlen( $str );
2686 for ( $i = 0; $i < $len; $i++ ) {
2687 $c = $str[$i];
2688
2689 switch ( $state ) {
2690 # (Using the number is a performance hack for common cases)
2691 case 0: # self::COLON_STATE_TEXT:
2692 switch ( $c ) {
2693 case "<":
2694 # Could be either a <start> tag or an </end> tag
2695 $state = self::COLON_STATE_TAGSTART;
2696 break;
2697 case ":":
2698 if ( $stack == 0 ) {
2699 # We found it!
2700 $before = substr( $str, 0, $i );
2701 $after = substr( $str, $i + 1 );
2702 wfProfileOut( __METHOD__ );
2703 return $i;
2704 }
2705 # Embedded in a tag; don't break it.
2706 break;
2707 default:
2708 # Skip ahead looking for something interesting
2709 $colon = strpos( $str, ':', $i );
2710 if ( $colon === false ) {
2711 # Nothing else interesting
2712 wfProfileOut( __METHOD__ );
2713 return false;
2714 }
2715 $lt = strpos( $str, '<', $i );
2716 if ( $stack === 0 ) {
2717 if ( $lt === false || $colon < $lt ) {
2718 # We found it!
2719 $before = substr( $str, 0, $colon );
2720 $after = substr( $str, $colon + 1 );
2721 wfProfileOut( __METHOD__ );
2722 return $i;
2723 }
2724 }
2725 if ( $lt === false ) {
2726 # Nothing else interesting to find; abort!
2727 # We're nested, but there's no close tags left. Abort!
2728 break 2;
2729 }
2730 # Skip ahead to next tag start
2731 $i = $lt;
2732 $state = self::COLON_STATE_TAGSTART;
2733 }
2734 break;
2735 case 1: # self::COLON_STATE_TAG:
2736 # In a <tag>
2737 switch ( $c ) {
2738 case ">":
2739 $stack++;
2740 $state = self::COLON_STATE_TEXT;
2741 break;
2742 case "/":
2743 # Slash may be followed by >?
2744 $state = self::COLON_STATE_TAGSLASH;
2745 break;
2746 default:
2747 # ignore
2748 }
2749 break;
2750 case 2: # self::COLON_STATE_TAGSTART:
2751 switch ( $c ) {
2752 case "/":
2753 $state = self::COLON_STATE_CLOSETAG;
2754 break;
2755 case "!":
2756 $state = self::COLON_STATE_COMMENT;
2757 break;
2758 case ">":
2759 # Illegal early close? This shouldn't happen D:
2760 $state = self::COLON_STATE_TEXT;
2761 break;
2762 default:
2763 $state = self::COLON_STATE_TAG;
2764 }
2765 break;
2766 case 3: # self::COLON_STATE_CLOSETAG:
2767 # In a </tag>
2768 if ( $c === ">" ) {
2769 $stack--;
2770 if ( $stack < 0 ) {
2771 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2772 wfProfileOut( __METHOD__ );
2773 return false;
2774 }
2775 $state = self::COLON_STATE_TEXT;
2776 }
2777 break;
2778 case self::COLON_STATE_TAGSLASH:
2779 if ( $c === ">" ) {
2780 # Yes, a self-closed tag <blah/>
2781 $state = self::COLON_STATE_TEXT;
2782 } else {
2783 # Probably we're jumping the gun, and this is an attribute
2784 $state = self::COLON_STATE_TAG;
2785 }
2786 break;
2787 case 5: # self::COLON_STATE_COMMENT:
2788 if ( $c === "-" ) {
2789 $state = self::COLON_STATE_COMMENTDASH;
2790 }
2791 break;
2792 case self::COLON_STATE_COMMENTDASH:
2793 if ( $c === "-" ) {
2794 $state = self::COLON_STATE_COMMENTDASHDASH;
2795 } else {
2796 $state = self::COLON_STATE_COMMENT;
2797 }
2798 break;
2799 case self::COLON_STATE_COMMENTDASHDASH:
2800 if ( $c === ">" ) {
2801 $state = self::COLON_STATE_TEXT;
2802 } else {
2803 $state = self::COLON_STATE_COMMENT;
2804 }
2805 break;
2806 default:
2807 wfProfileOut( __METHOD__ );
2808 throw new MWException( "State machine error in " . __METHOD__ );
2809 }
2810 }
2811 if ( $stack > 0 ) {
2812 wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2813 wfProfileOut( __METHOD__ );
2814 return false;
2815 }
2816 wfProfileOut( __METHOD__ );
2817 return false;
2818 }
2819
2820 /**
2821 * Return value of a magic variable (like PAGENAME)
2822 *
2823 * @private
2824 *
2825 * @param int $index
2826 * @param bool|PPFrame $frame
2827 *
2828 * @throws MWException
2829 * @return string
2830 */
2831 public function getVariableValue( $index, $frame = false ) {
2832 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2833 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2834
2835 if ( is_null( $this->mTitle ) ) {
2836 // If no title set, bad things are going to happen
2837 // later. Title should always be set since this
2838 // should only be called in the middle of a parse
2839 // operation (but the unit-tests do funky stuff)
2840 throw new MWException( __METHOD__ . ' Should only be '
2841 . ' called while parsing (no title set)' );
2842 }
2843
2844 /**
2845 * Some of these require message or data lookups and can be
2846 * expensive to check many times.
2847 */
2848 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2849 if ( isset( $this->mVarCache[$index] ) ) {
2850 return $this->mVarCache[$index];
2851 }
2852 }
2853
2854 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2855 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2856
2857 $pageLang = $this->getFunctionLang();
2858
2859 switch ( $index ) {
2860 case '!':
2861 $value = '|';
2862 break;
2863 case 'currentmonth':
2864 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2865 break;
2866 case 'currentmonth1':
2867 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2868 break;
2869 case 'currentmonthname':
2870 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2871 break;
2872 case 'currentmonthnamegen':
2873 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2874 break;
2875 case 'currentmonthabbrev':
2876 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2877 break;
2878 case 'currentday':
2879 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2880 break;
2881 case 'currentday2':
2882 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2883 break;
2884 case 'localmonth':
2885 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2886 break;
2887 case 'localmonth1':
2888 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2889 break;
2890 case 'localmonthname':
2891 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2892 break;
2893 case 'localmonthnamegen':
2894 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2895 break;
2896 case 'localmonthabbrev':
2897 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2898 break;
2899 case 'localday':
2900 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2901 break;
2902 case 'localday2':
2903 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2904 break;
2905 case 'pagename':
2906 $value = wfEscapeWikiText( $this->mTitle->getText() );
2907 break;
2908 case 'pagenamee':
2909 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2910 break;
2911 case 'fullpagename':
2912 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2913 break;
2914 case 'fullpagenamee':
2915 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2916 break;
2917 case 'subpagename':
2918 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2919 break;
2920 case 'subpagenamee':
2921 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2922 break;
2923 case 'rootpagename':
2924 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2925 break;
2926 case 'rootpagenamee':
2927 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2928 ' ',
2929 '_',
2930 $this->mTitle->getRootText()
2931 ) ) );
2932 break;
2933 case 'basepagename':
2934 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2935 break;
2936 case 'basepagenamee':
2937 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2938 ' ',
2939 '_',
2940 $this->mTitle->getBaseText()
2941 ) ) );
2942 break;
2943 case 'talkpagename':
2944 if ( $this->mTitle->canTalk() ) {
2945 $talkPage = $this->mTitle->getTalkPage();
2946 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2947 } else {
2948 $value = '';
2949 }
2950 break;
2951 case 'talkpagenamee':
2952 if ( $this->mTitle->canTalk() ) {
2953 $talkPage = $this->mTitle->getTalkPage();
2954 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2955 } else {
2956 $value = '';
2957 }
2958 break;
2959 case 'subjectpagename':
2960 $subjPage = $this->mTitle->getSubjectPage();
2961 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2962 break;
2963 case 'subjectpagenamee':
2964 $subjPage = $this->mTitle->getSubjectPage();
2965 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2966 break;
2967 case 'pageid': // requested in bug 23427
2968 $pageid = $this->getTitle()->getArticleID();
2969 if ( $pageid == 0 ) {
2970 # 0 means the page doesn't exist in the database,
2971 # which means the user is previewing a new page.
2972 # The vary-revision flag must be set, because the magic word
2973 # will have a different value once the page is saved.
2974 $this->mOutput->setFlag( 'vary-revision' );
2975 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2976 }
2977 $value = $pageid ? $pageid : null;
2978 break;
2979 case 'revisionid':
2980 # Let the edit saving system know we should parse the page
2981 # *after* a revision ID has been assigned.
2982 $this->mOutput->setFlag( 'vary-revision' );
2983 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2984 $value = $this->mRevisionId;
2985 break;
2986 case 'revisionday':
2987 # Let the edit saving system know we should parse the page
2988 # *after* a revision ID has been assigned. This is for null edits.
2989 $this->mOutput->setFlag( 'vary-revision' );
2990 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2991 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2992 break;
2993 case 'revisionday2':
2994 # Let the edit saving system know we should parse the page
2995 # *after* a revision ID has been assigned. This is for null edits.
2996 $this->mOutput->setFlag( 'vary-revision' );
2997 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2998 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2999 break;
3000 case 'revisionmonth':
3001 # Let the edit saving system know we should parse the page
3002 # *after* a revision ID has been assigned. This is for null edits.
3003 $this->mOutput->setFlag( 'vary-revision' );
3004 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
3005 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
3006 break;
3007 case 'revisionmonth1':
3008 # Let the edit saving system know we should parse the page
3009 # *after* a revision ID has been assigned. This is for null edits.
3010 $this->mOutput->setFlag( 'vary-revision' );
3011 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
3012 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
3013 break;
3014 case 'revisionyear':
3015 # Let the edit saving system know we should parse the page
3016 # *after* a revision ID has been assigned. This is for null edits.
3017 $this->mOutput->setFlag( 'vary-revision' );
3018 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
3019 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
3020 break;
3021 case 'revisiontimestamp':
3022 # Let the edit saving system know we should parse the page
3023 # *after* a revision ID has been assigned. This is for null edits.
3024 $this->mOutput->setFlag( 'vary-revision' );
3025 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
3026 $value = $this->getRevisionTimestamp();
3027 break;
3028 case 'revisionuser':
3029 # Let the edit saving system know we should parse the page
3030 # *after* a revision ID has been assigned. This is for null edits.
3031 $this->mOutput->setFlag( 'vary-revision' );
3032 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
3033 $value = $this->getRevisionUser();
3034 break;
3035 case 'revisionsize':
3036 # Let the edit saving system know we should parse the page
3037 # *after* a revision ID has been assigned. This is for null edits.
3038 $this->mOutput->setFlag( 'vary-revision' );
3039 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
3040 $value = $this->getRevisionSize();
3041 break;
3042 case 'namespace':
3043 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3044 break;
3045 case 'namespacee':
3046 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3047 break;
3048 case 'namespacenumber':
3049 $value = $this->mTitle->getNamespace();
3050 break;
3051 case 'talkspace':
3052 $value = $this->mTitle->canTalk()
3053 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
3054 : '';
3055 break;
3056 case 'talkspacee':
3057 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
3058 break;
3059 case 'subjectspace':
3060 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
3061 break;
3062 case 'subjectspacee':
3063 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
3064 break;
3065 case 'currentdayname':
3066 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
3067 break;
3068 case 'currentyear':
3069 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
3070 break;
3071 case 'currenttime':
3072 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
3073 break;
3074 case 'currenthour':
3075 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
3076 break;
3077 case 'currentweek':
3078 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3079 # int to remove the padding
3080 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
3081 break;
3082 case 'currentdow':
3083 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
3084 break;
3085 case 'localdayname':
3086 $value = $pageLang->getWeekdayName(
3087 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
3088 );
3089 break;
3090 case 'localyear':
3091 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
3092 break;
3093 case 'localtime':
3094 $value = $pageLang->time(
3095 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
3096 false,
3097 false
3098 );
3099 break;
3100 case 'localhour':
3101 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
3102 break;
3103 case 'localweek':
3104 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3105 # int to remove the padding
3106 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
3107 break;
3108 case 'localdow':
3109 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
3110 break;
3111 case 'numberofarticles':
3112 $value = $pageLang->formatNum( SiteStats::articles() );
3113 break;
3114 case 'numberoffiles':
3115 $value = $pageLang->formatNum( SiteStats::images() );
3116 break;
3117 case 'numberofusers':
3118 $value = $pageLang->formatNum( SiteStats::users() );
3119 break;
3120 case 'numberofactiveusers':
3121 $value = $pageLang->formatNum( SiteStats::activeUsers() );
3122 break;
3123 case 'numberofpages':
3124 $value = $pageLang->formatNum( SiteStats::pages() );
3125 break;
3126 case 'numberofadmins':
3127 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
3128 break;
3129 case 'numberofedits':
3130 $value = $pageLang->formatNum( SiteStats::edits() );
3131 break;
3132 case 'numberofviews':
3133 global $wgDisableCounters;
3134 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
3135 break;
3136 case 'currenttimestamp':
3137 $value = wfTimestamp( TS_MW, $ts );
3138 break;
3139 case 'localtimestamp':
3140 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3141 break;
3142 case 'currentversion':
3143 $value = SpecialVersion::getVersion();
3144 break;
3145 case 'articlepath':
3146 return $wgArticlePath;
3147 case 'sitename':
3148 return $wgSitename;
3149 case 'server':
3150 return $wgServer;
3151 case 'servername':
3152 return $wgServerName;
3153 case 'scriptpath':
3154 return $wgScriptPath;
3155 case 'stylepath':
3156 return $wgStylePath;
3157 case 'directionmark':
3158 return $pageLang->getDirMark();
3159 case 'contentlanguage':
3160 global $wgLanguageCode;
3161 return $wgLanguageCode;
3162 case 'cascadingsources':
3163 $value = CoreParserFunctions::cascadingsources( $this );
3164 break;
3165 default:
3166 $ret = null;
3167 wfRunHooks(
3168 'ParserGetVariableValueSwitch',
3169 array( &$this, &$this->mVarCache, &$index, &$ret, &$frame )
3170 );
3171
3172 return $ret;
3173 }
3174
3175 if ( $index ) {
3176 $this->mVarCache[$index] = $value;
3177 }
3178
3179 return $value;
3180 }
3181
3182 /**
3183 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3184 *
3185 * @private
3186 */
3187 public function initialiseVariables() {
3188 wfProfileIn( __METHOD__ );
3189 $variableIDs = MagicWord::getVariableIDs();
3190 $substIDs = MagicWord::getSubstIDs();
3191
3192 $this->mVariables = new MagicWordArray( $variableIDs );
3193 $this->mSubstWords = new MagicWordArray( $substIDs );
3194 wfProfileOut( __METHOD__ );
3195 }
3196
3197 /**
3198 * Preprocess some wikitext and return the document tree.
3199 * This is the ghost of replace_variables().
3200 *
3201 * @param string $text The text to parse
3202 * @param int $flags Bitwise combination of:
3203 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3204 * included. Default is to assume a direct page view.
3205 *
3206 * The generated DOM tree must depend only on the input text and the flags.
3207 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3208 *
3209 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3210 * change in the DOM tree for a given text, must be passed through the section identifier
3211 * in the section edit link and thus back to extractSections().
3212 *
3213 * The output of this function is currently only cached in process memory, but a persistent
3214 * cache may be implemented at a later date which takes further advantage of these strict
3215 * dependency requirements.
3216 *
3217 * @return PPNode
3218 */
3219 public function preprocessToDom( $text, $flags = 0 ) {
3220 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3221 return $dom;
3222 }
3223
3224 /**
3225 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3226 *
3227 * @param string $s
3228 *
3229 * @return array
3230 */
3231 public static function splitWhitespace( $s ) {
3232 $ltrimmed = ltrim( $s );
3233 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3234 $trimmed = rtrim( $ltrimmed );
3235 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3236 if ( $diff > 0 ) {
3237 $w2 = substr( $ltrimmed, -$diff );
3238 } else {
3239 $w2 = '';
3240 }
3241 return array( $w1, $trimmed, $w2 );
3242 }
3243
3244 /**
3245 * Replace magic variables, templates, and template arguments
3246 * with the appropriate text. Templates are substituted recursively,
3247 * taking care to avoid infinite loops.
3248 *
3249 * Note that the substitution depends on value of $mOutputType:
3250 * self::OT_WIKI: only {{subst:}} templates
3251 * self::OT_PREPROCESS: templates but not extension tags
3252 * self::OT_HTML: all templates and extension tags
3253 *
3254 * @param string $text The text to transform
3255 * @param bool|PPFrame $frame Object describing the arguments passed to the
3256 * template. Arguments may also be provided as an associative array, as
3257 * was the usual case before MW1.12. Providing arguments this way may be
3258 * useful for extensions wishing to perform variable replacement
3259 * explicitly.
3260 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3261 * double-brace expansion.
3262 * @return string
3263 */
3264 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3265 # Is there any text? Also, Prevent too big inclusions!
3266 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3267 return $text;
3268 }
3269 wfProfileIn( __METHOD__ );
3270
3271 if ( $frame === false ) {
3272 $frame = $this->getPreprocessor()->newFrame();
3273 } elseif ( !( $frame instanceof PPFrame ) ) {
3274 wfDebug( __METHOD__ . " called using plain parameters instead of "
3275 . "a PPFrame instance. Creating custom frame.\n" );
3276 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3277 }
3278
3279 $dom = $this->preprocessToDom( $text );
3280 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3281 $text = $frame->expand( $dom, $flags );
3282
3283 wfProfileOut( __METHOD__ );
3284 return $text;
3285 }
3286
3287 /**
3288 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3289 *
3290 * @param array $args
3291 *
3292 * @return array
3293 */
3294 public static function createAssocArgs( $args ) {
3295 $assocArgs = array();
3296 $index = 1;
3297 foreach ( $args as $arg ) {
3298 $eqpos = strpos( $arg, '=' );
3299 if ( $eqpos === false ) {
3300 $assocArgs[$index++] = $arg;
3301 } else {
3302 $name = trim( substr( $arg, 0, $eqpos ) );
3303 $value = trim( substr( $arg, $eqpos + 1 ) );
3304 if ( $value === false ) {
3305 $value = '';
3306 }
3307 if ( $name !== false ) {
3308 $assocArgs[$name] = $value;
3309 }
3310 }
3311 }
3312
3313 return $assocArgs;
3314 }
3315
3316 /**
3317 * Warn the user when a parser limitation is reached
3318 * Will warn at most once the user per limitation type
3319 *
3320 * @param string $limitationType Should be one of:
3321 * 'expensive-parserfunction' (corresponding messages:
3322 * 'expensive-parserfunction-warning',
3323 * 'expensive-parserfunction-category')
3324 * 'post-expand-template-argument' (corresponding messages:
3325 * 'post-expand-template-argument-warning',
3326 * 'post-expand-template-argument-category')
3327 * 'post-expand-template-inclusion' (corresponding messages:
3328 * 'post-expand-template-inclusion-warning',
3329 * 'post-expand-template-inclusion-category')
3330 * 'node-count-exceeded' (corresponding messages:
3331 * 'node-count-exceeded-warning',
3332 * 'node-count-exceeded-category')
3333 * 'expansion-depth-exceeded' (corresponding messages:
3334 * 'expansion-depth-exceeded-warning',
3335 * 'expansion-depth-exceeded-category')
3336 * @param string|int|null $current Current value
3337 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3338 * exceeded, provide the values (optional)
3339 */
3340 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3341 # does no harm if $current and $max are present but are unnecessary for the message
3342 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3343 ->inLanguage( $this->mOptions->getUserLangObj() )->text();
3344 $this->mOutput->addWarning( $warning );
3345 $this->addTrackingCategory( "$limitationType-category" );
3346 }
3347
3348 /**
3349 * Return the text of a template, after recursively
3350 * replacing any variables or templates within the template.
3351 *
3352 * @param array $piece The parts of the template
3353 * $piece['title']: the title, i.e. the part before the |
3354 * $piece['parts']: the parameter array
3355 * $piece['lineStart']: whether the brace was at the start of a line
3356 * @param PPFrame $frame The current frame, contains template arguments
3357 * @throws Exception
3358 * @return string The text of the template
3359 */
3360 public function braceSubstitution( $piece, $frame ) {
3361 wfProfileIn( __METHOD__ );
3362 wfProfileIn( __METHOD__ . '-setup' );
3363
3364 // Flags
3365
3366 // $text has been filled
3367 $found = false;
3368 // wiki markup in $text should be escaped
3369 $nowiki = false;
3370 // $text is HTML, armour it against wikitext transformation
3371 $isHTML = false;
3372 // Force interwiki transclusion to be done in raw mode not rendered
3373 $forceRawInterwiki = false;
3374 // $text is a DOM node needing expansion in a child frame
3375 $isChildObj = false;
3376 // $text is a DOM node needing expansion in the current frame
3377 $isLocalObj = false;
3378
3379 # Title object, where $text came from
3380 $title = false;
3381
3382 # $part1 is the bit before the first |, and must contain only title characters.
3383 # Various prefixes will be stripped from it later.
3384 $titleWithSpaces = $frame->expand( $piece['title'] );
3385 $part1 = trim( $titleWithSpaces );
3386 $titleText = false;
3387
3388 # Original title text preserved for various purposes
3389 $originalTitle = $part1;
3390
3391 # $args is a list of argument nodes, starting from index 0, not including $part1
3392 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3393 # below won't work b/c this $args isn't an object
3394 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3395 wfProfileOut( __METHOD__ . '-setup' );
3396
3397 $titleProfileIn = null; // profile templates
3398
3399 # SUBST
3400 wfProfileIn( __METHOD__ . '-modifiers' );
3401 if ( !$found ) {
3402
3403 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3404
3405 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3406 # Decide whether to expand template or keep wikitext as-is.
3407 if ( $this->ot['wiki'] ) {
3408 if ( $substMatch === false ) {
3409 $literal = true; # literal when in PST with no prefix
3410 } else {
3411 $literal = false; # expand when in PST with subst: or safesubst:
3412 }
3413 } else {
3414 if ( $substMatch == 'subst' ) {
3415 $literal = true; # literal when not in PST with plain subst:
3416 } else {
3417 $literal = false; # expand when not in PST with safesubst: or no prefix
3418 }
3419 }
3420 if ( $literal ) {
3421 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3422 $isLocalObj = true;
3423 $found = true;
3424 }
3425 }
3426
3427 # Variables
3428 if ( !$found && $args->getLength() == 0 ) {
3429 $id = $this->mVariables->matchStartToEnd( $part1 );
3430 if ( $id !== false ) {
3431 $text = $this->getVariableValue( $id, $frame );
3432 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3433 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3434 }
3435 $found = true;
3436 }
3437 }
3438
3439 # MSG, MSGNW and RAW
3440 if ( !$found ) {
3441 # Check for MSGNW:
3442 $mwMsgnw = MagicWord::get( 'msgnw' );
3443 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3444 $nowiki = true;
3445 } else {
3446 # Remove obsolete MSG:
3447 $mwMsg = MagicWord::get( 'msg' );
3448 $mwMsg->matchStartAndRemove( $part1 );
3449 }
3450
3451 # Check for RAW:
3452 $mwRaw = MagicWord::get( 'raw' );
3453 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3454 $forceRawInterwiki = true;
3455 }
3456 }
3457 wfProfileOut( __METHOD__ . '-modifiers' );
3458
3459 # Parser functions
3460 if ( !$found ) {
3461 wfProfileIn( __METHOD__ . '-pfunc' );
3462
3463 $colonPos = strpos( $part1, ':' );
3464 if ( $colonPos !== false ) {
3465 $func = substr( $part1, 0, $colonPos );
3466 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3467 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3468 $funcArgs[] = $args->item( $i );
3469 }
3470 try {
3471 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3472 } catch ( Exception $ex ) {
3473 wfProfileOut( __METHOD__ . '-pfunc' );
3474 wfProfileOut( __METHOD__ );
3475 throw $ex;
3476 }
3477
3478 # The interface for parser functions allows for extracting
3479 # flags into the local scope. Extract any forwarded flags
3480 # here.
3481 extract( $result );
3482 }
3483 wfProfileOut( __METHOD__ . '-pfunc' );
3484 }
3485
3486 # Finish mangling title and then check for loops.
3487 # Set $title to a Title object and $titleText to the PDBK
3488 if ( !$found ) {
3489 $ns = NS_TEMPLATE;
3490 # Split the title into page and subpage
3491 $subpage = '';
3492 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3493 if ( $part1 !== $relative ) {
3494 $part1 = $relative;
3495 $ns = $this->mTitle->getNamespace();
3496 }
3497 $title = Title::newFromText( $part1, $ns );
3498 if ( $title ) {
3499 $titleText = $title->getPrefixedText();
3500 # Check for language variants if the template is not found
3501 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3502 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3503 }
3504 # Do recursion depth check
3505 $limit = $this->mOptions->getMaxTemplateDepth();
3506 if ( $frame->depth >= $limit ) {
3507 $found = true;
3508 $text = '<span class="error">'
3509 . wfMessage( 'parser-template-recursion-depth-warning' )
3510 ->numParams( $limit )->inContentLanguage()->text()
3511 . '</span>';
3512 }
3513 }
3514 }
3515
3516 # Load from database
3517 if ( !$found && $title ) {
3518 if ( !Profiler::instance()->isPersistent() ) {
3519 # Too many unique items can kill profiling DBs/collectors
3520 $titleProfileIn = __METHOD__ . "-title-" . $title->getPrefixedDBkey();
3521 wfProfileIn( $titleProfileIn ); // template in
3522 }
3523 wfProfileIn( __METHOD__ . '-loadtpl' );
3524 if ( !$title->isExternal() ) {
3525 if ( $title->isSpecialPage()
3526 && $this->mOptions->getAllowSpecialInclusion()
3527 && $this->ot['html']
3528 ) {
3529 // Pass the template arguments as URL parameters.
3530 // "uselang" will have no effect since the Language object
3531 // is forced to the one defined in ParserOptions.
3532 $pageArgs = array();
3533 $argsLength = $args->getLength();
3534 for ( $i = 0; $i < $argsLength; $i++ ) {
3535 $bits = $args->item( $i )->splitArg();
3536 if ( strval( $bits['index'] ) === '' ) {
3537 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3538 $value = trim( $frame->expand( $bits['value'] ) );
3539 $pageArgs[$name] = $value;
3540 }
3541 }
3542
3543 // Create a new context to execute the special page
3544 $context = new RequestContext;
3545 $context->setTitle( $title );
3546 $context->setRequest( new FauxRequest( $pageArgs ) );
3547 $context->setUser( $this->getUser() );
3548 $context->setLanguage( $this->mOptions->getUserLangObj() );
3549 $ret = SpecialPageFactory::capturePath( $title, $context );
3550 if ( $ret ) {
3551 $text = $context->getOutput()->getHTML();
3552 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3553 $found = true;
3554 $isHTML = true;
3555 $this->disableCache();
3556 }
3557 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3558 $found = false; # access denied
3559 wfDebug( __METHOD__ . ": template inclusion denied for " .
3560 $title->getPrefixedDBkey() . "\n" );
3561 } else {
3562 list( $text, $title ) = $this->getTemplateDom( $title );
3563 if ( $text !== false ) {
3564 $found = true;
3565 $isChildObj = true;
3566 }
3567 }
3568
3569 # If the title is valid but undisplayable, make a link to it
3570 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3571 $text = "[[:$titleText]]";
3572 $found = true;
3573 }
3574 } elseif ( $title->isTrans() ) {
3575 # Interwiki transclusion
3576 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3577 $text = $this->interwikiTransclude( $title, 'render' );
3578 $isHTML = true;
3579 } else {
3580 $text = $this->interwikiTransclude( $title, 'raw' );
3581 # Preprocess it like a template
3582 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3583 $isChildObj = true;
3584 }
3585 $found = true;
3586 }
3587
3588 # Do infinite loop check
3589 # This has to be done after redirect resolution to avoid infinite loops via redirects
3590 if ( !$frame->loopCheck( $title ) ) {
3591 $found = true;
3592 $text = '<span class="error">'
3593 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3594 . '</span>';
3595 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3596 }
3597 wfProfileOut( __METHOD__ . '-loadtpl' );
3598 }
3599
3600 # If we haven't found text to substitute by now, we're done
3601 # Recover the source wikitext and return it
3602 if ( !$found ) {
3603 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3604 if ( $titleProfileIn ) {
3605 wfProfileOut( $titleProfileIn ); // template out
3606 }
3607 wfProfileOut( __METHOD__ );
3608 return array( 'object' => $text );
3609 }
3610
3611 # Expand DOM-style return values in a child frame
3612 if ( $isChildObj ) {
3613 # Clean up argument array
3614 $newFrame = $frame->newChild( $args, $title );
3615
3616 if ( $nowiki ) {
3617 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3618 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3619 # Expansion is eligible for the empty-frame cache
3620 $text = $newFrame->cachedExpand( $titleText, $text );
3621 } else {
3622 # Uncached expansion
3623 $text = $newFrame->expand( $text );
3624 }
3625 }
3626 if ( $isLocalObj && $nowiki ) {
3627 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3628 $isLocalObj = false;
3629 }
3630
3631 if ( $titleProfileIn ) {
3632 wfProfileOut( $titleProfileIn ); // template out
3633 }
3634
3635 # Replace raw HTML by a placeholder
3636 if ( $isHTML ) {
3637 $text = $this->insertStripItem( $text );
3638 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3639 # Escape nowiki-style return values
3640 $text = wfEscapeWikiText( $text );
3641 } elseif ( is_string( $text )
3642 && !$piece['lineStart']
3643 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3644 ) {
3645 # Bug 529: if the template begins with a table or block-level
3646 # element, it should be treated as beginning a new line.
3647 # This behavior is somewhat controversial.
3648 $text = "\n" . $text;
3649 }
3650
3651 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3652 # Error, oversize inclusion
3653 if ( $titleText !== false ) {
3654 # Make a working, properly escaped link if possible (bug 23588)
3655 $text = "[[:$titleText]]";
3656 } else {
3657 # This will probably not be a working link, but at least it may
3658 # provide some hint of where the problem is
3659 preg_replace( '/^:/', '', $originalTitle );
3660 $text = "[[:$originalTitle]]";
3661 }
3662 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3663 . 'post-expand include size too large -->' );
3664 $this->limitationWarn( 'post-expand-template-inclusion' );
3665 }
3666
3667 if ( $isLocalObj ) {
3668 $ret = array( 'object' => $text );
3669 } else {
3670 $ret = array( 'text' => $text );
3671 }
3672
3673 wfProfileOut( __METHOD__ );
3674 return $ret;
3675 }
3676
3677 /**
3678 * Call a parser function and return an array with text and flags.
3679 *
3680 * The returned array will always contain a boolean 'found', indicating
3681 * whether the parser function was found or not. It may also contain the
3682 * following:
3683 * text: string|object, resulting wikitext or PP DOM object
3684 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3685 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3686 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3687 * nowiki: bool, wiki markup in $text should be escaped
3688 *
3689 * @since 1.21
3690 * @param PPFrame $frame The current frame, contains template arguments
3691 * @param string $function Function name
3692 * @param array $args Arguments to the function
3693 * @throws MWException
3694 * @return array
3695 */
3696 public function callParserFunction( $frame, $function, array $args = array() ) {
3697 global $wgContLang;
3698
3699 wfProfileIn( __METHOD__ );
3700
3701 # Case sensitive functions
3702 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3703 $function = $this->mFunctionSynonyms[1][$function];
3704 } else {
3705 # Case insensitive functions
3706 $function = $wgContLang->lc( $function );
3707 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3708 $function = $this->mFunctionSynonyms[0][$function];
3709 } else {
3710 wfProfileOut( __METHOD__ );
3711 return array( 'found' => false );
3712 }
3713 }
3714
3715 wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3716 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3717
3718 # Workaround for PHP bug 35229 and similar
3719 if ( !is_callable( $callback ) ) {
3720 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3721 wfProfileOut( __METHOD__ );
3722 throw new MWException( "Tag hook for $function is not callable\n" );
3723 }
3724
3725 $allArgs = array( &$this );
3726 if ( $flags & SFH_OBJECT_ARGS ) {
3727 # Convert arguments to PPNodes and collect for appending to $allArgs
3728 $funcArgs = array();
3729 foreach ( $args as $k => $v ) {
3730 if ( $v instanceof PPNode || $k === 0 ) {
3731 $funcArgs[] = $v;
3732 } else {
3733 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
3734 }
3735 }
3736
3737 # Add a frame parameter, and pass the arguments as an array
3738 $allArgs[] = $frame;
3739 $allArgs[] = $funcArgs;
3740 } else {
3741 # Convert arguments to plain text and append to $allArgs
3742 foreach ( $args as $k => $v ) {
3743 if ( $v instanceof PPNode ) {
3744 $allArgs[] = trim( $frame->expand( $v ) );
3745 } elseif ( is_int( $k ) && $k >= 0 ) {
3746 $allArgs[] = trim( $v );
3747 } else {
3748 $allArgs[] = trim( "$k=$v" );
3749 }
3750 }
3751 }
3752
3753 $result = call_user_func_array( $callback, $allArgs );
3754
3755 # The interface for function hooks allows them to return a wikitext
3756 # string or an array containing the string and any flags. This mungs
3757 # things around to match what this method should return.
3758 if ( !is_array( $result ) ) {
3759 $result = array(
3760 'found' => true,
3761 'text' => $result,
3762 );
3763 } else {
3764 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3765 $result['text'] = $result[0];
3766 }
3767 unset( $result[0] );
3768 $result += array(
3769 'found' => true,
3770 );
3771 }
3772
3773 $noparse = true;
3774 $preprocessFlags = 0;
3775 if ( isset( $result['noparse'] ) ) {
3776 $noparse = $result['noparse'];
3777 }
3778 if ( isset( $result['preprocessFlags'] ) ) {
3779 $preprocessFlags = $result['preprocessFlags'];
3780 }
3781
3782 if ( !$noparse ) {
3783 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3784 $result['isChildObj'] = true;
3785 }
3786 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3787 wfProfileOut( __METHOD__ );
3788
3789 return $result;
3790 }
3791
3792 /**
3793 * Get the semi-parsed DOM representation of a template with a given title,
3794 * and its redirect destination title. Cached.
3795 *
3796 * @param Title $title
3797 *
3798 * @return array
3799 */
3800 public function getTemplateDom( $title ) {
3801 $cacheTitle = $title;
3802 $titleText = $title->getPrefixedDBkey();
3803
3804 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3805 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3806 $title = Title::makeTitle( $ns, $dbk );
3807 $titleText = $title->getPrefixedDBkey();
3808 }
3809 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3810 return array( $this->mTplDomCache[$titleText], $title );
3811 }
3812
3813 # Cache miss, go to the database
3814 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3815
3816 if ( $text === false ) {
3817 $this->mTplDomCache[$titleText] = false;
3818 return array( false, $title );
3819 }
3820
3821 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3822 $this->mTplDomCache[$titleText] = $dom;
3823
3824 if ( !$title->equals( $cacheTitle ) ) {
3825 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3826 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3827 }
3828
3829 return array( $dom, $title );
3830 }
3831
3832 /**
3833 * Fetch the unparsed text of a template and register a reference to it.
3834 * @param Title $title
3835 * @return array ( string or false, Title )
3836 */
3837 public function fetchTemplateAndTitle( $title ) {
3838 // Defaults to Parser::statelessFetchTemplate()
3839 $templateCb = $this->mOptions->getTemplateCallback();
3840 $stuff = call_user_func( $templateCb, $title, $this );
3841 $text = $stuff['text'];
3842 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3843 if ( isset( $stuff['deps'] ) ) {
3844 foreach ( $stuff['deps'] as $dep ) {
3845 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3846 if ( $dep['title']->equals( $this->getTitle() ) ) {
3847 // If we transclude ourselves, the final result
3848 // will change based on the new version of the page
3849 $this->mOutput->setFlag( 'vary-revision' );
3850 }
3851 }
3852 }
3853 return array( $text, $finalTitle );
3854 }
3855
3856 /**
3857 * Fetch the unparsed text of a template and register a reference to it.
3858 * @param Title $title
3859 * @return string|bool
3860 */
3861 public function fetchTemplate( $title ) {
3862 $rv = $this->fetchTemplateAndTitle( $title );
3863 return $rv[0];
3864 }
3865
3866 /**
3867 * Static function to get a template
3868 * Can be overridden via ParserOptions::setTemplateCallback().
3869 *
3870 * @param Title $title
3871 * @param bool|Parser $parser
3872 *
3873 * @return array
3874 */
3875 public static function statelessFetchTemplate( $title, $parser = false ) {
3876 $text = $skip = false;
3877 $finalTitle = $title;
3878 $deps = array();
3879
3880 # Loop to fetch the article, with up to 1 redirect
3881 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3882 # Give extensions a chance to select the revision instead
3883 $id = false; # Assume current
3884 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3885 array( $parser, $title, &$skip, &$id ) );
3886
3887 if ( $skip ) {
3888 $text = false;
3889 $deps[] = array(
3890 'title' => $title,
3891 'page_id' => $title->getArticleID(),
3892 'rev_id' => null
3893 );
3894 break;
3895 }
3896 # Get the revision
3897 $rev = $id
3898 ? Revision::newFromId( $id )
3899 : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
3900 $rev_id = $rev ? $rev->getId() : 0;
3901 # If there is no current revision, there is no page
3902 if ( $id === false && !$rev ) {
3903 $linkCache = LinkCache::singleton();
3904 $linkCache->addBadLinkObj( $title );
3905 }
3906
3907 $deps[] = array(
3908 'title' => $title,
3909 'page_id' => $title->getArticleID(),
3910 'rev_id' => $rev_id );
3911 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3912 # We fetched a rev from a different title; register it too...
3913 $deps[] = array(
3914 'title' => $rev->getTitle(),
3915 'page_id' => $rev->getPage(),
3916 'rev_id' => $rev_id );
3917 }
3918
3919 if ( $rev ) {
3920 $content = $rev->getContent();
3921 $text = $content ? $content->getWikitextForTransclusion() : null;
3922
3923 if ( $text === false || $text === null ) {
3924 $text = false;
3925 break;
3926 }
3927 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3928 global $wgContLang;
3929 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3930 if ( !$message->exists() ) {
3931 $text = false;
3932 break;
3933 }
3934 $content = $message->content();
3935 $text = $message->plain();
3936 } else {
3937 break;
3938 }
3939 if ( !$content ) {
3940 break;
3941 }
3942 # Redirect?
3943 $finalTitle = $title;
3944 $title = $content->getRedirectTarget();
3945 }
3946 return array(
3947 'text' => $text,
3948 'finalTitle' => $finalTitle,
3949 'deps' => $deps );
3950 }
3951
3952 /**
3953 * Fetch a file and its title and register a reference to it.
3954 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3955 * @param Title $title
3956 * @param array $options Array of options to RepoGroup::findFile
3957 * @return File|bool
3958 */
3959 public function fetchFile( $title, $options = array() ) {
3960 $res = $this->fetchFileAndTitle( $title, $options );
3961 return $res[0];
3962 }
3963
3964 /**
3965 * Fetch a file and its title and register a reference to it.
3966 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3967 * @param Title $title
3968 * @param array $options Array of options to RepoGroup::findFile
3969 * @return array ( File or false, Title of file )
3970 */
3971 public function fetchFileAndTitle( $title, $options = array() ) {
3972 $file = $this->fetchFileNoRegister( $title, $options );
3973
3974 $time = $file ? $file->getTimestamp() : false;
3975 $sha1 = $file ? $file->getSha1() : false;
3976 # Register the file as a dependency...
3977 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3978 if ( $file && !$title->equals( $file->getTitle() ) ) {
3979 # Update fetched file title
3980 $title = $file->getTitle();
3981 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3982 }
3983 return array( $file, $title );
3984 }
3985
3986 /**
3987 * Helper function for fetchFileAndTitle.
3988 *
3989 * Also useful if you need to fetch a file but not use it yet,
3990 * for example to get the file's handler.
3991 *
3992 * @param Title $title
3993 * @param array $options Array of options to RepoGroup::findFile
3994 * @return File|bool
3995 */
3996 protected function fetchFileNoRegister( $title, $options = array() ) {
3997 if ( isset( $options['broken'] ) ) {
3998 $file = false; // broken thumbnail forced by hook
3999 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
4000 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
4001 } else { // get by (name,timestamp)
4002 $file = wfFindFile( $title, $options );
4003 }
4004 return $file;
4005 }
4006
4007 /**
4008 * Transclude an interwiki link.
4009 *
4010 * @param Title $title
4011 * @param string $action
4012 *
4013 * @return string
4014 */
4015 public function interwikiTransclude( $title, $action ) {
4016 global $wgEnableScaryTranscluding;
4017
4018 if ( !$wgEnableScaryTranscluding ) {
4019 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
4020 }
4021
4022 $url = $title->getFullURL( array( 'action' => $action ) );
4023
4024 if ( strlen( $url ) > 255 ) {
4025 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
4026 }
4027 return $this->fetchScaryTemplateMaybeFromCache( $url );
4028 }
4029
4030 /**
4031 * @param string $url
4032 * @return mixed|string
4033 */
4034 public function fetchScaryTemplateMaybeFromCache( $url ) {
4035 global $wgTranscludeCacheExpiry;
4036 $dbr = wfGetDB( DB_SLAVE );
4037 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
4038 $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
4039 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
4040 if ( $obj ) {
4041 return $obj->tc_contents;
4042 }
4043
4044 $req = MWHttpRequest::factory( $url );
4045 $status = $req->execute(); // Status object
4046 if ( $status->isOK() ) {
4047 $text = $req->getContent();
4048 } elseif ( $req->getStatus() != 200 ) {
4049 // Though we failed to fetch the content, this status is useless.
4050 return wfMessage( 'scarytranscludefailed-httpstatus' )
4051 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
4052 } else {
4053 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4054 }
4055
4056 $dbw = wfGetDB( DB_MASTER );
4057 $dbw->replace( 'transcache', array( 'tc_url' ), array(
4058 'tc_url' => $url,
4059 'tc_time' => $dbw->timestamp( time() ),
4060 'tc_contents' => $text
4061 ) );
4062 return $text;
4063 }
4064
4065 /**
4066 * Triple brace replacement -- used for template arguments
4067 * @private
4068 *
4069 * @param array $piece
4070 * @param PPFrame $frame
4071 *
4072 * @return array
4073 */
4074 public function argSubstitution( $piece, $frame ) {
4075 wfProfileIn( __METHOD__ );
4076
4077 $error = false;
4078 $parts = $piece['parts'];
4079 $nameWithSpaces = $frame->expand( $piece['title'] );
4080 $argName = trim( $nameWithSpaces );
4081 $object = false;
4082 $text = $frame->getArgument( $argName );
4083 if ( $text === false && $parts->getLength() > 0
4084 && ( $this->ot['html']
4085 || $this->ot['pre']
4086 || ( $this->ot['wiki'] && $frame->isTemplate() )
4087 )
4088 ) {
4089 # No match in frame, use the supplied default
4090 $object = $parts->item( 0 )->getChildren();
4091 }
4092 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4093 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4094 $this->limitationWarn( 'post-expand-template-argument' );
4095 }
4096
4097 if ( $text === false && $object === false ) {
4098 # No match anywhere
4099 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4100 }
4101 if ( $error !== false ) {
4102 $text .= $error;
4103 }
4104 if ( $object !== false ) {
4105 $ret = array( 'object' => $object );
4106 } else {
4107 $ret = array( 'text' => $text );
4108 }
4109
4110 wfProfileOut( __METHOD__ );
4111 return $ret;
4112 }
4113
4114 /**
4115 * Return the text to be used for a given extension tag.
4116 * This is the ghost of strip().
4117 *
4118 * @param array $params Associative array of parameters:
4119 * name PPNode for the tag name
4120 * attr PPNode for unparsed text where tag attributes are thought to be
4121 * attributes Optional associative array of parsed attributes
4122 * inner Contents of extension element
4123 * noClose Original text did not have a close tag
4124 * @param PPFrame $frame
4125 *
4126 * @throws MWException
4127 * @return string
4128 */
4129 public function extensionSubstitution( $params, $frame ) {
4130 $name = $frame->expand( $params['name'] );
4131 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4132 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4133 $marker = "{$this->mUniqPrefix}-$name-"
4134 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4135
4136 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4137 ( $this->ot['html'] || $this->ot['pre'] );
4138 if ( $isFunctionTag ) {
4139 $markerType = 'none';
4140 } else {
4141 $markerType = 'general';
4142 }
4143 if ( $this->ot['html'] || $isFunctionTag ) {
4144 $name = strtolower( $name );
4145 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4146 if ( isset( $params['attributes'] ) ) {
4147 $attributes = $attributes + $params['attributes'];
4148 }
4149
4150 if ( isset( $this->mTagHooks[$name] ) ) {
4151 # Workaround for PHP bug 35229 and similar
4152 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4153 throw new MWException( "Tag hook for $name is not callable\n" );
4154 }
4155 $output = call_user_func_array( $this->mTagHooks[$name],
4156 array( $content, $attributes, $this, $frame ) );
4157 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4158 list( $callback, ) = $this->mFunctionTagHooks[$name];
4159 if ( !is_callable( $callback ) ) {
4160 throw new MWException( "Tag hook for $name is not callable\n" );
4161 }
4162
4163 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
4164 } else {
4165 $output = '<span class="error">Invalid tag extension name: ' .
4166 htmlspecialchars( $name ) . '</span>';
4167 }
4168
4169 if ( is_array( $output ) ) {
4170 # Extract flags to local scope (to override $markerType)
4171 $flags = $output;
4172 $output = $flags[0];
4173 unset( $flags[0] );
4174 extract( $flags );
4175 }
4176 } else {
4177 if ( is_null( $attrText ) ) {
4178 $attrText = '';
4179 }
4180 if ( isset( $params['attributes'] ) ) {
4181 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4182 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4183 htmlspecialchars( $attrValue ) . '"';
4184 }
4185 }
4186 if ( $content === null ) {
4187 $output = "<$name$attrText/>";
4188 } else {
4189 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4190 $output = "<$name$attrText>$content$close";
4191 }
4192 }
4193
4194 if ( $markerType === 'none' ) {
4195 return $output;
4196 } elseif ( $markerType === 'nowiki' ) {
4197 $this->mStripState->addNoWiki( $marker, $output );
4198 } elseif ( $markerType === 'general' ) {
4199 $this->mStripState->addGeneral( $marker, $output );
4200 } else {
4201 throw new MWException( __METHOD__ . ': invalid marker type' );
4202 }
4203 return $marker;
4204 }
4205
4206 /**
4207 * Increment an include size counter
4208 *
4209 * @param string $type The type of expansion
4210 * @param int $size The size of the text
4211 * @return bool False if this inclusion would take it over the maximum, true otherwise
4212 */
4213 public function incrementIncludeSize( $type, $size ) {
4214 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4215 return false;
4216 } else {
4217 $this->mIncludeSizes[$type] += $size;
4218 return true;
4219 }
4220 }
4221
4222 /**
4223 * Increment the expensive function count
4224 *
4225 * @return bool False if the limit has been exceeded
4226 */
4227 public function incrementExpensiveFunctionCount() {
4228 $this->mExpensiveFunctionCount++;
4229 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4230 }
4231
4232 /**
4233 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4234 * Fills $this->mDoubleUnderscores, returns the modified text
4235 *
4236 * @param string $text
4237 *
4238 * @return string
4239 */
4240 public function doDoubleUnderscore( $text ) {
4241 wfProfileIn( __METHOD__ );
4242
4243 # The position of __TOC__ needs to be recorded
4244 $mw = MagicWord::get( 'toc' );
4245 if ( $mw->match( $text ) ) {
4246 $this->mShowToc = true;
4247 $this->mForceTocPosition = true;
4248
4249 # Set a placeholder. At the end we'll fill it in with the TOC.
4250 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4251
4252 # Only keep the first one.
4253 $text = $mw->replace( '', $text );
4254 }
4255
4256 # Now match and remove the rest of them
4257 $mwa = MagicWord::getDoubleUnderscoreArray();
4258 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4259
4260 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4261 $this->mOutput->mNoGallery = true;
4262 }
4263 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4264 $this->mShowToc = false;
4265 }
4266 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4267 && $this->mTitle->getNamespace() == NS_CATEGORY
4268 ) {
4269 $this->addTrackingCategory( 'hidden-category-category' );
4270 }
4271 # (bug 8068) Allow control over whether robots index a page.
4272 #
4273 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
4274 # is not desirable, the last one on the page should win.
4275 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4276 $this->mOutput->setIndexPolicy( 'noindex' );
4277 $this->addTrackingCategory( 'noindex-category' );
4278 }
4279 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4280 $this->mOutput->setIndexPolicy( 'index' );
4281 $this->addTrackingCategory( 'index-category' );
4282 }
4283
4284 # Cache all double underscores in the database
4285 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4286 $this->mOutput->setProperty( $key, '' );
4287 }
4288
4289 wfProfileOut( __METHOD__ );
4290 return $text;
4291 }
4292
4293 /**
4294 * Add a tracking category, getting the title from a system message,
4295 * or print a debug message if the title is invalid.
4296 *
4297 * Please add any message that you use with this function to
4298 * $wgTrackingCategories. That way they will be listed on
4299 * Special:TrackingCategories.
4300 *
4301 * @param string $msg Message key
4302 * @return bool Whether the addition was successful
4303 */
4304 public function addTrackingCategory( $msg ) {
4305 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
4306 wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
4307 return false;
4308 }
4309 // Important to parse with correct title (bug 31469)
4310 $cat = wfMessage( $msg )
4311 ->title( $this->getTitle() )
4312 ->inContentLanguage()
4313 ->text();
4314
4315 # Allow tracking categories to be disabled by setting them to "-"
4316 if ( $cat === '-' ) {
4317 return false;
4318 }
4319
4320 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
4321 if ( $containerCategory ) {
4322 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
4323 return true;
4324 } else {
4325 wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
4326 return false;
4327 }
4328 }
4329
4330 /**
4331 * This function accomplishes several tasks:
4332 * 1) Auto-number headings if that option is enabled
4333 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4334 * 3) Add a Table of contents on the top for users who have enabled the option
4335 * 4) Auto-anchor headings
4336 *
4337 * It loops through all headlines, collects the necessary data, then splits up the
4338 * string and re-inserts the newly formatted headlines.
4339 *
4340 * @param string $text
4341 * @param string $origText Original, untouched wikitext
4342 * @param bool $isMain
4343 * @return mixed|string
4344 * @private
4345 */
4346 public function formatHeadings( $text, $origText, $isMain = true ) {
4347 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4348
4349 # Inhibit editsection links if requested in the page
4350 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4351 $maybeShowEditLink = $showEditLink = false;
4352 } else {
4353 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4354 $showEditLink = $this->mOptions->getEditSection();
4355 }
4356 if ( $showEditLink ) {
4357 $this->mOutput->setEditSectionTokens( true );
4358 }
4359
4360 # Get all headlines for numbering them and adding funky stuff like [edit]
4361 # links - this is for later, but we need the number of headlines right now
4362 $matches = array();
4363 $numMatches = preg_match_all(
4364 '/<H(?P<level>[1-6])(?P<attrib>.*?' . '>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4365 $text,
4366 $matches
4367 );
4368
4369 # if there are fewer than 4 headlines in the article, do not show TOC
4370 # unless it's been explicitly enabled.
4371 $enoughToc = $this->mShowToc &&
4372 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4373
4374 # Allow user to stipulate that a page should have a "new section"
4375 # link added via __NEWSECTIONLINK__
4376 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4377 $this->mOutput->setNewSection( true );
4378 }
4379
4380 # Allow user to remove the "new section"
4381 # link via __NONEWSECTIONLINK__
4382 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4383 $this->mOutput->hideNewSection( true );
4384 }
4385
4386 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4387 # override above conditions and always show TOC above first header
4388 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4389 $this->mShowToc = true;
4390 $enoughToc = true;
4391 }
4392
4393 # headline counter
4394 $headlineCount = 0;
4395 $numVisible = 0;
4396
4397 # Ugh .. the TOC should have neat indentation levels which can be
4398 # passed to the skin functions. These are determined here
4399 $toc = '';
4400 $full = '';
4401 $head = array();
4402 $sublevelCount = array();
4403 $levelCount = array();
4404 $level = 0;
4405 $prevlevel = 0;
4406 $toclevel = 0;
4407 $prevtoclevel = 0;
4408 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4409 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4410 $oldType = $this->mOutputType;
4411 $this->setOutputType( self::OT_WIKI );
4412 $frame = $this->getPreprocessor()->newFrame();
4413 $root = $this->preprocessToDom( $origText );
4414 $node = $root->getFirstChild();
4415 $byteOffset = 0;
4416 $tocraw = array();
4417 $refers = array();
4418
4419 foreach ( $matches[3] as $headline ) {
4420 $isTemplate = false;
4421 $titleText = false;
4422 $sectionIndex = false;
4423 $numbering = '';
4424 $markerMatches = array();
4425 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4426 $serial = $markerMatches[1];
4427 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4428 $isTemplate = ( $titleText != $baseTitleText );
4429 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4430 }
4431
4432 if ( $toclevel ) {
4433 $prevlevel = $level;
4434 }
4435 $level = $matches[1][$headlineCount];
4436
4437 if ( $level > $prevlevel ) {
4438 # Increase TOC level
4439 $toclevel++;
4440 $sublevelCount[$toclevel] = 0;
4441 if ( $toclevel < $wgMaxTocLevel ) {
4442 $prevtoclevel = $toclevel;
4443 $toc .= Linker::tocIndent();
4444 $numVisible++;
4445 }
4446 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4447 # Decrease TOC level, find level to jump to
4448
4449 for ( $i = $toclevel; $i > 0; $i-- ) {
4450 if ( $levelCount[$i] == $level ) {
4451 # Found last matching level
4452 $toclevel = $i;
4453 break;
4454 } elseif ( $levelCount[$i] < $level ) {
4455 # Found first matching level below current level
4456 $toclevel = $i + 1;
4457 break;
4458 }
4459 }
4460 if ( $i == 0 ) {
4461 $toclevel = 1;
4462 }
4463 if ( $toclevel < $wgMaxTocLevel ) {
4464 if ( $prevtoclevel < $wgMaxTocLevel ) {
4465 # Unindent only if the previous toc level was shown :p
4466 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4467 $prevtoclevel = $toclevel;
4468 } else {
4469 $toc .= Linker::tocLineEnd();
4470 }
4471 }
4472 } else {
4473 # No change in level, end TOC line
4474 if ( $toclevel < $wgMaxTocLevel ) {
4475 $toc .= Linker::tocLineEnd();
4476 }
4477 }
4478
4479 $levelCount[$toclevel] = $level;
4480
4481 # count number of headlines for each level
4482 $sublevelCount[$toclevel]++;
4483 $dot = 0;
4484 for ( $i = 1; $i <= $toclevel; $i++ ) {
4485 if ( !empty( $sublevelCount[$i] ) ) {
4486 if ( $dot ) {
4487 $numbering .= '.';
4488 }
4489 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4490 $dot = 1;
4491 }
4492 }
4493
4494 # The safe header is a version of the header text safe to use for links
4495
4496 # Remove link placeholders by the link text.
4497 # <!--LINK number-->
4498 # turns into
4499 # link text with suffix
4500 # Do this before unstrip since link text can contain strip markers
4501 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4502
4503 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4504 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4505
4506 # Strip out HTML (first regex removes any tag not allowed)
4507 # Allowed tags are:
4508 # * <sup> and <sub> (bug 8393)
4509 # * <i> (bug 26375)
4510 # * <b> (r105284)
4511 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4512 #
4513 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4514 # to allow setting directionality in toc items.
4515 $tocline = preg_replace(
4516 array(
4517 '#<(?!/?(span|sup|sub|i|b)(?: [^>]*)?>).*?' . '>#',
4518 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|i|b))(?: .*?)?' . '>#'
4519 ),
4520 array( '', '<$1>' ),
4521 $safeHeadline
4522 );
4523 $tocline = trim( $tocline );
4524
4525 # For the anchor, strip out HTML-y stuff period
4526 $safeHeadline = preg_replace( '/<.*?' . '>/', '', $safeHeadline );
4527 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4528
4529 # Save headline for section edit hint before it's escaped
4530 $headlineHint = $safeHeadline;
4531
4532 if ( $wgExperimentalHtmlIds ) {
4533 # For reverse compatibility, provide an id that's
4534 # HTML4-compatible, like we used to.
4535 #
4536 # It may be worth noting, academically, that it's possible for
4537 # the legacy anchor to conflict with a non-legacy headline
4538 # anchor on the page. In this case likely the "correct" thing
4539 # would be to either drop the legacy anchors or make sure
4540 # they're numbered first. However, this would require people
4541 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4542 # manually, so let's not bother worrying about it.
4543 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4544 array( 'noninitial', 'legacy' ) );
4545 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4546
4547 if ( $legacyHeadline == $safeHeadline ) {
4548 # No reason to have both (in fact, we can't)
4549 $legacyHeadline = false;
4550 }
4551 } else {
4552 $legacyHeadline = false;
4553 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4554 'noninitial' );
4555 }
4556
4557 # HTML names must be case-insensitively unique (bug 10721).
4558 # This does not apply to Unicode characters per
4559 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4560 # @todo FIXME: We may be changing them depending on the current locale.
4561 $arrayKey = strtolower( $safeHeadline );
4562 if ( $legacyHeadline === false ) {
4563 $legacyArrayKey = false;
4564 } else {
4565 $legacyArrayKey = strtolower( $legacyHeadline );
4566 }
4567
4568 # count how many in assoc. array so we can track dupes in anchors
4569 if ( isset( $refers[$arrayKey] ) ) {
4570 $refers[$arrayKey]++;
4571 } else {
4572 $refers[$arrayKey] = 1;
4573 }
4574 if ( isset( $refers[$legacyArrayKey] ) ) {
4575 $refers[$legacyArrayKey]++;
4576 } else {
4577 $refers[$legacyArrayKey] = 1;
4578 }
4579
4580 # Don't number the heading if it is the only one (looks silly)
4581 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4582 # the two are different if the line contains a link
4583 $headline = Html::element(
4584 'span',
4585 array( 'class' => 'mw-headline-number' ),
4586 $numbering
4587 ) . ' ' . $headline;
4588 }
4589
4590 # Create the anchor for linking from the TOC to the section
4591 $anchor = $safeHeadline;
4592 $legacyAnchor = $legacyHeadline;
4593 if ( $refers[$arrayKey] > 1 ) {
4594 $anchor .= '_' . $refers[$arrayKey];
4595 }
4596 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4597 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4598 }
4599 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4600 $toc .= Linker::tocLine( $anchor, $tocline,
4601 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4602 }
4603
4604 # Add the section to the section tree
4605 # Find the DOM node for this header
4606 $noOffset = ( $isTemplate || $sectionIndex === false );
4607 while ( $node && !$noOffset ) {
4608 if ( $node->getName() === 'h' ) {
4609 $bits = $node->splitHeading();
4610 if ( $bits['i'] == $sectionIndex ) {
4611 break;
4612 }
4613 }
4614 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4615 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4616 $node = $node->getNextSibling();
4617 }
4618 $tocraw[] = array(
4619 'toclevel' => $toclevel,
4620 'level' => $level,
4621 'line' => $tocline,
4622 'number' => $numbering,
4623 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4624 'fromtitle' => $titleText,
4625 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4626 'anchor' => $anchor,
4627 );
4628
4629 # give headline the correct <h#> tag
4630 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4631 // Output edit section links as markers with styles that can be customized by skins
4632 if ( $isTemplate ) {
4633 # Put a T flag in the section identifier, to indicate to extractSections()
4634 # that sections inside <includeonly> should be counted.
4635 $editsectionPage = $titleText;
4636 $editsectionSection = "T-$sectionIndex";
4637 $editsectionContent = null;
4638 } else {
4639 $editsectionPage = $this->mTitle->getPrefixedText();
4640 $editsectionSection = $sectionIndex;
4641 $editsectionContent = $headlineHint;
4642 }
4643 // We use a bit of pesudo-xml for editsection markers. The
4644 // language converter is run later on. Using a UNIQ style marker
4645 // leads to the converter screwing up the tokens when it
4646 // converts stuff. And trying to insert strip tags fails too. At
4647 // this point all real inputted tags have already been escaped,
4648 // so we don't have to worry about a user trying to input one of
4649 // these markers directly. We use a page and section attribute
4650 // to stop the language converter from converting these
4651 // important bits of data, but put the headline hint inside a
4652 // content block because the language converter is supposed to
4653 // be able to convert that piece of data.
4654 // Gets replaced with html in ParserOutput::getText
4655 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4656 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4657 if ( $editsectionContent !== null ) {
4658 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4659 } else {
4660 $editlink .= '/>';
4661 }
4662 } else {
4663 $editlink = '';
4664 }
4665 $head[$headlineCount] = Linker::makeHeadline( $level,
4666 $matches['attrib'][$headlineCount], $anchor, $headline,
4667 $editlink, $legacyAnchor );
4668
4669 $headlineCount++;
4670 }
4671
4672 $this->setOutputType( $oldType );
4673
4674 # Never ever show TOC if no headers
4675 if ( $numVisible < 1 ) {
4676 $enoughToc = false;
4677 }
4678
4679 if ( $enoughToc ) {
4680 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4681 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4682 }
4683 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4684 $this->mOutput->setTOCHTML( $toc );
4685 $toc = self::TOC_START . $toc . self::TOC_END;
4686 $this->mOutput->addModules( 'mediawiki.toc' );
4687 }
4688
4689 if ( $isMain ) {
4690 $this->mOutput->setSections( $tocraw );
4691 }
4692
4693 # split up and insert constructed headlines
4694 $blocks = preg_split( '/<H[1-6].*?' . '>[\s\S]*?<\/H[1-6]>/i', $text );
4695 $i = 0;
4696
4697 // build an array of document sections
4698 $sections = array();
4699 foreach ( $blocks as $block ) {
4700 // $head is zero-based, sections aren't.
4701 if ( empty( $head[$i - 1] ) ) {
4702 $sections[$i] = $block;
4703 } else {
4704 $sections[$i] = $head[$i - 1] . $block;
4705 }
4706
4707 /**
4708 * Send a hook, one per section.
4709 * The idea here is to be able to make section-level DIVs, but to do so in a
4710 * lower-impact, more correct way than r50769
4711 *
4712 * $this : caller
4713 * $section : the section number
4714 * &$sectionContent : ref to the content of the section
4715 * $showEditLinks : boolean describing whether this section has an edit link
4716 */
4717 wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4718
4719 $i++;
4720 }
4721
4722 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4723 // append the TOC at the beginning
4724 // Top anchor now in skin
4725 $sections[0] = $sections[0] . $toc . "\n";
4726 }
4727
4728 $full .= join( '', $sections );
4729
4730 if ( $this->mForceTocPosition ) {
4731 return str_replace( '<!--MWTOC-->', $toc, $full );
4732 } else {
4733 return $full;
4734 }
4735 }
4736
4737 /**
4738 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4739 * conversion, substitting signatures, {{subst:}} templates, etc.
4740 *
4741 * @param string $text The text to transform
4742 * @param Title $title The Title object for the current article
4743 * @param User $user The User object describing the current user
4744 * @param ParserOptions $options Parsing options
4745 * @param bool $clearState Whether to clear the parser state first
4746 * @return string The altered wiki markup
4747 */
4748 public function preSaveTransform( $text, Title $title, User $user,
4749 ParserOptions $options, $clearState = true
4750 ) {
4751 if ( $clearState ) {
4752 $magicScopeVariable = $this->lock();
4753 }
4754 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4755 $this->setUser( $user );
4756
4757 $pairs = array(
4758 "\r\n" => "\n",
4759 );
4760 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4761 if ( $options->getPreSaveTransform() ) {
4762 $text = $this->pstPass2( $text, $user );
4763 }
4764 $text = $this->mStripState->unstripBoth( $text );
4765
4766 $this->setUser( null ); #Reset
4767
4768 return $text;
4769 }
4770
4771 /**
4772 * Pre-save transform helper function
4773 *
4774 * @param string $text
4775 * @param User $user
4776 *
4777 * @return string
4778 */
4779 private function pstPass2( $text, $user ) {
4780 global $wgContLang;
4781
4782 # Note: This is the timestamp saved as hardcoded wikitext to
4783 # the database, we use $wgContLang here in order to give
4784 # everyone the same signature and use the default one rather
4785 # than the one selected in each user's preferences.
4786 # (see also bug 12815)
4787 $ts = $this->mOptions->getTimestamp();
4788 $timestamp = MWTimestamp::getLocalInstance( $ts );
4789 $ts = $timestamp->format( 'YmdHis' );
4790 $tzMsg = $timestamp->format( 'T' ); # might vary on DST changeover!
4791
4792 # Allow translation of timezones through wiki. format() can return
4793 # whatever crap the system uses, localised or not, so we cannot
4794 # ship premade translations.
4795 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4796 $msg = wfMessage( $key )->inContentLanguage();
4797 if ( $msg->exists() ) {
4798 $tzMsg = $msg->text();
4799 }
4800
4801 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4802
4803 # Variable replacement
4804 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4805 $text = $this->replaceVariables( $text );
4806
4807 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4808 # which may corrupt this parser instance via its wfMessage()->text() call-
4809
4810 # Signatures
4811 $sigText = $this->getUserSig( $user );
4812 $text = strtr( $text, array(
4813 '~~~~~' => $d,
4814 '~~~~' => "$sigText $d",
4815 '~~~' => $sigText
4816 ) );
4817
4818 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4819 $tc = '[' . Title::legalChars() . ']';
4820 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4821
4822 // [[ns:page (context)|]]
4823 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4824 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4825 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4826 // [[ns:page (context), context|]] (using either single or double-width comma)
4827 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4828 // [[|page]] (reverse pipe trick: add context from page title)
4829 $p2 = "/\[\[\\|($tc+)]]/";
4830
4831 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4832 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4833 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4834 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4835
4836 $t = $this->mTitle->getText();
4837 $m = array();
4838 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4839 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4840 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4841 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4842 } else {
4843 # if there's no context, don't bother duplicating the title
4844 $text = preg_replace( $p2, '[[\\1]]', $text );
4845 }
4846
4847 # Trim trailing whitespace
4848 $text = rtrim( $text );
4849
4850 return $text;
4851 }
4852
4853 /**
4854 * Fetch the user's signature text, if any, and normalize to
4855 * validated, ready-to-insert wikitext.
4856 * If you have pre-fetched the nickname or the fancySig option, you can
4857 * specify them here to save a database query.
4858 * Do not reuse this parser instance after calling getUserSig(),
4859 * as it may have changed if it's the $wgParser.
4860 *
4861 * @param User $user
4862 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4863 * @param bool|null $fancySig whether the nicknname is the complete signature
4864 * or null to use default value
4865 * @return string
4866 */
4867 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4868 global $wgMaxSigChars;
4869
4870 $username = $user->getName();
4871
4872 # If not given, retrieve from the user object.
4873 if ( $nickname === false ) {
4874 $nickname = $user->getOption( 'nickname' );
4875 }
4876
4877 if ( is_null( $fancySig ) ) {
4878 $fancySig = $user->getBoolOption( 'fancysig' );
4879 }
4880
4881 $nickname = $nickname == null ? $username : $nickname;
4882
4883 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4884 $nickname = $username;
4885 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4886 } elseif ( $fancySig !== false ) {
4887 # Sig. might contain markup; validate this
4888 if ( $this->validateSig( $nickname ) !== false ) {
4889 # Validated; clean up (if needed) and return it
4890 return $this->cleanSig( $nickname, true );
4891 } else {
4892 # Failed to validate; fall back to the default
4893 $nickname = $username;
4894 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4895 }
4896 }
4897
4898 # Make sure nickname doesnt get a sig in a sig
4899 $nickname = self::cleanSigInSig( $nickname );
4900
4901 # If we're still here, make it a link to the user page
4902 $userText = wfEscapeWikiText( $username );
4903 $nickText = wfEscapeWikiText( $nickname );
4904 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4905
4906 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4907 ->title( $this->getTitle() )->text();
4908 }
4909
4910 /**
4911 * Check that the user's signature contains no bad XML
4912 *
4913 * @param string $text
4914 * @return string|bool An expanded string, or false if invalid.
4915 */
4916 public function validateSig( $text ) {
4917 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4918 }
4919
4920 /**
4921 * Clean up signature text
4922 *
4923 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4924 * 2) Substitute all transclusions
4925 *
4926 * @param string $text
4927 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4928 * @return string Signature text
4929 */
4930 public function cleanSig( $text, $parsing = false ) {
4931 if ( !$parsing ) {
4932 global $wgTitle;
4933 $magicScopeVariable = $this->lock();
4934 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4935 }
4936
4937 # Option to disable this feature
4938 if ( !$this->mOptions->getCleanSignatures() ) {
4939 return $text;
4940 }
4941
4942 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4943 # => Move this logic to braceSubstitution()
4944 $substWord = MagicWord::get( 'subst' );
4945 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4946 $substText = '{{' . $substWord->getSynonym( 0 );
4947
4948 $text = preg_replace( $substRegex, $substText, $text );
4949 $text = self::cleanSigInSig( $text );
4950 $dom = $this->preprocessToDom( $text );
4951 $frame = $this->getPreprocessor()->newFrame();
4952 $text = $frame->expand( $dom );
4953
4954 if ( !$parsing ) {
4955 $text = $this->mStripState->unstripBoth( $text );
4956 }
4957
4958 return $text;
4959 }
4960
4961 /**
4962 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4963 *
4964 * @param string $text
4965 * @return string Signature text with /~{3,5}/ removed
4966 */
4967 public static function cleanSigInSig( $text ) {
4968 $text = preg_replace( '/~{3,5}/', '', $text );
4969 return $text;
4970 }
4971
4972 /**
4973 * Set up some variables which are usually set up in parse()
4974 * so that an external function can call some class members with confidence
4975 *
4976 * @param Title|null $title
4977 * @param ParserOptions $options
4978 * @param int $outputType
4979 * @param bool $clearState
4980 */
4981 public function startExternalParse( Title $title = null, ParserOptions $options,
4982 $outputType, $clearState = true
4983 ) {
4984 $this->startParse( $title, $options, $outputType, $clearState );
4985 }
4986
4987 /**
4988 * @param Title|null $title
4989 * @param ParserOptions $options
4990 * @param int $outputType
4991 * @param bool $clearState
4992 */
4993 private function startParse( Title $title = null, ParserOptions $options,
4994 $outputType, $clearState = true
4995 ) {
4996 $this->setTitle( $title );
4997 $this->mOptions = $options;
4998 $this->setOutputType( $outputType );
4999 if ( $clearState ) {
5000 $this->clearState();
5001 }
5002 }
5003
5004 /**
5005 * Wrapper for preprocess()
5006 *
5007 * @param string $text The text to preprocess
5008 * @param ParserOptions $options Options
5009 * @param Title|null $title Title object or null to use $wgTitle
5010 * @return string
5011 */
5012 public function transformMsg( $text, $options, $title = null ) {
5013 static $executing = false;
5014
5015 # Guard against infinite recursion
5016 if ( $executing ) {
5017 return $text;
5018 }
5019 $executing = true;
5020
5021 wfProfileIn( __METHOD__ );
5022 if ( !$title ) {
5023 global $wgTitle;
5024 $title = $wgTitle;
5025 }
5026
5027 $text = $this->preprocess( $text, $title, $options );
5028
5029 $executing = false;
5030 wfProfileOut( __METHOD__ );
5031 return $text;
5032 }
5033
5034 /**
5035 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
5036 * The callback should have the following form:
5037 * function myParserHook( $text, $params, $parser, $frame ) { ... }
5038 *
5039 * Transform and return $text. Use $parser for any required context, e.g. use
5040 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
5041 *
5042 * Hooks may return extended information by returning an array, of which the
5043 * first numbered element (index 0) must be the return string, and all other
5044 * entries are extracted into local variables within an internal function
5045 * in the Parser class.
5046 *
5047 * This interface (introduced r61913) appears to be undocumented, but
5048 * 'markerName' is used by some core tag hooks to override which strip
5049 * array their results are placed in. **Use great caution if attempting
5050 * this interface, as it is not documented and injudicious use could smash
5051 * private variables.**
5052 *
5053 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5054 * @param callable $callback The callback function (and object) to use for the tag
5055 * @throws MWException
5056 * @return callable|null The old value of the mTagHooks array associated with the hook
5057 */
5058 public function setHook( $tag, $callback ) {
5059 $tag = strtolower( $tag );
5060 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5061 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5062 }
5063 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
5064 $this->mTagHooks[$tag] = $callback;
5065 if ( !in_array( $tag, $this->mStripList ) ) {
5066 $this->mStripList[] = $tag;
5067 }
5068
5069 return $oldVal;
5070 }
5071
5072 /**
5073 * As setHook(), but letting the contents be parsed.
5074 *
5075 * Transparent tag hooks are like regular XML-style tag hooks, except they
5076 * operate late in the transformation sequence, on HTML instead of wikitext.
5077 *
5078 * This is probably obsoleted by things dealing with parser frames?
5079 * The only extension currently using it is geoserver.
5080 *
5081 * @since 1.10
5082 * @todo better document or deprecate this
5083 *
5084 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5085 * @param callable $callback The callback function (and object) to use for the tag
5086 * @throws MWException
5087 * @return callable|null The old value of the mTagHooks array associated with the hook
5088 */
5089 public function setTransparentTagHook( $tag, $callback ) {
5090 $tag = strtolower( $tag );
5091 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5092 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5093 }
5094 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
5095 $this->mTransparentTagHooks[$tag] = $callback;
5096
5097 return $oldVal;
5098 }
5099
5100 /**
5101 * Remove all tag hooks
5102 */
5103 public function clearTagHooks() {
5104 $this->mTagHooks = array();
5105 $this->mFunctionTagHooks = array();
5106 $this->mStripList = $this->mDefaultStripList;
5107 }
5108
5109 /**
5110 * Create a function, e.g. {{sum:1|2|3}}
5111 * The callback function should have the form:
5112 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5113 *
5114 * Or with SFH_OBJECT_ARGS:
5115 * function myParserFunction( $parser, $frame, $args ) { ... }
5116 *
5117 * The callback may either return the text result of the function, or an array with the text
5118 * in element 0, and a number of flags in the other elements. The names of the flags are
5119 * specified in the keys. Valid flags are:
5120 * found The text returned is valid, stop processing the template. This
5121 * is on by default.
5122 * nowiki Wiki markup in the return value should be escaped
5123 * isHTML The returned text is HTML, armour it against wikitext transformation
5124 *
5125 * @param string $id The magic word ID
5126 * @param callable $callback The callback function (and object) to use
5127 * @param int $flags A combination of the following flags:
5128 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5129 *
5130 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
5131 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
5132 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5133 * the arguments, and to control the way they are expanded.
5134 *
5135 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5136 * arguments, for instance:
5137 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5138 *
5139 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5140 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5141 * working if/when this is changed.
5142 *
5143 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5144 * expansion.
5145 *
5146 * Please read the documentation in includes/parser/Preprocessor.php for more information
5147 * about the methods available in PPFrame and PPNode.
5148 *
5149 * @throws MWException
5150 * @return string|callable The old callback function for this name, if any
5151 */
5152 public function setFunctionHook( $id, $callback, $flags = 0 ) {
5153 global $wgContLang;
5154
5155 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5156 $this->mFunctionHooks[$id] = array( $callback, $flags );
5157
5158 # Add to function cache
5159 $mw = MagicWord::get( $id );
5160 if ( !$mw ) {
5161 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5162 }
5163
5164 $synonyms = $mw->getSynonyms();
5165 $sensitive = intval( $mw->isCaseSensitive() );
5166
5167 foreach ( $synonyms as $syn ) {
5168 # Case
5169 if ( !$sensitive ) {
5170 $syn = $wgContLang->lc( $syn );
5171 }
5172 # Add leading hash
5173 if ( !( $flags & SFH_NO_HASH ) ) {
5174 $syn = '#' . $syn;
5175 }
5176 # Remove trailing colon
5177 if ( substr( $syn, -1, 1 ) === ':' ) {
5178 $syn = substr( $syn, 0, -1 );
5179 }
5180 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5181 }
5182 return $oldVal;
5183 }
5184
5185 /**
5186 * Get all registered function hook identifiers
5187 *
5188 * @return array
5189 */
5190 public function getFunctionHooks() {
5191 return array_keys( $this->mFunctionHooks );
5192 }
5193
5194 /**
5195 * Create a tag function, e.g. "<test>some stuff</test>".
5196 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5197 * Unlike parser functions, their content is not preprocessed.
5198 * @param string $tag
5199 * @param callable $callback
5200 * @param int $flags
5201 * @throws MWException
5202 * @return null
5203 */
5204 public function setFunctionTagHook( $tag, $callback, $flags ) {
5205 $tag = strtolower( $tag );
5206 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5207 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5208 }
5209 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5210 $this->mFunctionTagHooks[$tag] : null;
5211 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
5212
5213 if ( !in_array( $tag, $this->mStripList ) ) {
5214 $this->mStripList[] = $tag;
5215 }
5216
5217 return $old;
5218 }
5219
5220 /**
5221 * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
5222 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5223 * Placeholders created in Skin::makeLinkObj()
5224 *
5225 * @param string $text
5226 * @param int $options
5227 *
5228 * @return array Array of link CSS classes, indexed by PDBK.
5229 */
5230 public function replaceLinkHolders( &$text, $options = 0 ) {
5231 return $this->mLinkHolders->replace( $text );
5232 }
5233
5234 /**
5235 * Replace "<!--LINK-->" link placeholders with plain text of links
5236 * (not HTML-formatted).
5237 *
5238 * @param string $text
5239 * @return string
5240 */
5241 public function replaceLinkHoldersText( $text ) {
5242 return $this->mLinkHolders->replaceText( $text );
5243 }
5244
5245 /**
5246 * Renders an image gallery from a text with one line per image.
5247 * text labels may be given by using |-style alternative text. E.g.
5248 * Image:one.jpg|The number "1"
5249 * Image:tree.jpg|A tree
5250 * given as text will return the HTML of a gallery with two images,
5251 * labeled 'The number "1"' and
5252 * 'A tree'.
5253 *
5254 * @param string $text
5255 * @param array $params
5256 * @return string HTML
5257 */
5258 public function renderImageGallery( $text, $params ) {
5259 wfProfileIn( __METHOD__ );
5260
5261 $mode = false;
5262 if ( isset( $params['mode'] ) ) {
5263 $mode = $params['mode'];
5264 }
5265
5266 try {
5267 $ig = ImageGalleryBase::factory( $mode );
5268 } catch ( MWException $e ) {
5269 // If invalid type set, fallback to default.
5270 $ig = ImageGalleryBase::factory( false );
5271 }
5272
5273 $ig->setContextTitle( $this->mTitle );
5274 $ig->setShowBytes( false );
5275 $ig->setShowFilename( false );
5276 $ig->setParser( $this );
5277 $ig->setHideBadImages();
5278 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5279
5280 if ( isset( $params['showfilename'] ) ) {
5281 $ig->setShowFilename( true );
5282 } else {
5283 $ig->setShowFilename( false );
5284 }
5285 if ( isset( $params['caption'] ) ) {
5286 $caption = $params['caption'];
5287 $caption = htmlspecialchars( $caption );
5288 $caption = $this->replaceInternalLinks( $caption );
5289 $ig->setCaptionHtml( $caption );
5290 }
5291 if ( isset( $params['perrow'] ) ) {
5292 $ig->setPerRow( $params['perrow'] );
5293 }
5294 if ( isset( $params['widths'] ) ) {
5295 $ig->setWidths( $params['widths'] );
5296 }
5297 if ( isset( $params['heights'] ) ) {
5298 $ig->setHeights( $params['heights'] );
5299 }
5300 $ig->setAdditionalOptions( $params );
5301
5302 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
5303
5304 $lines = StringUtils::explode( "\n", $text );
5305 foreach ( $lines as $line ) {
5306 # match lines like these:
5307 # Image:someimage.jpg|This is some image
5308 $matches = array();
5309 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5310 # Skip empty lines
5311 if ( count( $matches ) == 0 ) {
5312 continue;
5313 }
5314
5315 if ( strpos( $matches[0], '%' ) !== false ) {
5316 $matches[1] = rawurldecode( $matches[1] );
5317 }
5318 $title = Title::newFromText( $matches[1], NS_FILE );
5319 if ( is_null( $title ) ) {
5320 # Bogus title. Ignore these so we don't bomb out later.
5321 continue;
5322 }
5323
5324 # We need to get what handler the file uses, to figure out parameters.
5325 # Note, a hook can overide the file name, and chose an entirely different
5326 # file (which potentially could be of a different type and have different handler).
5327 $options = array();
5328 $descQuery = false;
5329 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5330 array( $this, $title, &$options, &$descQuery ) );
5331 # Don't register it now, as ImageGallery does that later.
5332 $file = $this->fetchFileNoRegister( $title, $options );
5333 $handler = $file ? $file->getHandler() : false;
5334
5335 wfProfileIn( __METHOD__ . '-getMagicWord' );
5336 $paramMap = array(
5337 'img_alt' => 'gallery-internal-alt',
5338 'img_link' => 'gallery-internal-link',
5339 );
5340 if ( $handler ) {
5341 $paramMap = $paramMap + $handler->getParamMap();
5342 // We don't want people to specify per-image widths.
5343 // Additionally the width parameter would need special casing anyhow.
5344 unset( $paramMap['img_width'] );
5345 }
5346
5347 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5348 wfProfileOut( __METHOD__ . '-getMagicWord' );
5349
5350 $label = '';
5351 $alt = '';
5352 $link = '';
5353 $handlerOptions = array();
5354 if ( isset( $matches[3] ) ) {
5355 // look for an |alt= definition while trying not to break existing
5356 // captions with multiple pipes (|) in it, until a more sensible grammar
5357 // is defined for images in galleries
5358
5359 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5360 // splitting on '|' is a bit odd, and different from makeImage.
5361 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5362 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5363
5364 foreach ( $parameterMatches as $parameterMatch ) {
5365 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5366 if ( $magicName ) {
5367 $paramName = $paramMap[$magicName];
5368
5369 switch ( $paramName ) {
5370 case 'gallery-internal-alt':
5371 $alt = $this->stripAltText( $match, false );
5372 break;
5373 case 'gallery-internal-link':
5374 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5375 $chars = self::EXT_LINK_URL_CLASS;
5376 $prots = $this->mUrlProtocols;
5377 //check to see if link matches an absolute url, if not then it must be a wiki link.
5378 if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
5379 $link = $linkValue;
5380 } else {
5381 $localLinkTitle = Title::newFromText( $linkValue );
5382 if ( $localLinkTitle !== null ) {
5383 $link = $localLinkTitle->getLinkURL();
5384 }
5385 }
5386 break;
5387 default:
5388 // Must be a handler specific parameter.
5389 if ( $handler->validateParam( $paramName, $match ) ) {
5390 $handlerOptions[$paramName] = $match;
5391 } else {
5392 // Guess not. Append it to the caption.
5393 wfDebug( "$parameterMatch failed parameter validation\n" );
5394 $label .= '|' . $parameterMatch;
5395 }
5396 }
5397
5398 } else {
5399 // concatenate all other pipes
5400 $label .= '|' . $parameterMatch;
5401 }
5402 }
5403 // remove the first pipe
5404 $label = substr( $label, 1 );
5405 }
5406
5407 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5408 }
5409 $html = $ig->toHTML();
5410 wfRunHooks( 'AfterParserFetchFileAndTitle', array( $this, $ig, &$html ) );
5411 wfProfileOut( __METHOD__ );
5412 return $html;
5413 }
5414
5415 /**
5416 * @param string $handler
5417 * @return array
5418 */
5419 public function getImageParams( $handler ) {
5420 if ( $handler ) {
5421 $handlerClass = get_class( $handler );
5422 } else {
5423 $handlerClass = '';
5424 }
5425 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5426 # Initialise static lists
5427 static $internalParamNames = array(
5428 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
5429 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5430 'bottom', 'text-bottom' ),
5431 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
5432 'upright', 'border', 'link', 'alt', 'class' ),
5433 );
5434 static $internalParamMap;
5435 if ( !$internalParamMap ) {
5436 $internalParamMap = array();
5437 foreach ( $internalParamNames as $type => $names ) {
5438 foreach ( $names as $name ) {
5439 $magicName = str_replace( '-', '_', "img_$name" );
5440 $internalParamMap[$magicName] = array( $type, $name );
5441 }
5442 }
5443 }
5444
5445 # Add handler params
5446 $paramMap = $internalParamMap;
5447 if ( $handler ) {
5448 $handlerParamMap = $handler->getParamMap();
5449 foreach ( $handlerParamMap as $magic => $paramName ) {
5450 $paramMap[$magic] = array( 'handler', $paramName );
5451 }
5452 }
5453 $this->mImageParams[$handlerClass] = $paramMap;
5454 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5455 }
5456 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5457 }
5458
5459 /**
5460 * Parse image options text and use it to make an image
5461 *
5462 * @param Title $title
5463 * @param string $options
5464 * @param LinkHolderArray|bool $holders
5465 * @return string HTML
5466 */
5467 public function makeImage( $title, $options, $holders = false ) {
5468 # Check if the options text is of the form "options|alt text"
5469 # Options are:
5470 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5471 # * left no resizing, just left align. label is used for alt= only
5472 # * right same, but right aligned
5473 # * none same, but not aligned
5474 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5475 # * center center the image
5476 # * frame Keep original image size, no magnify-button.
5477 # * framed Same as "frame"
5478 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5479 # * upright reduce width for upright images, rounded to full __0 px
5480 # * border draw a 1px border around the image
5481 # * alt Text for HTML alt attribute (defaults to empty)
5482 # * class Set a class for img node
5483 # * link Set the target of the image link. Can be external, interwiki, or local
5484 # vertical-align values (no % or length right now):
5485 # * baseline
5486 # * sub
5487 # * super
5488 # * top
5489 # * text-top
5490 # * middle
5491 # * bottom
5492 # * text-bottom
5493
5494 $parts = StringUtils::explode( "|", $options );
5495
5496 # Give extensions a chance to select the file revision for us
5497 $options = array();
5498 $descQuery = false;
5499 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5500 array( $this, $title, &$options, &$descQuery ) );
5501 # Fetch and register the file (file title may be different via hooks)
5502 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5503
5504 # Get parameter map
5505 $handler = $file ? $file->getHandler() : false;
5506
5507 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5508
5509 if ( !$file ) {
5510 $this->addTrackingCategory( 'broken-file-category' );
5511 }
5512
5513 # Process the input parameters
5514 $caption = '';
5515 $params = array( 'frame' => array(), 'handler' => array(),
5516 'horizAlign' => array(), 'vertAlign' => array() );
5517 $seenformat = false;
5518 foreach ( $parts as $part ) {
5519 $part = trim( $part );
5520 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5521 $validated = false;
5522 if ( isset( $paramMap[$magicName] ) ) {
5523 list( $type, $paramName ) = $paramMap[$magicName];
5524
5525 # Special case; width and height come in one variable together
5526 if ( $type === 'handler' && $paramName === 'width' ) {
5527 $parsedWidthParam = $this->parseWidthParam( $value );
5528 if ( isset( $parsedWidthParam['width'] ) ) {
5529 $width = $parsedWidthParam['width'];
5530 if ( $handler->validateParam( 'width', $width ) ) {
5531 $params[$type]['width'] = $width;
5532 $validated = true;
5533 }
5534 }
5535 if ( isset( $parsedWidthParam['height'] ) ) {
5536 $height = $parsedWidthParam['height'];
5537 if ( $handler->validateParam( 'height', $height ) ) {
5538 $params[$type]['height'] = $height;
5539 $validated = true;
5540 }
5541 }
5542 # else no validation -- bug 13436
5543 } else {
5544 if ( $type === 'handler' ) {
5545 # Validate handler parameter
5546 $validated = $handler->validateParam( $paramName, $value );
5547 } else {
5548 # Validate internal parameters
5549 switch ( $paramName ) {
5550 case 'manualthumb':
5551 case 'alt':
5552 case 'class':
5553 # @todo FIXME: Possibly check validity here for
5554 # manualthumb? downstream behavior seems odd with
5555 # missing manual thumbs.
5556 $validated = true;
5557 $value = $this->stripAltText( $value, $holders );
5558 break;
5559 case 'link':
5560 $chars = self::EXT_LINK_URL_CLASS;
5561 $prots = $this->mUrlProtocols;
5562 if ( $value === '' ) {
5563 $paramName = 'no-link';
5564 $value = true;
5565 $validated = true;
5566 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5567 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5568 $paramName = 'link-url';
5569 $this->mOutput->addExternalLink( $value );
5570 if ( $this->mOptions->getExternalLinkTarget() ) {
5571 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5572 }
5573 $validated = true;
5574 }
5575 } else {
5576 $linkTitle = Title::newFromText( $value );
5577 if ( $linkTitle ) {
5578 $paramName = 'link-title';
5579 $value = $linkTitle;
5580 $this->mOutput->addLink( $linkTitle );
5581 $validated = true;
5582 }
5583 }
5584 break;
5585 case 'frameless':
5586 case 'framed':
5587 case 'thumbnail':
5588 // use first appearing option, discard others.
5589 $validated = ! $seenformat;
5590 $seenformat = true;
5591 break;
5592 default:
5593 # Most other things appear to be empty or numeric...
5594 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5595 }
5596 }
5597
5598 if ( $validated ) {
5599 $params[$type][$paramName] = $value;
5600 }
5601 }
5602 }
5603 if ( !$validated ) {
5604 $caption = $part;
5605 }
5606 }
5607
5608 # Process alignment parameters
5609 if ( $params['horizAlign'] ) {
5610 $params['frame']['align'] = key( $params['horizAlign'] );
5611 }
5612 if ( $params['vertAlign'] ) {
5613 $params['frame']['valign'] = key( $params['vertAlign'] );
5614 }
5615
5616 $params['frame']['caption'] = $caption;
5617
5618 # Will the image be presented in a frame, with the caption below?
5619 $imageIsFramed = isset( $params['frame']['frame'] )
5620 || isset( $params['frame']['framed'] )
5621 || isset( $params['frame']['thumbnail'] )
5622 || isset( $params['frame']['manualthumb'] );
5623
5624 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5625 # came to also set the caption, ordinary text after the image -- which
5626 # makes no sense, because that just repeats the text multiple times in
5627 # screen readers. It *also* came to set the title attribute.
5628 #
5629 # Now that we have an alt attribute, we should not set the alt text to
5630 # equal the caption: that's worse than useless, it just repeats the
5631 # text. This is the framed/thumbnail case. If there's no caption, we
5632 # use the unnamed parameter for alt text as well, just for the time be-
5633 # ing, if the unnamed param is set and the alt param is not.
5634 #
5635 # For the future, we need to figure out if we want to tweak this more,
5636 # e.g., introducing a title= parameter for the title; ignoring the un-
5637 # named parameter entirely for images without a caption; adding an ex-
5638 # plicit caption= parameter and preserving the old magic unnamed para-
5639 # meter for BC; ...
5640 if ( $imageIsFramed ) { # Framed image
5641 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5642 # No caption or alt text, add the filename as the alt text so
5643 # that screen readers at least get some description of the image
5644 $params['frame']['alt'] = $title->getText();
5645 }
5646 # Do not set $params['frame']['title'] because tooltips don't make sense
5647 # for framed images
5648 } else { # Inline image
5649 if ( !isset( $params['frame']['alt'] ) ) {
5650 # No alt text, use the "caption" for the alt text
5651 if ( $caption !== '' ) {
5652 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5653 } else {
5654 # No caption, fall back to using the filename for the
5655 # alt text
5656 $params['frame']['alt'] = $title->getText();
5657 }
5658 }
5659 # Use the "caption" for the tooltip text
5660 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5661 }
5662
5663 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5664
5665 # Linker does the rest
5666 $time = isset( $options['time'] ) ? $options['time'] : false;
5667 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5668 $time, $descQuery, $this->mOptions->getThumbSize() );
5669
5670 # Give the handler a chance to modify the parser object
5671 if ( $handler ) {
5672 $handler->parserTransformHook( $this, $file );
5673 }
5674
5675 return $ret;
5676 }
5677
5678 /**
5679 * @param string $caption
5680 * @param LinkHolderArray|bool $holders
5681 * @return mixed|string
5682 */
5683 protected function stripAltText( $caption, $holders ) {
5684 # Strip bad stuff out of the title (tooltip). We can't just use
5685 # replaceLinkHoldersText() here, because if this function is called
5686 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5687 if ( $holders ) {
5688 $tooltip = $holders->replaceText( $caption );
5689 } else {
5690 $tooltip = $this->replaceLinkHoldersText( $caption );
5691 }
5692
5693 # make sure there are no placeholders in thumbnail attributes
5694 # that are later expanded to html- so expand them now and
5695 # remove the tags
5696 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5697 $tooltip = Sanitizer::stripAllTags( $tooltip );
5698
5699 return $tooltip;
5700 }
5701
5702 /**
5703 * Set a flag in the output object indicating that the content is dynamic and
5704 * shouldn't be cached.
5705 */
5706 public function disableCache() {
5707 wfDebug( "Parser output marked as uncacheable.\n" );
5708 if ( !$this->mOutput ) {
5709 throw new MWException( __METHOD__ .
5710 " can only be called when actually parsing something" );
5711 }
5712 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5713 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5714 }
5715
5716 /**
5717 * Callback from the Sanitizer for expanding items found in HTML attribute
5718 * values, so they can be safely tested and escaped.
5719 *
5720 * @param string $text
5721 * @param bool|PPFrame $frame
5722 * @return string
5723 */
5724 public function attributeStripCallback( &$text, $frame = false ) {
5725 $text = $this->replaceVariables( $text, $frame );
5726 $text = $this->mStripState->unstripBoth( $text );
5727 return $text;
5728 }
5729
5730 /**
5731 * Accessor
5732 *
5733 * @return array
5734 */
5735 public function getTags() {
5736 return array_merge(
5737 array_keys( $this->mTransparentTagHooks ),
5738 array_keys( $this->mTagHooks ),
5739 array_keys( $this->mFunctionTagHooks )
5740 );
5741 }
5742
5743 /**
5744 * Replace transparent tags in $text with the values given by the callbacks.
5745 *
5746 * Transparent tag hooks are like regular XML-style tag hooks, except they
5747 * operate late in the transformation sequence, on HTML instead of wikitext.
5748 *
5749 * @param string $text
5750 *
5751 * @return string
5752 */
5753 public function replaceTransparentTags( $text ) {
5754 $matches = array();
5755 $elements = array_keys( $this->mTransparentTagHooks );
5756 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5757 $replacements = array();
5758
5759 foreach ( $matches as $marker => $data ) {
5760 list( $element, $content, $params, $tag ) = $data;
5761 $tagName = strtolower( $element );
5762 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5763 $output = call_user_func_array(
5764 $this->mTransparentTagHooks[$tagName],
5765 array( $content, $params, $this )
5766 );
5767 } else {
5768 $output = $tag;
5769 }
5770 $replacements[$marker] = $output;
5771 }
5772 return strtr( $text, $replacements );
5773 }
5774
5775 /**
5776 * Break wikitext input into sections, and either pull or replace
5777 * some particular section's text.
5778 *
5779 * External callers should use the getSection and replaceSection methods.
5780 *
5781 * @param string $text Page wikitext
5782 * @param string|number $sectionId A section identifier string of the form:
5783 * "<flag1> - <flag2> - ... - <section number>"
5784 *
5785 * Currently the only recognised flag is "T", which means the target section number
5786 * was derived during a template inclusion parse, in other words this is a template
5787 * section edit link. If no flags are given, it was an ordinary section edit link.
5788 * This flag is required to avoid a section numbering mismatch when a section is
5789 * enclosed by "<includeonly>" (bug 6563).
5790 *
5791 * The section number 0 pulls the text before the first heading; other numbers will
5792 * pull the given section along with its lower-level subsections. If the section is
5793 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5794 *
5795 * Section 0 is always considered to exist, even if it only contains the empty
5796 * string. If $text is the empty string and section 0 is replaced, $newText is
5797 * returned.
5798 *
5799 * @param string $mode One of "get" or "replace"
5800 * @param string $newText Replacement text for section data.
5801 * @return string For "get", the extracted section text.
5802 * for "replace", the whole page with the section replaced.
5803 */
5804 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5805 global $wgTitle; # not generally used but removes an ugly failure mode
5806
5807 $magicScopeVariable = $this->lock();
5808 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5809 $outText = '';
5810 $frame = $this->getPreprocessor()->newFrame();
5811
5812 # Process section extraction flags
5813 $flags = 0;
5814 $sectionParts = explode( '-', $sectionId );
5815 $sectionIndex = array_pop( $sectionParts );
5816 foreach ( $sectionParts as $part ) {
5817 if ( $part === 'T' ) {
5818 $flags |= self::PTD_FOR_INCLUSION;
5819 }
5820 }
5821
5822 # Check for empty input
5823 if ( strval( $text ) === '' ) {
5824 # Only sections 0 and T-0 exist in an empty document
5825 if ( $sectionIndex == 0 ) {
5826 if ( $mode === 'get' ) {
5827 return '';
5828 } else {
5829 return $newText;
5830 }
5831 } else {
5832 if ( $mode === 'get' ) {
5833 return $newText;
5834 } else {
5835 return $text;
5836 }
5837 }
5838 }
5839
5840 # Preprocess the text
5841 $root = $this->preprocessToDom( $text, $flags );
5842
5843 # <h> nodes indicate section breaks
5844 # They can only occur at the top level, so we can find them by iterating the root's children
5845 $node = $root->getFirstChild();
5846
5847 # Find the target section
5848 if ( $sectionIndex == 0 ) {
5849 # Section zero doesn't nest, level=big
5850 $targetLevel = 1000;
5851 } else {
5852 while ( $node ) {
5853 if ( $node->getName() === 'h' ) {
5854 $bits = $node->splitHeading();
5855 if ( $bits['i'] == $sectionIndex ) {
5856 $targetLevel = $bits['level'];
5857 break;
5858 }
5859 }
5860 if ( $mode === 'replace' ) {
5861 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5862 }
5863 $node = $node->getNextSibling();
5864 }
5865 }
5866
5867 if ( !$node ) {
5868 # Not found
5869 if ( $mode === 'get' ) {
5870 return $newText;
5871 } else {
5872 return $text;
5873 }
5874 }
5875
5876 # Find the end of the section, including nested sections
5877 do {
5878 if ( $node->getName() === 'h' ) {
5879 $bits = $node->splitHeading();
5880 $curLevel = $bits['level'];
5881 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5882 break;
5883 }
5884 }
5885 if ( $mode === 'get' ) {
5886 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5887 }
5888 $node = $node->getNextSibling();
5889 } while ( $node );
5890
5891 # Write out the remainder (in replace mode only)
5892 if ( $mode === 'replace' ) {
5893 # Output the replacement text
5894 # Add two newlines on -- trailing whitespace in $newText is conventionally
5895 # stripped by the editor, so we need both newlines to restore the paragraph gap
5896 # Only add trailing whitespace if there is newText
5897 if ( $newText != "" ) {
5898 $outText .= $newText . "\n\n";
5899 }
5900
5901 while ( $node ) {
5902 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5903 $node = $node->getNextSibling();
5904 }
5905 }
5906
5907 if ( is_string( $outText ) ) {
5908 # Re-insert stripped tags
5909 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5910 }
5911
5912 return $outText;
5913 }
5914
5915 /**
5916 * This function returns the text of a section, specified by a number ($section).
5917 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5918 * the first section before any such heading (section 0).
5919 *
5920 * If a section contains subsections, these are also returned.
5921 *
5922 * @param string $text Text to look in
5923 * @param string|number $sectionId Section identifier as a number or string
5924 * (e.g. 0, 1 or 'T-1').
5925 * @param string $defaultText Default to return if section is not found
5926 *
5927 * @return string Text of the requested section
5928 */
5929 public function getSection( $text, $sectionId, $defaultText = '' ) {
5930 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5931 }
5932
5933 /**
5934 * This function returns $oldtext after the content of the section
5935 * specified by $section has been replaced with $text. If the target
5936 * section does not exist, $oldtext is returned unchanged.
5937 *
5938 * @param string $oldText Former text of the article
5939 * @param string|number $sectionId Section identifier as a number or string
5940 * (e.g. 0, 1 or 'T-1').
5941 * @param string $newText Replacing text
5942 *
5943 * @return string Modified text
5944 */
5945 public function replaceSection( $oldText, $sectionId, $newText ) {
5946 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5947 }
5948
5949 /**
5950 * Get the ID of the revision we are parsing
5951 *
5952 * @return int|null
5953 */
5954 public function getRevisionId() {
5955 return $this->mRevisionId;
5956 }
5957
5958 /**
5959 * Get the revision object for $this->mRevisionId
5960 *
5961 * @return Revision|null Either a Revision object or null
5962 * @since 1.23 (public since 1.23)
5963 */
5964 public function getRevisionObject() {
5965 if ( !is_null( $this->mRevisionObject ) ) {
5966 return $this->mRevisionObject;
5967 }
5968 if ( is_null( $this->mRevisionId ) ) {
5969 return null;
5970 }
5971
5972 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5973 return $this->mRevisionObject;
5974 }
5975
5976 /**
5977 * Get the timestamp associated with the current revision, adjusted for
5978 * the default server-local timestamp
5979 * @return string
5980 */
5981 public function getRevisionTimestamp() {
5982 if ( is_null( $this->mRevisionTimestamp ) ) {
5983 wfProfileIn( __METHOD__ );
5984
5985 global $wgContLang;
5986
5987 $revObject = $this->getRevisionObject();
5988 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5989
5990 # The cryptic '' timezone parameter tells to use the site-default
5991 # timezone offset instead of the user settings.
5992 #
5993 # Since this value will be saved into the parser cache, served
5994 # to other users, and potentially even used inside links and such,
5995 # it needs to be consistent for all visitors.
5996 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5997
5998 wfProfileOut( __METHOD__ );
5999 }
6000 return $this->mRevisionTimestamp;
6001 }
6002
6003 /**
6004 * Get the name of the user that edited the last revision
6005 *
6006 * @return string User name
6007 */
6008 public function getRevisionUser() {
6009 if ( is_null( $this->mRevisionUser ) ) {
6010 $revObject = $this->getRevisionObject();
6011
6012 # if this template is subst: the revision id will be blank,
6013 # so just use the current user's name
6014 if ( $revObject ) {
6015 $this->mRevisionUser = $revObject->getUserText();
6016 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6017 $this->mRevisionUser = $this->getUser()->getName();
6018 }
6019 }
6020 return $this->mRevisionUser;
6021 }
6022
6023 /**
6024 * Get the size of the revision
6025 *
6026 * @return int|null Revision size
6027 */
6028 public function getRevisionSize() {
6029 if ( is_null( $this->mRevisionSize ) ) {
6030 $revObject = $this->getRevisionObject();
6031
6032 # if this variable is subst: the revision id will be blank,
6033 # so just use the parser input size, because the own substituation
6034 # will change the size.
6035 if ( $revObject ) {
6036 $this->mRevisionSize = $revObject->getSize();
6037 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6038 $this->mRevisionSize = $this->mInputSize;
6039 }
6040 }
6041 return $this->mRevisionSize;
6042 }
6043
6044 /**
6045 * Mutator for $mDefaultSort
6046 *
6047 * @param string $sort New value
6048 */
6049 public function setDefaultSort( $sort ) {
6050 $this->mDefaultSort = $sort;
6051 $this->mOutput->setProperty( 'defaultsort', $sort );
6052 }
6053
6054 /**
6055 * Accessor for $mDefaultSort
6056 * Will use the empty string if none is set.
6057 *
6058 * This value is treated as a prefix, so the
6059 * empty string is equivalent to sorting by
6060 * page name.
6061 *
6062 * @return string
6063 */
6064 public function getDefaultSort() {
6065 if ( $this->mDefaultSort !== false ) {
6066 return $this->mDefaultSort;
6067 } else {
6068 return '';
6069 }
6070 }
6071
6072 /**
6073 * Accessor for $mDefaultSort
6074 * Unlike getDefaultSort(), will return false if none is set
6075 *
6076 * @return string|bool
6077 */
6078 public function getCustomDefaultSort() {
6079 return $this->mDefaultSort;
6080 }
6081
6082 /**
6083 * Try to guess the section anchor name based on a wikitext fragment
6084 * presumably extracted from a heading, for example "Header" from
6085 * "== Header ==".
6086 *
6087 * @param string $text
6088 *
6089 * @return string
6090 */
6091 public function guessSectionNameFromWikiText( $text ) {
6092 # Strip out wikitext links(they break the anchor)
6093 $text = $this->stripSectionName( $text );
6094 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6095 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
6096 }
6097
6098 /**
6099 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6100 * instead. For use in redirects, since IE6 interprets Redirect: headers
6101 * as something other than UTF-8 (apparently?), resulting in breakage.
6102 *
6103 * @param string $text The section name
6104 * @return string An anchor
6105 */
6106 public function guessLegacySectionNameFromWikiText( $text ) {
6107 # Strip out wikitext links(they break the anchor)
6108 $text = $this->stripSectionName( $text );
6109 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6110 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
6111 }
6112
6113 /**
6114 * Strips a text string of wikitext for use in a section anchor
6115 *
6116 * Accepts a text string and then removes all wikitext from the
6117 * string and leaves only the resultant text (i.e. the result of
6118 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6119 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6120 * to create valid section anchors by mimicing the output of the
6121 * parser when headings are parsed.
6122 *
6123 * @param string $text Text string to be stripped of wikitext
6124 * for use in a Section anchor
6125 * @return string Filtered text string
6126 */
6127 public function stripSectionName( $text ) {
6128 # Strip internal link markup
6129 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6130 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6131
6132 # Strip external link markup
6133 # @todo FIXME: Not tolerant to blank link text
6134 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6135 # on how many empty links there are on the page - need to figure that out.
6136 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6137
6138 # Parse wikitext quotes (italics & bold)
6139 $text = $this->doQuotes( $text );
6140
6141 # Strip HTML tags
6142 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6143 return $text;
6144 }
6145
6146 /**
6147 * strip/replaceVariables/unstrip for preprocessor regression testing
6148 *
6149 * @param string $text
6150 * @param Title $title
6151 * @param ParserOptions $options
6152 * @param int $outputType
6153 *
6154 * @return string
6155 */
6156 public function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
6157 $magicScopeVariable = $this->lock();
6158 $this->startParse( $title, $options, $outputType, true );
6159
6160 $text = $this->replaceVariables( $text );
6161 $text = $this->mStripState->unstripBoth( $text );
6162 $text = Sanitizer::removeHTMLtags( $text );
6163 return $text;
6164 }
6165
6166 /**
6167 * @param string $text
6168 * @param Title $title
6169 * @param ParserOptions $options
6170 * @return string
6171 */
6172 public function testPst( $text, Title $title, ParserOptions $options ) {
6173 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6174 }
6175
6176 /**
6177 * @param string $text
6178 * @param Title $title
6179 * @param ParserOptions $options
6180 * @return string
6181 */
6182 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6183 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6184 }
6185
6186 /**
6187 * Call a callback function on all regions of the given text that are not
6188 * inside strip markers, and replace those regions with the return value
6189 * of the callback. For example, with input:
6190 *
6191 * aaa<MARKER>bbb
6192 *
6193 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6194 * two strings will be replaced with the value returned by the callback in
6195 * each case.
6196 *
6197 * @param string $s
6198 * @param callable $callback
6199 *
6200 * @return string
6201 */
6202 public function markerSkipCallback( $s, $callback ) {
6203 $i = 0;
6204 $out = '';
6205 while ( $i < strlen( $s ) ) {
6206 $markerStart = strpos( $s, $this->mUniqPrefix, $i );
6207 if ( $markerStart === false ) {
6208 $out .= call_user_func( $callback, substr( $s, $i ) );
6209 break;
6210 } else {
6211 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6212 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6213 if ( $markerEnd === false ) {
6214 $out .= substr( $s, $markerStart );
6215 break;
6216 } else {
6217 $markerEnd += strlen( self::MARKER_SUFFIX );
6218 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6219 $i = $markerEnd;
6220 }
6221 }
6222 }
6223 return $out;
6224 }
6225
6226 /**
6227 * Remove any strip markers found in the given text.
6228 *
6229 * @param string $text Input string
6230 * @return string
6231 */
6232 public function killMarkers( $text ) {
6233 return $this->mStripState->killMarkers( $text );
6234 }
6235
6236 /**
6237 * Save the parser state required to convert the given half-parsed text to
6238 * HTML. "Half-parsed" in this context means the output of
6239 * recursiveTagParse() or internalParse(). This output has strip markers
6240 * from replaceVariables (extensionSubstitution() etc.), and link
6241 * placeholders from replaceLinkHolders().
6242 *
6243 * Returns an array which can be serialized and stored persistently. This
6244 * array can later be loaded into another parser instance with
6245 * unserializeHalfParsedText(). The text can then be safely incorporated into
6246 * the return value of a parser hook.
6247 *
6248 * @param string $text
6249 *
6250 * @return array
6251 */
6252 public function serializeHalfParsedText( $text ) {
6253 wfProfileIn( __METHOD__ );
6254 $data = array(
6255 'text' => $text,
6256 'version' => self::HALF_PARSED_VERSION,
6257 'stripState' => $this->mStripState->getSubState( $text ),
6258 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6259 );
6260 wfProfileOut( __METHOD__ );
6261 return $data;
6262 }
6263
6264 /**
6265 * Load the parser state given in the $data array, which is assumed to
6266 * have been generated by serializeHalfParsedText(). The text contents is
6267 * extracted from the array, and its markers are transformed into markers
6268 * appropriate for the current Parser instance. This transformed text is
6269 * returned, and can be safely included in the return value of a parser
6270 * hook.
6271 *
6272 * If the $data array has been stored persistently, the caller should first
6273 * check whether it is still valid, by calling isValidHalfParsedText().
6274 *
6275 * @param array $data Serialized data
6276 * @throws MWException
6277 * @return string
6278 */
6279 public function unserializeHalfParsedText( $data ) {
6280 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6281 throw new MWException( __METHOD__ . ': invalid version' );
6282 }
6283
6284 # First, extract the strip state.
6285 $texts = array( $data['text'] );
6286 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6287
6288 # Now renumber links
6289 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6290
6291 # Should be good to go.
6292 return $texts[0];
6293 }
6294
6295 /**
6296 * Returns true if the given array, presumed to be generated by
6297 * serializeHalfParsedText(), is compatible with the current version of the
6298 * parser.
6299 *
6300 * @param array $data
6301 *
6302 * @return bool
6303 */
6304 public function isValidHalfParsedText( $data ) {
6305 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6306 }
6307
6308 /**
6309 * Parsed a width param of imagelink like 300px or 200x300px
6310 *
6311 * @param string $value
6312 *
6313 * @return array
6314 * @since 1.20
6315 */
6316 public function parseWidthParam( $value ) {
6317 $parsedWidthParam = array();
6318 if ( $value === '' ) {
6319 return $parsedWidthParam;
6320 }
6321 $m = array();
6322 # (bug 13500) In both cases (width/height and width only),
6323 # permit trailing "px" for backward compatibility.
6324 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6325 $width = intval( $m[1] );
6326 $height = intval( $m[2] );
6327 $parsedWidthParam['width'] = $width;
6328 $parsedWidthParam['height'] = $height;
6329 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6330 $width = intval( $value );
6331 $parsedWidthParam['width'] = $width;
6332 }
6333 return $parsedWidthParam;
6334 }
6335
6336 /**
6337 * Lock the current instance of the parser.
6338 *
6339 * This is meant to stop someone from calling the parser
6340 * recursively and messing up all the strip state.
6341 *
6342 * @throws MWException If parser is in a parse
6343 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6344 */
6345 protected function lock() {
6346 if ( $this->mInParse ) {
6347 throw new MWException( "Parser state cleared while parsing. "
6348 . "Did you call Parser::parse recursively?" );
6349 }
6350 $this->mInParse = true;
6351
6352 $that = $this;
6353 $recursiveCheck = new ScopedCallback( function() use ( $that ) {
6354 $that->mInParse = false;
6355 } );
6356
6357 return $recursiveCheck;
6358 }
6359
6360 /**
6361 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6362 *
6363 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6364 * or if there is more than one <p/> tag in the input HTML.
6365 *
6366 * @param string $html
6367 * @return string
6368 * @since 1.24
6369 */
6370 public static function stripOuterParagraph( $html ) {
6371 $m = array();
6372 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6373 if ( strpos( $m[1], '</p>' ) === false ) {
6374 $html = $m[1];
6375 }
6376 }
6377
6378 return $html;
6379 }
6380
6381 /**
6382 * Return this parser if it is not doing anything, otherwise
6383 * get a fresh parser. You can use this method by doing
6384 * $myParser = $wgParser->getFreshParser(), or more simply
6385 * $wgParser->getFreshParser()->parse( ... );
6386 * if you're unsure if $wgParser is safe to use.
6387 *
6388 * @since 1.24
6389 * @return Parser A parser object that is not parsing anything
6390 */
6391 public function getFreshParser() {
6392 global $wgParserConf;
6393 if ( $this->mInParse ) {
6394 return new $wgParserConf['class']( $wgParserConf );
6395 } else {
6396 return $this;
6397 }
6398 }
6399 }
Wiklou, le Wiki du Wiklou