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