dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Replace deprecated wfMsg* calls with Message class calls.
[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 XHTML 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 * $wgLocaltimezone
58 * $wgNamespacesWithSubpages
59 *
60 * @par Settings only within ParserOptions:
61 * $wgAllowExternalImages
62 * $wgAllowSpecialInclusion
63 * $wgInterwikiMagic
64 * $wgMaxArticleSize
65 * $wgUseDynamicDates
66 *
67 * @ingroup Parser
68 */
69 class Parser {
70 /**
71 * Update this version number when the ParserOutput format
72 * changes in an incompatible way, so the parser cache
73 * can automatically discard old data.
74 */
75 const VERSION = '1.6.4';
76
77 /**
78 * Update this version number when the output of serialiseHalfParsedText()
79 * changes in an incompatible way
80 */
81 const HALF_PARSED_VERSION = 2;
82
83 # Flags for Parser::setFunctionHook
84 # Also available as global constants from Defines.php
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
92 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
93 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
94 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
95
96 # State constants for the definition list colon extraction
97 const COLON_STATE_TEXT = 0;
98 const COLON_STATE_TAG = 1;
99 const COLON_STATE_TAGSTART = 2;
100 const COLON_STATE_CLOSETAG = 3;
101 const COLON_STATE_TAGSLASH = 4;
102 const COLON_STATE_COMMENT = 5;
103 const COLON_STATE_COMMENTDASH = 6;
104 const COLON_STATE_COMMENTDASHDASH = 7;
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 # Marker Suffix needs to be accessible staticly.
118 const MARKER_SUFFIX = "-QINU\x7f";
119
120 # Persistent:
121 var $mTagHooks = array();
122 var $mTransparentTagHooks = array();
123 var $mFunctionHooks = array();
124 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
125 var $mFunctionTagHooks = array();
126 var $mStripList = array();
127 var $mDefaultStripList = array();
128 var $mVarCache = array();
129 var $mImageParams = array();
130 var $mImageParamsMagicArray = array();
131 var $mMarkerIndex = 0;
132 var $mFirstCall = true;
133
134 # Initialised by initialiseVariables()
135
136 /**
137 * @var MagicWordArray
138 */
139 var $mVariables;
140
141 /**
142 * @var MagicWordArray
143 */
144 var $mSubstWords;
145 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
146
147 # Cleared with clearState():
148 /**
149 * @var ParserOutput
150 */
151 var $mOutput;
152 var $mAutonumber, $mDTopen;
153
154 /**
155 * @var StripState
156 */
157 var $mStripState;
158
159 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
160 /**
161 * @var LinkHolderArray
162 */
163 var $mLinkHolders;
164
165 var $mLinkID;
166 var $mIncludeSizes, $mPPNodeCount, $mHighestExpansionDepth, $mDefaultSort;
167 var $mTplExpandCache; # empty-frame expansion cache
168 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
169 var $mExpensiveFunctionCount; # number of expensive parser function calls
170 var $mShowToc, $mForceTocPosition;
171
172 /**
173 * @var User
174 */
175 var $mUser; # User object; only used when doing pre-save transform
176
177 # Temporary
178 # These are variables reset at least once per parse regardless of $clearState
179
180 /**
181 * @var ParserOptions
182 */
183 var $mOptions;
184
185 /**
186 * @var Title
187 */
188 var $mTitle; # Title context, used for self-link rendering and similar things
189 var $mOutputType; # Output type, one of the OT_xxx constants
190 var $ot; # Shortcut alias, see setOutputType()
191 var $mRevisionObject; # The revision object of the specified revision ID
192 var $mRevisionId; # ID to display in {{REVISIONID}} tags
193 var $mRevisionTimestamp; # The timestamp of the specified revision ID
194 var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
195 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
196
197 /**
198 * @var string
199 */
200 var $mUniqPrefix;
201
202 /**
203 * Constructor
204 *
205 * @param $conf array
206 */
207 public function __construct( $conf = array() ) {
208 $this->mConf = $conf;
209 $this->mUrlProtocols = wfUrlProtocols();
210 $this->mExtLinkBracketedRegex = '/\[((' . $this->mUrlProtocols . ')'.
211 self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
212 if ( isset( $conf['preprocessorClass'] ) ) {
213 $this->mPreprocessorClass = $conf['preprocessorClass'];
214 } elseif ( defined( 'MW_COMPILED' ) ) {
215 # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
216 $this->mPreprocessorClass = 'Preprocessor_Hash';
217 } elseif ( extension_loaded( 'domxml' ) ) {
218 # PECL extension that conflicts with the core DOM extension (bug 13770)
219 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
220 $this->mPreprocessorClass = 'Preprocessor_Hash';
221 } elseif ( extension_loaded( 'dom' ) ) {
222 $this->mPreprocessorClass = 'Preprocessor_DOM';
223 } else {
224 $this->mPreprocessorClass = 'Preprocessor_Hash';
225 }
226 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
227 }
228
229 /**
230 * Reduce memory usage to reduce the impact of circular references
231 */
232 function __destruct() {
233 if ( isset( $this->mLinkHolders ) ) {
234 unset( $this->mLinkHolders );
235 }
236 foreach ( $this as $name => $value ) {
237 unset( $this->$name );
238 }
239 }
240
241 /**
242 * Do various kinds of initialisation on the first call of the parser
243 */
244 function firstCallInit() {
245 if ( !$this->mFirstCall ) {
246 return;
247 }
248 $this->mFirstCall = false;
249
250 wfProfileIn( __METHOD__ );
251
252 CoreParserFunctions::register( $this );
253 CoreTagHooks::register( $this );
254 $this->initialiseVariables();
255
256 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
257 wfProfileOut( __METHOD__ );
258 }
259
260 /**
261 * Clear Parser state
262 *
263 * @private
264 */
265 function clearState() {
266 wfProfileIn( __METHOD__ );
267 if ( $this->mFirstCall ) {
268 $this->firstCallInit();
269 }
270 $this->mOutput = new ParserOutput;
271 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
272 $this->mAutonumber = 0;
273 $this->mLastSection = '';
274 $this->mDTopen = false;
275 $this->mIncludeCount = array();
276 $this->mArgStack = false;
277 $this->mInPre = false;
278 $this->mLinkHolders = new LinkHolderArray( $this );
279 $this->mLinkID = 0;
280 $this->mRevisionObject = $this->mRevisionTimestamp =
281 $this->mRevisionId = $this->mRevisionUser = null;
282 $this->mVarCache = array();
283 $this->mUser = null;
284
285 /**
286 * Prefix for temporary replacement strings for the multipass parser.
287 * \x07 should never appear in input as it's disallowed in XML.
288 * Using it at the front also gives us a little extra robustness
289 * since it shouldn't match when butted up against identifier-like
290 * string constructs.
291 *
292 * Must not consist of all title characters, or else it will change
293 * the behaviour of <nowiki> in a link.
294 */
295 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
296 $this->mStripState = new StripState( $this->mUniqPrefix );
297
298
299 # Clear these on every parse, bug 4549
300 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
301
302 $this->mShowToc = true;
303 $this->mForceTocPosition = false;
304 $this->mIncludeSizes = array(
305 'post-expand' => 0,
306 'arg' => 0,
307 );
308 $this->mPPNodeCount = 0;
309 $this->mHighestExpansionDepth = 0;
310 $this->mDefaultSort = false;
311 $this->mHeadings = array();
312 $this->mDoubleUnderscores = array();
313 $this->mExpensiveFunctionCount = 0;
314
315 # Fix cloning
316 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
317 $this->mPreprocessor = null;
318 }
319
320 wfRunHooks( 'ParserClearState', array( &$this ) );
321 wfProfileOut( __METHOD__ );
322 }
323
324 /**
325 * Convert wikitext to HTML
326 * Do not call this function recursively.
327 *
328 * @param $text String: text we want to parse
329 * @param $title Title object
330 * @param $options ParserOptions
331 * @param $linestart boolean
332 * @param $clearState boolean
333 * @param $revid Int: number to pass in {{REVISIONID}}
334 * @return ParserOutput a ParserOutput
335 */
336 public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
337 /**
338 * First pass--just handle <nowiki> sections, pass the rest off
339 * to internalParse() which does all the real work.
340 */
341
342 global $wgUseTidy, $wgAlwaysUseTidy, $wgDisableLangConversion, $wgDisableTitleConversion;
343 $fname = __METHOD__.'-' . wfGetCaller();
344 wfProfileIn( __METHOD__ );
345 wfProfileIn( $fname );
346
347 $this->startParse( $title, $options, self::OT_HTML, $clearState );
348
349 # Remove the strip marker tag prefix from the input, if present.
350 if ( $clearState ) {
351 $text = str_replace( $this->mUniqPrefix, '', $text );
352 }
353
354 $oldRevisionId = $this->mRevisionId;
355 $oldRevisionObject = $this->mRevisionObject;
356 $oldRevisionTimestamp = $this->mRevisionTimestamp;
357 $oldRevisionUser = $this->mRevisionUser;
358 if ( $revid !== null ) {
359 $this->mRevisionId = $revid;
360 $this->mRevisionObject = null;
361 $this->mRevisionTimestamp = null;
362 $this->mRevisionUser = null;
363 }
364
365 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
366 # No more strip!
367 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
368 $text = $this->internalParse( $text );
369 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
370
371 $text = $this->mStripState->unstripGeneral( $text );
372
373 # Clean up special characters, only run once, next-to-last before doBlockLevels
374 $fixtags = array(
375 # french spaces, last one Guillemet-left
376 # only if there is something before the space
377 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
378 # french spaces, Guillemet-right
379 '/(\\302\\253) /' => '\\1&#160;',
380 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
381 );
382 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
383
384 $text = $this->doBlockLevels( $text, $linestart );
385
386 $this->replaceLinkHolders( $text );
387
388 /**
389 * The input doesn't get language converted if
390 * a) It's disabled
391 * b) Content isn't converted
392 * c) It's a conversion table
393 * d) it is an interface message (which is in the user language)
394 */
395 if ( !( $wgDisableLangConversion
396 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
397 || $this->mTitle->isConversionTable() ) )
398 {
399 # Run convert unconditionally in 1.18-compatible mode
400 global $wgBug34832TransitionalRollback;
401 if ( $wgBug34832TransitionalRollback || !$this->mOptions->getInterfaceMessage() ) {
402 # The position of the convert() call should not be changed. it
403 # assumes that the links are all replaced and the only thing left
404 # is the <nowiki> mark.
405 $text = $this->getConverterLanguage()->convert( $text );
406 }
407 }
408
409 /**
410 * A converted title will be provided in the output object if title and
411 * content conversion are enabled, the article text does not contain
412 * a conversion-suppressing double-underscore tag, and no
413 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
414 * automatic link conversion.
415 */
416 if ( !( $wgDisableLangConversion
417 || $wgDisableTitleConversion
418 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
419 || isset( $this->mDoubleUnderscores['notitleconvert'] )
420 || $this->mOutput->getDisplayTitle() !== false ) )
421 {
422 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
423 if ( $convruletitle ) {
424 $this->mOutput->setTitleText( $convruletitle );
425 } else {
426 $titleText = $this->getConverterLanguage()->convertTitle( $title );
427 $this->mOutput->setTitleText( $titleText );
428 }
429 }
430
431 $text = $this->mStripState->unstripNoWiki( $text );
432
433 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
434
435 $text = $this->replaceTransparentTags( $text );
436 $text = $this->mStripState->unstripGeneral( $text );
437
438 $text = Sanitizer::normalizeCharReferences( $text );
439
440 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
441 $text = MWTidy::tidy( $text );
442 } else {
443 # attempt to sanitize at least some nesting problems
444 # (bug #2702 and quite a few others)
445 $tidyregs = array(
446 # ''Something [http://www.cool.com cool''] -->
447 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
448 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
449 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
450 # fix up an anchor inside another anchor, only
451 # at least for a single single nested link (bug 3695)
452 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
453 '\\1\\2</a>\\3</a>\\1\\4</a>',
454 # fix div inside inline elements- doBlockLevels won't wrap a line which
455 # contains a div, so fix it up here; replace
456 # div with escaped text
457 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
458 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
459 # remove empty italic or bold tag pairs, some
460 # introduced by rules above
461 '/<([bi])><\/\\1>/' => '',
462 );
463
464 $text = preg_replace(
465 array_keys( $tidyregs ),
466 array_values( $tidyregs ),
467 $text );
468 }
469
470 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
471 $this->limitationWarn( 'expensive-parserfunction',
472 $this->mExpensiveFunctionCount,
473 $this->mOptions->getExpensiveParserFunctionLimit()
474 );
475 }
476
477 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
478
479 # Information on include size limits, for the benefit of users who try to skirt them
480 if ( $this->mOptions->getEnableLimitReport() ) {
481 $max = $this->mOptions->getMaxIncludeSize();
482 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/{$this->mOptions->getExpensiveParserFunctionLimit()}\n";
483 $limitReport =
484 "NewPP limit report\n" .
485 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
486 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
487 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
488 "Highest expansion depth: {$this->mHighestExpansionDepth}/{$this->mOptions->getMaxPPExpandDepth()}\n".
489 $PFreport;
490 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
491 $text .= "\n<!-- \n$limitReport-->\n";
492 }
493 $this->mOutput->setText( $text );
494
495 $this->mRevisionId = $oldRevisionId;
496 $this->mRevisionObject = $oldRevisionObject;
497 $this->mRevisionTimestamp = $oldRevisionTimestamp;
498 $this->mRevisionUser = $oldRevisionUser;
499 wfProfileOut( $fname );
500 wfProfileOut( __METHOD__ );
501
502 return $this->mOutput;
503 }
504
505 /**
506 * Recursive parser entry point that can be called from an extension tag
507 * hook.
508 *
509 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
510 *
511 * @param $text String: text extension wants to have parsed
512 * @param $frame PPFrame: The frame to use for expanding any template variables
513 *
514 * @return string
515 */
516 function recursiveTagParse( $text, $frame=false ) {
517 wfProfileIn( __METHOD__ );
518 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
519 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
520 $text = $this->internalParse( $text, false, $frame );
521 wfProfileOut( __METHOD__ );
522 return $text;
523 }
524
525 /**
526 * Expand templates and variables in the text, producing valid, static wikitext.
527 * Also removes comments.
528 * @return mixed|string
529 */
530 function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
531 wfProfileIn( __METHOD__ );
532 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
533 if ( $revid !== null ) {
534 $this->mRevisionId = $revid;
535 }
536 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
537 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
538 $text = $this->replaceVariables( $text );
539 $text = $this->mStripState->unstripBoth( $text );
540 wfProfileOut( __METHOD__ );
541 return $text;
542 }
543
544 /**
545 * Recursive parser entry point that can be called from an extension tag
546 * hook.
547 *
548 * @param $text String: text to be expanded
549 * @param $frame PPFrame: The frame to use for expanding any template variables
550 * @return String
551 * @since 1.19
552 */
553 public function recursivePreprocess( $text, $frame = false ) {
554 wfProfileIn( __METHOD__ );
555 $text = $this->replaceVariables( $text, $frame );
556 $text = $this->mStripState->unstripBoth( $text );
557 wfProfileOut( __METHOD__ );
558 return $text;
559 }
560
561 /**
562 * Process the wikitext for the "?preload=" feature. (bug 5210)
563 *
564 * "<noinclude>", "<includeonly>" etc. are parsed as for template
565 * transclusion, comments, templates, arguments, tags hooks and parser
566 * functions are untouched.
567 *
568 * @param $text String
569 * @param $title Title
570 * @param $options ParserOptions
571 * @return String
572 */
573 public function getPreloadText( $text, Title $title, ParserOptions $options ) {
574 # Parser (re)initialisation
575 $this->startParse( $title, $options, self::OT_PLAIN, true );
576
577 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
578 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
579 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
580 $text = $this->mStripState->unstripBoth( $text );
581 return $text;
582 }
583
584 /**
585 * Get a random string
586 *
587 * @return string
588 */
589 static public function getRandomString() {
590 return wfRandomString( 16 );
591 }
592
593 /**
594 * Set the current user.
595 * Should only be used when doing pre-save transform.
596 *
597 * @param $user Mixed: User object or null (to reset)
598 */
599 function setUser( $user ) {
600 $this->mUser = $user;
601 }
602
603 /**
604 * Accessor for mUniqPrefix.
605 *
606 * @return String
607 */
608 public function uniqPrefix() {
609 if ( !isset( $this->mUniqPrefix ) ) {
610 # @todo FIXME: This is probably *horribly wrong*
611 # LanguageConverter seems to want $wgParser's uniqPrefix, however
612 # if this is called for a parser cache hit, the parser may not
613 # have ever been initialized in the first place.
614 # Not really sure what the heck is supposed to be going on here.
615 return '';
616 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
617 }
618 return $this->mUniqPrefix;
619 }
620
621 /**
622 * Set the context title
623 *
624 * @param $t Title
625 */
626 function setTitle( $t ) {
627 if ( !$t || $t instanceof FakeTitle ) {
628 $t = Title::newFromText( 'NO TITLE' );
629 }
630
631 if ( strval( $t->getFragment() ) !== '' ) {
632 # Strip the fragment to avoid various odd effects
633 $this->mTitle = clone $t;
634 $this->mTitle->setFragment( '' );
635 } else {
636 $this->mTitle = $t;
637 }
638 }
639
640 /**
641 * Accessor for the Title object
642 *
643 * @return Title object
644 */
645 function getTitle() {
646 return $this->mTitle;
647 }
648
649 /**
650 * Accessor/mutator for the Title object
651 *
652 * @param $x Title object or null to just get the current one
653 * @return Title object
654 */
655 function Title( $x = null ) {
656 return wfSetVar( $this->mTitle, $x );
657 }
658
659 /**
660 * Set the output type
661 *
662 * @param $ot Integer: new value
663 */
664 function setOutputType( $ot ) {
665 $this->mOutputType = $ot;
666 # Shortcut alias
667 $this->ot = array(
668 'html' => $ot == self::OT_HTML,
669 'wiki' => $ot == self::OT_WIKI,
670 'pre' => $ot == self::OT_PREPROCESS,
671 'plain' => $ot == self::OT_PLAIN,
672 );
673 }
674
675 /**
676 * Accessor/mutator for the output type
677 *
678 * @param $x int|null New value or null to just get the current one
679 * @return Integer
680 */
681 function OutputType( $x = null ) {
682 return wfSetVar( $this->mOutputType, $x );
683 }
684
685 /**
686 * Get the ParserOutput object
687 *
688 * @return ParserOutput object
689 */
690 function getOutput() {
691 return $this->mOutput;
692 }
693
694 /**
695 * Get the ParserOptions object
696 *
697 * @return ParserOptions object
698 */
699 function getOptions() {
700 return $this->mOptions;
701 }
702
703 /**
704 * Accessor/mutator for the ParserOptions object
705 *
706 * @param $x ParserOptions New value or null to just get the current one
707 * @return ParserOptions Current ParserOptions object
708 */
709 function Options( $x = null ) {
710 return wfSetVar( $this->mOptions, $x );
711 }
712
713 /**
714 * @return int
715 */
716 function nextLinkID() {
717 return $this->mLinkID++;
718 }
719
720 /**
721 * @param $id int
722 */
723 function setLinkID( $id ) {
724 $this->mLinkID = $id;
725 }
726
727 /**
728 * Get a language object for use in parser functions such as {{FORMATNUM:}}
729 * @return Language
730 */
731 function getFunctionLang() {
732 return $this->getTargetLanguage();
733 }
734
735 /**
736 * Get the target language for the content being parsed. This is usually the
737 * language that the content is in.
738 *
739 * @since 1.19
740 *
741 * @return Language|null
742 */
743 public function getTargetLanguage() {
744 $target = $this->mOptions->getTargetLanguage();
745
746 if ( $target !== null ) {
747 return $target;
748 } elseif( $this->mOptions->getInterfaceMessage() ) {
749 return $this->mOptions->getUserLangObj();
750 } elseif( is_null( $this->mTitle ) ) {
751 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
752 }
753
754 return $this->mTitle->getPageLanguage();
755 }
756
757 /**
758 * Get the language object for language conversion
759 */
760 function getConverterLanguage() {
761 global $wgBug34832TransitionalRollback, $wgContLang;
762 if ( $wgBug34832TransitionalRollback ) {
763 return $wgContLang;
764 } else {
765 return $this->getTargetLanguage();
766 }
767 }
768
769 /**
770 * Get a User object either from $this->mUser, if set, or from the
771 * ParserOptions object otherwise
772 *
773 * @return User object
774 */
775 function getUser() {
776 if ( !is_null( $this->mUser ) ) {
777 return $this->mUser;
778 }
779 return $this->mOptions->getUser();
780 }
781
782 /**
783 * Get a preprocessor object
784 *
785 * @return Preprocessor instance
786 */
787 function getPreprocessor() {
788 if ( !isset( $this->mPreprocessor ) ) {
789 $class = $this->mPreprocessorClass;
790 $this->mPreprocessor = new $class( $this );
791 }
792 return $this->mPreprocessor;
793 }
794
795 /**
796 * Replaces all occurrences of HTML-style comments and the given tags
797 * in the text with a random marker and returns the next text. The output
798 * parameter $matches will be an associative array filled with data in
799 * the form:
800 *
801 * @code
802 * 'UNIQ-xxxxx' => array(
803 * 'element',
804 * 'tag content',
805 * array( 'param' => 'x' ),
806 * '<element param="x">tag content</element>' ) )
807 * @endcode
808 *
809 * @param $elements array list of element names. Comments are always extracted.
810 * @param $text string Source text string.
811 * @param $matches array Out parameter, Array: extracted tags
812 * @param $uniq_prefix string
813 * @return String: stripped text
814 */
815 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
816 static $n = 1;
817 $stripped = '';
818 $matches = array();
819
820 $taglist = implode( '|', $elements );
821 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
822
823 while ( $text != '' ) {
824 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
825 $stripped .= $p[0];
826 if ( count( $p ) < 5 ) {
827 break;
828 }
829 if ( count( $p ) > 5 ) {
830 # comment
831 $element = $p[4];
832 $attributes = '';
833 $close = '';
834 $inside = $p[5];
835 } else {
836 # tag
837 $element = $p[1];
838 $attributes = $p[2];
839 $close = $p[3];
840 $inside = $p[4];
841 }
842
843 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
844 $stripped .= $marker;
845
846 if ( $close === '/>' ) {
847 # Empty element tag, <tag />
848 $content = null;
849 $text = $inside;
850 $tail = null;
851 } else {
852 if ( $element === '!--' ) {
853 $end = '/(-->)/';
854 } else {
855 $end = "/(<\\/$element\\s*>)/i";
856 }
857 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
858 $content = $q[0];
859 if ( count( $q ) < 3 ) {
860 # No end tag -- let it run out to the end of the text.
861 $tail = '';
862 $text = '';
863 } else {
864 $tail = $q[1];
865 $text = $q[2];
866 }
867 }
868
869 $matches[$marker] = array( $element,
870 $content,
871 Sanitizer::decodeTagAttributes( $attributes ),
872 "<$element$attributes$close$content$tail" );
873 }
874 return $stripped;
875 }
876
877 /**
878 * Get a list of strippable XML-like elements
879 *
880 * @return array
881 */
882 function getStripList() {
883 return $this->mStripList;
884 }
885
886 /**
887 * Add an item to the strip state
888 * Returns the unique tag which must be inserted into the stripped text
889 * The tag will be replaced with the original text in unstrip()
890 *
891 * @param $text string
892 *
893 * @return string
894 */
895 function insertStripItem( $text ) {
896 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
897 $this->mMarkerIndex++;
898 $this->mStripState->addGeneral( $rnd, $text );
899 return $rnd;
900 }
901
902 /**
903 * parse the wiki syntax used to render tables
904 *
905 * @private
906 * @return string
907 */
908 function doTableStuff( $text ) {
909 wfProfileIn( __METHOD__ );
910
911 $lines = StringUtils::explode( "\n", $text );
912 $out = '';
913 $td_history = array(); # Is currently a td tag open?
914 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
915 $tr_history = array(); # Is currently a tr tag open?
916 $tr_attributes = array(); # history of tr attributes
917 $has_opened_tr = array(); # Did this table open a <tr> element?
918 $indent_level = 0; # indent level of the table
919
920 foreach ( $lines as $outLine ) {
921 $line = trim( $outLine );
922
923 if ( $line === '' ) { # empty line, go to next line
924 $out .= $outLine."\n";
925 continue;
926 }
927
928 $first_character = $line[0];
929 $matches = array();
930
931 if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
932 # First check if we are starting a new table
933 $indent_level = strlen( $matches[1] );
934
935 $attributes = $this->mStripState->unstripBoth( $matches[2] );
936 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
937
938 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
939 array_push( $td_history , false );
940 array_push( $last_tag_history , '' );
941 array_push( $tr_history , false );
942 array_push( $tr_attributes , '' );
943 array_push( $has_opened_tr , false );
944 } elseif ( count( $td_history ) == 0 ) {
945 # Don't do any of the following
946 $out .= $outLine."\n";
947 continue;
948 } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
949 # We are ending a table
950 $line = '</table>' . substr( $line , 2 );
951 $last_tag = array_pop( $last_tag_history );
952
953 if ( !array_pop( $has_opened_tr ) ) {
954 $line = "<tr><td></td></tr>{$line}";
955 }
956
957 if ( array_pop( $tr_history ) ) {
958 $line = "</tr>{$line}";
959 }
960
961 if ( array_pop( $td_history ) ) {
962 $line = "</{$last_tag}>{$line}";
963 }
964 array_pop( $tr_attributes );
965 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
966 } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
967 # Now we have a table row
968 $line = preg_replace( '#^\|-+#', '', $line );
969
970 # Whats after the tag is now only attributes
971 $attributes = $this->mStripState->unstripBoth( $line );
972 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
973 array_pop( $tr_attributes );
974 array_push( $tr_attributes, $attributes );
975
976 $line = '';
977 $last_tag = array_pop( $last_tag_history );
978 array_pop( $has_opened_tr );
979 array_push( $has_opened_tr , true );
980
981 if ( array_pop( $tr_history ) ) {
982 $line = '</tr>';
983 }
984
985 if ( array_pop( $td_history ) ) {
986 $line = "</{$last_tag}>{$line}";
987 }
988
989 $outLine = $line;
990 array_push( $tr_history , false );
991 array_push( $td_history , false );
992 array_push( $last_tag_history , '' );
993 } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 ) === '|+' ) {
994 # This might be cell elements, td, th or captions
995 if ( substr( $line , 0 , 2 ) === '|+' ) {
996 $first_character = '+';
997 $line = substr( $line , 1 );
998 }
999
1000 $line = substr( $line , 1 );
1001
1002 if ( $first_character === '!' ) {
1003 $line = str_replace( '!!' , '||' , $line );
1004 }
1005
1006 # Split up multiple cells on the same line.
1007 # FIXME : This can result in improper nesting of tags processed
1008 # by earlier parser steps, but should avoid splitting up eg
1009 # attribute values containing literal "||".
1010 $cells = StringUtils::explodeMarkup( '||' , $line );
1011
1012 $outLine = '';
1013
1014 # Loop through each table cell
1015 foreach ( $cells as $cell ) {
1016 $previous = '';
1017 if ( $first_character !== '+' ) {
1018 $tr_after = array_pop( $tr_attributes );
1019 if ( !array_pop( $tr_history ) ) {
1020 $previous = "<tr{$tr_after}>\n";
1021 }
1022 array_push( $tr_history , true );
1023 array_push( $tr_attributes , '' );
1024 array_pop( $has_opened_tr );
1025 array_push( $has_opened_tr , true );
1026 }
1027
1028 $last_tag = array_pop( $last_tag_history );
1029
1030 if ( array_pop( $td_history ) ) {
1031 $previous = "</{$last_tag}>\n{$previous}";
1032 }
1033
1034 if ( $first_character === '|' ) {
1035 $last_tag = 'td';
1036 } elseif ( $first_character === '!' ) {
1037 $last_tag = 'th';
1038 } elseif ( $first_character === '+' ) {
1039 $last_tag = 'caption';
1040 } else {
1041 $last_tag = '';
1042 }
1043
1044 array_push( $last_tag_history , $last_tag );
1045
1046 # A cell could contain both parameters and data
1047 $cell_data = explode( '|' , $cell , 2 );
1048
1049 # Bug 553: Note that a '|' inside an invalid link should not
1050 # be mistaken as delimiting cell parameters
1051 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1052 $cell = "{$previous}<{$last_tag}>{$cell}";
1053 } elseif ( count( $cell_data ) == 1 ) {
1054 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1055 } else {
1056 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1057 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
1058 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1059 }
1060
1061 $outLine .= $cell;
1062 array_push( $td_history , true );
1063 }
1064 }
1065 $out .= $outLine . "\n";
1066 }
1067
1068 # Closing open td, tr && table
1069 while ( count( $td_history ) > 0 ) {
1070 if ( array_pop( $td_history ) ) {
1071 $out .= "</td>\n";
1072 }
1073 if ( array_pop( $tr_history ) ) {
1074 $out .= "</tr>\n";
1075 }
1076 if ( !array_pop( $has_opened_tr ) ) {
1077 $out .= "<tr><td></td></tr>\n" ;
1078 }
1079
1080 $out .= "</table>\n";
1081 }
1082
1083 # Remove trailing line-ending (b/c)
1084 if ( substr( $out, -1 ) === "\n" ) {
1085 $out = substr( $out, 0, -1 );
1086 }
1087
1088 # special case: don't return empty table
1089 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1090 $out = '';
1091 }
1092
1093 wfProfileOut( __METHOD__ );
1094
1095 return $out;
1096 }
1097
1098 /**
1099 * Helper function for parse() that transforms wiki markup into
1100 * HTML. Only called for $mOutputType == self::OT_HTML.
1101 *
1102 * @private
1103 *
1104 * @param $text string
1105 * @param $isMain bool
1106 * @param $frame bool
1107 *
1108 * @return string
1109 */
1110 function internalParse( $text, $isMain = true, $frame = false ) {
1111 wfProfileIn( __METHOD__ );
1112
1113 $origText = $text;
1114
1115 # Hook to suspend the parser in this state
1116 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1117 wfProfileOut( __METHOD__ );
1118 return $text ;
1119 }
1120
1121 # if $frame is provided, then use $frame for replacing any variables
1122 if ( $frame ) {
1123 # use frame depth to infer how include/noinclude tags should be handled
1124 # depth=0 means this is the top-level document; otherwise it's an included document
1125 if ( !$frame->depth ) {
1126 $flag = 0;
1127 } else {
1128 $flag = Parser::PTD_FOR_INCLUSION;
1129 }
1130 $dom = $this->preprocessToDom( $text, $flag );
1131 $text = $frame->expand( $dom );
1132 } else {
1133 # if $frame is not provided, then use old-style replaceVariables
1134 $text = $this->replaceVariables( $text );
1135 }
1136
1137 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1138 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1139 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1140
1141 # Tables need to come after variable replacement for things to work
1142 # properly; putting them before other transformations should keep
1143 # exciting things like link expansions from showing up in surprising
1144 # places.
1145 $text = $this->doTableStuff( $text );
1146
1147 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1148
1149 $text = $this->doDoubleUnderscore( $text );
1150
1151 $text = $this->doHeadings( $text );
1152 if ( $this->mOptions->getUseDynamicDates() ) {
1153 $df = DateFormatter::getInstance();
1154 $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1155 }
1156 $text = $this->replaceInternalLinks( $text );
1157 $text = $this->doAllQuotes( $text );
1158 $text = $this->replaceExternalLinks( $text );
1159
1160 # replaceInternalLinks may sometimes leave behind
1161 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1162 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1163
1164 $text = $this->doMagicLinks( $text );
1165 $text = $this->formatHeadings( $text, $origText, $isMain );
1166
1167 wfProfileOut( __METHOD__ );
1168 return $text;
1169 }
1170
1171 /**
1172 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1173 * magic external links.
1174 *
1175 * DML
1176 * @private
1177 *
1178 * @param $text string
1179 *
1180 * @return string
1181 */
1182 function doMagicLinks( $text ) {
1183 wfProfileIn( __METHOD__ );
1184 $prots = wfUrlProtocolsWithoutProtRel();
1185 $urlChar = self::EXT_LINK_URL_CLASS;
1186 $text = preg_replace_callback(
1187 '!(?: # Start cases
1188 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1189 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1190 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1191 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1192 ISBN\s+(\b # m[5]: ISBN, capture number
1193 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1194 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1195 [0-9Xx] # check digit
1196 \b)
1197 )!xu', array( &$this, 'magicLinkCallback' ), $text );
1198 wfProfileOut( __METHOD__ );
1199 return $text;
1200 }
1201
1202 /**
1203 * @throws MWException
1204 * @param $m array
1205 * @return HTML|string
1206 */
1207 function magicLinkCallback( $m ) {
1208 if ( isset( $m[1] ) && $m[1] !== '' ) {
1209 # Skip anchor
1210 return $m[0];
1211 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1212 # Skip HTML element
1213 return $m[0];
1214 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1215 # Free external link
1216 return $this->makeFreeExternalLink( $m[0] );
1217 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1218 # RFC or PMID
1219 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1220 $keyword = 'RFC';
1221 $urlmsg = 'rfcurl';
1222 $CssClass = 'mw-magiclink-rfc';
1223 $id = $m[4];
1224 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1225 $keyword = 'PMID';
1226 $urlmsg = 'pubmedurl';
1227 $CssClass = 'mw-magiclink-pmid';
1228 $id = $m[4];
1229 } else {
1230 throw new MWException( __METHOD__.': unrecognised match type "' .
1231 substr( $m[0], 0, 20 ) . '"' );
1232 }
1233 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1234 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1235 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1236 # ISBN
1237 $isbn = $m[5];
1238 $num = strtr( $isbn, array(
1239 '-' => '',
1240 ' ' => '',
1241 'x' => 'X',
1242 ));
1243 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1244 return'<a href="' .
1245 htmlspecialchars( $titleObj->getLocalUrl() ) .
1246 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1247 } else {
1248 return $m[0];
1249 }
1250 }
1251
1252 /**
1253 * Make a free external link, given a user-supplied URL
1254 *
1255 * @param $url string
1256 *
1257 * @return string HTML
1258 * @private
1259 */
1260 function makeFreeExternalLink( $url ) {
1261 wfProfileIn( __METHOD__ );
1262
1263 $trail = '';
1264
1265 # The characters '<' and '>' (which were escaped by
1266 # removeHTMLtags()) should not be included in
1267 # URLs, per RFC 2396.
1268 $m2 = array();
1269 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1270 $trail = substr( $url, $m2[0][1] ) . $trail;
1271 $url = substr( $url, 0, $m2[0][1] );
1272 }
1273
1274 # Move trailing punctuation to $trail
1275 $sep = ',;\.:!?';
1276 # If there is no left bracket, then consider right brackets fair game too
1277 if ( strpos( $url, '(' ) === false ) {
1278 $sep .= ')';
1279 }
1280
1281 $numSepChars = strspn( strrev( $url ), $sep );
1282 if ( $numSepChars ) {
1283 $trail = substr( $url, -$numSepChars ) . $trail;
1284 $url = substr( $url, 0, -$numSepChars );
1285 }
1286
1287 $url = Sanitizer::cleanUrl( $url );
1288
1289 # Is this an external image?
1290 $text = $this->maybeMakeExternalImage( $url );
1291 if ( $text === false ) {
1292 # Not an image, make a link
1293 $text = Linker::makeExternalLink( $url,
1294 $this->getConverterLanguage()->markNoConversion($url), true, 'free',
1295 $this->getExternalLinkAttribs( $url ) );
1296 # Register it in the output object...
1297 # Replace unnecessary URL escape codes with their equivalent characters
1298 $pasteurized = self::replaceUnusualEscapes( $url );
1299 $this->mOutput->addExternalLink( $pasteurized );
1300 }
1301 wfProfileOut( __METHOD__ );
1302 return $text . $trail;
1303 }
1304
1305
1306 /**
1307 * Parse headers and return html
1308 *
1309 * @private
1310 *
1311 * @param $text string
1312 *
1313 * @return string
1314 */
1315 function doHeadings( $text ) {
1316 wfProfileIn( __METHOD__ );
1317 for ( $i = 6; $i >= 1; --$i ) {
1318 $h = str_repeat( '=', $i );
1319 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1320 "<h$i>\\1</h$i>", $text );
1321 }
1322 wfProfileOut( __METHOD__ );
1323 return $text;
1324 }
1325
1326 /**
1327 * Replace single quotes with HTML markup
1328 * @private
1329 *
1330 * @param $text string
1331 *
1332 * @return string the altered text
1333 */
1334 function doAllQuotes( $text ) {
1335 wfProfileIn( __METHOD__ );
1336 $outtext = '';
1337 $lines = StringUtils::explode( "\n", $text );
1338 foreach ( $lines as $line ) {
1339 $outtext .= $this->doQuotes( $line ) . "\n";
1340 }
1341 $outtext = substr( $outtext, 0,-1 );
1342 wfProfileOut( __METHOD__ );
1343 return $outtext;
1344 }
1345
1346 /**
1347 * Helper function for doAllQuotes()
1348 *
1349 * @param $text string
1350 *
1351 * @return string
1352 */
1353 public function doQuotes( $text ) {
1354 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1355 if ( count( $arr ) == 1 ) {
1356 return $text;
1357 } else {
1358 # First, do some preliminary work. This may shift some apostrophes from
1359 # being mark-up to being text. It also counts the number of occurrences
1360 # of bold and italics mark-ups.
1361 $numbold = 0;
1362 $numitalics = 0;
1363 for ( $i = 0; $i < count( $arr ); $i++ ) {
1364 if ( ( $i % 2 ) == 1 ) {
1365 # If there are ever four apostrophes, assume the first is supposed to
1366 # be text, and the remaining three constitute mark-up for bold text.
1367 if ( strlen( $arr[$i] ) == 4 ) {
1368 $arr[$i-1] .= "'";
1369 $arr[$i] = "'''";
1370 } elseif ( strlen( $arr[$i] ) > 5 ) {
1371 # If there are more than 5 apostrophes in a row, assume they're all
1372 # text except for the last 5.
1373 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1374 $arr[$i] = "'''''";
1375 }
1376 # Count the number of occurrences of bold and italics mark-ups.
1377 # We are not counting sequences of five apostrophes.
1378 if ( strlen( $arr[$i] ) == 2 ) {
1379 $numitalics++;
1380 } elseif ( strlen( $arr[$i] ) == 3 ) {
1381 $numbold++;
1382 } elseif ( strlen( $arr[$i] ) == 5 ) {
1383 $numitalics++;
1384 $numbold++;
1385 }
1386 }
1387 }
1388
1389 # If there is an odd number of both bold and italics, it is likely
1390 # that one of the bold ones was meant to be an apostrophe followed
1391 # by italics. Which one we cannot know for certain, but it is more
1392 # likely to be one that has a single-letter word before it.
1393 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1394 $i = 0;
1395 $firstsingleletterword = -1;
1396 $firstmultiletterword = -1;
1397 $firstspace = -1;
1398 foreach ( $arr as $r ) {
1399 if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1400 $x1 = substr( $arr[$i-1], -1 );
1401 $x2 = substr( $arr[$i-1], -2, 1 );
1402 if ( $x1 === ' ' ) {
1403 if ( $firstspace == -1 ) {
1404 $firstspace = $i;
1405 }
1406 } elseif ( $x2 === ' ') {
1407 if ( $firstsingleletterword == -1 ) {
1408 $firstsingleletterword = $i;
1409 }
1410 } else {
1411 if ( $firstmultiletterword == -1 ) {
1412 $firstmultiletterword = $i;
1413 }
1414 }
1415 }
1416 $i++;
1417 }
1418
1419 # If there is a single-letter word, use it!
1420 if ( $firstsingleletterword > -1 ) {
1421 $arr[$firstsingleletterword] = "''";
1422 $arr[$firstsingleletterword-1] .= "'";
1423 } elseif ( $firstmultiletterword > -1 ) {
1424 # If not, but there's a multi-letter word, use that one.
1425 $arr[$firstmultiletterword] = "''";
1426 $arr[$firstmultiletterword-1] .= "'";
1427 } elseif ( $firstspace > -1 ) {
1428 # ... otherwise use the first one that has neither.
1429 # (notice that it is possible for all three to be -1 if, for example,
1430 # there is only one pentuple-apostrophe in the line)
1431 $arr[$firstspace] = "''";
1432 $arr[$firstspace-1] .= "'";
1433 }
1434 }
1435
1436 # Now let's actually convert our apostrophic mush to HTML!
1437 $output = '';
1438 $buffer = '';
1439 $state = '';
1440 $i = 0;
1441 foreach ( $arr as $r ) {
1442 if ( ( $i % 2 ) == 0 ) {
1443 if ( $state === 'both' ) {
1444 $buffer .= $r;
1445 } else {
1446 $output .= $r;
1447 }
1448 } else {
1449 if ( strlen( $r ) == 2 ) {
1450 if ( $state === 'i' ) {
1451 $output .= '</i>'; $state = '';
1452 } elseif ( $state === 'bi' ) {
1453 $output .= '</i>'; $state = 'b';
1454 } elseif ( $state === 'ib' ) {
1455 $output .= '</b></i><b>'; $state = 'b';
1456 } elseif ( $state === 'both' ) {
1457 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1458 } else { # $state can be 'b' or ''
1459 $output .= '<i>'; $state .= 'i';
1460 }
1461 } elseif ( strlen( $r ) == 3 ) {
1462 if ( $state === 'b' ) {
1463 $output .= '</b>'; $state = '';
1464 } elseif ( $state === 'bi' ) {
1465 $output .= '</i></b><i>'; $state = 'i';
1466 } elseif ( $state === 'ib' ) {
1467 $output .= '</b>'; $state = 'i';
1468 } elseif ( $state === 'both' ) {
1469 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1470 } else { # $state can be 'i' or ''
1471 $output .= '<b>'; $state .= 'b';
1472 }
1473 } elseif ( strlen( $r ) == 5 ) {
1474 if ( $state === 'b' ) {
1475 $output .= '</b><i>'; $state = 'i';
1476 } elseif ( $state === 'i' ) {
1477 $output .= '</i><b>'; $state = 'b';
1478 } elseif ( $state === 'bi' ) {
1479 $output .= '</i></b>'; $state = '';
1480 } elseif ( $state === 'ib' ) {
1481 $output .= '</b></i>'; $state = '';
1482 } elseif ( $state === 'both' ) {
1483 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1484 } else { # ($state == '')
1485 $buffer = ''; $state = 'both';
1486 }
1487 }
1488 }
1489 $i++;
1490 }
1491 # Now close all remaining tags. Notice that the order is important.
1492 if ( $state === 'b' || $state === 'ib' ) {
1493 $output .= '</b>';
1494 }
1495 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1496 $output .= '</i>';
1497 }
1498 if ( $state === 'bi' ) {
1499 $output .= '</b>';
1500 }
1501 # There might be lonely ''''', so make sure we have a buffer
1502 if ( $state === 'both' && $buffer ) {
1503 $output .= '<b><i>'.$buffer.'</i></b>';
1504 }
1505 return $output;
1506 }
1507 }
1508
1509 /**
1510 * Replace external links (REL)
1511 *
1512 * Note: this is all very hackish and the order of execution matters a lot.
1513 * Make sure to run maintenance/parserTests.php if you change this code.
1514 *
1515 * @private
1516 *
1517 * @param $text string
1518 *
1519 * @return string
1520 */
1521 function replaceExternalLinks( $text ) {
1522 wfProfileIn( __METHOD__ );
1523
1524 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1525 $s = array_shift( $bits );
1526
1527 $i = 0;
1528 while ( $i<count( $bits ) ) {
1529 $url = $bits[$i++];
1530 $protocol = $bits[$i++];
1531 $text = $bits[$i++];
1532 $trail = $bits[$i++];
1533
1534 # The characters '<' and '>' (which were escaped by
1535 # removeHTMLtags()) should not be included in
1536 # URLs, per RFC 2396.
1537 $m2 = array();
1538 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1539 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1540 $url = substr( $url, 0, $m2[0][1] );
1541 }
1542
1543 # If the link text is an image URL, replace it with an <img> tag
1544 # This happened by accident in the original parser, but some people used it extensively
1545 $img = $this->maybeMakeExternalImage( $text );
1546 if ( $img !== false ) {
1547 $text = $img;
1548 }
1549
1550 $dtrail = '';
1551
1552 # Set linktype for CSS - if URL==text, link is essentially free
1553 $linktype = ( $text === $url ) ? 'free' : 'text';
1554
1555 # No link text, e.g. [http://domain.tld/some.link]
1556 if ( $text == '' ) {
1557 # Autonumber
1558 $langObj = $this->getTargetLanguage();
1559 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1560 $linktype = 'autonumber';
1561 } else {
1562 # Have link text, e.g. [http://domain.tld/some.link text]s
1563 # Check for trail
1564 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1565 }
1566
1567 $text = $this->getConverterLanguage()->markNoConversion( $text );
1568
1569 $url = Sanitizer::cleanUrl( $url );
1570
1571 # Use the encoded URL
1572 # This means that users can paste URLs directly into the text
1573 # Funny characters like ö aren't valid in URLs anyway
1574 # This was changed in August 2004
1575 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1576 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1577
1578 # Register link in the output object.
1579 # Replace unnecessary URL escape codes with the referenced character
1580 # This prevents spammers from hiding links from the filters
1581 $pasteurized = self::replaceUnusualEscapes( $url );
1582 $this->mOutput->addExternalLink( $pasteurized );
1583 }
1584
1585 wfProfileOut( __METHOD__ );
1586 return $s;
1587 }
1588
1589 /**
1590 * Get an associative array of additional HTML attributes appropriate for a
1591 * particular external link. This currently may include rel => nofollow
1592 * (depending on configuration, namespace, and the URL's domain) and/or a
1593 * target attribute (depending on configuration).
1594 *
1595 * @param $url String|bool optional URL, to extract the domain from for rel =>
1596 * nofollow if appropriate
1597 * @return Array associative array of HTML attributes
1598 */
1599 function getExternalLinkAttribs( $url = false ) {
1600 $attribs = array();
1601 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1602 $ns = $this->mTitle->getNamespace();
1603 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1604 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1605 {
1606 $attribs['rel'] = 'nofollow';
1607 }
1608 if ( $this->mOptions->getExternalLinkTarget() ) {
1609 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1610 }
1611 return $attribs;
1612 }
1613
1614 /**
1615 * Replace unusual URL escape codes with their equivalent characters
1616 *
1617 * @param $url String
1618 * @return String
1619 *
1620 * @todo This can merge genuinely required bits in the path or query string,
1621 * breaking legit URLs. A proper fix would treat the various parts of
1622 * the URL differently; as a workaround, just use the output for
1623 * statistical records, not for actual linking/output.
1624 */
1625 static function replaceUnusualEscapes( $url ) {
1626 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1627 array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1628 }
1629
1630 /**
1631 * Callback function used in replaceUnusualEscapes().
1632 * Replaces unusual URL escape codes with their equivalent character
1633 *
1634 * @param $matches array
1635 *
1636 * @return string
1637 */
1638 private static function replaceUnusualEscapesCallback( $matches ) {
1639 $char = urldecode( $matches[0] );
1640 $ord = ord( $char );
1641 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1642 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1643 # No, shouldn't be escaped
1644 return $char;
1645 } else {
1646 # Yes, leave it escaped
1647 return $matches[0];
1648 }
1649 }
1650
1651 /**
1652 * make an image if it's allowed, either through the global
1653 * option, through the exception, or through the on-wiki whitelist
1654 * @private
1655 *
1656 * $param $url string
1657 *
1658 * @return string
1659 */
1660 function maybeMakeExternalImage( $url ) {
1661 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1662 $imagesexception = !empty( $imagesfrom );
1663 $text = false;
1664 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1665 if ( $imagesexception && is_array( $imagesfrom ) ) {
1666 $imagematch = false;
1667 foreach ( $imagesfrom as $match ) {
1668 if ( strpos( $url, $match ) === 0 ) {
1669 $imagematch = true;
1670 break;
1671 }
1672 }
1673 } elseif ( $imagesexception ) {
1674 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1675 } else {
1676 $imagematch = false;
1677 }
1678 if ( $this->mOptions->getAllowExternalImages()
1679 || ( $imagesexception && $imagematch ) ) {
1680 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1681 # Image found
1682 $text = Linker::makeExternalImage( $url );
1683 }
1684 }
1685 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1686 && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1687 $whitelist = explode( "\n", wfMessage( 'external_image_whitelist' )->inContentLanguage()->text() );
1688 foreach ( $whitelist as $entry ) {
1689 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1690 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1691 continue;
1692 }
1693 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1694 # Image matches a whitelist entry
1695 $text = Linker::makeExternalImage( $url );
1696 break;
1697 }
1698 }
1699 }
1700 return $text;
1701 }
1702
1703 /**
1704 * Process [[ ]] wikilinks
1705 *
1706 * @param $s string
1707 *
1708 * @return String: processed text
1709 *
1710 * @private
1711 */
1712 function replaceInternalLinks( $s ) {
1713 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1714 return $s;
1715 }
1716
1717 /**
1718 * Process [[ ]] wikilinks (RIL)
1719 * @return LinkHolderArray
1720 *
1721 * @private
1722 */
1723 function replaceInternalLinks2( &$s ) {
1724 wfProfileIn( __METHOD__ );
1725
1726 wfProfileIn( __METHOD__.'-setup' );
1727 static $tc = FALSE, $e1, $e1_img;
1728 # the % is needed to support urlencoded titles as well
1729 if ( !$tc ) {
1730 $tc = Title::legalChars() . '#%';
1731 # Match a link having the form [[namespace:link|alternate]]trail
1732 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1733 # Match cases where there is no "]]", which might still be images
1734 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1735 }
1736
1737 $holders = new LinkHolderArray( $this );
1738
1739 # split the entire text string on occurrences of [[
1740 $a = StringUtils::explode( '[[', ' ' . $s );
1741 # get the first element (all text up to first [[), and remove the space we added
1742 $s = $a->current();
1743 $a->next();
1744 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1745 $s = substr( $s, 1 );
1746
1747 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1748 $e2 = null;
1749 if ( $useLinkPrefixExtension ) {
1750 # Match the end of a line for a word that's not followed by whitespace,
1751 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1752 $e2 = wfMessage( 'linkprefix' )->inContentLanguage()->text();
1753 }
1754
1755 if ( is_null( $this->mTitle ) ) {
1756 wfProfileOut( __METHOD__.'-setup' );
1757 wfProfileOut( __METHOD__ );
1758 throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1759 }
1760 $nottalk = !$this->mTitle->isTalkPage();
1761
1762 if ( $useLinkPrefixExtension ) {
1763 $m = array();
1764 if ( preg_match( $e2, $s, $m ) ) {
1765 $first_prefix = $m[2];
1766 } else {
1767 $first_prefix = false;
1768 }
1769 } else {
1770 $prefix = '';
1771 }
1772
1773 if ( $this->getConverterLanguage()->hasVariants() ) {
1774 $selflink = $this->getConverterLanguage()->autoConvertToAllVariants(
1775 $this->mTitle->getPrefixedText() );
1776 } else {
1777 $selflink = array( $this->mTitle->getPrefixedText() );
1778 }
1779 $useSubpages = $this->areSubpagesAllowed();
1780 wfProfileOut( __METHOD__.'-setup' );
1781
1782 # Loop for each link
1783 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1784 # Check for excessive memory usage
1785 if ( $holders->isBig() ) {
1786 # Too big
1787 # Do the existence check, replace the link holders and clear the array
1788 $holders->replace( $s );
1789 $holders->clear();
1790 }
1791
1792 if ( $useLinkPrefixExtension ) {
1793 wfProfileIn( __METHOD__.'-prefixhandling' );
1794 if ( preg_match( $e2, $s, $m ) ) {
1795 $prefix = $m[2];
1796 $s = $m[1];
1797 } else {
1798 $prefix='';
1799 }
1800 # first link
1801 if ( $first_prefix ) {
1802 $prefix = $first_prefix;
1803 $first_prefix = false;
1804 }
1805 wfProfileOut( __METHOD__.'-prefixhandling' );
1806 }
1807
1808 $might_be_img = false;
1809
1810 wfProfileIn( __METHOD__."-e1" );
1811 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1812 $text = $m[2];
1813 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1814 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1815 # the real problem is with the $e1 regex
1816 # See bug 1300.
1817 #
1818 # Still some problems for cases where the ] is meant to be outside punctuation,
1819 # and no image is in sight. See bug 2095.
1820 #
1821 if ( $text !== '' &&
1822 substr( $m[3], 0, 1 ) === ']' &&
1823 strpos( $text, '[' ) !== false
1824 )
1825 {
1826 $text .= ']'; # so that replaceExternalLinks($text) works later
1827 $m[3] = substr( $m[3], 1 );
1828 }
1829 # fix up urlencoded title texts
1830 if ( strpos( $m[1], '%' ) !== false ) {
1831 # Should anchors '#' also be rejected?
1832 $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1833 }
1834 $trail = $m[3];
1835 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1836 $might_be_img = true;
1837 $text = $m[2];
1838 if ( strpos( $m[1], '%' ) !== false ) {
1839 $m[1] = rawurldecode( $m[1] );
1840 }
1841 $trail = "";
1842 } else { # Invalid form; output directly
1843 $s .= $prefix . '[[' . $line ;
1844 wfProfileOut( __METHOD__."-e1" );
1845 continue;
1846 }
1847 wfProfileOut( __METHOD__."-e1" );
1848 wfProfileIn( __METHOD__."-misc" );
1849
1850 # Don't allow internal links to pages containing
1851 # PROTO: where PROTO is a valid URL protocol; these
1852 # should be external links.
1853 if ( preg_match( '/^(?:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
1854 $s .= $prefix . '[[' . $line ;
1855 wfProfileOut( __METHOD__."-misc" );
1856 continue;
1857 }
1858
1859 # Make subpage if necessary
1860 if ( $useSubpages ) {
1861 $link = $this->maybeDoSubpageLink( $m[1], $text );
1862 } else {
1863 $link = $m[1];
1864 }
1865
1866 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1867 if ( !$noforce ) {
1868 # Strip off leading ':'
1869 $link = substr( $link, 1 );
1870 }
1871
1872 wfProfileOut( __METHOD__."-misc" );
1873 wfProfileIn( __METHOD__."-title" );
1874 $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1875 if ( $nt === null ) {
1876 $s .= $prefix . '[[' . $line;
1877 wfProfileOut( __METHOD__."-title" );
1878 continue;
1879 }
1880
1881 $ns = $nt->getNamespace();
1882 $iw = $nt->getInterWiki();
1883 wfProfileOut( __METHOD__."-title" );
1884
1885 if ( $might_be_img ) { # if this is actually an invalid link
1886 wfProfileIn( __METHOD__."-might_be_img" );
1887 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1888 $found = false;
1889 while ( true ) {
1890 # look at the next 'line' to see if we can close it there
1891 $a->next();
1892 $next_line = $a->current();
1893 if ( $next_line === false || $next_line === null ) {
1894 break;
1895 }
1896 $m = explode( ']]', $next_line, 3 );
1897 if ( count( $m ) == 3 ) {
1898 # the first ]] closes the inner link, the second the image
1899 $found = true;
1900 $text .= "[[{$m[0]}]]{$m[1]}";
1901 $trail = $m[2];
1902 break;
1903 } elseif ( count( $m ) == 2 ) {
1904 # if there's exactly one ]] that's fine, we'll keep looking
1905 $text .= "[[{$m[0]}]]{$m[1]}";
1906 } else {
1907 # if $next_line is invalid too, we need look no further
1908 $text .= '[[' . $next_line;
1909 break;
1910 }
1911 }
1912 if ( !$found ) {
1913 # we couldn't find the end of this imageLink, so output it raw
1914 # but don't ignore what might be perfectly normal links in the text we've examined
1915 $holders->merge( $this->replaceInternalLinks2( $text ) );
1916 $s .= "{$prefix}[[$link|$text";
1917 # note: no $trail, because without an end, there *is* no trail
1918 wfProfileOut( __METHOD__."-might_be_img" );
1919 continue;
1920 }
1921 } else { # it's not an image, so output it raw
1922 $s .= "{$prefix}[[$link|$text";
1923 # note: no $trail, because without an end, there *is* no trail
1924 wfProfileOut( __METHOD__."-might_be_img" );
1925 continue;
1926 }
1927 wfProfileOut( __METHOD__."-might_be_img" );
1928 }
1929
1930 $wasblank = ( $text == '' );
1931 if ( $wasblank ) {
1932 $text = $link;
1933 } else {
1934 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1935 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1936 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1937 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1938 $text = $this->doQuotes( $text );
1939 }
1940
1941 # Link not escaped by : , create the various objects
1942 if ( $noforce ) {
1943 # Interwikis
1944 wfProfileIn( __METHOD__."-interwiki" );
1945 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' ) ) {
1946 $this->mOutput->addLanguageLink( $nt->getFullText() );
1947 $s = rtrim( $s . $prefix );
1948 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1949 wfProfileOut( __METHOD__."-interwiki" );
1950 continue;
1951 }
1952 wfProfileOut( __METHOD__."-interwiki" );
1953
1954 if ( $ns == NS_FILE ) {
1955 wfProfileIn( __METHOD__."-image" );
1956 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1957 if ( $wasblank ) {
1958 # if no parameters were passed, $text
1959 # becomes something like "File:Foo.png",
1960 # which we don't want to pass on to the
1961 # image generator
1962 $text = '';
1963 } else {
1964 # recursively parse links inside the image caption
1965 # actually, this will parse them in any other parameters, too,
1966 # but it might be hard to fix that, and it doesn't matter ATM
1967 $text = $this->replaceExternalLinks( $text );
1968 $holders->merge( $this->replaceInternalLinks2( $text ) );
1969 }
1970 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1971 $s .= $prefix . $this->armorLinks(
1972 $this->makeImage( $nt, $text, $holders ) ) . $trail;
1973 } else {
1974 $s .= $prefix . $trail;
1975 }
1976 wfProfileOut( __METHOD__."-image" );
1977 continue;
1978 }
1979
1980 if ( $ns == NS_CATEGORY ) {
1981 wfProfileIn( __METHOD__."-category" );
1982 $s = rtrim( $s . "\n" ); # bug 87
1983
1984 if ( $wasblank ) {
1985 $sortkey = $this->getDefaultSort();
1986 } else {
1987 $sortkey = $text;
1988 }
1989 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1990 $sortkey = str_replace( "\n", '', $sortkey );
1991 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
1992 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1993
1994 /**
1995 * Strip the whitespace Category links produce, see bug 87
1996 * @todo We might want to use trim($tmp, "\n") here.
1997 */
1998 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1999
2000 wfProfileOut( __METHOD__."-category" );
2001 continue;
2002 }
2003 }
2004
2005 # Self-link checking
2006 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2007 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2008 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2009 continue;
2010 }
2011 }
2012
2013 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2014 # @todo FIXME: Should do batch file existence checks, see comment below
2015 if ( $ns == NS_MEDIA ) {
2016 wfProfileIn( __METHOD__."-media" );
2017 # Give extensions a chance to select the file revision for us
2018 $options = array();
2019 $descQuery = false;
2020 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2021 array( $this, $nt, &$options, &$descQuery ) );
2022 # Fetch and register the file (file title may be different via hooks)
2023 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2024 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2025 $s .= $prefix . $this->armorLinks(
2026 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2027 wfProfileOut( __METHOD__."-media" );
2028 continue;
2029 }
2030
2031 wfProfileIn( __METHOD__."-always_known" );
2032 # Some titles, such as valid special pages or files in foreign repos, should
2033 # be shown as bluelinks even though they're not included in the page table
2034 #
2035 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2036 # batch file existence checks for NS_FILE and NS_MEDIA
2037 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2038 $this->mOutput->addLink( $nt );
2039 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2040 } else {
2041 # Links will be added to the output link list after checking
2042 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2043 }
2044 wfProfileOut( __METHOD__."-always_known" );
2045 }
2046 wfProfileOut( __METHOD__ );
2047 return $holders;
2048 }
2049
2050 /**
2051 * Render a forced-blue link inline; protect against double expansion of
2052 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2053 * Since this little disaster has to split off the trail text to avoid
2054 * breaking URLs in the following text without breaking trails on the
2055 * wiki links, it's been made into a horrible function.
2056 *
2057 * @param $nt Title
2058 * @param $text String
2059 * @param $query Array or String
2060 * @param $trail String
2061 * @param $prefix String
2062 * @return String: HTML-wikitext mix oh yuck
2063 */
2064 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2065 list( $inside, $trail ) = Linker::splitTrail( $trail );
2066
2067 if ( is_string( $query ) ) {
2068 $query = wfCgiToArray( $query );
2069 }
2070 if ( $text == '' ) {
2071 $text = htmlspecialchars( $nt->getPrefixedText() );
2072 }
2073
2074 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2075
2076 return $this->armorLinks( $link ) . $trail;
2077 }
2078
2079 /**
2080 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2081 * going to go through further parsing steps before inline URL expansion.
2082 *
2083 * Not needed quite as much as it used to be since free links are a bit
2084 * more sensible these days. But bracketed links are still an issue.
2085 *
2086 * @param $text String: more-or-less HTML
2087 * @return String: less-or-more HTML with NOPARSE bits
2088 */
2089 function armorLinks( $text ) {
2090 return preg_replace( '/\b(' . $this->mUrlProtocols . ')/',
2091 "{$this->mUniqPrefix}NOPARSE$1", $text );
2092 }
2093
2094 /**
2095 * Return true if subpage links should be expanded on this page.
2096 * @return Boolean
2097 */
2098 function areSubpagesAllowed() {
2099 # Some namespaces don't allow subpages
2100 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2101 }
2102
2103 /**
2104 * Handle link to subpage if necessary
2105 *
2106 * @param $target String: the source of the link
2107 * @param &$text String: the link text, modified as necessary
2108 * @return string the full name of the link
2109 * @private
2110 */
2111 function maybeDoSubpageLink( $target, &$text ) {
2112 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2113 }
2114
2115 /**#@+
2116 * Used by doBlockLevels()
2117 * @private
2118 *
2119 * @return string
2120 */
2121 function closeParagraph() {
2122 $result = '';
2123 if ( $this->mLastSection != '' ) {
2124 $result = '</' . $this->mLastSection . ">\n";
2125 }
2126 $this->mInPre = false;
2127 $this->mLastSection = '';
2128 return $result;
2129 }
2130
2131 /**
2132 * getCommon() returns the length of the longest common substring
2133 * of both arguments, starting at the beginning of both.
2134 * @private
2135 *
2136 * @param $st1 string
2137 * @param $st2 string
2138 *
2139 * @return int
2140 */
2141 function getCommon( $st1, $st2 ) {
2142 $fl = strlen( $st1 );
2143 $shorter = strlen( $st2 );
2144 if ( $fl < $shorter ) {
2145 $shorter = $fl;
2146 }
2147
2148 for ( $i = 0; $i < $shorter; ++$i ) {
2149 if ( $st1[$i] != $st2[$i] ) {
2150 break;
2151 }
2152 }
2153 return $i;
2154 }
2155
2156 /**
2157 * These next three functions open, continue, and close the list
2158 * element appropriate to the prefix character passed into them.
2159 * @private
2160 *
2161 * @param $char string
2162 *
2163 * @return string
2164 */
2165 function openList( $char ) {
2166 $result = $this->closeParagraph();
2167
2168 if ( '*' === $char ) {
2169 $result .= '<ul><li>';
2170 } elseif ( '#' === $char ) {
2171 $result .= '<ol><li>';
2172 } elseif ( ':' === $char ) {
2173 $result .= '<dl><dd>';
2174 } elseif ( ';' === $char ) {
2175 $result .= '<dl><dt>';
2176 $this->mDTopen = true;
2177 } else {
2178 $result = '<!-- ERR 1 -->';
2179 }
2180
2181 return $result;
2182 }
2183
2184 /**
2185 * TODO: document
2186 * @param $char String
2187 * @private
2188 *
2189 * @return string
2190 */
2191 function nextItem( $char ) {
2192 if ( '*' === $char || '#' === $char ) {
2193 return '</li><li>';
2194 } elseif ( ':' === $char || ';' === $char ) {
2195 $close = '</dd>';
2196 if ( $this->mDTopen ) {
2197 $close = '</dt>';
2198 }
2199 if ( ';' === $char ) {
2200 $this->mDTopen = true;
2201 return $close . '<dt>';
2202 } else {
2203 $this->mDTopen = false;
2204 return $close . '<dd>';
2205 }
2206 }
2207 return '<!-- ERR 2 -->';
2208 }
2209
2210 /**
2211 * TODO: document
2212 * @param $char String
2213 * @private
2214 *
2215 * @return string
2216 */
2217 function closeList( $char ) {
2218 if ( '*' === $char ) {
2219 $text = '</li></ul>';
2220 } elseif ( '#' === $char ) {
2221 $text = '</li></ol>';
2222 } elseif ( ':' === $char ) {
2223 if ( $this->mDTopen ) {
2224 $this->mDTopen = false;
2225 $text = '</dt></dl>';
2226 } else {
2227 $text = '</dd></dl>';
2228 }
2229 } else {
2230 return '<!-- ERR 3 -->';
2231 }
2232 return $text."\n";
2233 }
2234 /**#@-*/
2235
2236 /**
2237 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2238 *
2239 * @param $text String
2240 * @param $linestart Boolean: whether or not this is at the start of a line.
2241 * @private
2242 * @return string the lists rendered as HTML
2243 */
2244 function doBlockLevels( $text, $linestart ) {
2245 wfProfileIn( __METHOD__ );
2246
2247 # Parsing through the text line by line. The main thing
2248 # happening here is handling of block-level elements p, pre,
2249 # and making lists from lines starting with * # : etc.
2250 #
2251 $textLines = StringUtils::explode( "\n", $text );
2252
2253 $lastPrefix = $output = '';
2254 $this->mDTopen = $inBlockElem = false;
2255 $prefixLength = 0;
2256 $paragraphStack = false;
2257
2258 foreach ( $textLines as $oLine ) {
2259 # Fix up $linestart
2260 if ( !$linestart ) {
2261 $output .= $oLine;
2262 $linestart = true;
2263 continue;
2264 }
2265 # * = ul
2266 # # = ol
2267 # ; = dt
2268 # : = dd
2269
2270 $lastPrefixLength = strlen( $lastPrefix );
2271 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2272 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2273 # If not in a <pre> element, scan for and figure out what prefixes are there.
2274 if ( !$this->mInPre ) {
2275 # Multiple prefixes may abut each other for nested lists.
2276 $prefixLength = strspn( $oLine, '*#:;' );
2277 $prefix = substr( $oLine, 0, $prefixLength );
2278
2279 # eh?
2280 # ; and : are both from definition-lists, so they're equivalent
2281 # for the purposes of determining whether or not we need to open/close
2282 # elements.
2283 $prefix2 = str_replace( ';', ':', $prefix );
2284 $t = substr( $oLine, $prefixLength );
2285 $this->mInPre = (bool)$preOpenMatch;
2286 } else {
2287 # Don't interpret any other prefixes in preformatted text
2288 $prefixLength = 0;
2289 $prefix = $prefix2 = '';
2290 $t = $oLine;
2291 }
2292
2293 # List generation
2294 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2295 # Same as the last item, so no need to deal with nesting or opening stuff
2296 $output .= $this->nextItem( substr( $prefix, -1 ) );
2297 $paragraphStack = false;
2298
2299 if ( substr( $prefix, -1 ) === ';') {
2300 # The one nasty exception: definition lists work like this:
2301 # ; title : definition text
2302 # So we check for : in the remainder text to split up the
2303 # title and definition, without b0rking links.
2304 $term = $t2 = '';
2305 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2306 $t = $t2;
2307 $output .= $term . $this->nextItem( ':' );
2308 }
2309 }
2310 } elseif ( $prefixLength || $lastPrefixLength ) {
2311 # We need to open or close prefixes, or both.
2312
2313 # Either open or close a level...
2314 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2315 $paragraphStack = false;
2316
2317 # Close all the prefixes which aren't shared.
2318 while ( $commonPrefixLength < $lastPrefixLength ) {
2319 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2320 --$lastPrefixLength;
2321 }
2322
2323 # Continue the current prefix if appropriate.
2324 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2325 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2326 }
2327
2328 # Open prefixes where appropriate.
2329 while ( $prefixLength > $commonPrefixLength ) {
2330 $char = substr( $prefix, $commonPrefixLength, 1 );
2331 $output .= $this->openList( $char );
2332
2333 if ( ';' === $char ) {
2334 # @todo FIXME: This is dupe of code above
2335 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2336 $t = $t2;
2337 $output .= $term . $this->nextItem( ':' );
2338 }
2339 }
2340 ++$commonPrefixLength;
2341 }
2342 $lastPrefix = $prefix2;
2343 }
2344
2345 # If we have no prefixes, go to paragraph mode.
2346 if ( 0 == $prefixLength ) {
2347 wfProfileIn( __METHOD__."-paragraph" );
2348 # No prefix (not in list)--go to paragraph mode
2349 # XXX: use a stack for nestable elements like span, table and div
2350 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2351 $closematch = preg_match(
2352 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2353 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2354 if ( $openmatch or $closematch ) {
2355 $paragraphStack = false;
2356 # TODO bug 5718: paragraph closed
2357 $output .= $this->closeParagraph();
2358 if ( $preOpenMatch and !$preCloseMatch ) {
2359 $this->mInPre = true;
2360 }
2361 $inBlockElem = !$closematch;
2362 } elseif ( !$inBlockElem && !$this->mInPre ) {
2363 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2364 # pre
2365 if ( $this->mLastSection !== 'pre' ) {
2366 $paragraphStack = false;
2367 $output .= $this->closeParagraph().'<pre>';
2368 $this->mLastSection = 'pre';
2369 }
2370 $t = substr( $t, 1 );
2371 } else {
2372 # paragraph
2373 if ( trim( $t ) === '' ) {
2374 if ( $paragraphStack ) {
2375 $output .= $paragraphStack.'<br />';
2376 $paragraphStack = false;
2377 $this->mLastSection = 'p';
2378 } else {
2379 if ( $this->mLastSection !== 'p' ) {
2380 $output .= $this->closeParagraph();
2381 $this->mLastSection = '';
2382 $paragraphStack = '<p>';
2383 } else {
2384 $paragraphStack = '</p><p>';
2385 }
2386 }
2387 } else {
2388 if ( $paragraphStack ) {
2389 $output .= $paragraphStack;
2390 $paragraphStack = false;
2391 $this->mLastSection = 'p';
2392 } elseif ( $this->mLastSection !== 'p' ) {
2393 $output .= $this->closeParagraph().'<p>';
2394 $this->mLastSection = 'p';
2395 }
2396 }
2397 }
2398 }
2399 wfProfileOut( __METHOD__."-paragraph" );
2400 }
2401 # somewhere above we forget to get out of pre block (bug 785)
2402 if ( $preCloseMatch && $this->mInPre ) {
2403 $this->mInPre = false;
2404 }
2405 if ( $paragraphStack === false ) {
2406 $output .= $t."\n";
2407 }
2408 }
2409 while ( $prefixLength ) {
2410 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2411 --$prefixLength;
2412 }
2413 if ( $this->mLastSection != '' ) {
2414 $output .= '</' . $this->mLastSection . '>';
2415 $this->mLastSection = '';
2416 }
2417
2418 wfProfileOut( __METHOD__ );
2419 return $output;
2420 }
2421
2422 /**
2423 * Split up a string on ':', ignoring any occurrences inside tags
2424 * to prevent illegal overlapping.
2425 *
2426 * @param $str String the string to split
2427 * @param &$before String set to everything before the ':'
2428 * @param &$after String set to everything after the ':'
2429 * @return String the position of the ':', or false if none found
2430 */
2431 function findColonNoLinks( $str, &$before, &$after ) {
2432 wfProfileIn( __METHOD__ );
2433
2434 $pos = strpos( $str, ':' );
2435 if ( $pos === false ) {
2436 # Nothing to find!
2437 wfProfileOut( __METHOD__ );
2438 return false;
2439 }
2440
2441 $lt = strpos( $str, '<' );
2442 if ( $lt === false || $lt > $pos ) {
2443 # Easy; no tag nesting to worry about
2444 $before = substr( $str, 0, $pos );
2445 $after = substr( $str, $pos+1 );
2446 wfProfileOut( __METHOD__ );
2447 return $pos;
2448 }
2449
2450 # Ugly state machine to walk through avoiding tags.
2451 $state = self::COLON_STATE_TEXT;
2452 $stack = 0;
2453 $len = strlen( $str );
2454 for( $i = 0; $i < $len; $i++ ) {
2455 $c = $str[$i];
2456
2457 switch( $state ) {
2458 # (Using the number is a performance hack for common cases)
2459 case 0: # self::COLON_STATE_TEXT:
2460 switch( $c ) {
2461 case "<":
2462 # Could be either a <start> tag or an </end> tag
2463 $state = self::COLON_STATE_TAGSTART;
2464 break;
2465 case ":":
2466 if ( $stack == 0 ) {
2467 # We found it!
2468 $before = substr( $str, 0, $i );
2469 $after = substr( $str, $i + 1 );
2470 wfProfileOut( __METHOD__ );
2471 return $i;
2472 }
2473 # Embedded in a tag; don't break it.
2474 break;
2475 default:
2476 # Skip ahead looking for something interesting
2477 $colon = strpos( $str, ':', $i );
2478 if ( $colon === false ) {
2479 # Nothing else interesting
2480 wfProfileOut( __METHOD__ );
2481 return false;
2482 }
2483 $lt = strpos( $str, '<', $i );
2484 if ( $stack === 0 ) {
2485 if ( $lt === false || $colon < $lt ) {
2486 # We found it!
2487 $before = substr( $str, 0, $colon );
2488 $after = substr( $str, $colon + 1 );
2489 wfProfileOut( __METHOD__ );
2490 return $i;
2491 }
2492 }
2493 if ( $lt === false ) {
2494 # Nothing else interesting to find; abort!
2495 # We're nested, but there's no close tags left. Abort!
2496 break 2;
2497 }
2498 # Skip ahead to next tag start
2499 $i = $lt;
2500 $state = self::COLON_STATE_TAGSTART;
2501 }
2502 break;
2503 case 1: # self::COLON_STATE_TAG:
2504 # In a <tag>
2505 switch( $c ) {
2506 case ">":
2507 $stack++;
2508 $state = self::COLON_STATE_TEXT;
2509 break;
2510 case "/":
2511 # Slash may be followed by >?
2512 $state = self::COLON_STATE_TAGSLASH;
2513 break;
2514 default:
2515 # ignore
2516 }
2517 break;
2518 case 2: # self::COLON_STATE_TAGSTART:
2519 switch( $c ) {
2520 case "/":
2521 $state = self::COLON_STATE_CLOSETAG;
2522 break;
2523 case "!":
2524 $state = self::COLON_STATE_COMMENT;
2525 break;
2526 case ">":
2527 # Illegal early close? This shouldn't happen D:
2528 $state = self::COLON_STATE_TEXT;
2529 break;
2530 default:
2531 $state = self::COLON_STATE_TAG;
2532 }
2533 break;
2534 case 3: # self::COLON_STATE_CLOSETAG:
2535 # In a </tag>
2536 if ( $c === ">" ) {
2537 $stack--;
2538 if ( $stack < 0 ) {
2539 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2540 wfProfileOut( __METHOD__ );
2541 return false;
2542 }
2543 $state = self::COLON_STATE_TEXT;
2544 }
2545 break;
2546 case self::COLON_STATE_TAGSLASH:
2547 if ( $c === ">" ) {
2548 # Yes, a self-closed tag <blah/>
2549 $state = self::COLON_STATE_TEXT;
2550 } else {
2551 # Probably we're jumping the gun, and this is an attribute
2552 $state = self::COLON_STATE_TAG;
2553 }
2554 break;
2555 case 5: # self::COLON_STATE_COMMENT:
2556 if ( $c === "-" ) {
2557 $state = self::COLON_STATE_COMMENTDASH;
2558 }
2559 break;
2560 case self::COLON_STATE_COMMENTDASH:
2561 if ( $c === "-" ) {
2562 $state = self::COLON_STATE_COMMENTDASHDASH;
2563 } else {
2564 $state = self::COLON_STATE_COMMENT;
2565 }
2566 break;
2567 case self::COLON_STATE_COMMENTDASHDASH:
2568 if ( $c === ">" ) {
2569 $state = self::COLON_STATE_TEXT;
2570 } else {
2571 $state = self::COLON_STATE_COMMENT;
2572 }
2573 break;
2574 default:
2575 throw new MWException( "State machine error in " . __METHOD__ );
2576 }
2577 }
2578 if ( $stack > 0 ) {
2579 wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2580 wfProfileOut( __METHOD__ );
2581 return false;
2582 }
2583 wfProfileOut( __METHOD__ );
2584 return false;
2585 }
2586
2587 /**
2588 * Return value of a magic variable (like PAGENAME)
2589 *
2590 * @private
2591 *
2592 * @param $index integer
2593 * @param $frame PPFrame
2594 *
2595 * @return string
2596 */
2597 function getVariableValue( $index, $frame = false ) {
2598 global $wgContLang, $wgSitename, $wgServer;
2599 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2600
2601 if ( is_null( $this->mTitle ) ) {
2602 // If no title set, bad things are going to happen
2603 // later. Title should always be set since this
2604 // should only be called in the middle of a parse
2605 // operation (but the unit-tests do funky stuff)
2606 throw new MWException( __METHOD__ . ' Should only be '
2607 . ' called while parsing (no title set)' );
2608 }
2609
2610 /**
2611 * Some of these require message or data lookups and can be
2612 * expensive to check many times.
2613 */
2614 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2615 if ( isset( $this->mVarCache[$index] ) ) {
2616 return $this->mVarCache[$index];
2617 }
2618 }
2619
2620 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2621 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2622
2623 # Use the time zone
2624 global $wgLocaltimezone;
2625 if ( isset( $wgLocaltimezone ) ) {
2626 $oldtz = date_default_timezone_get();
2627 date_default_timezone_set( $wgLocaltimezone );
2628 }
2629
2630 $localTimestamp = date( 'YmdHis', $ts );
2631 $localMonth = date( 'm', $ts );
2632 $localMonth1 = date( 'n', $ts );
2633 $localMonthName = date( 'n', $ts );
2634 $localDay = date( 'j', $ts );
2635 $localDay2 = date( 'd', $ts );
2636 $localDayOfWeek = date( 'w', $ts );
2637 $localWeek = date( 'W', $ts );
2638 $localYear = date( 'Y', $ts );
2639 $localHour = date( 'H', $ts );
2640 if ( isset( $wgLocaltimezone ) ) {
2641 date_default_timezone_set( $oldtz );
2642 }
2643
2644 $pageLang = $this->getFunctionLang();
2645
2646 switch ( $index ) {
2647 case 'currentmonth':
2648 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2649 break;
2650 case 'currentmonth1':
2651 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2652 break;
2653 case 'currentmonthname':
2654 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2655 break;
2656 case 'currentmonthnamegen':
2657 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2658 break;
2659 case 'currentmonthabbrev':
2660 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2661 break;
2662 case 'currentday':
2663 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2664 break;
2665 case 'currentday2':
2666 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2667 break;
2668 case 'localmonth':
2669 $value = $pageLang->formatNum( $localMonth );
2670 break;
2671 case 'localmonth1':
2672 $value = $pageLang->formatNum( $localMonth1 );
2673 break;
2674 case 'localmonthname':
2675 $value = $pageLang->getMonthName( $localMonthName );
2676 break;
2677 case 'localmonthnamegen':
2678 $value = $pageLang->getMonthNameGen( $localMonthName );
2679 break;
2680 case 'localmonthabbrev':
2681 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2682 break;
2683 case 'localday':
2684 $value = $pageLang->formatNum( $localDay );
2685 break;
2686 case 'localday2':
2687 $value = $pageLang->formatNum( $localDay2 );
2688 break;
2689 case 'pagename':
2690 $value = wfEscapeWikiText( $this->mTitle->getText() );
2691 break;
2692 case 'pagenamee':
2693 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2694 break;
2695 case 'fullpagename':
2696 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2697 break;
2698 case 'fullpagenamee':
2699 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2700 break;
2701 case 'subpagename':
2702 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2703 break;
2704 case 'subpagenamee':
2705 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2706 break;
2707 case 'basepagename':
2708 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2709 break;
2710 case 'basepagenamee':
2711 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2712 break;
2713 case 'talkpagename':
2714 if ( $this->mTitle->canTalk() ) {
2715 $talkPage = $this->mTitle->getTalkPage();
2716 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2717 } else {
2718 $value = '';
2719 }
2720 break;
2721 case 'talkpagenamee':
2722 if ( $this->mTitle->canTalk() ) {
2723 $talkPage = $this->mTitle->getTalkPage();
2724 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2725 } else {
2726 $value = '';
2727 }
2728 break;
2729 case 'subjectpagename':
2730 $subjPage = $this->mTitle->getSubjectPage();
2731 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2732 break;
2733 case 'subjectpagenamee':
2734 $subjPage = $this->mTitle->getSubjectPage();
2735 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2736 break;
2737 case 'pageid': // requested in bug 23427
2738 $pageid = $this->getTitle()->getArticleId();
2739 if( $pageid == 0 ) {
2740 # 0 means the page doesn't exist in the database,
2741 # which means the user is previewing a new page.
2742 # The vary-revision flag must be set, because the magic word
2743 # will have a different value once the page is saved.
2744 $this->mOutput->setFlag( 'vary-revision' );
2745 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2746 }
2747 $value = $pageid ? $pageid : null;
2748 break;
2749 case 'revisionid':
2750 # Let the edit saving system know we should parse the page
2751 # *after* a revision ID has been assigned.
2752 $this->mOutput->setFlag( 'vary-revision' );
2753 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2754 $value = $this->mRevisionId;
2755 break;
2756 case 'revisionday':
2757 # Let the edit saving system know we should parse the page
2758 # *after* a revision ID has been assigned. This is for null edits.
2759 $this->mOutput->setFlag( 'vary-revision' );
2760 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2761 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2762 break;
2763 case 'revisionday2':
2764 # Let the edit saving system know we should parse the page
2765 # *after* a revision ID has been assigned. This is for null edits.
2766 $this->mOutput->setFlag( 'vary-revision' );
2767 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2768 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2769 break;
2770 case 'revisionmonth':
2771 # Let the edit saving system know we should parse the page
2772 # *after* a revision ID has been assigned. This is for null edits.
2773 $this->mOutput->setFlag( 'vary-revision' );
2774 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2775 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2776 break;
2777 case 'revisionmonth1':
2778 # Let the edit saving system know we should parse the page
2779 # *after* a revision ID has been assigned. This is for null edits.
2780 $this->mOutput->setFlag( 'vary-revision' );
2781 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2782 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2783 break;
2784 case 'revisionyear':
2785 # Let the edit saving system know we should parse the page
2786 # *after* a revision ID has been assigned. This is for null edits.
2787 $this->mOutput->setFlag( 'vary-revision' );
2788 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2789 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2790 break;
2791 case 'revisiontimestamp':
2792 # Let the edit saving system know we should parse the page
2793 # *after* a revision ID has been assigned. This is for null edits.
2794 $this->mOutput->setFlag( 'vary-revision' );
2795 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2796 $value = $this->getRevisionTimestamp();
2797 break;
2798 case 'revisionuser':
2799 # Let the edit saving system know we should parse the page
2800 # *after* a revision ID has been assigned. This is for null edits.
2801 $this->mOutput->setFlag( 'vary-revision' );
2802 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2803 $value = $this->getRevisionUser();
2804 break;
2805 case 'namespace':
2806 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2807 break;
2808 case 'namespacee':
2809 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2810 break;
2811 case 'namespacenumber':
2812 $value = $this->mTitle->getNamespace();
2813 break;
2814 case 'talkspace':
2815 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2816 break;
2817 case 'talkspacee':
2818 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2819 break;
2820 case 'subjectspace':
2821 $value = $this->mTitle->getSubjectNsText();
2822 break;
2823 case 'subjectspacee':
2824 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2825 break;
2826 case 'currentdayname':
2827 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2828 break;
2829 case 'currentyear':
2830 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2831 break;
2832 case 'currenttime':
2833 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2834 break;
2835 case 'currenthour':
2836 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2837 break;
2838 case 'currentweek':
2839 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2840 # int to remove the padding
2841 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2842 break;
2843 case 'currentdow':
2844 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2845 break;
2846 case 'localdayname':
2847 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2848 break;
2849 case 'localyear':
2850 $value = $pageLang->formatNum( $localYear, true );
2851 break;
2852 case 'localtime':
2853 $value = $pageLang->time( $localTimestamp, false, false );
2854 break;
2855 case 'localhour':
2856 $value = $pageLang->formatNum( $localHour, true );
2857 break;
2858 case 'localweek':
2859 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2860 # int to remove the padding
2861 $value = $pageLang->formatNum( (int)$localWeek );
2862 break;
2863 case 'localdow':
2864 $value = $pageLang->formatNum( $localDayOfWeek );
2865 break;
2866 case 'numberofarticles':
2867 $value = $pageLang->formatNum( SiteStats::articles() );
2868 break;
2869 case 'numberoffiles':
2870 $value = $pageLang->formatNum( SiteStats::images() );
2871 break;
2872 case 'numberofusers':
2873 $value = $pageLang->formatNum( SiteStats::users() );
2874 break;
2875 case 'numberofactiveusers':
2876 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2877 break;
2878 case 'numberofpages':
2879 $value = $pageLang->formatNum( SiteStats::pages() );
2880 break;
2881 case 'numberofadmins':
2882 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2883 break;
2884 case 'numberofedits':
2885 $value = $pageLang->formatNum( SiteStats::edits() );
2886 break;
2887 case 'numberofviews':
2888 global $wgDisableCounters;
2889 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
2890 break;
2891 case 'currenttimestamp':
2892 $value = wfTimestamp( TS_MW, $ts );
2893 break;
2894 case 'localtimestamp':
2895 $value = $localTimestamp;
2896 break;
2897 case 'currentversion':
2898 $value = SpecialVersion::getVersion();
2899 break;
2900 case 'articlepath':
2901 return $wgArticlePath;
2902 case 'sitename':
2903 return $wgSitename;
2904 case 'server':
2905 return $wgServer;
2906 case 'servername':
2907 $serverParts = wfParseUrl( $wgServer );
2908 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
2909 case 'scriptpath':
2910 return $wgScriptPath;
2911 case 'stylepath':
2912 return $wgStylePath;
2913 case 'directionmark':
2914 return $pageLang->getDirMark();
2915 case 'contentlanguage':
2916 global $wgLanguageCode;
2917 return $wgLanguageCode;
2918 default:
2919 $ret = null;
2920 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2921 return $ret;
2922 } else {
2923 return null;
2924 }
2925 }
2926
2927 if ( $index ) {
2928 $this->mVarCache[$index] = $value;
2929 }
2930
2931 return $value;
2932 }
2933
2934 /**
2935 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2936 *
2937 * @private
2938 */
2939 function initialiseVariables() {
2940 wfProfileIn( __METHOD__ );
2941 $variableIDs = MagicWord::getVariableIDs();
2942 $substIDs = MagicWord::getSubstIDs();
2943
2944 $this->mVariables = new MagicWordArray( $variableIDs );
2945 $this->mSubstWords = new MagicWordArray( $substIDs );
2946 wfProfileOut( __METHOD__ );
2947 }
2948
2949 /**
2950 * Preprocess some wikitext and return the document tree.
2951 * This is the ghost of replace_variables().
2952 *
2953 * @param $text String: The text to parse
2954 * @param $flags Integer: bitwise combination of:
2955 * self::PTD_FOR_INCLUSION Handle "<noinclude>" and "<includeonly>" as if the text is being
2956 * included. Default is to assume a direct page view.
2957 *
2958 * The generated DOM tree must depend only on the input text and the flags.
2959 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2960 *
2961 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2962 * change in the DOM tree for a given text, must be passed through the section identifier
2963 * in the section edit link and thus back to extractSections().
2964 *
2965 * The output of this function is currently only cached in process memory, but a persistent
2966 * cache may be implemented at a later date which takes further advantage of these strict
2967 * dependency requirements.
2968 *
2969 * @private
2970 *
2971 * @return PPNode
2972 */
2973 function preprocessToDom( $text, $flags = 0 ) {
2974 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2975 return $dom;
2976 }
2977
2978 /**
2979 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2980 *
2981 * @param $s string
2982 *
2983 * @return array
2984 */
2985 public static function splitWhitespace( $s ) {
2986 $ltrimmed = ltrim( $s );
2987 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2988 $trimmed = rtrim( $ltrimmed );
2989 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2990 if ( $diff > 0 ) {
2991 $w2 = substr( $ltrimmed, -$diff );
2992 } else {
2993 $w2 = '';
2994 }
2995 return array( $w1, $trimmed, $w2 );
2996 }
2997
2998 /**
2999 * Replace magic variables, templates, and template arguments
3000 * with the appropriate text. Templates are substituted recursively,
3001 * taking care to avoid infinite loops.
3002 *
3003 * Note that the substitution depends on value of $mOutputType:
3004 * self::OT_WIKI: only {{subst:}} templates
3005 * self::OT_PREPROCESS: templates but not extension tags
3006 * self::OT_HTML: all templates and extension tags
3007 *
3008 * @param $text String the text to transform
3009 * @param $frame PPFrame Object describing the arguments passed to the template.
3010 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3011 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3012 * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3013 * @private
3014 *
3015 * @return string
3016 */
3017 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3018 # Is there any text? Also, Prevent too big inclusions!
3019 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3020 return $text;
3021 }
3022 wfProfileIn( __METHOD__ );
3023
3024 if ( $frame === false ) {
3025 $frame = $this->getPreprocessor()->newFrame();
3026 } elseif ( !( $frame instanceof PPFrame ) ) {
3027 wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3028 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3029 }
3030
3031 $dom = $this->preprocessToDom( $text );
3032 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3033 $text = $frame->expand( $dom, $flags );
3034
3035 wfProfileOut( __METHOD__ );
3036 return $text;
3037 }
3038
3039 /**
3040 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3041 *
3042 * @param $args array
3043 *
3044 * @return array
3045 */
3046 static function createAssocArgs( $args ) {
3047 $assocArgs = array();
3048 $index = 1;
3049 foreach ( $args as $arg ) {
3050 $eqpos = strpos( $arg, '=' );
3051 if ( $eqpos === false ) {
3052 $assocArgs[$index++] = $arg;
3053 } else {
3054 $name = trim( substr( $arg, 0, $eqpos ) );
3055 $value = trim( substr( $arg, $eqpos+1 ) );
3056 if ( $value === false ) {
3057 $value = '';
3058 }
3059 if ( $name !== false ) {
3060 $assocArgs[$name] = $value;
3061 }
3062 }
3063 }
3064
3065 return $assocArgs;
3066 }
3067
3068 /**
3069 * Warn the user when a parser limitation is reached
3070 * Will warn at most once the user per limitation type
3071 *
3072 * @param $limitationType String: should be one of:
3073 * 'expensive-parserfunction' (corresponding messages:
3074 * 'expensive-parserfunction-warning',
3075 * 'expensive-parserfunction-category')
3076 * 'post-expand-template-argument' (corresponding messages:
3077 * 'post-expand-template-argument-warning',
3078 * 'post-expand-template-argument-category')
3079 * 'post-expand-template-inclusion' (corresponding messages:
3080 * 'post-expand-template-inclusion-warning',
3081 * 'post-expand-template-inclusion-category')
3082 * @param $current int|null Current value
3083 * @param $max int|null Maximum allowed, when an explicit limit has been
3084 * exceeded, provide the values (optional)
3085 */
3086 function limitationWarn( $limitationType, $current = null, $max = null) {
3087 # does no harm if $current and $max are present but are unnecessary for the message
3088 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3089 ->inContentLanguage()->escaped();
3090 $this->mOutput->addWarning( $warning );
3091 $this->addTrackingCategory( "$limitationType-category" );
3092 }
3093
3094 /**
3095 * Return the text of a template, after recursively
3096 * replacing any variables or templates within the template.
3097 *
3098 * @param $piece Array: the parts of the template
3099 * $piece['title']: the title, i.e. the part before the |
3100 * $piece['parts']: the parameter array
3101 * $piece['lineStart']: whether the brace was at the start of a line
3102 * @param $frame PPFrame The current frame, contains template arguments
3103 * @return String: the text of the template
3104 * @private
3105 */
3106 function braceSubstitution( $piece, $frame ) {
3107 global $wgContLang;
3108 wfProfileIn( __METHOD__ );
3109 wfProfileIn( __METHOD__.'-setup' );
3110
3111 # Flags
3112 $found = false; # $text has been filled
3113 $nowiki = false; # wiki markup in $text should be escaped
3114 $isHTML = false; # $text is HTML, armour it against wikitext transformation
3115 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3116 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
3117 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
3118
3119 # Title object, where $text came from
3120 $title = false;
3121
3122 # $part1 is the bit before the first |, and must contain only title characters.
3123 # Various prefixes will be stripped from it later.
3124 $titleWithSpaces = $frame->expand( $piece['title'] );
3125 $part1 = trim( $titleWithSpaces );
3126 $titleText = false;
3127
3128 # Original title text preserved for various purposes
3129 $originalTitle = $part1;
3130
3131 # $args is a list of argument nodes, starting from index 0, not including $part1
3132 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3133 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3134 wfProfileOut( __METHOD__.'-setup' );
3135
3136 $titleProfileIn = null; // profile templates
3137
3138 # SUBST
3139 wfProfileIn( __METHOD__.'-modifiers' );
3140 if ( !$found ) {
3141
3142 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3143
3144 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3145 # Decide whether to expand template or keep wikitext as-is.
3146 if ( $this->ot['wiki'] ) {
3147 if ( $substMatch === false ) {
3148 $literal = true; # literal when in PST with no prefix
3149 } else {
3150 $literal = false; # expand when in PST with subst: or safesubst:
3151 }
3152 } else {
3153 if ( $substMatch == 'subst' ) {
3154 $literal = true; # literal when not in PST with plain subst:
3155 } else {
3156 $literal = false; # expand when not in PST with safesubst: or no prefix
3157 }
3158 }
3159 if ( $literal ) {
3160 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3161 $isLocalObj = true;
3162 $found = true;
3163 }
3164 }
3165
3166 # Variables
3167 if ( !$found && $args->getLength() == 0 ) {
3168 $id = $this->mVariables->matchStartToEnd( $part1 );
3169 if ( $id !== false ) {
3170 $text = $this->getVariableValue( $id, $frame );
3171 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3172 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3173 }
3174 $found = true;
3175 }
3176 }
3177
3178 # MSG, MSGNW and RAW
3179 if ( !$found ) {
3180 # Check for MSGNW:
3181 $mwMsgnw = MagicWord::get( 'msgnw' );
3182 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3183 $nowiki = true;
3184 } else {
3185 # Remove obsolete MSG:
3186 $mwMsg = MagicWord::get( 'msg' );
3187 $mwMsg->matchStartAndRemove( $part1 );
3188 }
3189
3190 # Check for RAW:
3191 $mwRaw = MagicWord::get( 'raw' );
3192 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3193 $forceRawInterwiki = true;
3194 }
3195 }
3196 wfProfileOut( __METHOD__.'-modifiers' );
3197
3198 # Parser functions
3199 if ( !$found ) {
3200 wfProfileIn( __METHOD__ . '-pfunc' );
3201
3202 $colonPos = strpos( $part1, ':' );
3203 if ( $colonPos !== false ) {
3204 # Case sensitive functions
3205 $function = substr( $part1, 0, $colonPos );
3206 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3207 $function = $this->mFunctionSynonyms[1][$function];
3208 } else {
3209 # Case insensitive functions
3210 $function = $wgContLang->lc( $function );
3211 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3212 $function = $this->mFunctionSynonyms[0][$function];
3213 } else {
3214 $function = false;
3215 }
3216 }
3217 if ( $function ) {
3218 wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3219 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3220 $initialArgs = array( &$this );
3221 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3222 if ( $flags & SFH_OBJECT_ARGS ) {
3223 # Add a frame parameter, and pass the arguments as an array
3224 $allArgs = $initialArgs;
3225 $allArgs[] = $frame;
3226 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3227 $funcArgs[] = $args->item( $i );
3228 }
3229 $allArgs[] = $funcArgs;
3230 } else {
3231 # Convert arguments to plain text
3232 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3233 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3234 }
3235 $allArgs = array_merge( $initialArgs, $funcArgs );
3236 }
3237
3238 # Workaround for PHP bug 35229 and similar
3239 if ( !is_callable( $callback ) ) {
3240 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3241 wfProfileOut( __METHOD__ . '-pfunc' );
3242 wfProfileOut( __METHOD__ );
3243 throw new MWException( "Tag hook for $function is not callable\n" );
3244 }
3245 $result = call_user_func_array( $callback, $allArgs );
3246 $found = true;
3247 $noparse = true;
3248 $preprocessFlags = 0;
3249
3250 if ( is_array( $result ) ) {
3251 if ( isset( $result[0] ) ) {
3252 $text = $result[0];
3253 unset( $result[0] );
3254 }
3255
3256 # Extract flags into the local scope
3257 # This allows callers to set flags such as nowiki, found, etc.
3258 extract( $result );
3259 } else {
3260 $text = $result;
3261 }
3262 if ( !$noparse ) {
3263 $text = $this->preprocessToDom( $text, $preprocessFlags );
3264 $isChildObj = true;
3265 }
3266 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3267 }
3268 }
3269 wfProfileOut( __METHOD__ . '-pfunc' );
3270 }
3271
3272 # Finish mangling title and then check for loops.
3273 # Set $title to a Title object and $titleText to the PDBK
3274 if ( !$found ) {
3275 $ns = NS_TEMPLATE;
3276 # Split the title into page and subpage
3277 $subpage = '';
3278 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3279 if ( $subpage !== '' ) {
3280 $ns = $this->mTitle->getNamespace();
3281 }
3282 $title = Title::newFromText( $part1, $ns );
3283 if ( $title ) {
3284 $titleText = $title->getPrefixedText();
3285 # Check for language variants if the template is not found
3286 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3287 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3288 }
3289 # Do recursion depth check
3290 $limit = $this->mOptions->getMaxTemplateDepth();
3291 if ( $frame->depth >= $limit ) {
3292 $found = true;
3293 $text = '<span class="error">'
3294 . wfMessage( 'parser-template-recursion-depth-warning' )
3295 ->numParams( $limit )->inContentLanguage()->text()
3296 . '</span>';
3297 }
3298 }
3299 }
3300
3301 # Load from database
3302 if ( !$found && $title ) {
3303 if ( !Profiler::instance()->isPersistent() ) {
3304 # Too many unique items can kill profiling DBs/collectors
3305 $titleProfileIn = __METHOD__ . "-title-" . $title->getDBKey();
3306 wfProfileIn( $titleProfileIn ); // template in
3307 }
3308 wfProfileIn( __METHOD__ . '-loadtpl' );
3309 if ( !$title->isExternal() ) {
3310 if ( $title->isSpecialPage()
3311 && $this->mOptions->getAllowSpecialInclusion()
3312 && $this->ot['html'] )
3313 {
3314 // Pass the template arguments as URL parameters.
3315 // "uselang" will have no effect since the Language object
3316 // is forced to the one defined in ParserOptions.
3317 $pageArgs = array();
3318 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3319 $bits = $args->item( $i )->splitArg();
3320 if ( strval( $bits['index'] ) === '' ) {
3321 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3322 $value = trim( $frame->expand( $bits['value'] ) );
3323 $pageArgs[$name] = $value;
3324 }
3325 }
3326
3327 // Create a new context to execute the special page
3328 $context = new RequestContext;
3329 $context->setTitle( $title );
3330 $context->setRequest( new FauxRequest( $pageArgs ) );
3331 $context->setUser( $this->getUser() );
3332 $context->setLanguage( $this->mOptions->getUserLangObj() );
3333 $ret = SpecialPageFactory::capturePath( $title, $context );
3334 if ( $ret ) {
3335 $text = $context->getOutput()->getHTML();
3336 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3337 $found = true;
3338 $isHTML = true;
3339 $this->disableCache();
3340 }
3341 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3342 $found = false; # access denied
3343 wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3344 } else {
3345 list( $text, $title ) = $this->getTemplateDom( $title );
3346 if ( $text !== false ) {
3347 $found = true;
3348 $isChildObj = true;
3349 }
3350 }
3351
3352 # If the title is valid but undisplayable, make a link to it
3353 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3354 $text = "[[:$titleText]]";
3355 $found = true;
3356 }
3357 } elseif ( $title->isTrans() ) {
3358 # Interwiki transclusion
3359 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3360 $text = $this->interwikiTransclude( $title, 'render' );
3361 $isHTML = true;
3362 } else {
3363 $text = $this->interwikiTransclude( $title, 'raw' );
3364 # Preprocess it like a template
3365 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3366 $isChildObj = true;
3367 }
3368 $found = true;
3369 }
3370
3371 # Do infinite loop check
3372 # This has to be done after redirect resolution to avoid infinite loops via redirects
3373 if ( !$frame->loopCheck( $title ) ) {
3374 $found = true;
3375 $text = '<span class="error">'
3376 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3377 . '</span>';
3378 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3379 }
3380 wfProfileOut( __METHOD__ . '-loadtpl' );
3381 }
3382
3383 # If we haven't found text to substitute by now, we're done
3384 # Recover the source wikitext and return it
3385 if ( !$found ) {
3386 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3387 if ( $titleProfileIn ) {
3388 wfProfileOut( $titleProfileIn ); // template out
3389 }
3390 wfProfileOut( __METHOD__ );
3391 return array( 'object' => $text );
3392 }
3393
3394 # Expand DOM-style return values in a child frame
3395 if ( $isChildObj ) {
3396 # Clean up argument array
3397 $newFrame = $frame->newChild( $args, $title );
3398
3399 if ( $nowiki ) {
3400 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3401 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3402 # Expansion is eligible for the empty-frame cache
3403 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3404 $text = $this->mTplExpandCache[$titleText];
3405 } else {
3406 $text = $newFrame->expand( $text );
3407 $this->mTplExpandCache[$titleText] = $text;
3408 }
3409 } else {
3410 # Uncached expansion
3411 $text = $newFrame->expand( $text );
3412 }
3413 }
3414 if ( $isLocalObj && $nowiki ) {
3415 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3416 $isLocalObj = false;
3417 }
3418
3419 if ( $titleProfileIn ) {
3420 wfProfileOut( $titleProfileIn ); // template out
3421 }
3422
3423 # Replace raw HTML by a placeholder
3424 if ( $isHTML ) {
3425 $text = $this->insertStripItem( $text );
3426 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3427 # Escape nowiki-style return values
3428 $text = wfEscapeWikiText( $text );
3429 } elseif ( is_string( $text )
3430 && !$piece['lineStart']
3431 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3432 {
3433 # Bug 529: if the template begins with a table or block-level
3434 # element, it should be treated as beginning a new line.
3435 # This behaviour is somewhat controversial.
3436 $text = "\n" . $text;
3437 }
3438
3439 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3440 # Error, oversize inclusion
3441 if ( $titleText !== false ) {
3442 # Make a working, properly escaped link if possible (bug 23588)
3443 $text = "[[:$titleText]]";
3444 } else {
3445 # This will probably not be a working link, but at least it may
3446 # provide some hint of where the problem is
3447 preg_replace( '/^:/', '', $originalTitle );
3448 $text = "[[:$originalTitle]]";
3449 }
3450 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3451 $this->limitationWarn( 'post-expand-template-inclusion' );
3452 }
3453
3454 if ( $isLocalObj ) {
3455 $ret = array( 'object' => $text );
3456 } else {
3457 $ret = array( 'text' => $text );
3458 }
3459
3460 wfProfileOut( __METHOD__ );
3461 return $ret;
3462 }
3463
3464 /**
3465 * Get the semi-parsed DOM representation of a template with a given title,
3466 * and its redirect destination title. Cached.
3467 *
3468 * @param $title Title
3469 *
3470 * @return array
3471 */
3472 function getTemplateDom( $title ) {
3473 $cacheTitle = $title;
3474 $titleText = $title->getPrefixedDBkey();
3475
3476 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3477 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3478 $title = Title::makeTitle( $ns, $dbk );
3479 $titleText = $title->getPrefixedDBkey();
3480 }
3481 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3482 return array( $this->mTplDomCache[$titleText], $title );
3483 }
3484
3485 # Cache miss, go to the database
3486 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3487
3488 if ( $text === false ) {
3489 $this->mTplDomCache[$titleText] = false;
3490 return array( false, $title );
3491 }
3492
3493 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3494 $this->mTplDomCache[ $titleText ] = $dom;
3495
3496 if ( !$title->equals( $cacheTitle ) ) {
3497 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3498 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3499 }
3500
3501 return array( $dom, $title );
3502 }
3503
3504 /**
3505 * Fetch the unparsed text of a template and register a reference to it.
3506 * @param Title $title
3507 * @return Array ( string or false, Title )
3508 */
3509 function fetchTemplateAndTitle( $title ) {
3510 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3511 $stuff = call_user_func( $templateCb, $title, $this );
3512 $text = $stuff['text'];
3513 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3514 if ( isset( $stuff['deps'] ) ) {
3515 foreach ( $stuff['deps'] as $dep ) {
3516 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3517 }
3518 }
3519 return array( $text, $finalTitle );
3520 }
3521
3522 /**
3523 * Fetch the unparsed text of a template and register a reference to it.
3524 * @param Title $title
3525 * @return mixed string or false
3526 */
3527 function fetchTemplate( $title ) {
3528 $rv = $this->fetchTemplateAndTitle( $title );
3529 return $rv[0];
3530 }
3531
3532 /**
3533 * Static function to get a template
3534 * Can be overridden via ParserOptions::setTemplateCallback().
3535 *
3536 * @param $title Title
3537 * @param $parser Parser
3538 *
3539 * @return array
3540 */
3541 static function statelessFetchTemplate( $title, $parser = false ) {
3542 $text = $skip = false;
3543 $finalTitle = $title;
3544 $deps = array();
3545
3546 # Loop to fetch the article, with up to 1 redirect
3547 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3548 # Give extensions a chance to select the revision instead
3549 $id = false; # Assume current
3550 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3551 array( $parser, $title, &$skip, &$id ) );
3552
3553 if ( $skip ) {
3554 $text = false;
3555 $deps[] = array(
3556 'title' => $title,
3557 'page_id' => $title->getArticleID(),
3558 'rev_id' => null
3559 );
3560 break;
3561 }
3562 # Get the revision
3563 $rev = $id
3564 ? Revision::newFromId( $id )
3565 : Revision::newFromTitle( $title, 0, Revision::READ_NORMAL );
3566 $rev_id = $rev ? $rev->getId() : 0;
3567 # If there is no current revision, there is no page
3568 if ( $id === false && !$rev ) {
3569 $linkCache = LinkCache::singleton();
3570 $linkCache->addBadLinkObj( $title );
3571 }
3572
3573 $deps[] = array(
3574 'title' => $title,
3575 'page_id' => $title->getArticleID(),
3576 'rev_id' => $rev_id );
3577 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3578 # We fetched a rev from a different title; register it too...
3579 $deps[] = array(
3580 'title' => $rev->getTitle(),
3581 'page_id' => $rev->getPage(),
3582 'rev_id' => $rev_id );
3583 }
3584
3585 if ( $rev ) {
3586 $text = $rev->getText();
3587 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3588 global $wgContLang;
3589 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3590 if ( !$message->exists() ) {
3591 $text = false;
3592 break;
3593 }
3594 $text = $message->plain();
3595 } else {
3596 break;
3597 }
3598 if ( $text === false ) {
3599 break;
3600 }
3601 # Redirect?
3602 $finalTitle = $title;
3603 $title = Title::newFromRedirect( $text );
3604 }
3605 return array(
3606 'text' => $text,
3607 'finalTitle' => $finalTitle,
3608 'deps' => $deps );
3609 }
3610
3611 /**
3612 * Fetch a file and its title and register a reference to it.
3613 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3614 * @param Title $title
3615 * @param Array $options Array of options to RepoGroup::findFile
3616 * @return File|bool
3617 */
3618 function fetchFile( $title, $options = array() ) {
3619 $res = $this->fetchFileAndTitle( $title, $options );
3620 return $res[0];
3621 }
3622
3623 /**
3624 * Fetch a file and its title and register a reference to it.
3625 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3626 * @param Title $title
3627 * @param Array $options Array of options to RepoGroup::findFile
3628 * @return Array ( File or false, Title of file )
3629 */
3630 function fetchFileAndTitle( $title, $options = array() ) {
3631 if ( isset( $options['broken'] ) ) {
3632 $file = false; // broken thumbnail forced by hook
3633 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3634 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3635 } else { // get by (name,timestamp)
3636 $file = wfFindFile( $title, $options );
3637 }
3638 $time = $file ? $file->getTimestamp() : false;
3639 $sha1 = $file ? $file->getSha1() : false;
3640 # Register the file as a dependency...
3641 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3642 if ( $file && !$title->equals( $file->getTitle() ) ) {
3643 # Update fetched file title
3644 $title = $file->getTitle();
3645 if ( is_null( $file->getRedirectedTitle() ) ) {
3646 # This file was not a redirect, but the title does not match.
3647 # Register under the new name because otherwise the link will
3648 # get lost.
3649 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3650 }
3651 }
3652 return array( $file, $title );
3653 }
3654
3655 /**
3656 * Transclude an interwiki link.
3657 *
3658 * @param $title Title
3659 * @param $action
3660 *
3661 * @return string
3662 */
3663 function interwikiTransclude( $title, $action ) {
3664 global $wgEnableScaryTranscluding;
3665
3666 if ( !$wgEnableScaryTranscluding ) {
3667 return wfMessage('scarytranscludedisabled')->inContentLanguage()->text();
3668 }
3669
3670 $url = $title->getFullUrl( "action=$action" );
3671
3672 if ( strlen( $url ) > 255 ) {
3673 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3674 }
3675 return $this->fetchScaryTemplateMaybeFromCache( $url );
3676 }
3677
3678 /**
3679 * @param $url string
3680 * @return Mixed|String
3681 */
3682 function fetchScaryTemplateMaybeFromCache( $url ) {
3683 global $wgTranscludeCacheExpiry;
3684 $dbr = wfGetDB( DB_SLAVE );
3685 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3686 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3687 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3688 if ( $obj ) {
3689 return $obj->tc_contents;
3690 }
3691
3692 $text = Http::get( $url );
3693 if ( !$text ) {
3694 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3695 }
3696
3697 $dbw = wfGetDB( DB_MASTER );
3698 $dbw->replace( 'transcache', array('tc_url'), array(
3699 'tc_url' => $url,
3700 'tc_time' => $dbw->timestamp( time() ),
3701 'tc_contents' => $text)
3702 );
3703 return $text;
3704 }
3705
3706 /**
3707 * Triple brace replacement -- used for template arguments
3708 * @private
3709 *
3710 * @param $piece array
3711 * @param $frame PPFrame
3712 *
3713 * @return array
3714 */
3715 function argSubstitution( $piece, $frame ) {
3716 wfProfileIn( __METHOD__ );
3717
3718 $error = false;
3719 $parts = $piece['parts'];
3720 $nameWithSpaces = $frame->expand( $piece['title'] );
3721 $argName = trim( $nameWithSpaces );
3722 $object = false;
3723 $text = $frame->getArgument( $argName );
3724 if ( $text === false && $parts->getLength() > 0
3725 && (
3726 $this->ot['html']
3727 || $this->ot['pre']
3728 || ( $this->ot['wiki'] && $frame->isTemplate() )
3729 )
3730 ) {
3731 # No match in frame, use the supplied default
3732 $object = $parts->item( 0 )->getChildren();
3733 }
3734 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3735 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3736 $this->limitationWarn( 'post-expand-template-argument' );
3737 }
3738
3739 if ( $text === false && $object === false ) {
3740 # No match anywhere
3741 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3742 }
3743 if ( $error !== false ) {
3744 $text .= $error;
3745 }
3746 if ( $object !== false ) {
3747 $ret = array( 'object' => $object );
3748 } else {
3749 $ret = array( 'text' => $text );
3750 }
3751
3752 wfProfileOut( __METHOD__ );
3753 return $ret;
3754 }
3755
3756 /**
3757 * Return the text to be used for a given extension tag.
3758 * This is the ghost of strip().
3759 *
3760 * @param $params array Associative array of parameters:
3761 * name PPNode for the tag name
3762 * attr PPNode for unparsed text where tag attributes are thought to be
3763 * attributes Optional associative array of parsed attributes
3764 * inner Contents of extension element
3765 * noClose Original text did not have a close tag
3766 * @param $frame PPFrame
3767 *
3768 * @return string
3769 */
3770 function extensionSubstitution( $params, $frame ) {
3771 $name = $frame->expand( $params['name'] );
3772 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3773 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3774 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3775
3776 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3777 ( $this->ot['html'] || $this->ot['pre'] );
3778 if ( $isFunctionTag ) {
3779 $markerType = 'none';
3780 } else {
3781 $markerType = 'general';
3782 }
3783 if ( $this->ot['html'] || $isFunctionTag ) {
3784 $name = strtolower( $name );
3785 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3786 if ( isset( $params['attributes'] ) ) {
3787 $attributes = $attributes + $params['attributes'];
3788 }
3789
3790 if ( isset( $this->mTagHooks[$name] ) ) {
3791 # Workaround for PHP bug 35229 and similar
3792 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3793 throw new MWException( "Tag hook for $name is not callable\n" );
3794 }
3795 $output = call_user_func_array( $this->mTagHooks[$name],
3796 array( $content, $attributes, $this, $frame ) );
3797 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3798 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3799 if ( !is_callable( $callback ) ) {
3800 throw new MWException( "Tag hook for $name is not callable\n" );
3801 }
3802
3803 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3804 } else {
3805 $output = '<span class="error">Invalid tag extension name: ' .
3806 htmlspecialchars( $name ) . '</span>';
3807 }
3808
3809 if ( is_array( $output ) ) {
3810 # Extract flags to local scope (to override $markerType)
3811 $flags = $output;
3812 $output = $flags[0];
3813 unset( $flags[0] );
3814 extract( $flags );
3815 }
3816 } else {
3817 if ( is_null( $attrText ) ) {
3818 $attrText = '';
3819 }
3820 if ( isset( $params['attributes'] ) ) {
3821 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3822 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3823 htmlspecialchars( $attrValue ) . '"';
3824 }
3825 }
3826 if ( $content === null ) {
3827 $output = "<$name$attrText/>";
3828 } else {
3829 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3830 $output = "<$name$attrText>$content$close";
3831 }
3832 }
3833
3834 if ( $markerType === 'none' ) {
3835 return $output;
3836 } elseif ( $markerType === 'nowiki' ) {
3837 $this->mStripState->addNoWiki( $marker, $output );
3838 } elseif ( $markerType === 'general' ) {
3839 $this->mStripState->addGeneral( $marker, $output );
3840 } else {
3841 throw new MWException( __METHOD__.': invalid marker type' );
3842 }
3843 return $marker;
3844 }
3845
3846 /**
3847 * Increment an include size counter
3848 *
3849 * @param $type String: the type of expansion
3850 * @param $size Integer: the size of the text
3851 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3852 */
3853 function incrementIncludeSize( $type, $size ) {
3854 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3855 return false;
3856 } else {
3857 $this->mIncludeSizes[$type] += $size;
3858 return true;
3859 }
3860 }
3861
3862 /**
3863 * Increment the expensive function count
3864 *
3865 * @return Boolean: false if the limit has been exceeded
3866 */
3867 function incrementExpensiveFunctionCount() {
3868 $this->mExpensiveFunctionCount++;
3869 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3870 }
3871
3872 /**
3873 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3874 * Fills $this->mDoubleUnderscores, returns the modified text
3875 *
3876 * @param $text string
3877 *
3878 * @return string
3879 */
3880 function doDoubleUnderscore( $text ) {
3881 wfProfileIn( __METHOD__ );
3882
3883 # The position of __TOC__ needs to be recorded
3884 $mw = MagicWord::get( 'toc' );
3885 if ( $mw->match( $text ) ) {
3886 $this->mShowToc = true;
3887 $this->mForceTocPosition = true;
3888
3889 # Set a placeholder. At the end we'll fill it in with the TOC.
3890 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3891
3892 # Only keep the first one.
3893 $text = $mw->replace( '', $text );
3894 }
3895
3896 # Now match and remove the rest of them
3897 $mwa = MagicWord::getDoubleUnderscoreArray();
3898 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3899
3900 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3901 $this->mOutput->mNoGallery = true;
3902 }
3903 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3904 $this->mShowToc = false;
3905 }
3906 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3907 $this->addTrackingCategory( 'hidden-category-category' );
3908 }
3909 # (bug 8068) Allow control over whether robots index a page.
3910 #
3911 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3912 # is not desirable, the last one on the page should win.
3913 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3914 $this->mOutput->setIndexPolicy( 'noindex' );
3915 $this->addTrackingCategory( 'noindex-category' );
3916 }
3917 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3918 $this->mOutput->setIndexPolicy( 'index' );
3919 $this->addTrackingCategory( 'index-category' );
3920 }
3921
3922 # Cache all double underscores in the database
3923 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3924 $this->mOutput->setProperty( $key, '' );
3925 }
3926
3927 wfProfileOut( __METHOD__ );
3928 return $text;
3929 }
3930
3931 /**
3932 * Add a tracking category, getting the title from a system message,
3933 * or print a debug message if the title is invalid.
3934 *
3935 * @param $msg String: message key
3936 * @return Boolean: whether the addition was successful
3937 */
3938 public function addTrackingCategory( $msg ) {
3939 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3940 wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3941 return false;
3942 }
3943 // Important to parse with correct title (bug 31469)
3944 $cat = wfMessage( $msg )
3945 ->title( $this->getTitle() )
3946 ->inContentLanguage()
3947 ->text();
3948
3949 # Allow tracking categories to be disabled by setting them to "-"
3950 if ( $cat === '-' ) {
3951 return false;
3952 }
3953
3954 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3955 if ( $containerCategory ) {
3956 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3957 return true;
3958 } else {
3959 wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3960 return false;
3961 }
3962 }
3963
3964 /**
3965 * This function accomplishes several tasks:
3966 * 1) Auto-number headings if that option is enabled
3967 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3968 * 3) Add a Table of contents on the top for users who have enabled the option
3969 * 4) Auto-anchor headings
3970 *
3971 * It loops through all headlines, collects the necessary data, then splits up the
3972 * string and re-inserts the newly formatted headlines.
3973 *
3974 * @param $text String
3975 * @param $origText String: original, untouched wikitext
3976 * @param $isMain Boolean
3977 * @return mixed|string
3978 * @private
3979 */
3980 function formatHeadings( $text, $origText, $isMain=true ) {
3981 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3982
3983 # Inhibit editsection links if requested in the page
3984 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3985 $maybeShowEditLink = $showEditLink = false;
3986 } else {
3987 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3988 $showEditLink = $this->mOptions->getEditSection();
3989 }
3990 if ( $showEditLink ) {
3991 $this->mOutput->setEditSectionTokens( true );
3992 }
3993
3994 # Get all headlines for numbering them and adding funky stuff like [edit]
3995 # links - this is for later, but we need the number of headlines right now
3996 $matches = array();
3997 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3998
3999 # if there are fewer than 4 headlines in the article, do not show TOC
4000 # unless it's been explicitly enabled.
4001 $enoughToc = $this->mShowToc &&
4002 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4003
4004 # Allow user to stipulate that a page should have a "new section"
4005 # link added via __NEWSECTIONLINK__
4006 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4007 $this->mOutput->setNewSection( true );
4008 }
4009
4010 # Allow user to remove the "new section"
4011 # link via __NONEWSECTIONLINK__
4012 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4013 $this->mOutput->hideNewSection( true );
4014 }
4015
4016 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4017 # override above conditions and always show TOC above first header
4018 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4019 $this->mShowToc = true;
4020 $enoughToc = true;
4021 }
4022
4023 # headline counter
4024 $headlineCount = 0;
4025 $numVisible = 0;
4026
4027 # Ugh .. the TOC should have neat indentation levels which can be
4028 # passed to the skin functions. These are determined here
4029 $toc = '';
4030 $full = '';
4031 $head = array();
4032 $sublevelCount = array();
4033 $levelCount = array();
4034 $level = 0;
4035 $prevlevel = 0;
4036 $toclevel = 0;
4037 $prevtoclevel = 0;
4038 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4039 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4040 $oldType = $this->mOutputType;
4041 $this->setOutputType( self::OT_WIKI );
4042 $frame = $this->getPreprocessor()->newFrame();
4043 $root = $this->preprocessToDom( $origText );
4044 $node = $root->getFirstChild();
4045 $byteOffset = 0;
4046 $tocraw = array();
4047 $refers = array();
4048
4049 foreach ( $matches[3] as $headline ) {
4050 $isTemplate = false;
4051 $titleText = false;
4052 $sectionIndex = false;
4053 $numbering = '';
4054 $markerMatches = array();
4055 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
4056 $serial = $markerMatches[1];
4057 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4058 $isTemplate = ( $titleText != $baseTitleText );
4059 $headline = preg_replace( "/^$markerRegex/", "", $headline );
4060 }
4061
4062 if ( $toclevel ) {
4063 $prevlevel = $level;
4064 }
4065 $level = $matches[1][$headlineCount];
4066
4067 if ( $level > $prevlevel ) {
4068 # Increase TOC level
4069 $toclevel++;
4070 $sublevelCount[$toclevel] = 0;
4071 if ( $toclevel<$wgMaxTocLevel ) {
4072 $prevtoclevel = $toclevel;
4073 $toc .= Linker::tocIndent();
4074 $numVisible++;
4075 }
4076 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4077 # Decrease TOC level, find level to jump to
4078
4079 for ( $i = $toclevel; $i > 0; $i-- ) {
4080 if ( $levelCount[$i] == $level ) {
4081 # Found last matching level
4082 $toclevel = $i;
4083 break;
4084 } elseif ( $levelCount[$i] < $level ) {
4085 # Found first matching level below current level
4086 $toclevel = $i + 1;
4087 break;
4088 }
4089 }
4090 if ( $i == 0 ) {
4091 $toclevel = 1;
4092 }
4093 if ( $toclevel<$wgMaxTocLevel ) {
4094 if ( $prevtoclevel < $wgMaxTocLevel ) {
4095 # Unindent only if the previous toc level was shown :p
4096 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4097 $prevtoclevel = $toclevel;
4098 } else {
4099 $toc .= Linker::tocLineEnd();
4100 }
4101 }
4102 } else {
4103 # No change in level, end TOC line
4104 if ( $toclevel<$wgMaxTocLevel ) {
4105 $toc .= Linker::tocLineEnd();
4106 }
4107 }
4108
4109 $levelCount[$toclevel] = $level;
4110
4111 # count number of headlines for each level
4112 @$sublevelCount[$toclevel]++;
4113 $dot = 0;
4114 for( $i = 1; $i <= $toclevel; $i++ ) {
4115 if ( !empty( $sublevelCount[$i] ) ) {
4116 if ( $dot ) {
4117 $numbering .= '.';
4118 }
4119 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4120 $dot = 1;
4121 }
4122 }
4123
4124 # The safe header is a version of the header text safe to use for links
4125
4126 # Remove link placeholders by the link text.
4127 # <!--LINK number-->
4128 # turns into
4129 # link text with suffix
4130 # Do this before unstrip since link text can contain strip markers
4131 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4132
4133 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4134 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4135
4136 # Strip out HTML (first regex removes any tag not allowed)
4137 # Allowed tags are <sup> and <sub> (bug 8393), <i> (bug 26375) and <b> (r105284)
4138 # We strip any parameter from accepted tags (second regex)
4139 $tocline = preg_replace(
4140 array( '#<(?!/?(sup|sub|i|b)(?: [^>]*)?>).*?'.'>#', '#<(/?(sup|sub|i|b))(?: .*?)?'.'>#' ),
4141 array( '', '<$1>' ),
4142 $safeHeadline
4143 );
4144 $tocline = trim( $tocline );
4145
4146 # For the anchor, strip out HTML-y stuff period
4147 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4148 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4149
4150 # Save headline for section edit hint before it's escaped
4151 $headlineHint = $safeHeadline;
4152
4153 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4154 # For reverse compatibility, provide an id that's
4155 # HTML4-compatible, like we used to.
4156 #
4157 # It may be worth noting, academically, that it's possible for
4158 # the legacy anchor to conflict with a non-legacy headline
4159 # anchor on the page. In this case likely the "correct" thing
4160 # would be to either drop the legacy anchors or make sure
4161 # they're numbered first. However, this would require people
4162 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4163 # manually, so let's not bother worrying about it.
4164 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4165 array( 'noninitial', 'legacy' ) );
4166 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4167
4168 if ( $legacyHeadline == $safeHeadline ) {
4169 # No reason to have both (in fact, we can't)
4170 $legacyHeadline = false;
4171 }
4172 } else {
4173 $legacyHeadline = false;
4174 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4175 'noninitial' );
4176 }
4177
4178 # HTML names must be case-insensitively unique (bug 10721).
4179 # This does not apply to Unicode characters per
4180 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4181 # @todo FIXME: We may be changing them depending on the current locale.
4182 $arrayKey = strtolower( $safeHeadline );
4183 if ( $legacyHeadline === false ) {
4184 $legacyArrayKey = false;
4185 } else {
4186 $legacyArrayKey = strtolower( $legacyHeadline );
4187 }
4188
4189 # count how many in assoc. array so we can track dupes in anchors
4190 if ( isset( $refers[$arrayKey] ) ) {
4191 $refers[$arrayKey]++;
4192 } else {
4193 $refers[$arrayKey] = 1;
4194 }
4195 if ( isset( $refers[$legacyArrayKey] ) ) {
4196 $refers[$legacyArrayKey]++;
4197 } else {
4198 $refers[$legacyArrayKey] = 1;
4199 }
4200
4201 # Don't number the heading if it is the only one (looks silly)
4202 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4203 # the two are different if the line contains a link
4204 $headline = Html::element( 'span', array( 'class' => 'mw-headline-number' ), $numbering ) . ' ' . $headline;
4205 }
4206
4207 # Create the anchor for linking from the TOC to the section
4208 $anchor = $safeHeadline;
4209 $legacyAnchor = $legacyHeadline;
4210 if ( $refers[$arrayKey] > 1 ) {
4211 $anchor .= '_' . $refers[$arrayKey];
4212 }
4213 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4214 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4215 }
4216 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4217 $toc .= Linker::tocLine( $anchor, $tocline,
4218 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4219 }
4220
4221 # Add the section to the section tree
4222 # Find the DOM node for this header
4223 while ( $node && !$isTemplate ) {
4224 if ( $node->getName() === 'h' ) {
4225 $bits = $node->splitHeading();
4226 if ( $bits['i'] == $sectionIndex ) {
4227 break;
4228 }
4229 }
4230 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4231 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4232 $node = $node->getNextSibling();
4233 }
4234 $tocraw[] = array(
4235 'toclevel' => $toclevel,
4236 'level' => $level,
4237 'line' => $tocline,
4238 'number' => $numbering,
4239 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4240 'fromtitle' => $titleText,
4241 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4242 'anchor' => $anchor,
4243 );
4244
4245 # give headline the correct <h#> tag
4246 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4247 // Output edit section links as markers with styles that can be customized by skins
4248 if ( $isTemplate ) {
4249 # Put a T flag in the section identifier, to indicate to extractSections()
4250 # that sections inside <includeonly> should be counted.
4251 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4252 } else {
4253 $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4254 }
4255 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4256 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4257 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4258 // so we don't have to worry about a user trying to input one of these markers directly.
4259 // We use a page and section attribute to stop the language converter from converting these important bits
4260 // of data, but put the headline hint inside a content block because the language converter is supposed to
4261 // be able to convert that piece of data.
4262 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4263 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4264 if ( isset($editlinkArgs[2]) ) {
4265 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4266 } else {
4267 $editlink .= '/>';
4268 }
4269 } else {
4270 $editlink = '';
4271 }
4272 $head[$headlineCount] = Linker::makeHeadline( $level,
4273 $matches['attrib'][$headlineCount], $anchor, $headline,
4274 $editlink, $legacyAnchor );
4275
4276 $headlineCount++;
4277 }
4278
4279 $this->setOutputType( $oldType );
4280
4281 # Never ever show TOC if no headers
4282 if ( $numVisible < 1 ) {
4283 $enoughToc = false;
4284 }
4285
4286 if ( $enoughToc ) {
4287 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4288 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4289 }
4290 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4291 $this->mOutput->setTOCHTML( $toc );
4292 }
4293
4294 if ( $isMain ) {
4295 $this->mOutput->setSections( $tocraw );
4296 }
4297
4298 # split up and insert constructed headlines
4299 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4300 $i = 0;
4301
4302 // build an array of document sections
4303 $sections = array();
4304 foreach ( $blocks as $block ) {
4305 // $head is zero-based, sections aren't.
4306 if ( empty( $head[$i - 1] ) ) {
4307 $sections[$i] = $block;
4308 } else {
4309 $sections[$i] = $head[$i - 1] . $block;
4310 }
4311
4312 /**
4313 * Send a hook, one per section.
4314 * The idea here is to be able to make section-level DIVs, but to do so in a
4315 * lower-impact, more correct way than r50769
4316 *
4317 * $this : caller
4318 * $section : the section number
4319 * &$sectionContent : ref to the content of the section
4320 * $showEditLinks : boolean describing whether this section has an edit link
4321 */
4322 wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4323
4324 $i++;
4325 }
4326
4327 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4328 // append the TOC at the beginning
4329 // Top anchor now in skin
4330 $sections[0] = $sections[0] . $toc . "\n";
4331 }
4332
4333 $full .= join( '', $sections );
4334
4335 if ( $this->mForceTocPosition ) {
4336 return str_replace( '<!--MWTOC-->', $toc, $full );
4337 } else {
4338 return $full;
4339 }
4340 }
4341
4342 /**
4343 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4344 * conversion, substitting signatures, {{subst:}} templates, etc.
4345 *
4346 * @param $text String: the text to transform
4347 * @param $title Title: the Title object for the current article
4348 * @param $user User: the User object describing the current user
4349 * @param $options ParserOptions: parsing options
4350 * @param $clearState Boolean: whether to clear the parser state first
4351 * @return String: the altered wiki markup
4352 */
4353 public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4354 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4355 $this->setUser( $user );
4356
4357 $pairs = array(
4358 "\r\n" => "\n",
4359 );
4360 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4361 if( $options->getPreSaveTransform() ) {
4362 $text = $this->pstPass2( $text, $user );
4363 }
4364 $text = $this->mStripState->unstripBoth( $text );
4365
4366 $this->setUser( null ); #Reset
4367
4368 return $text;
4369 }
4370
4371 /**
4372 * Pre-save transform helper function
4373 * @private
4374 *
4375 * @param $text string
4376 * @param $user User
4377 *
4378 * @return string
4379 */
4380 function pstPass2( $text, $user ) {
4381 global $wgContLang, $wgLocaltimezone;
4382
4383 # Note: This is the timestamp saved as hardcoded wikitext to
4384 # the database, we use $wgContLang here in order to give
4385 # everyone the same signature and use the default one rather
4386 # than the one selected in each user's preferences.
4387 # (see also bug 12815)
4388 $ts = $this->mOptions->getTimestamp();
4389 if ( isset( $wgLocaltimezone ) ) {
4390 $tz = $wgLocaltimezone;
4391 } else {
4392 $tz = date_default_timezone_get();
4393 }
4394
4395 $unixts = wfTimestamp( TS_UNIX, $ts );
4396 $oldtz = date_default_timezone_get();
4397 date_default_timezone_set( $tz );
4398 $ts = date( 'YmdHis', $unixts );
4399 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4400
4401 # Allow translation of timezones through wiki. date() can return
4402 # whatever crap the system uses, localised or not, so we cannot
4403 # ship premade translations.
4404 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4405 $msg = wfMessage( $key )->inContentLanguage();
4406 if ( $msg->exists() ) {
4407 $tzMsg = $msg->text();
4408 }
4409
4410 date_default_timezone_set( $oldtz );
4411
4412 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4413
4414 # Variable replacement
4415 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4416 $text = $this->replaceVariables( $text );
4417
4418 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4419 # which may corrupt this parser instance via its wfMessage()->text() call-
4420
4421 # Signatures
4422 $sigText = $this->getUserSig( $user );
4423 $text = strtr( $text, array(
4424 '~~~~~' => $d,
4425 '~~~~' => "$sigText $d",
4426 '~~~' => $sigText
4427 ) );
4428
4429 # Context links: [[|name]] and [[name (context)|]]
4430 $tc = '[' . Title::legalChars() . ']';
4431 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4432
4433 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4434 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/"; # [[ns:page（context）|]]
4435 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/"; # [[ns:page (context), context|]]
4436 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4437
4438 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4439 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4440 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4441 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4442
4443 $t = $this->mTitle->getText();
4444 $m = array();
4445 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4446 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4447 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4448 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4449 } else {
4450 # if there's no context, don't bother duplicating the title
4451 $text = preg_replace( $p2, '[[\\1]]', $text );
4452 }
4453
4454 # Trim trailing whitespace
4455 $text = rtrim( $text );
4456
4457 return $text;
4458 }
4459
4460 /**
4461 * Fetch the user's signature text, if any, and normalize to
4462 * validated, ready-to-insert wikitext.
4463 * If you have pre-fetched the nickname or the fancySig option, you can
4464 * specify them here to save a database query.
4465 * Do not reuse this parser instance after calling getUserSig(),
4466 * as it may have changed if it's the $wgParser.
4467 *
4468 * @param $user User
4469 * @param $nickname String|bool nickname to use or false to use user's default nickname
4470 * @param $fancySig Boolean|null whether the nicknname is the complete signature
4471 * or null to use default value
4472 * @return string
4473 */
4474 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4475 global $wgMaxSigChars;
4476
4477 $username = $user->getName();
4478
4479 # If not given, retrieve from the user object.
4480 if ( $nickname === false )
4481 $nickname = $user->getOption( 'nickname' );
4482
4483 if ( is_null( $fancySig ) ) {
4484 $fancySig = $user->getBoolOption( 'fancysig' );
4485 }
4486
4487 $nickname = $nickname == null ? $username : $nickname;
4488
4489 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4490 $nickname = $username;
4491 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4492 } elseif ( $fancySig !== false ) {
4493 # Sig. might contain markup; validate this
4494 if ( $this->validateSig( $nickname ) !== false ) {
4495 # Validated; clean up (if needed) and return it
4496 return $this->cleanSig( $nickname, true );
4497 } else {
4498 # Failed to validate; fall back to the default
4499 $nickname = $username;
4500 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4501 }
4502 }
4503
4504 # Make sure nickname doesnt get a sig in a sig
4505 $nickname = self::cleanSigInSig( $nickname );
4506
4507 # If we're still here, make it a link to the user page
4508 $userText = wfEscapeWikiText( $username );
4509 $nickText = wfEscapeWikiText( $nickname );
4510 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4511
4512 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4513 }
4514
4515 /**
4516 * Check that the user's signature contains no bad XML
4517 *
4518 * @param $text String
4519 * @return mixed An expanded string, or false if invalid.
4520 */
4521 function validateSig( $text ) {
4522 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4523 }
4524
4525 /**
4526 * Clean up signature text
4527 *
4528 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4529 * 2) Substitute all transclusions
4530 *
4531 * @param $text String
4532 * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4533 * @return String: signature text
4534 */
4535 public function cleanSig( $text, $parsing = false ) {
4536 if ( !$parsing ) {
4537 global $wgTitle;
4538 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4539 }
4540
4541 # Option to disable this feature
4542 if ( !$this->mOptions->getCleanSignatures() ) {
4543 return $text;
4544 }
4545
4546 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4547 # => Move this logic to braceSubstitution()
4548 $substWord = MagicWord::get( 'subst' );
4549 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4550 $substText = '{{' . $substWord->getSynonym( 0 );
4551
4552 $text = preg_replace( $substRegex, $substText, $text );
4553 $text = self::cleanSigInSig( $text );
4554 $dom = $this->preprocessToDom( $text );
4555 $frame = $this->getPreprocessor()->newFrame();
4556 $text = $frame->expand( $dom );
4557
4558 if ( !$parsing ) {
4559 $text = $this->mStripState->unstripBoth( $text );
4560 }
4561
4562 return $text;
4563 }
4564
4565 /**
4566 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4567 *
4568 * @param $text String
4569 * @return String: signature text with /~{3,5}/ removed
4570 */
4571 public static function cleanSigInSig( $text ) {
4572 $text = preg_replace( '/~{3,5}/', '', $text );
4573 return $text;
4574 }
4575
4576 /**
4577 * Set up some variables which are usually set up in parse()
4578 * so that an external function can call some class members with confidence
4579 *
4580 * @param $title Title|null
4581 * @param $options ParserOptions
4582 * @param $outputType
4583 * @param $clearState bool
4584 */
4585 public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4586 $this->startParse( $title, $options, $outputType, $clearState );
4587 }
4588
4589 /**
4590 * @param $title Title|null
4591 * @param $options ParserOptions
4592 * @param $outputType
4593 * @param $clearState bool
4594 */
4595 private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4596 $this->setTitle( $title );
4597 $this->mOptions = $options;
4598 $this->setOutputType( $outputType );
4599 if ( $clearState ) {
4600 $this->clearState();
4601 }
4602 }
4603
4604 /**
4605 * Wrapper for preprocess()
4606 *
4607 * @param $text String: the text to preprocess
4608 * @param $options ParserOptions: options
4609 * @param $title Title object or null to use $wgTitle
4610 * @return String
4611 */
4612 public function transformMsg( $text, $options, $title = null ) {
4613 static $executing = false;
4614
4615 # Guard against infinite recursion
4616 if ( $executing ) {
4617 return $text;
4618 }
4619 $executing = true;
4620
4621 wfProfileIn( __METHOD__ );
4622 if ( !$title ) {
4623 global $wgTitle;
4624 $title = $wgTitle;
4625 }
4626 if ( !$title ) {
4627 # It's not uncommon having a null $wgTitle in scripts. See r80898
4628 # Create a ghost title in such case
4629 $title = Title::newFromText( 'Dwimmerlaik' );
4630 }
4631 $text = $this->preprocess( $text, $title, $options );
4632
4633 $executing = false;
4634 wfProfileOut( __METHOD__ );
4635 return $text;
4636 }
4637
4638 /**
4639 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4640 * The callback should have the following form:
4641 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4642 *
4643 * Transform and return $text. Use $parser for any required context, e.g. use
4644 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4645 *
4646 * Hooks may return extended information by returning an array, of which the
4647 * first numbered element (index 0) must be the return string, and all other
4648 * entries are extracted into local variables within an internal function
4649 * in the Parser class.
4650 *
4651 * This interface (introduced r61913) appears to be undocumented, but
4652 * 'markerName' is used by some core tag hooks to override which strip
4653 * array their results are placed in. **Use great caution if attempting
4654 * this interface, as it is not documented and injudicious use could smash
4655 * private variables.**
4656 *
4657 * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4658 * @param $callback Mixed: the callback function (and object) to use for the tag
4659 * @return Mixed|null The old value of the mTagHooks array associated with the hook
4660 */
4661 public function setHook( $tag, $callback ) {
4662 $tag = strtolower( $tag );
4663 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4664 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4665 }
4666 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4667 $this->mTagHooks[$tag] = $callback;
4668 if ( !in_array( $tag, $this->mStripList ) ) {
4669 $this->mStripList[] = $tag;
4670 }
4671
4672 return $oldVal;
4673 }
4674
4675 /**
4676 * As setHook(), but letting the contents be parsed.
4677 *
4678 * Transparent tag hooks are like regular XML-style tag hooks, except they
4679 * operate late in the transformation sequence, on HTML instead of wikitext.
4680 *
4681 * This is probably obsoleted by things dealing with parser frames?
4682 * The only extension currently using it is geoserver.
4683 *
4684 * @since 1.10
4685 * @todo better document or deprecate this
4686 *
4687 * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4688 * @param $callback Mixed: the callback function (and object) to use for the tag
4689 * @return Mixed|null The old value of the mTagHooks array associated with the hook
4690 */
4691 function setTransparentTagHook( $tag, $callback ) {
4692 $tag = strtolower( $tag );
4693 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4694 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4695 }
4696 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4697 $this->mTransparentTagHooks[$tag] = $callback;
4698
4699 return $oldVal;
4700 }
4701
4702 /**
4703 * Remove all tag hooks
4704 */
4705 function clearTagHooks() {
4706 $this->mTagHooks = array();
4707 $this->mFunctionTagHooks = array();
4708 $this->mStripList = $this->mDefaultStripList;
4709 }
4710
4711 /**
4712 * Create a function, e.g. {{sum:1|2|3}}
4713 * The callback function should have the form:
4714 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4715 *
4716 * Or with SFH_OBJECT_ARGS:
4717 * function myParserFunction( $parser, $frame, $args ) { ... }
4718 *
4719 * The callback may either return the text result of the function, or an array with the text
4720 * in element 0, and a number of flags in the other elements. The names of the flags are
4721 * specified in the keys. Valid flags are:
4722 * found The text returned is valid, stop processing the template. This
4723 * is on by default.
4724 * nowiki Wiki markup in the return value should be escaped
4725 * isHTML The returned text is HTML, armour it against wikitext transformation
4726 *
4727 * @param $id String: The magic word ID
4728 * @param $callback Mixed: the callback function (and object) to use
4729 * @param $flags Integer: a combination of the following flags:
4730 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4731 *
4732 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4733 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4734 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4735 * the arguments, and to control the way they are expanded.
4736 *
4737 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4738 * arguments, for instance:
4739 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4740 *
4741 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4742 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4743 * working if/when this is changed.
4744 *
4745 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4746 * expansion.
4747 *
4748 * Please read the documentation in includes/parser/Preprocessor.php for more information
4749 * about the methods available in PPFrame and PPNode.
4750 *
4751 * @return string|callback The old callback function for this name, if any
4752 */
4753 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4754 global $wgContLang;
4755
4756 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4757 $this->mFunctionHooks[$id] = array( $callback, $flags );
4758
4759 # Add to function cache
4760 $mw = MagicWord::get( $id );
4761 if ( !$mw )
4762 throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4763
4764 $synonyms = $mw->getSynonyms();
4765 $sensitive = intval( $mw->isCaseSensitive() );
4766
4767 foreach ( $synonyms as $syn ) {
4768 # Case
4769 if ( !$sensitive ) {
4770 $syn = $wgContLang->lc( $syn );
4771 }
4772 # Add leading hash
4773 if ( !( $flags & SFH_NO_HASH ) ) {
4774 $syn = '#' . $syn;
4775 }
4776 # Remove trailing colon
4777 if ( substr( $syn, -1, 1 ) === ':' ) {
4778 $syn = substr( $syn, 0, -1 );
4779 }
4780 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4781 }
4782 return $oldVal;
4783 }
4784
4785 /**
4786 * Get all registered function hook identifiers
4787 *
4788 * @return Array
4789 */
4790 function getFunctionHooks() {
4791 return array_keys( $this->mFunctionHooks );
4792 }
4793
4794 /**
4795 * Create a tag function, e.g. "<test>some stuff</test>".
4796 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4797 * Unlike parser functions, their content is not preprocessed.
4798 * @return null
4799 */
4800 function setFunctionTagHook( $tag, $callback, $flags ) {
4801 $tag = strtolower( $tag );
4802 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4803 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4804 $this->mFunctionTagHooks[$tag] : null;
4805 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4806
4807 if ( !in_array( $tag, $this->mStripList ) ) {
4808 $this->mStripList[] = $tag;
4809 }
4810
4811 return $old;
4812 }
4813
4814 /**
4815 * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4816 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4817 * Placeholders created in Skin::makeLinkObj()
4818 *
4819 * @param $text string
4820 * @param $options int
4821 *
4822 * @return array of link CSS classes, indexed by PDBK.
4823 */
4824 function replaceLinkHolders( &$text, $options = 0 ) {
4825 return $this->mLinkHolders->replace( $text );
4826 }
4827
4828 /**
4829 * Replace "<!--LINK-->" link placeholders with plain text of links
4830 * (not HTML-formatted).
4831 *
4832 * @param $text String
4833 * @return String
4834 */
4835 function replaceLinkHoldersText( $text ) {
4836 return $this->mLinkHolders->replaceText( $text );
4837 }
4838
4839 /**
4840 * Renders an image gallery from a text with one line per image.
4841 * text labels may be given by using |-style alternative text. E.g.
4842 * Image:one.jpg|The number "1"
4843 * Image:tree.jpg|A tree
4844 * given as text will return the HTML of a gallery with two images,
4845 * labeled 'The number "1"' and
4846 * 'A tree'.
4847 *
4848 * @param string $text
4849 * @param array $params
4850 * @return string HTML
4851 */
4852 function renderImageGallery( $text, $params ) {
4853 $ig = new ImageGallery();
4854 $ig->setContextTitle( $this->mTitle );
4855 $ig->setShowBytes( false );
4856 $ig->setShowFilename( false );
4857 $ig->setParser( $this );
4858 $ig->setHideBadImages();
4859 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4860
4861 if ( isset( $params['showfilename'] ) ) {
4862 $ig->setShowFilename( true );
4863 } else {
4864 $ig->setShowFilename( false );
4865 }
4866 if ( isset( $params['caption'] ) ) {
4867 $caption = $params['caption'];
4868 $caption = htmlspecialchars( $caption );
4869 $caption = $this->replaceInternalLinks( $caption );
4870 $ig->setCaptionHtml( $caption );
4871 }
4872 if ( isset( $params['perrow'] ) ) {
4873 $ig->setPerRow( $params['perrow'] );
4874 }
4875 if ( isset( $params['widths'] ) ) {
4876 $ig->setWidths( $params['widths'] );
4877 }
4878 if ( isset( $params['heights'] ) ) {
4879 $ig->setHeights( $params['heights'] );
4880 }
4881
4882 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4883
4884 $lines = StringUtils::explode( "\n", $text );
4885 foreach ( $lines as $line ) {
4886 # match lines like these:
4887 # Image:someimage.jpg|This is some image
4888 $matches = array();
4889 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4890 # Skip empty lines
4891 if ( count( $matches ) == 0 ) {
4892 continue;
4893 }
4894
4895 if ( strpos( $matches[0], '%' ) !== false ) {
4896 $matches[1] = rawurldecode( $matches[1] );
4897 }
4898 $title = Title::newFromText( $matches[1], NS_FILE );
4899 if ( is_null( $title ) ) {
4900 # Bogus title. Ignore these so we don't bomb out later.
4901 continue;
4902 }
4903
4904 $label = '';
4905 $alt = '';
4906 $link = '';
4907 if ( isset( $matches[3] ) ) {
4908 // look for an |alt= definition while trying not to break existing
4909 // captions with multiple pipes (|) in it, until a more sensible grammar
4910 // is defined for images in galleries
4911
4912 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4913 $parameterMatches = StringUtils::explode('|', $matches[3]);
4914 $magicWordAlt = MagicWord::get( 'img_alt' );
4915 $magicWordLink = MagicWord::get( 'img_link' );
4916
4917 foreach ( $parameterMatches as $parameterMatch ) {
4918 if ( $match = $magicWordAlt->matchVariableStartToEnd( $parameterMatch ) ) {
4919 $alt = $this->stripAltText( $match, false );
4920 }
4921 elseif( $match = $magicWordLink->matchVariableStartToEnd( $parameterMatch ) ){
4922 $link = strip_tags($this->replaceLinkHoldersText($match));
4923 $chars = self::EXT_LINK_URL_CLASS;
4924 $prots = $this->mUrlProtocols;
4925 //check to see if link matches an absolute url, if not then it must be a wiki link.
4926 if(!preg_match( "/^($prots)$chars+$/u", $link)){
4927 $localLinkTitle = Title::newFromText($link);
4928 $link = $localLinkTitle->getLocalURL();
4929 }
4930 }
4931 else {
4932 // concatenate all other pipes
4933 $label .= '|' . $parameterMatch;
4934 }
4935 }
4936 // remove the first pipe
4937 $label = substr( $label, 1 );
4938 }
4939
4940 $ig->add( $title, $label, $alt ,$link);
4941 }
4942 return $ig->toHTML();
4943 }
4944
4945 /**
4946 * @param $handler
4947 * @return array
4948 */
4949 function getImageParams( $handler ) {
4950 if ( $handler ) {
4951 $handlerClass = get_class( $handler );
4952 } else {
4953 $handlerClass = '';
4954 }
4955 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
4956 # Initialise static lists
4957 static $internalParamNames = array(
4958 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4959 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4960 'bottom', 'text-bottom' ),
4961 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4962 'upright', 'border', 'link', 'alt' ),
4963 );
4964 static $internalParamMap;
4965 if ( !$internalParamMap ) {
4966 $internalParamMap = array();
4967 foreach ( $internalParamNames as $type => $names ) {
4968 foreach ( $names as $name ) {
4969 $magicName = str_replace( '-', '_', "img_$name" );
4970 $internalParamMap[$magicName] = array( $type, $name );
4971 }
4972 }
4973 }
4974
4975 # Add handler params
4976 $paramMap = $internalParamMap;
4977 if ( $handler ) {
4978 $handlerParamMap = $handler->getParamMap();
4979 foreach ( $handlerParamMap as $magic => $paramName ) {
4980 $paramMap[$magic] = array( 'handler', $paramName );
4981 }
4982 }
4983 $this->mImageParams[$handlerClass] = $paramMap;
4984 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4985 }
4986 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4987 }
4988
4989 /**
4990 * Parse image options text and use it to make an image
4991 *
4992 * @param $title Title
4993 * @param $options String
4994 * @param $holders LinkHolderArray|bool
4995 * @return string HTML
4996 */
4997 function makeImage( $title, $options, $holders = false ) {
4998 # Check if the options text is of the form "options|alt text"
4999 # Options are:
5000 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5001 # * left no resizing, just left align. label is used for alt= only
5002 # * right same, but right aligned
5003 # * none same, but not aligned
5004 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5005 # * center center the image
5006 # * frame Keep original image size, no magnify-button.
5007 # * framed Same as "frame"
5008 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5009 # * upright reduce width for upright images, rounded to full __0 px
5010 # * border draw a 1px border around the image
5011 # * alt Text for HTML alt attribute (defaults to empty)
5012 # * link Set the target of the image link. Can be external, interwiki, or local
5013 # vertical-align values (no % or length right now):
5014 # * baseline
5015 # * sub
5016 # * super
5017 # * top
5018 # * text-top
5019 # * middle
5020 # * bottom
5021 # * text-bottom
5022
5023 $parts = StringUtils::explode( "|", $options );
5024
5025 # Give extensions a chance to select the file revision for us
5026 $options = array();
5027 $descQuery = false;
5028 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5029 array( $this, $title, &$options, &$descQuery ) );
5030 # Fetch and register the file (file title may be different via hooks)
5031 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5032
5033 # Get parameter map
5034 $handler = $file ? $file->getHandler() : false;
5035
5036 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5037
5038 if ( !$file ) {
5039 $this->addTrackingCategory( 'broken-file-category' );
5040 }
5041
5042 # Process the input parameters
5043 $caption = '';
5044 $params = array( 'frame' => array(), 'handler' => array(),
5045 'horizAlign' => array(), 'vertAlign' => array() );
5046 foreach ( $parts as $part ) {
5047 $part = trim( $part );
5048 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5049 $validated = false;
5050 if ( isset( $paramMap[$magicName] ) ) {
5051 list( $type, $paramName ) = $paramMap[$magicName];
5052
5053 # Special case; width and height come in one variable together
5054 if ( $type === 'handler' && $paramName === 'width' ) {
5055 $parsedWidthParam = $this->parseWidthParam( $value );
5056 if( isset( $parsedWidthParam['width'] ) ) {
5057 $width = $parsedWidthParam['width'];
5058 if ( $handler->validateParam( 'width', $width ) ) {
5059 $params[$type]['width'] = $width;
5060 $validated = true;
5061 }
5062 }
5063 if( isset( $parsedWidthParam['height'] ) ) {
5064 $height = $parsedWidthParam['height'];
5065 if ( $handler->validateParam( 'height', $height ) ) {
5066 $params[$type]['height'] = $height;
5067 $validated = true;
5068 }
5069 }
5070 # else no validation -- bug 13436
5071 } else {
5072 if ( $type === 'handler' ) {
5073 # Validate handler parameter
5074 $validated = $handler->validateParam( $paramName, $value );
5075 } else {
5076 # Validate internal parameters
5077 switch( $paramName ) {
5078 case 'manualthumb':
5079 case 'alt':
5080 # @todo FIXME: Possibly check validity here for
5081 # manualthumb? downstream behavior seems odd with
5082 # missing manual thumbs.
5083 $validated = true;
5084 $value = $this->stripAltText( $value, $holders );
5085 break;
5086 case 'link':
5087 $chars = self::EXT_LINK_URL_CLASS;
5088 $prots = $this->mUrlProtocols;
5089 if ( $value === '' ) {
5090 $paramName = 'no-link';
5091 $value = true;
5092 $validated = true;
5093 } elseif ( preg_match( "/^$prots/", $value ) ) {
5094 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
5095 $paramName = 'link-url';
5096 $this->mOutput->addExternalLink( $value );
5097 if ( $this->mOptions->getExternalLinkTarget() ) {
5098 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5099 }
5100 $validated = true;
5101 }
5102 } else {
5103 $linkTitle = Title::newFromText( $value );
5104 if ( $linkTitle ) {
5105 $paramName = 'link-title';
5106 $value = $linkTitle;
5107 $this->mOutput->addLink( $linkTitle );
5108 $validated = true;
5109 }
5110 }
5111 break;
5112 default:
5113 # Most other things appear to be empty or numeric...
5114 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5115 }
5116 }
5117
5118 if ( $validated ) {
5119 $params[$type][$paramName] = $value;
5120 }
5121 }
5122 }
5123 if ( !$validated ) {
5124 $caption = $part;
5125 }
5126 }
5127
5128 # Process alignment parameters
5129 if ( $params['horizAlign'] ) {
5130 $params['frame']['align'] = key( $params['horizAlign'] );
5131 }
5132 if ( $params['vertAlign'] ) {
5133 $params['frame']['valign'] = key( $params['vertAlign'] );
5134 }
5135
5136 $params['frame']['caption'] = $caption;
5137
5138 # Will the image be presented in a frame, with the caption below?
5139 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5140 isset( $params['frame']['framed'] ) ||
5141 isset( $params['frame']['thumbnail'] ) ||
5142 isset( $params['frame']['manualthumb'] );
5143
5144 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5145 # came to also set the caption, ordinary text after the image -- which
5146 # makes no sense, because that just repeats the text multiple times in
5147 # screen readers. It *also* came to set the title attribute.
5148 #
5149 # Now that we have an alt attribute, we should not set the alt text to
5150 # equal the caption: that's worse than useless, it just repeats the
5151 # text. This is the framed/thumbnail case. If there's no caption, we
5152 # use the unnamed parameter for alt text as well, just for the time be-
5153 # ing, if the unnamed param is set and the alt param is not.
5154 #
5155 # For the future, we need to figure out if we want to tweak this more,
5156 # e.g., introducing a title= parameter for the title; ignoring the un-
5157 # named parameter entirely for images without a caption; adding an ex-
5158 # plicit caption= parameter and preserving the old magic unnamed para-
5159 # meter for BC; ...
5160 if ( $imageIsFramed ) { # Framed image
5161 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5162 # No caption or alt text, add the filename as the alt text so
5163 # that screen readers at least get some description of the image
5164 $params['frame']['alt'] = $title->getText();
5165 }
5166 # Do not set $params['frame']['title'] because tooltips don't make sense
5167 # for framed images
5168 } else { # Inline image
5169 if ( !isset( $params['frame']['alt'] ) ) {
5170 # No alt text, use the "caption" for the alt text
5171 if ( $caption !== '') {
5172 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5173 } else {
5174 # No caption, fall back to using the filename for the
5175 # alt text
5176 $params['frame']['alt'] = $title->getText();
5177 }
5178 }
5179 # Use the "caption" for the tooltip text
5180 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5181 }
5182
5183 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5184
5185 # Linker does the rest
5186 $time = isset( $options['time'] ) ? $options['time'] : false;
5187 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5188 $time, $descQuery, $this->mOptions->getThumbSize() );
5189
5190 # Give the handler a chance to modify the parser object
5191 if ( $handler ) {
5192 $handler->parserTransformHook( $this, $file );
5193 }
5194
5195 return $ret;
5196 }
5197
5198 /**
5199 * @param $caption
5200 * @param $holders LinkHolderArray
5201 * @return mixed|String
5202 */
5203 protected function stripAltText( $caption, $holders ) {
5204 # Strip bad stuff out of the title (tooltip). We can't just use
5205 # replaceLinkHoldersText() here, because if this function is called
5206 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5207 if ( $holders ) {
5208 $tooltip = $holders->replaceText( $caption );
5209 } else {
5210 $tooltip = $this->replaceLinkHoldersText( $caption );
5211 }
5212
5213 # make sure there are no placeholders in thumbnail attributes
5214 # that are later expanded to html- so expand them now and
5215 # remove the tags
5216 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5217 $tooltip = Sanitizer::stripAllTags( $tooltip );
5218
5219 return $tooltip;
5220 }
5221
5222 /**
5223 * Set a flag in the output object indicating that the content is dynamic and
5224 * shouldn't be cached.
5225 */
5226 function disableCache() {
5227 wfDebug( "Parser output marked as uncacheable.\n" );
5228 if ( !$this->mOutput ) {
5229 throw new MWException( __METHOD__ .
5230 " can only be called when actually parsing something" );
5231 }
5232 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5233 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5234 }
5235
5236 /**
5237 * Callback from the Sanitizer for expanding items found in HTML attribute
5238 * values, so they can be safely tested and escaped.
5239 *
5240 * @param $text String
5241 * @param $frame PPFrame
5242 * @return String
5243 */
5244 function attributeStripCallback( &$text, $frame = false ) {
5245 $text = $this->replaceVariables( $text, $frame );
5246 $text = $this->mStripState->unstripBoth( $text );
5247 return $text;
5248 }
5249
5250 /**
5251 * Accessor
5252 *
5253 * @return array
5254 */
5255 function getTags() {
5256 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ), array_keys( $this->mFunctionTagHooks ) );
5257 }
5258
5259 /**
5260 * Replace transparent tags in $text with the values given by the callbacks.
5261 *
5262 * Transparent tag hooks are like regular XML-style tag hooks, except they
5263 * operate late in the transformation sequence, on HTML instead of wikitext.
5264 *
5265 * @param $text string
5266 *
5267 * @return string
5268 */
5269 function replaceTransparentTags( $text ) {
5270 $matches = array();
5271 $elements = array_keys( $this->mTransparentTagHooks );
5272 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5273 $replacements = array();
5274
5275 foreach ( $matches as $marker => $data ) {
5276 list( $element, $content, $params, $tag ) = $data;
5277 $tagName = strtolower( $element );
5278 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5279 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5280 } else {
5281 $output = $tag;
5282 }
5283 $replacements[$marker] = $output;
5284 }
5285 return strtr( $text, $replacements );
5286 }
5287
5288 /**
5289 * Break wikitext input into sections, and either pull or replace
5290 * some particular section's text.
5291 *
5292 * External callers should use the getSection and replaceSection methods.
5293 *
5294 * @param $text String: Page wikitext
5295 * @param $section String: a section identifier string of the form:
5296 * "<flag1> - <flag2> - ... - <section number>"
5297 *
5298 * Currently the only recognised flag is "T", which means the target section number
5299 * was derived during a template inclusion parse, in other words this is a template
5300 * section edit link. If no flags are given, it was an ordinary section edit link.
5301 * This flag is required to avoid a section numbering mismatch when a section is
5302 * enclosed by "<includeonly>" (bug 6563).
5303 *
5304 * The section number 0 pulls the text before the first heading; other numbers will
5305 * pull the given section along with its lower-level subsections. If the section is
5306 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5307 *
5308 * Section 0 is always considered to exist, even if it only contains the empty
5309 * string. If $text is the empty string and section 0 is replaced, $newText is
5310 * returned.
5311 *
5312 * @param $mode String: one of "get" or "replace"
5313 * @param $newText String: replacement text for section data.
5314 * @return String: for "get", the extracted section text.
5315 * for "replace", the whole page with the section replaced.
5316 */
5317 private function extractSections( $text, $section, $mode, $newText='' ) {
5318 global $wgTitle; # not generally used but removes an ugly failure mode
5319 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5320 $outText = '';
5321 $frame = $this->getPreprocessor()->newFrame();
5322
5323 # Process section extraction flags
5324 $flags = 0;
5325 $sectionParts = explode( '-', $section );
5326 $sectionIndex = array_pop( $sectionParts );
5327 foreach ( $sectionParts as $part ) {
5328 if ( $part === 'T' ) {
5329 $flags |= self::PTD_FOR_INCLUSION;
5330 }
5331 }
5332
5333 # Check for empty input
5334 if ( strval( $text ) === '' ) {
5335 # Only sections 0 and T-0 exist in an empty document
5336 if ( $sectionIndex == 0 ) {
5337 if ( $mode === 'get' ) {
5338 return '';
5339 } else {
5340 return $newText;
5341 }
5342 } else {
5343 if ( $mode === 'get' ) {
5344 return $newText;
5345 } else {
5346 return $text;
5347 }
5348 }
5349 }
5350
5351 # Preprocess the text
5352 $root = $this->preprocessToDom( $text, $flags );
5353
5354 # <h> nodes indicate section breaks
5355 # They can only occur at the top level, so we can find them by iterating the root's children
5356 $node = $root->getFirstChild();
5357
5358 # Find the target section
5359 if ( $sectionIndex == 0 ) {
5360 # Section zero doesn't nest, level=big
5361 $targetLevel = 1000;
5362 } else {
5363 while ( $node ) {
5364 if ( $node->getName() === 'h' ) {
5365 $bits = $node->splitHeading();
5366 if ( $bits['i'] == $sectionIndex ) {
5367 $targetLevel = $bits['level'];
5368 break;
5369 }
5370 }
5371 if ( $mode === 'replace' ) {
5372 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5373 }
5374 $node = $node->getNextSibling();
5375 }
5376 }
5377
5378 if ( !$node ) {
5379 # Not found
5380 if ( $mode === 'get' ) {
5381 return $newText;
5382 } else {
5383 return $text;
5384 }
5385 }
5386
5387 # Find the end of the section, including nested sections
5388 do {
5389 if ( $node->getName() === 'h' ) {
5390 $bits = $node->splitHeading();
5391 $curLevel = $bits['level'];
5392 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5393 break;
5394 }
5395 }
5396 if ( $mode === 'get' ) {
5397 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5398 }
5399 $node = $node->getNextSibling();
5400 } while ( $node );
5401
5402 # Write out the remainder (in replace mode only)
5403 if ( $mode === 'replace' ) {
5404 # Output the replacement text
5405 # Add two newlines on -- trailing whitespace in $newText is conventionally
5406 # stripped by the editor, so we need both newlines to restore the paragraph gap
5407 # Only add trailing whitespace if there is newText
5408 if ( $newText != "" ) {
5409 $outText .= $newText . "\n\n";
5410 }
5411
5412 while ( $node ) {
5413 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5414 $node = $node->getNextSibling();
5415 }
5416 }
5417
5418 if ( is_string( $outText ) ) {
5419 # Re-insert stripped tags
5420 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5421 }
5422
5423 return $outText;
5424 }
5425
5426 /**
5427 * This function returns the text of a section, specified by a number ($section).
5428 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5429 * the first section before any such heading (section 0).
5430 *
5431 * If a section contains subsections, these are also returned.
5432 *
5433 * @param $text String: text to look in
5434 * @param $section String: section identifier
5435 * @param $deftext String: default to return if section is not found
5436 * @return string text of the requested section
5437 */
5438 public function getSection( $text, $section, $deftext='' ) {
5439 return $this->extractSections( $text, $section, "get", $deftext );
5440 }
5441
5442 /**
5443 * This function returns $oldtext after the content of the section
5444 * specified by $section has been replaced with $text. If the target
5445 * section does not exist, $oldtext is returned unchanged.
5446 *
5447 * @param $oldtext String: former text of the article
5448 * @param $section int section identifier
5449 * @param $text String: replacing text
5450 * @return String: modified text
5451 */
5452 public function replaceSection( $oldtext, $section, $text ) {
5453 return $this->extractSections( $oldtext, $section, "replace", $text );
5454 }
5455
5456 /**
5457 * Get the ID of the revision we are parsing
5458 *
5459 * @return Mixed: integer or null
5460 */
5461 function getRevisionId() {
5462 return $this->mRevisionId;
5463 }
5464
5465 /**
5466 * Get the revision object for $this->mRevisionId
5467 *
5468 * @return Revision|null either a Revision object or null
5469 */
5470 protected function getRevisionObject() {
5471 if ( !is_null( $this->mRevisionObject ) ) {
5472 return $this->mRevisionObject;
5473 }
5474 if ( is_null( $this->mRevisionId ) ) {
5475 return null;
5476 }
5477
5478 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5479 return $this->mRevisionObject;
5480 }
5481
5482 /**
5483 * Get the timestamp associated with the current revision, adjusted for
5484 * the default server-local timestamp
5485 */
5486 function getRevisionTimestamp() {
5487 if ( is_null( $this->mRevisionTimestamp ) ) {
5488 wfProfileIn( __METHOD__ );
5489
5490 global $wgContLang;
5491
5492 $revObject = $this->getRevisionObject();
5493 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5494
5495 # The cryptic '' timezone parameter tells to use the site-default
5496 # timezone offset instead of the user settings.
5497 #
5498 # Since this value will be saved into the parser cache, served
5499 # to other users, and potentially even used inside links and such,
5500 # it needs to be consistent for all visitors.
5501 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5502
5503 wfProfileOut( __METHOD__ );
5504 }
5505 return $this->mRevisionTimestamp;
5506 }
5507
5508 /**
5509 * Get the name of the user that edited the last revision
5510 *
5511 * @return String: user name
5512 */
5513 function getRevisionUser() {
5514 if( is_null( $this->mRevisionUser ) ) {
5515 $revObject = $this->getRevisionObject();
5516
5517 # if this template is subst: the revision id will be blank,
5518 # so just use the current user's name
5519 if( $revObject ) {
5520 $this->mRevisionUser = $revObject->getUserText();
5521 } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5522 $this->mRevisionUser = $this->getUser()->getName();
5523 }
5524 }
5525 return $this->mRevisionUser;
5526 }
5527
5528 /**
5529 * Mutator for $mDefaultSort
5530 *
5531 * @param $sort string New value
5532 */
5533 public function setDefaultSort( $sort ) {
5534 $this->mDefaultSort = $sort;
5535 $this->mOutput->setProperty( 'defaultsort', $sort );
5536 }
5537
5538 /**
5539 * Accessor for $mDefaultSort
5540 * Will use the empty string if none is set.
5541 *
5542 * This value is treated as a prefix, so the
5543 * empty string is equivalent to sorting by
5544 * page name.
5545 *
5546 * @return string
5547 */
5548 public function getDefaultSort() {
5549 if ( $this->mDefaultSort !== false ) {
5550 return $this->mDefaultSort;
5551 } else {
5552 return '';
5553 }
5554 }
5555
5556 /**
5557 * Accessor for $mDefaultSort
5558 * Unlike getDefaultSort(), will return false if none is set
5559 *
5560 * @return string or false
5561 */
5562 public function getCustomDefaultSort() {
5563 return $this->mDefaultSort;
5564 }
5565
5566 /**
5567 * Try to guess the section anchor name based on a wikitext fragment
5568 * presumably extracted from a heading, for example "Header" from
5569 * "== Header ==".
5570 *
5571 * @param $text string
5572 *
5573 * @return string
5574 */
5575 public function guessSectionNameFromWikiText( $text ) {
5576 # Strip out wikitext links(they break the anchor)
5577 $text = $this->stripSectionName( $text );
5578 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5579 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5580 }
5581
5582 /**
5583 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5584 * instead. For use in redirects, since IE6 interprets Redirect: headers
5585 * as something other than UTF-8 (apparently?), resulting in breakage.
5586 *
5587 * @param $text String: The section name
5588 * @return string An anchor
5589 */
5590 public function guessLegacySectionNameFromWikiText( $text ) {
5591 # Strip out wikitext links(they break the anchor)
5592 $text = $this->stripSectionName( $text );
5593 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5594 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5595 }
5596
5597 /**
5598 * Strips a text string of wikitext for use in a section anchor
5599 *
5600 * Accepts a text string and then removes all wikitext from the
5601 * string and leaves only the resultant text (i.e. the result of
5602 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5603 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5604 * to create valid section anchors by mimicing the output of the
5605 * parser when headings are parsed.
5606 *
5607 * @param $text String: text string to be stripped of wikitext
5608 * for use in a Section anchor
5609 * @return string Filtered text string
5610 */
5611 public function stripSectionName( $text ) {
5612 # Strip internal link markup
5613 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5614 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5615
5616 # Strip external link markup
5617 # @todo FIXME: Not tolerant to blank link text
5618 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5619 # on how many empty links there are on the page - need to figure that out.
5620 $text = preg_replace( '/\[(?:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5621
5622 # Parse wikitext quotes (italics & bold)
5623 $text = $this->doQuotes( $text );
5624
5625 # Strip HTML tags
5626 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5627 return $text;
5628 }
5629
5630 /**
5631 * strip/replaceVariables/unstrip for preprocessor regression testing
5632 *
5633 * @param $text string
5634 * @param $title Title
5635 * @param $options ParserOptions
5636 * @param $outputType int
5637 *
5638 * @return string
5639 */
5640 function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5641 $this->startParse( $title, $options, $outputType, true );
5642
5643 $text = $this->replaceVariables( $text );
5644 $text = $this->mStripState->unstripBoth( $text );
5645 $text = Sanitizer::removeHTMLtags( $text );
5646 return $text;
5647 }
5648
5649 /**
5650 * @param $text string
5651 * @param $title Title
5652 * @param $options ParserOptions
5653 * @return string
5654 */
5655 function testPst( $text, Title $title, ParserOptions $options ) {
5656 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5657 }
5658
5659 /**
5660 * @param $text
5661 * @param $title Title
5662 * @param $options ParserOptions
5663 * @return string
5664 */
5665 function testPreprocess( $text, Title $title, ParserOptions $options ) {
5666 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5667 }
5668
5669 /**
5670 * Call a callback function on all regions of the given text that are not
5671 * inside strip markers, and replace those regions with the return value
5672 * of the callback. For example, with input:
5673 *
5674 * aaa<MARKER>bbb
5675 *
5676 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5677 * two strings will be replaced with the value returned by the callback in
5678 * each case.
5679 *
5680 * @param $s string
5681 * @param $callback
5682 *
5683 * @return string
5684 */
5685 function markerSkipCallback( $s, $callback ) {
5686 $i = 0;
5687 $out = '';
5688 while ( $i < strlen( $s ) ) {
5689 $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5690 if ( $markerStart === false ) {
5691 $out .= call_user_func( $callback, substr( $s, $i ) );
5692 break;
5693 } else {
5694 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5695 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5696 if ( $markerEnd === false ) {
5697 $out .= substr( $s, $markerStart );
5698 break;
5699 } else {
5700 $markerEnd += strlen( self::MARKER_SUFFIX );
5701 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5702 $i = $markerEnd;
5703 }
5704 }
5705 }
5706 return $out;
5707 }
5708
5709 /**
5710 * Remove any strip markers found in the given text.
5711 *
5712 * @param $text Input string
5713 * @return string
5714 */
5715 function killMarkers( $text ) {
5716 return $this->mStripState->killMarkers( $text );
5717 }
5718
5719 /**
5720 * Save the parser state required to convert the given half-parsed text to
5721 * HTML. "Half-parsed" in this context means the output of
5722 * recursiveTagParse() or internalParse(). This output has strip markers
5723 * from replaceVariables (extensionSubstitution() etc.), and link
5724 * placeholders from replaceLinkHolders().
5725 *
5726 * Returns an array which can be serialized and stored persistently. This
5727 * array can later be loaded into another parser instance with
5728 * unserializeHalfParsedText(). The text can then be safely incorporated into
5729 * the return value of a parser hook.
5730 *
5731 * @param $text string
5732 *
5733 * @return array
5734 */
5735 function serializeHalfParsedText( $text ) {
5736 wfProfileIn( __METHOD__ );
5737 $data = array(
5738 'text' => $text,
5739 'version' => self::HALF_PARSED_VERSION,
5740 'stripState' => $this->mStripState->getSubState( $text ),
5741 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5742 );
5743 wfProfileOut( __METHOD__ );
5744 return $data;
5745 }
5746
5747 /**
5748 * Load the parser state given in the $data array, which is assumed to
5749 * have been generated by serializeHalfParsedText(). The text contents is
5750 * extracted from the array, and its markers are transformed into markers
5751 * appropriate for the current Parser instance. This transformed text is
5752 * returned, and can be safely included in the return value of a parser
5753 * hook.
5754 *
5755 * If the $data array has been stored persistently, the caller should first
5756 * check whether it is still valid, by calling isValidHalfParsedText().
5757 *
5758 * @param $data array Serialized data
5759 * @return String
5760 */
5761 function unserializeHalfParsedText( $data ) {
5762 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5763 throw new MWException( __METHOD__.': invalid version' );
5764 }
5765
5766 # First, extract the strip state.
5767 $texts = array( $data['text'] );
5768 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5769
5770 # Now renumber links
5771 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5772
5773 # Should be good to go.
5774 return $texts[0];
5775 }
5776
5777 /**
5778 * Returns true if the given array, presumed to be generated by
5779 * serializeHalfParsedText(), is compatible with the current version of the
5780 * parser.
5781 *
5782 * @param $data Array
5783 *
5784 * @return bool
5785 */
5786 function isValidHalfParsedText( $data ) {
5787 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5788 }
5789
5790 /**
5791 * Parsed a width param of imagelink like 300px or 200x300px
5792 *
5793 * @param $value String
5794 *
5795 * @return array
5796 * @since 1.20
5797 */
5798 public function parseWidthParam( $value ) {
5799 $parsedWidthParam = array();
5800 if( $value === '' ) {
5801 return $parsedWidthParam;
5802 }
5803 $m = array();
5804 # (bug 13500) In both cases (width/height and width only),
5805 # permit trailing "px" for backward compatibility.
5806 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5807 $width = intval( $m[1] );
5808 $height = intval( $m[2] );
5809 $parsedWidthParam['width'] = $width;
5810 $parsedWidthParam['height'] = $height;
5811 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5812 $width = intval( $value );
5813 $parsedWidthParam['width'] = $width;
5814 }
5815 return $parsedWidthParam;
5816 }
5817 }
Wiklou, le Wiki du Wiklou