dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Fix fatal from r83823
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * @defgroup Parser Parser
4 *
5 * @file
6 * @ingroup Parser
7 * File for Parser and related classes
8 */
9
10
11 /**
12 * PHP Parser - Processes wiki markup (which uses a more user-friendly
13 * syntax, such as "[[link]]" for making links), and provides a one-way
14 * transformation of that wiki markup it into XHTML output / markup
15 * (which in turn the browser understands, and can display).
16 *
17 * <pre>
18 * There are five main entry points into the Parser class:
19 * parse()
20 * produces HTML output
21 * preSaveTransform().
22 * produces altered wiki markup.
23 * preprocess()
24 * removes HTML comments and expands templates
25 * cleanSig() / cleanSigInSig()
26 * Cleans a signature before saving it to preferences
27 * getSection()
28 * Return the content of a section from an article for section editing
29 * replaceSection()
30 * Replaces a section by number inside an article
31 * getPreloadText()
32 * Removes <noinclude> sections, and <includeonly> tags.
33 *
34 * Globals used:
35 * objects: $wgLang, $wgContLang
36 *
37 * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
38 *
39 * settings:
40 * $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
41 * $wgNamespacesWithSubpages, $wgAllowExternalImages*,
42 * $wgLocaltimezone, $wgAllowSpecialInclusion*,
43 * $wgMaxArticleSize*
44 *
45 * * only within ParserOptions
46 * </pre>
47 *
48 * @ingroup Parser
49 */
50 class Parser {
51 /**
52 * Update this version number when the ParserOutput format
53 * changes in an incompatible way, so the parser cache
54 * can automatically discard old data.
55 */
56 const VERSION = '1.6.4';
57
58 /**
59 * Update this version number when the output of serialiseHalfParsedText()
60 * changes in an incompatible way
61 */
62 const HALF_PARSED_VERSION = 2;
63
64 # Flags for Parser::setFunctionHook
65 # Also available as global constants from Defines.php
66 const SFH_NO_HASH = 1;
67 const SFH_OBJECT_ARGS = 2;
68
69 # Constants needed for external link processing
70 # Everything except bracket, space, or control characters
71 const EXT_LINK_URL_CLASS = '(?:[^\]\[<>"\\x00-\\x20\\x7F]|(?:\[\]))';
72 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
73 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
74
75 # State constants for the definition list colon extraction
76 const COLON_STATE_TEXT = 0;
77 const COLON_STATE_TAG = 1;
78 const COLON_STATE_TAGSTART = 2;
79 const COLON_STATE_CLOSETAG = 3;
80 const COLON_STATE_TAGSLASH = 4;
81 const COLON_STATE_COMMENT = 5;
82 const COLON_STATE_COMMENTDASH = 6;
83 const COLON_STATE_COMMENTDASHDASH = 7;
84
85 # Flags for preprocessToDom
86 const PTD_FOR_INCLUSION = 1;
87
88 # Allowed values for $this->mOutputType
89 # Parameter to startExternalParse().
90 const OT_HTML = 1; # like parse()
91 const OT_WIKI = 2; # like preSaveTransform()
92 const OT_PREPROCESS = 3; # like preprocess()
93 const OT_MSG = 3;
94 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
95
96 # Marker Suffix needs to be accessible staticly.
97 const MARKER_SUFFIX = "-QINU\x7f";
98
99 # Persistent:
100 var $mTagHooks = array();
101 var $mTransparentTagHooks = array();
102 var $mFunctionHooks = array();
103 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
104 var $mFunctionTagHooks = array();
105 var $mStripList = array();
106 var $mDefaultStripList = array();
107 var $mVarCache = array();
108 var $mImageParams = array();
109 var $mImageParamsMagicArray = array();
110 var $mMarkerIndex = 0;
111 var $mFirstCall = true;
112 var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
113 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
114
115 # Cleared with clearState():
116 var $mOutput, $mAutonumber, $mDTopen;
117
118 /**
119 * @var StripState
120 */
121 var $mStripState;
122
123 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
124 var $mLinkHolders, $mLinkID;
125 var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
126 var $mTplExpandCache; # empty-frame expansion cache
127 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
128 var $mExpensiveFunctionCount; # number of expensive parser function calls
129
130 /**
131 * @var User
132 */
133 var $mUser; # User object; only used when doing pre-save transform
134
135 # Temporary
136 # These are variables reset at least once per parse regardless of $clearState
137
138 /**
139 * @var ParserOptions
140 */
141 var $mOptions;
142
143 /**
144 * @var Title
145 */
146 var $mTitle; # Title context, used for self-link rendering and similar things
147 var $mOutputType; # Output type, one of the OT_xxx constants
148 var $ot; # Shortcut alias, see setOutputType()
149 var $mRevisionObject; # The revision object of the specified revision ID
150 var $mRevisionId; # ID to display in {{REVISIONID}} tags
151 var $mRevisionTimestamp; # The timestamp of the specified revision ID
152 var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
153 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
154
155 /**
156 * Constructor
157 *
158 * @public
159 */
160 function __construct( $conf = array() ) {
161 $this->mConf = $conf;
162 $this->mUrlProtocols = wfUrlProtocols();
163 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
164 '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
165 if ( isset( $conf['preprocessorClass'] ) ) {
166 $this->mPreprocessorClass = $conf['preprocessorClass'];
167 } elseif ( extension_loaded( 'domxml' ) ) {
168 # PECL extension that conflicts with the core DOM extension (bug 13770)
169 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
170 $this->mPreprocessorClass = 'Preprocessor_Hash';
171 } elseif ( extension_loaded( 'dom' ) ) {
172 $this->mPreprocessorClass = 'Preprocessor_DOM';
173 } else {
174 $this->mPreprocessorClass = 'Preprocessor_Hash';
175 }
176 }
177
178 /**
179 * Reduce memory usage to reduce the impact of circular references
180 */
181 function __destruct() {
182 if ( isset( $this->mLinkHolders ) ) {
183 $this->mLinkHolders->__destruct();
184 }
185 foreach ( $this as $name => $value ) {
186 unset( $this->$name );
187 }
188 }
189
190 /**
191 * Do various kinds of initialisation on the first call of the parser
192 */
193 function firstCallInit() {
194 if ( !$this->mFirstCall ) {
195 return;
196 }
197 $this->mFirstCall = false;
198
199 wfProfileIn( __METHOD__ );
200
201 CoreParserFunctions::register( $this );
202 CoreTagHooks::register( $this );
203 $this->initialiseVariables();
204
205 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
206 wfProfileOut( __METHOD__ );
207 }
208
209 /**
210 * Clear Parser state
211 *
212 * @private
213 */
214 function clearState() {
215 wfProfileIn( __METHOD__ );
216 if ( $this->mFirstCall ) {
217 $this->firstCallInit();
218 }
219 $this->mOutput = new ParserOutput;
220 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
221 $this->mAutonumber = 0;
222 $this->mLastSection = '';
223 $this->mDTopen = false;
224 $this->mIncludeCount = array();
225 $this->mArgStack = false;
226 $this->mInPre = false;
227 $this->mLinkHolders = new LinkHolderArray( $this );
228 $this->mLinkID = 0;
229 $this->mRevisionObject = $this->mRevisionTimestamp =
230 $this->mRevisionId = $this->mRevisionUser = null;
231 $this->mVarCache = array();
232 $this->mUser = null;
233
234 /**
235 * Prefix for temporary replacement strings for the multipass parser.
236 * \x07 should never appear in input as it's disallowed in XML.
237 * Using it at the front also gives us a little extra robustness
238 * since it shouldn't match when butted up against identifier-like
239 * string constructs.
240 *
241 * Must not consist of all title characters, or else it will change
242 * the behaviour of <nowiki> in a link.
243 */
244 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
245 # Changed to \x7f to allow XML double-parsing -- TS
246 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
247 $this->mStripState = new StripState( $this->mUniqPrefix );
248
249
250 # Clear these on every parse, bug 4549
251 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
252
253 $this->mShowToc = true;
254 $this->mForceTocPosition = false;
255 $this->mIncludeSizes = array(
256 'post-expand' => 0,
257 'arg' => 0,
258 );
259 $this->mPPNodeCount = 0;
260 $this->mDefaultSort = false;
261 $this->mHeadings = array();
262 $this->mDoubleUnderscores = array();
263 $this->mExpensiveFunctionCount = 0;
264
265 # Fix cloning
266 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
267 $this->mPreprocessor = null;
268 }
269
270 wfRunHooks( 'ParserClearState', array( &$this ) );
271 wfProfileOut( __METHOD__ );
272 }
273
274 /**
275 * Convert wikitext to HTML
276 * Do not call this function recursively.
277 *
278 * @param $text String: text we want to parse
279 * @param $title A title object
280 * @param $options ParserOptions
281 * @param $linestart boolean
282 * @param $clearState boolean
283 * @param $revid Int: number to pass in {{REVISIONID}}
284 * @return ParserOutput a ParserOutput
285 */
286 public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
287 /**
288 * First pass--just handle <nowiki> sections, pass the rest off
289 * to internalParse() which does all the real work.
290 */
291
292 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
293 $fname = __METHOD__.'-' . wfGetCaller();
294 wfProfileIn( __METHOD__ );
295 wfProfileIn( $fname );
296
297 $this->startParse( $title, $options, self::OT_HTML, $clearState );
298
299 $oldRevisionId = $this->mRevisionId;
300 $oldRevisionObject = $this->mRevisionObject;
301 $oldRevisionTimestamp = $this->mRevisionTimestamp;
302 $oldRevisionUser = $this->mRevisionUser;
303 if ( $revid !== null ) {
304 $this->mRevisionId = $revid;
305 $this->mRevisionObject = null;
306 $this->mRevisionTimestamp = null;
307 $this->mRevisionUser = null;
308 }
309
310 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
311 # No more strip!
312 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
313 $text = $this->internalParse( $text );
314
315 $text = $this->mStripState->unstripGeneral( $text );
316
317 # Clean up special characters, only run once, next-to-last before doBlockLevels
318 $fixtags = array(
319 # french spaces, last one Guillemet-left
320 # only if there is something before the space
321 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
322 # french spaces, Guillemet-right
323 '/(\\302\\253) /' => '\\1&#160;',
324 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
325 );
326 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
327
328 $text = $this->doBlockLevels( $text, $linestart );
329
330 $this->replaceLinkHolders( $text );
331
332 /**
333 * The page doesn't get language converted if
334 * a) It's disabled
335 * b) Content isn't converted
336 * c) It's a conversion table
337 */
338 if ( !( $wgDisableLangConversion
339 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
340 || $this->mTitle->isConversionTable() ) ) {
341
342 # The position of the convert() call should not be changed. it
343 # assumes that the links are all replaced and the only thing left
344 # is the <nowiki> mark.
345
346 $text = $wgContLang->convert( $text );
347 }
348
349 /**
350 * A converted title will be provided in the output object if title and
351 * content conversion are enabled, the article text does not contain
352 * a conversion-suppressing double-underscore tag, and no
353 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
354 * automatic link conversion.
355 */
356 if ( !( $wgDisableLangConversion
357 || $wgDisableTitleConversion
358 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
359 || isset( $this->mDoubleUnderscores['notitleconvert'] )
360 || $this->mOutput->getDisplayTitle() !== false ) )
361 {
362 $convruletitle = $wgContLang->getConvRuleTitle();
363 if ( $convruletitle ) {
364 $this->mOutput->setTitleText( $convruletitle );
365 } else {
366 $titleText = $wgContLang->convertTitle( $title );
367 $this->mOutput->setTitleText( $titleText );
368 }
369 }
370
371 $text = $this->mStripState->unstripNoWiki( $text );
372
373 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
374
375 $text = $this->replaceTransparentTags( $text );
376 $text = $this->mStripState->unstripGeneral( $text );
377
378 $text = Sanitizer::normalizeCharReferences( $text );
379
380 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
381 $text = MWTidy::tidy( $text );
382 } else {
383 # attempt to sanitize at least some nesting problems
384 # (bug #2702 and quite a few others)
385 $tidyregs = array(
386 # ''Something [http://www.cool.com cool''] -->
387 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
388 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
389 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
390 # fix up an anchor inside another anchor, only
391 # at least for a single single nested link (bug 3695)
392 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
393 '\\1\\2</a>\\3</a>\\1\\4</a>',
394 # fix div inside inline elements- doBlockLevels won't wrap a line which
395 # contains a div, so fix it up here; replace
396 # div with escaped text
397 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
398 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
399 # remove empty italic or bold tag pairs, some
400 # introduced by rules above
401 '/<([bi])><\/\\1>/' => '',
402 );
403
404 $text = preg_replace(
405 array_keys( $tidyregs ),
406 array_values( $tidyregs ),
407 $text );
408 }
409 global $wgExpensiveParserFunctionLimit;
410 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
411 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
412 }
413
414 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
415
416 # Information on include size limits, for the benefit of users who try to skirt them
417 if ( $this->mOptions->getEnableLimitReport() ) {
418 $max = $this->mOptions->getMaxIncludeSize();
419 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
420 $limitReport =
421 "NewPP limit report\n" .
422 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
423 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
424 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
425 $PFreport;
426 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
427 $text .= "\n<!-- \n$limitReport-->\n";
428 }
429 $this->mOutput->setText( $text );
430
431 $this->mRevisionId = $oldRevisionId;
432 $this->mRevisionObject = $oldRevisionObject;
433 $this->mRevisionTimestamp = $oldRevisionTimestamp;
434 $this->mRevisionUser = $oldRevisionUser;
435 wfProfileOut( $fname );
436 wfProfileOut( __METHOD__ );
437
438 return $this->mOutput;
439 }
440
441 /**
442 * Recursive parser entry point that can be called from an extension tag
443 * hook.
444 *
445 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
446 *
447 * @param $text String: text extension wants to have parsed
448 * @param $frame PPFrame: The frame to use for expanding any template variables
449 */
450 function recursiveTagParse( $text, $frame=false ) {
451 wfProfileIn( __METHOD__ );
452 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
453 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
454 $text = $this->internalParse( $text, false, $frame );
455 wfProfileOut( __METHOD__ );
456 return $text;
457 }
458
459 /**
460 * Expand templates and variables in the text, producing valid, static wikitext.
461 * Also removes comments.
462 */
463 function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
464 wfProfileIn( __METHOD__ );
465 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
466 if ( $revid !== null ) {
467 $this->mRevisionId = $revid;
468 }
469 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
470 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
471 $text = $this->replaceVariables( $text );
472 $text = $this->mStripState->unstripBoth( $text );
473 wfProfileOut( __METHOD__ );
474 return $text;
475 }
476
477 /**
478 * Process the wikitext for the ?preload= feature. (bug 5210)
479 *
480 * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
481 * comments, templates, arguments, tags hooks and parser functions are untouched.
482 */
483 public function getPreloadText( $text, Title $title, ParserOptions $options ) {
484 # Parser (re)initialisation
485 $this->startParse( $title, $options, self::OT_PLAIN, true );
486
487 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
488 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
489 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
490 $text = $this->mStripState->unstripBoth( $text );
491 return $text;
492 }
493
494 /**
495 * Get a random string
496 *
497 * @static
498 */
499 static public function getRandomString() {
500 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
501 }
502
503 /**
504 * Set the current user.
505 * Should only be used when doing pre-save transform.
506 *
507 * @param $user Mixed: User object or null (to reset)
508 */
509 function setUser( $user ) {
510 $this->mUser = $user;
511 }
512
513 /**
514 * Accessor for mUniqPrefix.
515 *
516 * @return String
517 */
518 public function uniqPrefix() {
519 if ( !isset( $this->mUniqPrefix ) ) {
520 # @todo Fixme: this is probably *horribly wrong*
521 # LanguageConverter seems to want $wgParser's uniqPrefix, however
522 # if this is called for a parser cache hit, the parser may not
523 # have ever been initialized in the first place.
524 # Not really sure what the heck is supposed to be going on here.
525 return '';
526 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
527 }
528 return $this->mUniqPrefix;
529 }
530
531 /**
532 * Set the context title
533 */
534 function setTitle( $t ) {
535 if ( !$t || $t instanceof FakeTitle ) {
536 $t = Title::newFromText( 'NO TITLE' );
537 }
538
539 if ( strval( $t->getFragment() ) !== '' ) {
540 # Strip the fragment to avoid various odd effects
541 $this->mTitle = clone $t;
542 $this->mTitle->setFragment( '' );
543 } else {
544 $this->mTitle = $t;
545 }
546 }
547
548 /**
549 * Accessor for the Title object
550 *
551 * @return Title object
552 */
553 function getTitle() {
554 return $this->mTitle;
555 }
556
557 /**
558 * Accessor/mutator for the Title object
559 *
560 * @param $x New Title object or null to just get the current one
561 * @return Title object
562 */
563 function Title( $x = null ) {
564 return wfSetVar( $this->mTitle, $x );
565 }
566
567 /**
568 * Set the output type
569 *
570 * @param $ot Integer: new value
571 */
572 function setOutputType( $ot ) {
573 $this->mOutputType = $ot;
574 # Shortcut alias
575 $this->ot = array(
576 'html' => $ot == self::OT_HTML,
577 'wiki' => $ot == self::OT_WIKI,
578 'pre' => $ot == self::OT_PREPROCESS,
579 'plain' => $ot == self::OT_PLAIN,
580 );
581 }
582
583 /**
584 * Accessor/mutator for the output type
585 *
586 * @param $x New value or null to just get the current one
587 * @return Integer
588 */
589 function OutputType( $x = null ) {
590 return wfSetVar( $this->mOutputType, $x );
591 }
592
593 /**
594 * Get the ParserOutput object
595 *
596 * @return ParserOutput object
597 */
598 function getOutput() {
599 return $this->mOutput;
600 }
601
602 /**
603 * Get the ParserOptions object
604 *
605 * @return ParserOptions object
606 */
607 function getOptions() {
608 return $this->mOptions;
609 }
610
611 /**
612 * Accessor/mutator for the ParserOptions object
613 *
614 * @param $x New value or null to just get the current one
615 * @return Current ParserOptions object
616 */
617 function Options( $x = null ) {
618 return wfSetVar( $this->mOptions, $x );
619 }
620
621 function nextLinkID() {
622 return $this->mLinkID++;
623 }
624
625 function setLinkID( $id ) {
626 $this->mLinkID = $id;
627 }
628
629 /**
630 * @return Language
631 */
632 function getFunctionLang() {
633 global $wgLang, $wgContLang;
634
635 $target = $this->mOptions->getTargetLanguage();
636 if ( $target !== null ) {
637 return $target;
638 } else {
639 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
640 }
641 }
642
643 /**
644 * Get a User object either from $this->mUser, if set, or from the
645 * ParserOptions object otherwise
646 *
647 * @return User object
648 */
649 function getUser() {
650 if ( !is_null( $this->mUser ) ) {
651 return $this->mUser;
652 }
653 return $this->mOptions->getUser();
654 }
655
656 /**
657 * Get a preprocessor object
658 *
659 * @return Preprocessor instance
660 */
661 function getPreprocessor() {
662 if ( !isset( $this->mPreprocessor ) ) {
663 $class = $this->mPreprocessorClass;
664 $this->mPreprocessor = new $class( $this );
665 }
666 return $this->mPreprocessor;
667 }
668
669 /**
670 * Replaces all occurrences of HTML-style comments and the given tags
671 * in the text with a random marker and returns the next text. The output
672 * parameter $matches will be an associative array filled with data in
673 * the form:
674 * 'UNIQ-xxxxx' => array(
675 * 'element',
676 * 'tag content',
677 * array( 'param' => 'x' ),
678 * '<element param="x">tag content</element>' ) )
679 *
680 * @param $elements list of element names. Comments are always extracted.
681 * @param $text Source text string.
682 * @param $matches Out parameter, Array: extracted tags
683 * @param $uniq_prefix
684 * @return String: stripped text
685 *
686 * @static
687 */
688 public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
689 static $n = 1;
690 $stripped = '';
691 $matches = array();
692
693 $taglist = implode( '|', $elements );
694 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
695
696 while ( $text != '' ) {
697 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
698 $stripped .= $p[0];
699 if ( count( $p ) < 5 ) {
700 break;
701 }
702 if ( count( $p ) > 5 ) {
703 # comment
704 $element = $p[4];
705 $attributes = '';
706 $close = '';
707 $inside = $p[5];
708 } else {
709 # tag
710 $element = $p[1];
711 $attributes = $p[2];
712 $close = $p[3];
713 $inside = $p[4];
714 }
715
716 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
717 $stripped .= $marker;
718
719 if ( $close === '/>' ) {
720 # Empty element tag, <tag />
721 $content = null;
722 $text = $inside;
723 $tail = null;
724 } else {
725 if ( $element === '!--' ) {
726 $end = '/(-->)/';
727 } else {
728 $end = "/(<\\/$element\\s*>)/i";
729 }
730 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
731 $content = $q[0];
732 if ( count( $q ) < 3 ) {
733 # No end tag -- let it run out to the end of the text.
734 $tail = '';
735 $text = '';
736 } else {
737 $tail = $q[1];
738 $text = $q[2];
739 }
740 }
741
742 $matches[$marker] = array( $element,
743 $content,
744 Sanitizer::decodeTagAttributes( $attributes ),
745 "<$element$attributes$close$content$tail" );
746 }
747 return $stripped;
748 }
749
750 /**
751 * Get a list of strippable XML-like elements
752 */
753 function getStripList() {
754 return $this->mStripList;
755 }
756
757 /**
758 * @deprecated use replaceVariables
759 */
760 function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
761 return $text;
762 }
763
764 /**
765 * Restores pre, math, and other extensions removed by strip()
766 *
767 * always call unstripNoWiki() after this one
768 * @private
769 * @deprecated use $this->mStripState->unstrip()
770 */
771 function unstrip( $text, $state ) {
772 return $state->unstripGeneral( $text );
773 }
774
775 /**
776 * Always call this after unstrip() to preserve the order
777 *
778 * @private
779 * @deprecated use $this->mStripState->unstrip()
780 */
781 function unstripNoWiki( $text, $state ) {
782 return $state->unstripNoWiki( $text );
783 }
784
785 /**
786 * @deprecated use $this->mStripState->unstripBoth()
787 */
788 function unstripForHTML( $text ) {
789 return $this->mStripState->unstripBoth( $text );
790 }
791
792 /**
793 * Add an item to the strip state
794 * Returns the unique tag which must be inserted into the stripped text
795 * The tag will be replaced with the original text in unstrip()
796 *
797 * @private
798 */
799 function insertStripItem( $text ) {
800 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
801 $this->mMarkerIndex++;
802 $this->mStripState->addGeneral( $rnd, $text );
803 return $rnd;
804 }
805
806 /**
807 * Interface with html tidy
808 * @deprecated Use MWTidy::tidy()
809 */
810 public static function tidy( $text ) {
811 wfDeprecated( __METHOD__ );
812 return MWTidy::tidy( $text );
813 }
814
815 /**
816 * parse the wiki syntax used to render tables
817 *
818 * @private
819 */
820 function doTableStuff( $text ) {
821 wfProfileIn( __METHOD__ );
822
823 $lines = StringUtils::explode( "\n", $text );
824 $out = '';
825 $td_history = array(); # Is currently a td tag open?
826 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
827 $tr_history = array(); # Is currently a tr tag open?
828 $tr_attributes = array(); # history of tr attributes
829 $has_opened_tr = array(); # Did this table open a <tr> element?
830 $indent_level = 0; # indent level of the table
831
832 foreach ( $lines as $outLine ) {
833 $line = trim( $outLine );
834
835 if ( $line === '' ) { # empty line, go to next line
836 $out .= $outLine."\n";
837 continue;
838 }
839
840 $first_character = $line[0];
841 $matches = array();
842
843 if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
844 # First check if we are starting a new table
845 $indent_level = strlen( $matches[1] );
846
847 $attributes = $this->mStripState->unstripBoth( $matches[2] );
848 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
849
850 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
851 array_push( $td_history , false );
852 array_push( $last_tag_history , '' );
853 array_push( $tr_history , false );
854 array_push( $tr_attributes , '' );
855 array_push( $has_opened_tr , false );
856 } elseif ( count( $td_history ) == 0 ) {
857 # Don't do any of the following
858 $out .= $outLine."\n";
859 continue;
860 } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
861 # We are ending a table
862 $line = '</table>' . substr( $line , 2 );
863 $last_tag = array_pop( $last_tag_history );
864
865 if ( !array_pop( $has_opened_tr ) ) {
866 $line = "<tr><td></td></tr>{$line}";
867 }
868
869 if ( array_pop( $tr_history ) ) {
870 $line = "</tr>{$line}";
871 }
872
873 if ( array_pop( $td_history ) ) {
874 $line = "</{$last_tag}>{$line}";
875 }
876 array_pop( $tr_attributes );
877 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
878 } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
879 # Now we have a table row
880 $line = preg_replace( '#^\|-+#', '', $line );
881
882 # Whats after the tag is now only attributes
883 $attributes = $this->mStripState->unstripBoth( $line );
884 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
885 array_pop( $tr_attributes );
886 array_push( $tr_attributes, $attributes );
887
888 $line = '';
889 $last_tag = array_pop( $last_tag_history );
890 array_pop( $has_opened_tr );
891 array_push( $has_opened_tr , true );
892
893 if ( array_pop( $tr_history ) ) {
894 $line = '</tr>';
895 }
896
897 if ( array_pop( $td_history ) ) {
898 $line = "</{$last_tag}>{$line}";
899 }
900
901 $outLine = $line;
902 array_push( $tr_history , false );
903 array_push( $td_history , false );
904 array_push( $last_tag_history , '' );
905 } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 ) === '|+' ) {
906 # This might be cell elements, td, th or captions
907 if ( substr( $line , 0 , 2 ) === '|+' ) {
908 $first_character = '+';
909 $line = substr( $line , 1 );
910 }
911
912 $line = substr( $line , 1 );
913
914 if ( $first_character === '!' ) {
915 $line = str_replace( '!!' , '||' , $line );
916 }
917
918 # Split up multiple cells on the same line.
919 # FIXME : This can result in improper nesting of tags processed
920 # by earlier parser steps, but should avoid splitting up eg
921 # attribute values containing literal "||".
922 $cells = StringUtils::explodeMarkup( '||' , $line );
923
924 $outLine = '';
925
926 # Loop through each table cell
927 foreach ( $cells as $cell ) {
928 $previous = '';
929 if ( $first_character !== '+' ) {
930 $tr_after = array_pop( $tr_attributes );
931 if ( !array_pop( $tr_history ) ) {
932 $previous = "<tr{$tr_after}>\n";
933 }
934 array_push( $tr_history , true );
935 array_push( $tr_attributes , '' );
936 array_pop( $has_opened_tr );
937 array_push( $has_opened_tr , true );
938 }
939
940 $last_tag = array_pop( $last_tag_history );
941
942 if ( array_pop( $td_history ) ) {
943 $previous = "</{$last_tag}>\n{$previous}";
944 }
945
946 if ( $first_character === '|' ) {
947 $last_tag = 'td';
948 } elseif ( $first_character === '!' ) {
949 $last_tag = 'th';
950 } elseif ( $first_character === '+' ) {
951 $last_tag = 'caption';
952 } else {
953 $last_tag = '';
954 }
955
956 array_push( $last_tag_history , $last_tag );
957
958 # A cell could contain both parameters and data
959 $cell_data = explode( '|' , $cell , 2 );
960
961 # Bug 553: Note that a '|' inside an invalid link should not
962 # be mistaken as delimiting cell parameters
963 if ( strpos( $cell_data[0], '[[' ) !== false ) {
964 $cell = "{$previous}<{$last_tag}>{$cell}";
965 } elseif ( count( $cell_data ) == 1 ) {
966 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
967 } else {
968 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
969 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
970 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
971 }
972
973 $outLine .= $cell;
974 array_push( $td_history , true );
975 }
976 }
977 $out .= $outLine . "\n";
978 }
979
980 # Closing open td, tr && table
981 while ( count( $td_history ) > 0 ) {
982 if ( array_pop( $td_history ) ) {
983 $out .= "</td>\n";
984 }
985 if ( array_pop( $tr_history ) ) {
986 $out .= "</tr>\n";
987 }
988 if ( !array_pop( $has_opened_tr ) ) {
989 $out .= "<tr><td></td></tr>\n" ;
990 }
991
992 $out .= "</table>\n";
993 }
994
995 # Remove trailing line-ending (b/c)
996 if ( substr( $out, -1 ) === "\n" ) {
997 $out = substr( $out, 0, -1 );
998 }
999
1000 # special case: don't return empty table
1001 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1002 $out = '';
1003 }
1004
1005 wfProfileOut( __METHOD__ );
1006
1007 return $out;
1008 }
1009
1010 /**
1011 * Helper function for parse() that transforms wiki markup into
1012 * HTML. Only called for $mOutputType == self::OT_HTML.
1013 *
1014 * @private
1015 */
1016 function internalParse( $text, $isMain = true, $frame=false ) {
1017 wfProfileIn( __METHOD__ );
1018
1019 $origText = $text;
1020
1021 # Hook to suspend the parser in this state
1022 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1023 wfProfileOut( __METHOD__ );
1024 return $text ;
1025 }
1026
1027 # if $frame is provided, then use $frame for replacing any variables
1028 if ( $frame ) {
1029 # use frame depth to infer how include/noinclude tags should be handled
1030 # depth=0 means this is the top-level document; otherwise it's an included document
1031 if ( !$frame->depth ) {
1032 $flag = 0;
1033 } else {
1034 $flag = Parser::PTD_FOR_INCLUSION;
1035 }
1036 $dom = $this->preprocessToDom( $text, $flag );
1037 $text = $frame->expand( $dom );
1038 } else {
1039 # if $frame is not provided, then use old-style replaceVariables
1040 $text = $this->replaceVariables( $text );
1041 }
1042
1043 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1044 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1045
1046 # Tables need to come after variable replacement for things to work
1047 # properly; putting them before other transformations should keep
1048 # exciting things like link expansions from showing up in surprising
1049 # places.
1050 $text = $this->doTableStuff( $text );
1051
1052 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1053
1054 $text = $this->doDoubleUnderscore( $text );
1055
1056 $text = $this->doHeadings( $text );
1057 if ( $this->mOptions->getUseDynamicDates() ) {
1058 $df = DateFormatter::getInstance();
1059 $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1060 }
1061 $text = $this->replaceInternalLinks( $text );
1062 $text = $this->doAllQuotes( $text );
1063 $text = $this->replaceExternalLinks( $text );
1064
1065 # replaceInternalLinks may sometimes leave behind
1066 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1067 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1068
1069 $text = $this->doMagicLinks( $text );
1070 $text = $this->formatHeadings( $text, $origText, $isMain );
1071
1072 wfProfileOut( __METHOD__ );
1073 return $text;
1074 }
1075
1076 /**
1077 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1078 * magic external links.
1079 *
1080 * DML
1081 * @private
1082 */
1083 function doMagicLinks( $text ) {
1084 wfProfileIn( __METHOD__ );
1085 $prots = $this->mUrlProtocols;
1086 $urlChar = self::EXT_LINK_URL_CLASS;
1087 $text = preg_replace_callback(
1088 '!(?: # Start cases
1089 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1090 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1091 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1092 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1093 ISBN\s+(\b # m[5]: ISBN, capture number
1094 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1095 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1096 [0-9Xx] # check digit
1097 \b)
1098 )!x', array( &$this, 'magicLinkCallback' ), $text );
1099 wfProfileOut( __METHOD__ );
1100 return $text;
1101 }
1102
1103 function magicLinkCallback( $m ) {
1104 if ( isset( $m[1] ) && $m[1] !== '' ) {
1105 # Skip anchor
1106 return $m[0];
1107 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1108 # Skip HTML element
1109 return $m[0];
1110 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1111 # Free external link
1112 return $this->makeFreeExternalLink( $m[0] );
1113 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1114 # RFC or PMID
1115 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1116 $keyword = 'RFC';
1117 $urlmsg = 'rfcurl';
1118 $CssClass = 'mw-magiclink-rfc';
1119 $id = $m[4];
1120 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1121 $keyword = 'PMID';
1122 $urlmsg = 'pubmedurl';
1123 $CssClass = 'mw-magiclink-pmid';
1124 $id = $m[4];
1125 } else {
1126 throw new MWException( __METHOD__.': unrecognised match type "' .
1127 substr( $m[0], 0, 20 ) . '"' );
1128 }
1129 $url = wfMsgForContent( $urlmsg, $id );
1130 $sk = $this->mOptions->getSkin( $this->mTitle );
1131 return $sk->makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1132 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1133 # ISBN
1134 $isbn = $m[5];
1135 $num = strtr( $isbn, array(
1136 '-' => '',
1137 ' ' => '',
1138 'x' => 'X',
1139 ));
1140 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1141 return'<a href="' .
1142 $titleObj->escapeLocalUrl() .
1143 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1144 } else {
1145 return $m[0];
1146 }
1147 }
1148
1149 /**
1150 * Make a free external link, given a user-supplied URL
1151 * @return HTML
1152 * @private
1153 */
1154 function makeFreeExternalLink( $url ) {
1155 global $wgContLang;
1156 wfProfileIn( __METHOD__ );
1157
1158 $sk = $this->mOptions->getSkin( $this->mTitle );
1159 $trail = '';
1160
1161 # The characters '<' and '>' (which were escaped by
1162 # removeHTMLtags()) should not be included in
1163 # URLs, per RFC 2396.
1164 $m2 = array();
1165 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1166 $trail = substr( $url, $m2[0][1] ) . $trail;
1167 $url = substr( $url, 0, $m2[0][1] );
1168 }
1169
1170 # Move trailing punctuation to $trail
1171 $sep = ',;\.:!?';
1172 # If there is no left bracket, then consider right brackets fair game too
1173 if ( strpos( $url, '(' ) === false ) {
1174 $sep .= ')';
1175 }
1176
1177 $numSepChars = strspn( strrev( $url ), $sep );
1178 if ( $numSepChars ) {
1179 $trail = substr( $url, -$numSepChars ) . $trail;
1180 $url = substr( $url, 0, -$numSepChars );
1181 }
1182
1183 $url = Sanitizer::cleanUrl( $url );
1184
1185 # Is this an external image?
1186 $text = $this->maybeMakeExternalImage( $url );
1187 if ( $text === false ) {
1188 # Not an image, make a link
1189 $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1190 $this->getExternalLinkAttribs( $url ) );
1191 # Register it in the output object...
1192 # Replace unnecessary URL escape codes with their equivalent characters
1193 $pasteurized = self::replaceUnusualEscapes( $url );
1194 $this->mOutput->addExternalLink( $pasteurized );
1195 }
1196 wfProfileOut( __METHOD__ );
1197 return $text . $trail;
1198 }
1199
1200
1201 /**
1202 * Parse headers and return html
1203 *
1204 * @private
1205 */
1206 function doHeadings( $text ) {
1207 wfProfileIn( __METHOD__ );
1208 for ( $i = 6; $i >= 1; --$i ) {
1209 $h = str_repeat( '=', $i );
1210 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1211 "<h$i>\\1</h$i>", $text );
1212 }
1213 wfProfileOut( __METHOD__ );
1214 return $text;
1215 }
1216
1217 /**
1218 * Replace single quotes with HTML markup
1219 * @private
1220 * @return string the altered text
1221 */
1222 function doAllQuotes( $text ) {
1223 wfProfileIn( __METHOD__ );
1224 $outtext = '';
1225 $lines = StringUtils::explode( "\n", $text );
1226 foreach ( $lines as $line ) {
1227 $outtext .= $this->doQuotes( $line ) . "\n";
1228 }
1229 $outtext = substr( $outtext, 0,-1 );
1230 wfProfileOut( __METHOD__ );
1231 return $outtext;
1232 }
1233
1234 /**
1235 * Helper function for doAllQuotes()
1236 */
1237 public function doQuotes( $text ) {
1238 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1239 if ( count( $arr ) == 1 ) {
1240 return $text;
1241 } else {
1242 # First, do some preliminary work. This may shift some apostrophes from
1243 # being mark-up to being text. It also counts the number of occurrences
1244 # of bold and italics mark-ups.
1245 $numbold = 0;
1246 $numitalics = 0;
1247 for ( $i = 0; $i < count( $arr ); $i++ ) {
1248 if ( ( $i % 2 ) == 1 ) {
1249 # If there are ever four apostrophes, assume the first is supposed to
1250 # be text, and the remaining three constitute mark-up for bold text.
1251 if ( strlen( $arr[$i] ) == 4 ) {
1252 $arr[$i-1] .= "'";
1253 $arr[$i] = "'''";
1254 } elseif ( strlen( $arr[$i] ) > 5 ) {
1255 # If there are more than 5 apostrophes in a row, assume they're all
1256 # text except for the last 5.
1257 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1258 $arr[$i] = "'''''";
1259 }
1260 # Count the number of occurrences of bold and italics mark-ups.
1261 # We are not counting sequences of five apostrophes.
1262 if ( strlen( $arr[$i] ) == 2 ) {
1263 $numitalics++;
1264 } elseif ( strlen( $arr[$i] ) == 3 ) {
1265 $numbold++;
1266 } elseif ( strlen( $arr[$i] ) == 5 ) {
1267 $numitalics++;
1268 $numbold++;
1269 }
1270 }
1271 }
1272
1273 # If there is an odd number of both bold and italics, it is likely
1274 # that one of the bold ones was meant to be an apostrophe followed
1275 # by italics. Which one we cannot know for certain, but it is more
1276 # likely to be one that has a single-letter word before it.
1277 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1278 $i = 0;
1279 $firstsingleletterword = -1;
1280 $firstmultiletterword = -1;
1281 $firstspace = -1;
1282 foreach ( $arr as $r ) {
1283 if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1284 $x1 = substr( $arr[$i-1], -1 );
1285 $x2 = substr( $arr[$i-1], -2, 1 );
1286 if ( $x1 === ' ' ) {
1287 if ( $firstspace == -1 ) {
1288 $firstspace = $i;
1289 }
1290 } elseif ( $x2 === ' ') {
1291 if ( $firstsingleletterword == -1 ) {
1292 $firstsingleletterword = $i;
1293 }
1294 } else {
1295 if ( $firstmultiletterword == -1 ) {
1296 $firstmultiletterword = $i;
1297 }
1298 }
1299 }
1300 $i++;
1301 }
1302
1303 # If there is a single-letter word, use it!
1304 if ( $firstsingleletterword > -1 ) {
1305 $arr[$firstsingleletterword] = "''";
1306 $arr[$firstsingleletterword-1] .= "'";
1307 } elseif ( $firstmultiletterword > -1 ) {
1308 # If not, but there's a multi-letter word, use that one.
1309 $arr[$firstmultiletterword] = "''";
1310 $arr[$firstmultiletterword-1] .= "'";
1311 } elseif ( $firstspace > -1 ) {
1312 # ... otherwise use the first one that has neither.
1313 # (notice that it is possible for all three to be -1 if, for example,
1314 # there is only one pentuple-apostrophe in the line)
1315 $arr[$firstspace] = "''";
1316 $arr[$firstspace-1] .= "'";
1317 }
1318 }
1319
1320 # Now let's actually convert our apostrophic mush to HTML!
1321 $output = '';
1322 $buffer = '';
1323 $state = '';
1324 $i = 0;
1325 foreach ( $arr as $r ) {
1326 if ( ( $i % 2 ) == 0 ) {
1327 if ( $state === 'both' ) {
1328 $buffer .= $r;
1329 } else {
1330 $output .= $r;
1331 }
1332 } else {
1333 if ( strlen( $r ) == 2 ) {
1334 if ( $state === 'i' ) {
1335 $output .= '</i>'; $state = '';
1336 } elseif ( $state === 'bi' ) {
1337 $output .= '</i>'; $state = 'b';
1338 } elseif ( $state === 'ib' ) {
1339 $output .= '</b></i><b>'; $state = 'b';
1340 } elseif ( $state === 'both' ) {
1341 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1342 } else { # $state can be 'b' or ''
1343 $output .= '<i>'; $state .= 'i';
1344 }
1345 } elseif ( strlen( $r ) == 3 ) {
1346 if ( $state === 'b' ) {
1347 $output .= '</b>'; $state = '';
1348 } elseif ( $state === 'bi' ) {
1349 $output .= '</i></b><i>'; $state = 'i';
1350 } elseif ( $state === 'ib' ) {
1351 $output .= '</b>'; $state = 'i';
1352 } elseif ( $state === 'both' ) {
1353 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1354 } else { # $state can be 'i' or ''
1355 $output .= '<b>'; $state .= 'b';
1356 }
1357 } elseif ( strlen( $r ) == 5 ) {
1358 if ( $state === 'b' ) {
1359 $output .= '</b><i>'; $state = 'i';
1360 } elseif ( $state === 'i' ) {
1361 $output .= '</i><b>'; $state = 'b';
1362 } elseif ( $state === 'bi' ) {
1363 $output .= '</i></b>'; $state = '';
1364 } elseif ( $state === 'ib' ) {
1365 $output .= '</b></i>'; $state = '';
1366 } elseif ( $state === 'both' ) {
1367 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1368 } else { # ($state == '')
1369 $buffer = ''; $state = 'both';
1370 }
1371 }
1372 }
1373 $i++;
1374 }
1375 # Now close all remaining tags. Notice that the order is important.
1376 if ( $state === 'b' || $state === 'ib' ) {
1377 $output .= '</b>';
1378 }
1379 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1380 $output .= '</i>';
1381 }
1382 if ( $state === 'bi' ) {
1383 $output .= '</b>';
1384 }
1385 # There might be lonely ''''', so make sure we have a buffer
1386 if ( $state === 'both' && $buffer ) {
1387 $output .= '<b><i>'.$buffer.'</i></b>';
1388 }
1389 return $output;
1390 }
1391 }
1392
1393 /**
1394 * Replace external links (REL)
1395 *
1396 * Note: this is all very hackish and the order of execution matters a lot.
1397 * Make sure to run maintenance/parserTests.php if you change this code.
1398 *
1399 * @private
1400 */
1401 function replaceExternalLinks( $text ) {
1402 global $wgContLang;
1403 wfProfileIn( __METHOD__ );
1404
1405 $sk = $this->mOptions->getSkin( $this->mTitle );
1406
1407 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1408 $s = array_shift( $bits );
1409
1410 $i = 0;
1411 while ( $i<count( $bits ) ) {
1412 $url = $bits[$i++];
1413 $protocol = $bits[$i++];
1414 $text = $bits[$i++];
1415 $trail = $bits[$i++];
1416
1417 # The characters '<' and '>' (which were escaped by
1418 # removeHTMLtags()) should not be included in
1419 # URLs, per RFC 2396.
1420 $m2 = array();
1421 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1422 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1423 $url = substr( $url, 0, $m2[0][1] );
1424 }
1425
1426 # If the link text is an image URL, replace it with an <img> tag
1427 # This happened by accident in the original parser, but some people used it extensively
1428 $img = $this->maybeMakeExternalImage( $text );
1429 if ( $img !== false ) {
1430 $text = $img;
1431 }
1432
1433 $dtrail = '';
1434
1435 # Set linktype for CSS - if URL==text, link is essentially free
1436 $linktype = ( $text === $url ) ? 'free' : 'text';
1437
1438 # No link text, e.g. [http://domain.tld/some.link]
1439 if ( $text == '' ) {
1440 # Autonumber if allowed. See bug #5918
1441 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1442 $langObj = $this->getFunctionLang();
1443 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1444 $linktype = 'autonumber';
1445 } else {
1446 # Otherwise just use the URL
1447 $text = htmlspecialchars( $url );
1448 $linktype = 'free';
1449 }
1450 } else {
1451 # Have link text, e.g. [http://domain.tld/some.link text]s
1452 # Check for trail
1453 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1454 }
1455
1456 $text = $wgContLang->markNoConversion( $text );
1457
1458 $url = Sanitizer::cleanUrl( $url );
1459
1460 # Use the encoded URL
1461 # This means that users can paste URLs directly into the text
1462 # Funny characters like ö aren't valid in URLs anyway
1463 # This was changed in August 2004
1464 $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1465 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1466
1467 # Register link in the output object.
1468 # Replace unnecessary URL escape codes with the referenced character
1469 # This prevents spammers from hiding links from the filters
1470 $pasteurized = self::replaceUnusualEscapes( $url );
1471 $this->mOutput->addExternalLink( $pasteurized );
1472 }
1473
1474 wfProfileOut( __METHOD__ );
1475 return $s;
1476 }
1477
1478 /**
1479 * Get an associative array of additional HTML attributes appropriate for a
1480 * particular external link. This currently may include rel => nofollow
1481 * (depending on configuration, namespace, and the URL's domain) and/or a
1482 * target attribute (depending on configuration).
1483 *
1484 * @param $url String: optional URL, to extract the domain from for rel =>
1485 * nofollow if appropriate
1486 * @return Array: associative array of HTML attributes
1487 */
1488 function getExternalLinkAttribs( $url = false ) {
1489 $attribs = array();
1490 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1491 $ns = $this->mTitle->getNamespace();
1492 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1493 $attribs['rel'] = 'nofollow';
1494
1495 global $wgNoFollowDomainExceptions;
1496 if ( $wgNoFollowDomainExceptions ) {
1497 $bits = wfParseUrl( $url );
1498 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1499 foreach ( $wgNoFollowDomainExceptions as $domain ) {
1500 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1501 unset( $attribs['rel'] );
1502 break;
1503 }
1504 }
1505 }
1506 }
1507 }
1508 if ( $this->mOptions->getExternalLinkTarget() ) {
1509 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1510 }
1511 return $attribs;
1512 }
1513
1514
1515 /**
1516 * Replace unusual URL escape codes with their equivalent characters
1517 *
1518 * @param $url String
1519 * @return String
1520 *
1521 * @todo This can merge genuinely required bits in the path or query string,
1522 * breaking legit URLs. A proper fix would treat the various parts of
1523 * the URL differently; as a workaround, just use the output for
1524 * statistical records, not for actual linking/output.
1525 */
1526 static function replaceUnusualEscapes( $url ) {
1527 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1528 array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1529 }
1530
1531 /**
1532 * Callback function used in replaceUnusualEscapes().
1533 * Replaces unusual URL escape codes with their equivalent character
1534 */
1535 private static function replaceUnusualEscapesCallback( $matches ) {
1536 $char = urldecode( $matches[0] );
1537 $ord = ord( $char );
1538 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1539 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1540 # No, shouldn't be escaped
1541 return $char;
1542 } else {
1543 # Yes, leave it escaped
1544 return $matches[0];
1545 }
1546 }
1547
1548 /**
1549 * make an image if it's allowed, either through the global
1550 * option, through the exception, or through the on-wiki whitelist
1551 * @private
1552 */
1553 function maybeMakeExternalImage( $url ) {
1554 $sk = $this->mOptions->getSkin( $this->mTitle );
1555 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1556 $imagesexception = !empty( $imagesfrom );
1557 $text = false;
1558 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1559 if ( $imagesexception && is_array( $imagesfrom ) ) {
1560 $imagematch = false;
1561 foreach ( $imagesfrom as $match ) {
1562 if ( strpos( $url, $match ) === 0 ) {
1563 $imagematch = true;
1564 break;
1565 }
1566 }
1567 } elseif ( $imagesexception ) {
1568 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1569 } else {
1570 $imagematch = false;
1571 }
1572 if ( $this->mOptions->getAllowExternalImages()
1573 || ( $imagesexception && $imagematch ) ) {
1574 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1575 # Image found
1576 $text = $sk->makeExternalImage( $url );
1577 }
1578 }
1579 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1580 && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1581 $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1582 foreach ( $whitelist as $entry ) {
1583 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1584 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1585 continue;
1586 }
1587 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1588 # Image matches a whitelist entry
1589 $text = $sk->makeExternalImage( $url );
1590 break;
1591 }
1592 }
1593 }
1594 return $text;
1595 }
1596
1597 /**
1598 * Process [[ ]] wikilinks
1599 * @return String: processed text
1600 *
1601 * @private
1602 */
1603 function replaceInternalLinks( $s ) {
1604 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1605 return $s;
1606 }
1607
1608 /**
1609 * Process [[ ]] wikilinks (RIL)
1610 * @return LinkHolderArray
1611 *
1612 * @private
1613 */
1614 function replaceInternalLinks2( &$s ) {
1615 global $wgContLang;
1616
1617 wfProfileIn( __METHOD__ );
1618
1619 wfProfileIn( __METHOD__.'-setup' );
1620 static $tc = FALSE, $e1, $e1_img;
1621 # the % is needed to support urlencoded titles as well
1622 if ( !$tc ) {
1623 $tc = Title::legalChars() . '#%';
1624 # Match a link having the form [[namespace:link|alternate]]trail
1625 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1626 # Match cases where there is no "]]", which might still be images
1627 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1628 }
1629
1630 $sk = $this->mOptions->getSkin( $this->mTitle );
1631 $holders = new LinkHolderArray( $this );
1632
1633 # split the entire text string on occurences of [[
1634 $a = StringUtils::explode( '[[', ' ' . $s );
1635 # get the first element (all text up to first [[), and remove the space we added
1636 $s = $a->current();
1637 $a->next();
1638 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1639 $s = substr( $s, 1 );
1640
1641 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1642 $e2 = null;
1643 if ( $useLinkPrefixExtension ) {
1644 # Match the end of a line for a word that's not followed by whitespace,
1645 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1646 $e2 = wfMsgForContent( 'linkprefix' );
1647 }
1648
1649 if ( is_null( $this->mTitle ) ) {
1650 wfProfileOut( __METHOD__.'-setup' );
1651 wfProfileOut( __METHOD__ );
1652 throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1653 }
1654 $nottalk = !$this->mTitle->isTalkPage();
1655
1656 if ( $useLinkPrefixExtension ) {
1657 $m = array();
1658 if ( preg_match( $e2, $s, $m ) ) {
1659 $first_prefix = $m[2];
1660 } else {
1661 $first_prefix = false;
1662 }
1663 } else {
1664 $prefix = '';
1665 }
1666
1667 if ( $wgContLang->hasVariants() ) {
1668 $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1669 } else {
1670 $selflink = array( $this->mTitle->getPrefixedText() );
1671 }
1672 $useSubpages = $this->areSubpagesAllowed();
1673 wfProfileOut( __METHOD__.'-setup' );
1674
1675 # Loop for each link
1676 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1677 # Check for excessive memory usage
1678 if ( $holders->isBig() ) {
1679 # Too big
1680 # Do the existence check, replace the link holders and clear the array
1681 $holders->replace( $s );
1682 $holders->clear();
1683 }
1684
1685 if ( $useLinkPrefixExtension ) {
1686 wfProfileIn( __METHOD__.'-prefixhandling' );
1687 if ( preg_match( $e2, $s, $m ) ) {
1688 $prefix = $m[2];
1689 $s = $m[1];
1690 } else {
1691 $prefix='';
1692 }
1693 # first link
1694 if ( $first_prefix ) {
1695 $prefix = $first_prefix;
1696 $first_prefix = false;
1697 }
1698 wfProfileOut( __METHOD__.'-prefixhandling' );
1699 }
1700
1701 $might_be_img = false;
1702
1703 wfProfileIn( __METHOD__."-e1" );
1704 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1705 $text = $m[2];
1706 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1707 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1708 # the real problem is with the $e1 regex
1709 # See bug 1300.
1710 #
1711 # Still some problems for cases where the ] is meant to be outside punctuation,
1712 # and no image is in sight. See bug 2095.
1713 #
1714 if ( $text !== '' &&
1715 substr( $m[3], 0, 1 ) === ']' &&
1716 strpos( $text, '[' ) !== false
1717 )
1718 {
1719 $text .= ']'; # so that replaceExternalLinks($text) works later
1720 $m[3] = substr( $m[3], 1 );
1721 }
1722 # fix up urlencoded title texts
1723 if ( strpos( $m[1], '%' ) !== false ) {
1724 # Should anchors '#' also be rejected?
1725 $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1726 }
1727 $trail = $m[3];
1728 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1729 $might_be_img = true;
1730 $text = $m[2];
1731 if ( strpos( $m[1], '%' ) !== false ) {
1732 $m[1] = rawurldecode( $m[1] );
1733 }
1734 $trail = "";
1735 } else { # Invalid form; output directly
1736 $s .= $prefix . '[[' . $line ;
1737 wfProfileOut( __METHOD__."-e1" );
1738 continue;
1739 }
1740 wfProfileOut( __METHOD__."-e1" );
1741 wfProfileIn( __METHOD__."-misc" );
1742
1743 # Don't allow internal links to pages containing
1744 # PROTO: where PROTO is a valid URL protocol; these
1745 # should be external links.
1746 if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1747 $s .= $prefix . '[[' . $line ;
1748 wfProfileOut( __METHOD__."-misc" );
1749 continue;
1750 }
1751
1752 # Make subpage if necessary
1753 if ( $useSubpages ) {
1754 $link = $this->maybeDoSubpageLink( $m[1], $text );
1755 } else {
1756 $link = $m[1];
1757 }
1758
1759 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1760 if ( !$noforce ) {
1761 # Strip off leading ':'
1762 $link = substr( $link, 1 );
1763 }
1764
1765 wfProfileOut( __METHOD__."-misc" );
1766 wfProfileIn( __METHOD__."-title" );
1767 $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1768 if ( $nt === null ) {
1769 $s .= $prefix . '[[' . $line;
1770 wfProfileOut( __METHOD__."-title" );
1771 continue;
1772 }
1773
1774 $ns = $nt->getNamespace();
1775 $iw = $nt->getInterWiki();
1776 wfProfileOut( __METHOD__."-title" );
1777
1778 if ( $might_be_img ) { # if this is actually an invalid link
1779 wfProfileIn( __METHOD__."-might_be_img" );
1780 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1781 $found = false;
1782 while ( true ) {
1783 # look at the next 'line' to see if we can close it there
1784 $a->next();
1785 $next_line = $a->current();
1786 if ( $next_line === false || $next_line === null ) {
1787 break;
1788 }
1789 $m = explode( ']]', $next_line, 3 );
1790 if ( count( $m ) == 3 ) {
1791 # the first ]] closes the inner link, the second the image
1792 $found = true;
1793 $text .= "[[{$m[0]}]]{$m[1]}";
1794 $trail = $m[2];
1795 break;
1796 } elseif ( count( $m ) == 2 ) {
1797 # if there's exactly one ]] that's fine, we'll keep looking
1798 $text .= "[[{$m[0]}]]{$m[1]}";
1799 } else {
1800 # if $next_line is invalid too, we need look no further
1801 $text .= '[[' . $next_line;
1802 break;
1803 }
1804 }
1805 if ( !$found ) {
1806 # we couldn't find the end of this imageLink, so output it raw
1807 # but don't ignore what might be perfectly normal links in the text we've examined
1808 $holders->merge( $this->replaceInternalLinks2( $text ) );
1809 $s .= "{$prefix}[[$link|$text";
1810 # note: no $trail, because without an end, there *is* no trail
1811 wfProfileOut( __METHOD__."-might_be_img" );
1812 continue;
1813 }
1814 } else { # it's not an image, so output it raw
1815 $s .= "{$prefix}[[$link|$text";
1816 # note: no $trail, because without an end, there *is* no trail
1817 wfProfileOut( __METHOD__."-might_be_img" );
1818 continue;
1819 }
1820 wfProfileOut( __METHOD__."-might_be_img" );
1821 }
1822
1823 $wasblank = ( $text == '' );
1824 if ( $wasblank ) {
1825 $text = $link;
1826 } else {
1827 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1828 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1829 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1830 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1831 $text = $this->doQuotes( $text );
1832 }
1833
1834 # Link not escaped by : , create the various objects
1835 if ( $noforce ) {
1836
1837 # Interwikis
1838 wfProfileIn( __METHOD__."-interwiki" );
1839 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1840 $this->mOutput->addLanguageLink( $nt->getFullText() );
1841 $s = rtrim( $s . $prefix );
1842 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1843 wfProfileOut( __METHOD__."-interwiki" );
1844 continue;
1845 }
1846 wfProfileOut( __METHOD__."-interwiki" );
1847
1848 if ( $ns == NS_FILE ) {
1849 wfProfileIn( __METHOD__."-image" );
1850 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1851 if ( $wasblank ) {
1852 # if no parameters were passed, $text
1853 # becomes something like "File:Foo.png",
1854 # which we don't want to pass on to the
1855 # image generator
1856 $text = '';
1857 } else {
1858 # recursively parse links inside the image caption
1859 # actually, this will parse them in any other parameters, too,
1860 # but it might be hard to fix that, and it doesn't matter ATM
1861 $text = $this->replaceExternalLinks( $text );
1862 $holders->merge( $this->replaceInternalLinks2( $text ) );
1863 }
1864 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1865 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1866 } else {
1867 $s .= $prefix . $trail;
1868 }
1869 $this->mOutput->addImage( $nt->getDBkey() );
1870 wfProfileOut( __METHOD__."-image" );
1871 continue;
1872
1873 }
1874
1875 if ( $ns == NS_CATEGORY ) {
1876 wfProfileIn( __METHOD__."-category" );
1877 $s = rtrim( $s . "\n" ); # bug 87
1878
1879 if ( $wasblank ) {
1880 $sortkey = $this->getDefaultSort();
1881 } else {
1882 $sortkey = $text;
1883 }
1884 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1885 $sortkey = str_replace( "\n", '', $sortkey );
1886 $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1887 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1888
1889 /**
1890 * Strip the whitespace Category links produce, see bug 87
1891 * @todo We might want to use trim($tmp, "\n") here.
1892 */
1893 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1894
1895 wfProfileOut( __METHOD__."-category" );
1896 continue;
1897 }
1898 }
1899
1900 # Self-link checking
1901 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1902 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1903 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1904 continue;
1905 }
1906 }
1907
1908 # NS_MEDIA is a pseudo-namespace for linking directly to a file
1909 # FIXME: Should do batch file existence checks, see comment below
1910 if ( $ns == NS_MEDIA ) {
1911 wfProfileIn( __METHOD__."-media" );
1912 # Give extensions a chance to select the file revision for us
1913 $skip = $time = false;
1914 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1915 if ( $skip ) {
1916 $link = $sk->link( $nt );
1917 } else {
1918 $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1919 }
1920 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1921 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1922 $this->mOutput->addImage( $nt->getDBkey() );
1923 wfProfileOut( __METHOD__."-media" );
1924 continue;
1925 }
1926
1927 wfProfileIn( __METHOD__."-always_known" );
1928 # Some titles, such as valid special pages or files in foreign repos, should
1929 # be shown as bluelinks even though they're not included in the page table
1930 #
1931 # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1932 # batch file existence checks for NS_FILE and NS_MEDIA
1933 if ( $iw == '' && $nt->isAlwaysKnown() ) {
1934 $this->mOutput->addLink( $nt );
1935 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1936 } else {
1937 # Links will be added to the output link list after checking
1938 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1939 }
1940 wfProfileOut( __METHOD__."-always_known" );
1941 }
1942 wfProfileOut( __METHOD__ );
1943 return $holders;
1944 }
1945
1946 /**
1947 * Make a link placeholder. The text returned can be later resolved to a real link with
1948 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1949 * parsing of interwiki links, and secondly to allow all existence checks and
1950 * article length checks (for stub links) to be bundled into a single query.
1951 *
1952 * @deprecated
1953 */
1954 function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
1955 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1956 }
1957
1958 /**
1959 * Render a forced-blue link inline; protect against double expansion of
1960 * URLs if we're in a mode that prepends full URL prefixes to internal links.
1961 * Since this little disaster has to split off the trail text to avoid
1962 * breaking URLs in the following text without breaking trails on the
1963 * wiki links, it's been made into a horrible function.
1964 *
1965 * @param $nt Title
1966 * @param $text String
1967 * @param $query Array or String
1968 * @param $trail String
1969 * @param $prefix String
1970 * @return String: HTML-wikitext mix oh yuck
1971 */
1972 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
1973 list( $inside, $trail ) = Linker::splitTrail( $trail );
1974
1975 if ( is_string( $query ) ) {
1976 $query = wfCgiToArray( $query );
1977 }
1978 if ( $text == '' ) {
1979 $text = htmlspecialchars( $nt->getPrefixedText() );
1980 }
1981
1982 $sk = $this->mOptions->getSkin( $this->mTitle );
1983 $link = $sk->linkKnown( $nt, "$prefix$text$inside", array(), $query );
1984
1985 return $this->armorLinks( $link ) . $trail;
1986 }
1987
1988 /**
1989 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1990 * going to go through further parsing steps before inline URL expansion.
1991 *
1992 * Not needed quite as much as it used to be since free links are a bit
1993 * more sensible these days. But bracketed links are still an issue.
1994 *
1995 * @param $text String: more-or-less HTML
1996 * @return String: less-or-more HTML with NOPARSE bits
1997 */
1998 function armorLinks( $text ) {
1999 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2000 "{$this->mUniqPrefix}NOPARSE$1", $text );
2001 }
2002
2003 /**
2004 * Return true if subpage links should be expanded on this page.
2005 * @return Boolean
2006 */
2007 function areSubpagesAllowed() {
2008 # Some namespaces don't allow subpages
2009 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2010 }
2011
2012 /**
2013 * Handle link to subpage if necessary
2014 *
2015 * @param $target String: the source of the link
2016 * @param &$text String: the link text, modified as necessary
2017 * @return string the full name of the link
2018 * @private
2019 */
2020 function maybeDoSubpageLink( $target, &$text ) {
2021 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2022 }
2023
2024 /**#@+
2025 * Used by doBlockLevels()
2026 * @private
2027 */
2028 function closeParagraph() {
2029 $result = '';
2030 if ( $this->mLastSection != '' ) {
2031 $result = '</' . $this->mLastSection . ">\n";
2032 }
2033 $this->mInPre = false;
2034 $this->mLastSection = '';
2035 return $result;
2036 }
2037
2038 /**
2039 * getCommon() returns the length of the longest common substring
2040 * of both arguments, starting at the beginning of both.
2041 * @private
2042 */
2043 function getCommon( $st1, $st2 ) {
2044 $fl = strlen( $st1 );
2045 $shorter = strlen( $st2 );
2046 if ( $fl < $shorter ) {
2047 $shorter = $fl;
2048 }
2049
2050 for ( $i = 0; $i < $shorter; ++$i ) {
2051 if ( $st1{$i} != $st2{$i} ) {
2052 break;
2053 }
2054 }
2055 return $i;
2056 }
2057
2058 /**
2059 * These next three functions open, continue, and close the list
2060 * element appropriate to the prefix character passed into them.
2061 * @private
2062 */
2063 function openList( $char ) {
2064 $result = $this->closeParagraph();
2065
2066 if ( '*' === $char ) {
2067 $result .= '<ul><li>';
2068 } elseif ( '#' === $char ) {
2069 $result .= '<ol><li>';
2070 } elseif ( ':' === $char ) {
2071 $result .= '<dl><dd>';
2072 } elseif ( ';' === $char ) {
2073 $result .= '<dl><dt>';
2074 $this->mDTopen = true;
2075 } else {
2076 $result = '<!-- ERR 1 -->';
2077 }
2078
2079 return $result;
2080 }
2081
2082 /**
2083 * TODO: document
2084 * @param $char String
2085 * @private
2086 */
2087 function nextItem( $char ) {
2088 if ( '*' === $char || '#' === $char ) {
2089 return '</li><li>';
2090 } elseif ( ':' === $char || ';' === $char ) {
2091 $close = '</dd>';
2092 if ( $this->mDTopen ) {
2093 $close = '</dt>';
2094 }
2095 if ( ';' === $char ) {
2096 $this->mDTopen = true;
2097 return $close . '<dt>';
2098 } else {
2099 $this->mDTopen = false;
2100 return $close . '<dd>';
2101 }
2102 }
2103 return '<!-- ERR 2 -->';
2104 }
2105
2106 /**
2107 * TODO: document
2108 * @param $char String
2109 * @private
2110 */
2111 function closeList( $char ) {
2112 if ( '*' === $char ) {
2113 $text = '</li></ul>';
2114 } elseif ( '#' === $char ) {
2115 $text = '</li></ol>';
2116 } elseif ( ':' === $char ) {
2117 if ( $this->mDTopen ) {
2118 $this->mDTopen = false;
2119 $text = '</dt></dl>';
2120 } else {
2121 $text = '</dd></dl>';
2122 }
2123 } else {
2124 return '<!-- ERR 3 -->';
2125 }
2126 return $text."\n";
2127 }
2128 /**#@-*/
2129
2130 /**
2131 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2132 *
2133 * @param $text String
2134 * @param $linestart Boolean: whether or not this is at the start of a line.
2135 * @private
2136 * @return string the lists rendered as HTML
2137 */
2138 function doBlockLevels( $text, $linestart ) {
2139 wfProfileIn( __METHOD__ );
2140
2141 # Parsing through the text line by line. The main thing
2142 # happening here is handling of block-level elements p, pre,
2143 # and making lists from lines starting with * # : etc.
2144 #
2145 $textLines = StringUtils::explode( "\n", $text );
2146
2147 $lastPrefix = $output = '';
2148 $this->mDTopen = $inBlockElem = false;
2149 $prefixLength = 0;
2150 $paragraphStack = false;
2151
2152 foreach ( $textLines as $oLine ) {
2153 # Fix up $linestart
2154 if ( !$linestart ) {
2155 $output .= $oLine;
2156 $linestart = true;
2157 continue;
2158 }
2159 # * = ul
2160 # # = ol
2161 # ; = dt
2162 # : = dd
2163
2164 $lastPrefixLength = strlen( $lastPrefix );
2165 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2166 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2167 # If not in a <pre> element, scan for and figure out what prefixes are there.
2168 if ( !$this->mInPre ) {
2169 # Multiple prefixes may abut each other for nested lists.
2170 $prefixLength = strspn( $oLine, '*#:;' );
2171 $prefix = substr( $oLine, 0, $prefixLength );
2172
2173 # eh?
2174 # ; and : are both from definition-lists, so they're equivalent
2175 # for the purposes of determining whether or not we need to open/close
2176 # elements.
2177 $prefix2 = str_replace( ';', ':', $prefix );
2178 $t = substr( $oLine, $prefixLength );
2179 $this->mInPre = (bool)$preOpenMatch;
2180 } else {
2181 # Don't interpret any other prefixes in preformatted text
2182 $prefixLength = 0;
2183 $prefix = $prefix2 = '';
2184 $t = $oLine;
2185 }
2186
2187 # List generation
2188 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2189 # Same as the last item, so no need to deal with nesting or opening stuff
2190 $output .= $this->nextItem( substr( $prefix, -1 ) );
2191 $paragraphStack = false;
2192
2193 if ( substr( $prefix, -1 ) === ';') {
2194 # The one nasty exception: definition lists work like this:
2195 # ; title : definition text
2196 # So we check for : in the remainder text to split up the
2197 # title and definition, without b0rking links.
2198 $term = $t2 = '';
2199 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2200 $t = $t2;
2201 $output .= $term . $this->nextItem( ':' );
2202 }
2203 }
2204 } elseif ( $prefixLength || $lastPrefixLength ) {
2205 # We need to open or close prefixes, or both.
2206
2207 # Either open or close a level...
2208 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2209 $paragraphStack = false;
2210
2211 # Close all the prefixes which aren't shared.
2212 while ( $commonPrefixLength < $lastPrefixLength ) {
2213 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2214 --$lastPrefixLength;
2215 }
2216
2217 # Continue the current prefix if appropriate.
2218 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2219 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2220 }
2221
2222 # Open prefixes where appropriate.
2223 while ( $prefixLength > $commonPrefixLength ) {
2224 $char = substr( $prefix, $commonPrefixLength, 1 );
2225 $output .= $this->openList( $char );
2226
2227 if ( ';' === $char ) {
2228 # FIXME: This is dupe of code above
2229 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2230 $t = $t2;
2231 $output .= $term . $this->nextItem( ':' );
2232 }
2233 }
2234 ++$commonPrefixLength;
2235 }
2236 $lastPrefix = $prefix2;
2237 }
2238
2239 # If we have no prefixes, go to paragraph mode.
2240 if ( 0 == $prefixLength ) {
2241 wfProfileIn( __METHOD__."-paragraph" );
2242 # No prefix (not in list)--go to paragraph mode
2243 # XXX: use a stack for nestable elements like span, table and div
2244 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2245 $closematch = preg_match(
2246 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2247 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2248 if ( $openmatch or $closematch ) {
2249 $paragraphStack = false;
2250 # TODO bug 5718: paragraph closed
2251 $output .= $this->closeParagraph();
2252 if ( $preOpenMatch and !$preCloseMatch ) {
2253 $this->mInPre = true;
2254 }
2255 $inBlockElem = !$closematch;
2256 } elseif ( !$inBlockElem && !$this->mInPre ) {
2257 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2258 # pre
2259 if ( $this->mLastSection !== 'pre' ) {
2260 $paragraphStack = false;
2261 $output .= $this->closeParagraph().'<pre>';
2262 $this->mLastSection = 'pre';
2263 }
2264 $t = substr( $t, 1 );
2265 } else {
2266 # paragraph
2267 if ( trim( $t ) === '' ) {
2268 if ( $paragraphStack ) {
2269 $output .= $paragraphStack.'<br />';
2270 $paragraphStack = false;
2271 $this->mLastSection = 'p';
2272 } else {
2273 if ( $this->mLastSection !== 'p' ) {
2274 $output .= $this->closeParagraph();
2275 $this->mLastSection = '';
2276 $paragraphStack = '<p>';
2277 } else {
2278 $paragraphStack = '</p><p>';
2279 }
2280 }
2281 } else {
2282 if ( $paragraphStack ) {
2283 $output .= $paragraphStack;
2284 $paragraphStack = false;
2285 $this->mLastSection = 'p';
2286 } elseif ( $this->mLastSection !== 'p' ) {
2287 $output .= $this->closeParagraph().'<p>';
2288 $this->mLastSection = 'p';
2289 }
2290 }
2291 }
2292 }
2293 wfProfileOut( __METHOD__."-paragraph" );
2294 }
2295 # somewhere above we forget to get out of pre block (bug 785)
2296 if ( $preCloseMatch && $this->mInPre ) {
2297 $this->mInPre = false;
2298 }
2299 if ( $paragraphStack === false ) {
2300 $output .= $t."\n";
2301 }
2302 }
2303 while ( $prefixLength ) {
2304 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2305 --$prefixLength;
2306 }
2307 if ( $this->mLastSection != '' ) {
2308 $output .= '</' . $this->mLastSection . '>';
2309 $this->mLastSection = '';
2310 }
2311
2312 wfProfileOut( __METHOD__ );
2313 return $output;
2314 }
2315
2316 /**
2317 * Split up a string on ':', ignoring any occurences inside tags
2318 * to prevent illegal overlapping.
2319 *
2320 * @param $str String: the string to split
2321 * @param &$before String: set to everything before the ':'
2322 * @param &$after String: set to everything after the ':'
2323 * return String: the position of the ':', or false if none found
2324 */
2325 function findColonNoLinks( $str, &$before, &$after ) {
2326 wfProfileIn( __METHOD__ );
2327
2328 $pos = strpos( $str, ':' );
2329 if ( $pos === false ) {
2330 # Nothing to find!
2331 wfProfileOut( __METHOD__ );
2332 return false;
2333 }
2334
2335 $lt = strpos( $str, '<' );
2336 if ( $lt === false || $lt > $pos ) {
2337 # Easy; no tag nesting to worry about
2338 $before = substr( $str, 0, $pos );
2339 $after = substr( $str, $pos+1 );
2340 wfProfileOut( __METHOD__ );
2341 return $pos;
2342 }
2343
2344 # Ugly state machine to walk through avoiding tags.
2345 $state = self::COLON_STATE_TEXT;
2346 $stack = 0;
2347 $len = strlen( $str );
2348 for( $i = 0; $i < $len; $i++ ) {
2349 $c = $str{$i};
2350
2351 switch( $state ) {
2352 # (Using the number is a performance hack for common cases)
2353 case 0: # self::COLON_STATE_TEXT:
2354 switch( $c ) {
2355 case "<":
2356 # Could be either a <start> tag or an </end> tag
2357 $state = self::COLON_STATE_TAGSTART;
2358 break;
2359 case ":":
2360 if ( $stack == 0 ) {
2361 # We found it!
2362 $before = substr( $str, 0, $i );
2363 $after = substr( $str, $i + 1 );
2364 wfProfileOut( __METHOD__ );
2365 return $i;
2366 }
2367 # Embedded in a tag; don't break it.
2368 break;
2369 default:
2370 # Skip ahead looking for something interesting
2371 $colon = strpos( $str, ':', $i );
2372 if ( $colon === false ) {
2373 # Nothing else interesting
2374 wfProfileOut( __METHOD__ );
2375 return false;
2376 }
2377 $lt = strpos( $str, '<', $i );
2378 if ( $stack === 0 ) {
2379 if ( $lt === false || $colon < $lt ) {
2380 # We found it!
2381 $before = substr( $str, 0, $colon );
2382 $after = substr( $str, $colon + 1 );
2383 wfProfileOut( __METHOD__ );
2384 return $i;
2385 }
2386 }
2387 if ( $lt === false ) {
2388 # Nothing else interesting to find; abort!
2389 # We're nested, but there's no close tags left. Abort!
2390 break 2;
2391 }
2392 # Skip ahead to next tag start
2393 $i = $lt;
2394 $state = self::COLON_STATE_TAGSTART;
2395 }
2396 break;
2397 case 1: # self::COLON_STATE_TAG:
2398 # In a <tag>
2399 switch( $c ) {
2400 case ">":
2401 $stack++;
2402 $state = self::COLON_STATE_TEXT;
2403 break;
2404 case "/":
2405 # Slash may be followed by >?
2406 $state = self::COLON_STATE_TAGSLASH;
2407 break;
2408 default:
2409 # ignore
2410 }
2411 break;
2412 case 2: # self::COLON_STATE_TAGSTART:
2413 switch( $c ) {
2414 case "/":
2415 $state = self::COLON_STATE_CLOSETAG;
2416 break;
2417 case "!":
2418 $state = self::COLON_STATE_COMMENT;
2419 break;
2420 case ">":
2421 # Illegal early close? This shouldn't happen D:
2422 $state = self::COLON_STATE_TEXT;
2423 break;
2424 default:
2425 $state = self::COLON_STATE_TAG;
2426 }
2427 break;
2428 case 3: # self::COLON_STATE_CLOSETAG:
2429 # In a </tag>
2430 if ( $c === ">" ) {
2431 $stack--;
2432 if ( $stack < 0 ) {
2433 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2434 wfProfileOut( __METHOD__ );
2435 return false;
2436 }
2437 $state = self::COLON_STATE_TEXT;
2438 }
2439 break;
2440 case self::COLON_STATE_TAGSLASH:
2441 if ( $c === ">" ) {
2442 # Yes, a self-closed tag <blah/>
2443 $state = self::COLON_STATE_TEXT;
2444 } else {
2445 # Probably we're jumping the gun, and this is an attribute
2446 $state = self::COLON_STATE_TAG;
2447 }
2448 break;
2449 case 5: # self::COLON_STATE_COMMENT:
2450 if ( $c === "-" ) {
2451 $state = self::COLON_STATE_COMMENTDASH;
2452 }
2453 break;
2454 case self::COLON_STATE_COMMENTDASH:
2455 if ( $c === "-" ) {
2456 $state = self::COLON_STATE_COMMENTDASHDASH;
2457 } else {
2458 $state = self::COLON_STATE_COMMENT;
2459 }
2460 break;
2461 case self::COLON_STATE_COMMENTDASHDASH:
2462 if ( $c === ">" ) {
2463 $state = self::COLON_STATE_TEXT;
2464 } else {
2465 $state = self::COLON_STATE_COMMENT;
2466 }
2467 break;
2468 default:
2469 throw new MWException( "State machine error in " . __METHOD__ );
2470 }
2471 }
2472 if ( $stack > 0 ) {
2473 wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2474 wfProfileOut( __METHOD__ );
2475 return false;
2476 }
2477 wfProfileOut( __METHOD__ );
2478 return false;
2479 }
2480
2481 /**
2482 * Return value of a magic variable (like PAGENAME)
2483 *
2484 * @private
2485 */
2486 function getVariableValue( $index, $frame=false ) {
2487 global $wgContLang, $wgSitename, $wgServer;
2488 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2489
2490 /**
2491 * Some of these require message or data lookups and can be
2492 * expensive to check many times.
2493 */
2494 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2495 if ( isset( $this->mVarCache[$index] ) ) {
2496 return $this->mVarCache[$index];
2497 }
2498 }
2499
2500 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2501 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2502
2503 # Use the time zone
2504 global $wgLocaltimezone;
2505 if ( isset( $wgLocaltimezone ) ) {
2506 $oldtz = date_default_timezone_get();
2507 date_default_timezone_set( $wgLocaltimezone );
2508 }
2509
2510 $localTimestamp = date( 'YmdHis', $ts );
2511 $localMonth = date( 'm', $ts );
2512 $localMonth1 = date( 'n', $ts );
2513 $localMonthName = date( 'n', $ts );
2514 $localDay = date( 'j', $ts );
2515 $localDay2 = date( 'd', $ts );
2516 $localDayOfWeek = date( 'w', $ts );
2517 $localWeek = date( 'W', $ts );
2518 $localYear = date( 'Y', $ts );
2519 $localHour = date( 'H', $ts );
2520 if ( isset( $wgLocaltimezone ) ) {
2521 date_default_timezone_set( $oldtz );
2522 }
2523
2524 switch ( $index ) {
2525 case 'currentmonth':
2526 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2527 break;
2528 case 'currentmonth1':
2529 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2530 break;
2531 case 'currentmonthname':
2532 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2533 break;
2534 case 'currentmonthnamegen':
2535 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2536 break;
2537 case 'currentmonthabbrev':
2538 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2539 break;
2540 case 'currentday':
2541 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2542 break;
2543 case 'currentday2':
2544 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2545 break;
2546 case 'localmonth':
2547 $value = $wgContLang->formatNum( $localMonth );
2548 break;
2549 case 'localmonth1':
2550 $value = $wgContLang->formatNum( $localMonth1 );
2551 break;
2552 case 'localmonthname':
2553 $value = $wgContLang->getMonthName( $localMonthName );
2554 break;
2555 case 'localmonthnamegen':
2556 $value = $wgContLang->getMonthNameGen( $localMonthName );
2557 break;
2558 case 'localmonthabbrev':
2559 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2560 break;
2561 case 'localday':
2562 $value = $wgContLang->formatNum( $localDay );
2563 break;
2564 case 'localday2':
2565 $value = $wgContLang->formatNum( $localDay2 );
2566 break;
2567 case 'pagename':
2568 $value = wfEscapeWikiText( $this->mTitle->getText() );
2569 break;
2570 case 'pagenamee':
2571 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2572 break;
2573 case 'fullpagename':
2574 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2575 break;
2576 case 'fullpagenamee':
2577 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2578 break;
2579 case 'subpagename':
2580 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2581 break;
2582 case 'subpagenamee':
2583 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2584 break;
2585 case 'basepagename':
2586 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2587 break;
2588 case 'basepagenamee':
2589 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2590 break;
2591 case 'talkpagename':
2592 if ( $this->mTitle->canTalk() ) {
2593 $talkPage = $this->mTitle->getTalkPage();
2594 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2595 } else {
2596 $value = '';
2597 }
2598 break;
2599 case 'talkpagenamee':
2600 if ( $this->mTitle->canTalk() ) {
2601 $talkPage = $this->mTitle->getTalkPage();
2602 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2603 } else {
2604 $value = '';
2605 }
2606 break;
2607 case 'subjectpagename':
2608 $subjPage = $this->mTitle->getSubjectPage();
2609 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2610 break;
2611 case 'subjectpagenamee':
2612 $subjPage = $this->mTitle->getSubjectPage();
2613 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2614 break;
2615 case 'revisionid':
2616 # Let the edit saving system know we should parse the page
2617 # *after* a revision ID has been assigned.
2618 $this->mOutput->setFlag( 'vary-revision' );
2619 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2620 $value = $this->mRevisionId;
2621 break;
2622 case 'revisionday':
2623 # Let the edit saving system know we should parse the page
2624 # *after* a revision ID has been assigned. This is for null edits.
2625 $this->mOutput->setFlag( 'vary-revision' );
2626 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2627 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2628 break;
2629 case 'revisionday2':
2630 # Let the edit saving system know we should parse the page
2631 # *after* a revision ID has been assigned. This is for null edits.
2632 $this->mOutput->setFlag( 'vary-revision' );
2633 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2634 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2635 break;
2636 case 'revisionmonth':
2637 # Let the edit saving system know we should parse the page
2638 # *after* a revision ID has been assigned. This is for null edits.
2639 $this->mOutput->setFlag( 'vary-revision' );
2640 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2641 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2642 break;
2643 case 'revisionmonth1':
2644 # Let the edit saving system know we should parse the page
2645 # *after* a revision ID has been assigned. This is for null edits.
2646 $this->mOutput->setFlag( 'vary-revision' );
2647 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2648 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2649 break;
2650 case 'revisionyear':
2651 # Let the edit saving system know we should parse the page
2652 # *after* a revision ID has been assigned. This is for null edits.
2653 $this->mOutput->setFlag( 'vary-revision' );
2654 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2655 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2656 break;
2657 case 'revisiontimestamp':
2658 # Let the edit saving system know we should parse the page
2659 # *after* a revision ID has been assigned. This is for null edits.
2660 $this->mOutput->setFlag( 'vary-revision' );
2661 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2662 $value = $this->getRevisionTimestamp();
2663 break;
2664 case 'revisionuser':
2665 # Let the edit saving system know we should parse the page
2666 # *after* a revision ID has been assigned. This is for null edits.
2667 $this->mOutput->setFlag( 'vary-revision' );
2668 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2669 $value = $this->getRevisionUser();
2670 break;
2671 case 'namespace':
2672 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2673 break;
2674 case 'namespacee':
2675 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2676 break;
2677 case 'talkspace':
2678 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2679 break;
2680 case 'talkspacee':
2681 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2682 break;
2683 case 'subjectspace':
2684 $value = $this->mTitle->getSubjectNsText();
2685 break;
2686 case 'subjectspacee':
2687 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2688 break;
2689 case 'currentdayname':
2690 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2691 break;
2692 case 'currentyear':
2693 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2694 break;
2695 case 'currenttime':
2696 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2697 break;
2698 case 'currenthour':
2699 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2700 break;
2701 case 'currentweek':
2702 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2703 # int to remove the padding
2704 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2705 break;
2706 case 'currentdow':
2707 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2708 break;
2709 case 'localdayname':
2710 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2711 break;
2712 case 'localyear':
2713 $value = $wgContLang->formatNum( $localYear, true );
2714 break;
2715 case 'localtime':
2716 $value = $wgContLang->time( $localTimestamp, false, false );
2717 break;
2718 case 'localhour':
2719 $value = $wgContLang->formatNum( $localHour, true );
2720 break;
2721 case 'localweek':
2722 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2723 # int to remove the padding
2724 $value = $wgContLang->formatNum( (int)$localWeek );
2725 break;
2726 case 'localdow':
2727 $value = $wgContLang->formatNum( $localDayOfWeek );
2728 break;
2729 case 'numberofarticles':
2730 $value = $wgContLang->formatNum( SiteStats::articles() );
2731 break;
2732 case 'numberoffiles':
2733 $value = $wgContLang->formatNum( SiteStats::images() );
2734 break;
2735 case 'numberofusers':
2736 $value = $wgContLang->formatNum( SiteStats::users() );
2737 break;
2738 case 'numberofactiveusers':
2739 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2740 break;
2741 case 'numberofpages':
2742 $value = $wgContLang->formatNum( SiteStats::pages() );
2743 break;
2744 case 'numberofadmins':
2745 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2746 break;
2747 case 'numberofedits':
2748 $value = $wgContLang->formatNum( SiteStats::edits() );
2749 break;
2750 case 'numberofviews':
2751 $value = $wgContLang->formatNum( SiteStats::views() );
2752 break;
2753 case 'currenttimestamp':
2754 $value = wfTimestamp( TS_MW, $ts );
2755 break;
2756 case 'localtimestamp':
2757 $value = $localTimestamp;
2758 break;
2759 case 'currentversion':
2760 $value = SpecialVersion::getVersion();
2761 break;
2762 case 'articlepath':
2763 return $wgArticlePath;
2764 case 'sitename':
2765 return $wgSitename;
2766 case 'server':
2767 return $wgServer;
2768 case 'servername':
2769 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2770 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2771 wfRestoreWarnings();
2772 return $serverName ? $serverName : $wgServer;
2773 case 'scriptpath':
2774 return $wgScriptPath;
2775 case 'stylepath':
2776 return $wgStylePath;
2777 case 'directionmark':
2778 return $wgContLang->getDirMark();
2779 case 'contentlanguage':
2780 global $wgLanguageCode;
2781 return $wgLanguageCode;
2782 default:
2783 $ret = null;
2784 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2785 return $ret;
2786 } else {
2787 return null;
2788 }
2789 }
2790
2791 if ( $index )
2792 $this->mVarCache[$index] = $value;
2793
2794 return $value;
2795 }
2796
2797 /**
2798 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2799 *
2800 * @private
2801 */
2802 function initialiseVariables() {
2803 wfProfileIn( __METHOD__ );
2804 $variableIDs = MagicWord::getVariableIDs();
2805 $substIDs = MagicWord::getSubstIDs();
2806
2807 $this->mVariables = new MagicWordArray( $variableIDs );
2808 $this->mSubstWords = new MagicWordArray( $substIDs );
2809 wfProfileOut( __METHOD__ );
2810 }
2811
2812 /**
2813 * Preprocess some wikitext and return the document tree.
2814 * This is the ghost of replace_variables().
2815 *
2816 * @param $text String: The text to parse
2817 * @param $flags Integer: bitwise combination of:
2818 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2819 * included. Default is to assume a direct page view.
2820 *
2821 * The generated DOM tree must depend only on the input text and the flags.
2822 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2823 *
2824 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2825 * change in the DOM tree for a given text, must be passed through the section identifier
2826 * in the section edit link and thus back to extractSections().
2827 *
2828 * The output of this function is currently only cached in process memory, but a persistent
2829 * cache may be implemented at a later date which takes further advantage of these strict
2830 * dependency requirements.
2831 *
2832 * @private
2833 */
2834 function preprocessToDom( $text, $flags = 0 ) {
2835 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2836 return $dom;
2837 }
2838
2839 /**
2840 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2841 */
2842 public static function splitWhitespace( $s ) {
2843 $ltrimmed = ltrim( $s );
2844 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2845 $trimmed = rtrim( $ltrimmed );
2846 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2847 if ( $diff > 0 ) {
2848 $w2 = substr( $ltrimmed, -$diff );
2849 } else {
2850 $w2 = '';
2851 }
2852 return array( $w1, $trimmed, $w2 );
2853 }
2854
2855 /**
2856 * Replace magic variables, templates, and template arguments
2857 * with the appropriate text. Templates are substituted recursively,
2858 * taking care to avoid infinite loops.
2859 *
2860 * Note that the substitution depends on value of $mOutputType:
2861 * self::OT_WIKI: only {{subst:}} templates
2862 * self::OT_PREPROCESS: templates but not extension tags
2863 * self::OT_HTML: all templates and extension tags
2864 *
2865 * @param $text String: the text to transform
2866 * @param $frame PPFrame Object describing the arguments passed to the template.
2867 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2868 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2869 * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2870 * @private
2871 */
2872 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2873 # Is there any text? Also, Prevent too big inclusions!
2874 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2875 return $text;
2876 }
2877 wfProfileIn( __METHOD__ );
2878
2879 if ( $frame === false ) {
2880 $frame = $this->getPreprocessor()->newFrame();
2881 } elseif ( !( $frame instanceof PPFrame ) ) {
2882 wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2883 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2884 }
2885
2886 $dom = $this->preprocessToDom( $text );
2887 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2888 $text = $frame->expand( $dom, $flags );
2889
2890 wfProfileOut( __METHOD__ );
2891 return $text;
2892 }
2893
2894 # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2895 static function createAssocArgs( $args ) {
2896 $assocArgs = array();
2897 $index = 1;
2898 foreach ( $args as $arg ) {
2899 $eqpos = strpos( $arg, '=' );
2900 if ( $eqpos === false ) {
2901 $assocArgs[$index++] = $arg;
2902 } else {
2903 $name = trim( substr( $arg, 0, $eqpos ) );
2904 $value = trim( substr( $arg, $eqpos+1 ) );
2905 if ( $value === false ) {
2906 $value = '';
2907 }
2908 if ( $name !== false ) {
2909 $assocArgs[$name] = $value;
2910 }
2911 }
2912 }
2913
2914 return $assocArgs;
2915 }
2916
2917 /**
2918 * Warn the user when a parser limitation is reached
2919 * Will warn at most once the user per limitation type
2920 *
2921 * @param $limitationType String: should be one of:
2922 * 'expensive-parserfunction' (corresponding messages:
2923 * 'expensive-parserfunction-warning',
2924 * 'expensive-parserfunction-category')
2925 * 'post-expand-template-argument' (corresponding messages:
2926 * 'post-expand-template-argument-warning',
2927 * 'post-expand-template-argument-category')
2928 * 'post-expand-template-inclusion' (corresponding messages:
2929 * 'post-expand-template-inclusion-warning',
2930 * 'post-expand-template-inclusion-category')
2931 * @param $current Current value
2932 * @param $max Maximum allowed, when an explicit limit has been
2933 * exceeded, provide the values (optional)
2934 */
2935 function limitationWarn( $limitationType, $current=null, $max=null) {
2936 # does no harm if $current and $max are present but are unnecessary for the message
2937 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2938 $this->mOutput->addWarning( $warning );
2939 $this->addTrackingCategory( "$limitationType-category" );
2940 }
2941
2942 /**
2943 * Return the text of a template, after recursively
2944 * replacing any variables or templates within the template.
2945 *
2946 * @param $piece Array: the parts of the template
2947 * $piece['title']: the title, i.e. the part before the |
2948 * $piece['parts']: the parameter array
2949 * $piece['lineStart']: whether the brace was at the start of a line
2950 * @param $frame PPFrame The current frame, contains template arguments
2951 * @return String: the text of the template
2952 * @private
2953 */
2954 function braceSubstitution( $piece, $frame ) {
2955 global $wgContLang, $wgNonincludableNamespaces;
2956 wfProfileIn( __METHOD__ );
2957 wfProfileIn( __METHOD__.'-setup' );
2958
2959 # Flags
2960 $found = false; # $text has been filled
2961 $nowiki = false; # wiki markup in $text should be escaped
2962 $isHTML = false; # $text is HTML, armour it against wikitext transformation
2963 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2964 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
2965 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
2966
2967 # Title object, where $text came from
2968 $title = null;
2969
2970 # $part1 is the bit before the first |, and must contain only title characters.
2971 # Various prefixes will be stripped from it later.
2972 $titleWithSpaces = $frame->expand( $piece['title'] );
2973 $part1 = trim( $titleWithSpaces );
2974 $titleText = false;
2975
2976 # Original title text preserved for various purposes
2977 $originalTitle = $part1;
2978
2979 # $args is a list of argument nodes, starting from index 0, not including $part1
2980 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2981 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2982 wfProfileOut( __METHOD__.'-setup' );
2983
2984 # SUBST
2985 wfProfileIn( __METHOD__.'-modifiers' );
2986 if ( !$found ) {
2987
2988 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2989
2990 # Possibilities for substMatch: "subst", "safesubst" or FALSE
2991 # Decide whether to expand template or keep wikitext as-is.
2992 if ( $this->ot['wiki'] ) {
2993 if ( $substMatch === false ) {
2994 $literal = true; # literal when in PST with no prefix
2995 } else {
2996 $literal = false; # expand when in PST with subst: or safesubst:
2997 }
2998 } else {
2999 if ( $substMatch == 'subst' ) {
3000 $literal = true; # literal when not in PST with plain subst:
3001 } else {
3002 $literal = false; # expand when not in PST with safesubst: or no prefix
3003 }
3004 }
3005 if ( $literal ) {
3006 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3007 $isLocalObj = true;
3008 $found = true;
3009 }
3010 }
3011
3012 # Variables
3013 if ( !$found && $args->getLength() == 0 ) {
3014 $id = $this->mVariables->matchStartToEnd( $part1 );
3015 if ( $id !== false ) {
3016 $text = $this->getVariableValue( $id, $frame );
3017 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3018 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3019 }
3020 $found = true;
3021 }
3022 }
3023
3024 # MSG, MSGNW and RAW
3025 if ( !$found ) {
3026 # Check for MSGNW:
3027 $mwMsgnw = MagicWord::get( 'msgnw' );
3028 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3029 $nowiki = true;
3030 } else {
3031 # Remove obsolete MSG:
3032 $mwMsg = MagicWord::get( 'msg' );
3033 $mwMsg->matchStartAndRemove( $part1 );
3034 }
3035
3036 # Check for RAW:
3037 $mwRaw = MagicWord::get( 'raw' );
3038 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3039 $forceRawInterwiki = true;
3040 }
3041 }
3042 wfProfileOut( __METHOD__.'-modifiers' );
3043
3044 # Parser functions
3045 if ( !$found ) {
3046 wfProfileIn( __METHOD__ . '-pfunc' );
3047
3048 $colonPos = strpos( $part1, ':' );
3049 if ( $colonPos !== false ) {
3050 # Case sensitive functions
3051 $function = substr( $part1, 0, $colonPos );
3052 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3053 $function = $this->mFunctionSynonyms[1][$function];
3054 } else {
3055 # Case insensitive functions
3056 $function = $wgContLang->lc( $function );
3057 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3058 $function = $this->mFunctionSynonyms[0][$function];
3059 } else {
3060 $function = false;
3061 }
3062 }
3063 if ( $function ) {
3064 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3065 $initialArgs = array( &$this );
3066 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3067 if ( $flags & SFH_OBJECT_ARGS ) {
3068 # Add a frame parameter, and pass the arguments as an array
3069 $allArgs = $initialArgs;
3070 $allArgs[] = $frame;
3071 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3072 $funcArgs[] = $args->item( $i );
3073 }
3074 $allArgs[] = $funcArgs;
3075 } else {
3076 # Convert arguments to plain text
3077 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3078 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3079 }
3080 $allArgs = array_merge( $initialArgs, $funcArgs );
3081 }
3082
3083 # Workaround for PHP bug 35229 and similar
3084 if ( !is_callable( $callback ) ) {
3085 wfProfileOut( __METHOD__ . '-pfunc' );
3086 wfProfileOut( __METHOD__ );
3087 throw new MWException( "Tag hook for $function is not callable\n" );
3088 }
3089 $result = call_user_func_array( $callback, $allArgs );
3090 $found = true;
3091 $noparse = true;
3092 $preprocessFlags = 0;
3093
3094 if ( is_array( $result ) ) {
3095 if ( isset( $result[0] ) ) {
3096 $text = $result[0];
3097 unset( $result[0] );
3098 }
3099
3100 # Extract flags into the local scope
3101 # This allows callers to set flags such as nowiki, found, etc.
3102 extract( $result );
3103 } else {
3104 $text = $result;
3105 }
3106 if ( !$noparse ) {
3107 $text = $this->preprocessToDom( $text, $preprocessFlags );
3108 $isChildObj = true;
3109 }
3110 }
3111 }
3112 wfProfileOut( __METHOD__ . '-pfunc' );
3113 }
3114
3115 # Finish mangling title and then check for loops.
3116 # Set $title to a Title object and $titleText to the PDBK
3117 if ( !$found ) {
3118 $ns = NS_TEMPLATE;
3119 # Split the title into page and subpage
3120 $subpage = '';
3121 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3122 if ( $subpage !== '' ) {
3123 $ns = $this->mTitle->getNamespace();
3124 }
3125 $title = Title::newFromText( $part1, $ns );
3126 if ( $title ) {
3127 $titleText = $title->getPrefixedText();
3128 # Check for language variants if the template is not found
3129 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3130 $wgContLang->findVariantLink( $part1, $title, true );
3131 }
3132 # Do recursion depth check
3133 $limit = $this->mOptions->getMaxTemplateDepth();
3134 if ( $frame->depth >= $limit ) {
3135 $found = true;
3136 $text = '<span class="error">'
3137 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3138 . '</span>';
3139 }
3140 }
3141 }
3142
3143 # Load from database
3144 if ( !$found && $title ) {
3145 wfProfileIn( __METHOD__ . '-loadtpl' );
3146 if ( !$title->isExternal() ) {
3147 if ( $title->getNamespace() == NS_SPECIAL
3148 && $this->mOptions->getAllowSpecialInclusion()
3149 && $this->ot['html'] )
3150 {
3151 $text = SpecialPage::capturePath( $title );
3152 if ( is_string( $text ) ) {
3153 $found = true;
3154 $isHTML = true;
3155 $this->disableCache();
3156 }
3157 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3158 $found = false; # access denied
3159 wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3160 } else {
3161 list( $text, $title ) = $this->getTemplateDom( $title );
3162 if ( $text !== false ) {
3163 $found = true;
3164 $isChildObj = true;
3165 }
3166 }
3167
3168 # If the title is valid but undisplayable, make a link to it
3169 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3170 $text = "[[:$titleText]]";
3171 $found = true;
3172 }
3173 } elseif ( $title->isTrans() ) {
3174 # Interwiki transclusion
3175 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3176 $text = $this->interwikiTransclude( $title, 'render' );
3177 $isHTML = true;
3178 } else {
3179 $text = $this->interwikiTransclude( $title, 'raw' );
3180 # Preprocess it like a template
3181 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3182 $isChildObj = true;
3183 }
3184 $found = true;
3185 }
3186
3187 # Do infinite loop check
3188 # This has to be done after redirect resolution to avoid infinite loops via redirects
3189 if ( !$frame->loopCheck( $title ) ) {
3190 $found = true;
3191 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3192 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3193 }
3194 wfProfileOut( __METHOD__ . '-loadtpl' );
3195 }
3196
3197 # If we haven't found text to substitute by now, we're done
3198 # Recover the source wikitext and return it
3199 if ( !$found ) {
3200 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3201 wfProfileOut( __METHOD__ );
3202 return array( 'object' => $text );
3203 }
3204
3205 # Expand DOM-style return values in a child frame
3206 if ( $isChildObj ) {
3207 # Clean up argument array
3208 $newFrame = $frame->newChild( $args, $title );
3209
3210 if ( $nowiki ) {
3211 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3212 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3213 # Expansion is eligible for the empty-frame cache
3214 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3215 $text = $this->mTplExpandCache[$titleText];
3216 } else {
3217 $text = $newFrame->expand( $text );
3218 $this->mTplExpandCache[$titleText] = $text;
3219 }
3220 } else {
3221 # Uncached expansion
3222 $text = $newFrame->expand( $text );
3223 }
3224 }
3225 if ( $isLocalObj && $nowiki ) {
3226 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3227 $isLocalObj = false;
3228 }
3229
3230 # Replace raw HTML by a placeholder
3231 # Add a blank line preceding, to prevent it from mucking up
3232 # immediately preceding headings
3233 if ( $isHTML ) {
3234 $text = "\n\n" . $this->insertStripItem( $text );
3235 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3236 # Escape nowiki-style return values
3237 $text = wfEscapeWikiText( $text );
3238 } elseif ( is_string( $text )
3239 && !$piece['lineStart']
3240 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3241 {
3242 # Bug 529: if the template begins with a table or block-level
3243 # element, it should be treated as beginning a new line.
3244 # This behaviour is somewhat controversial.
3245 $text = "\n" . $text;
3246 }
3247
3248 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3249 # Error, oversize inclusion
3250 if ( $titleText !== false ) {
3251 # Make a working, properly escaped link if possible (bug 23588)
3252 $text = "[[:$titleText]]";
3253 } else {
3254 # This will probably not be a working link, but at least it may
3255 # provide some hint of where the problem is
3256 preg_replace( '/^:/', '', $originalTitle );
3257 $text = "[[:$originalTitle]]";
3258 }
3259 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3260 $this->limitationWarn( 'post-expand-template-inclusion' );
3261 }
3262
3263 if ( $isLocalObj ) {
3264 $ret = array( 'object' => $text );
3265 } else {
3266 $ret = array( 'text' => $text );
3267 }
3268
3269 wfProfileOut( __METHOD__ );
3270 return $ret;
3271 }
3272
3273 /**
3274 * Get the semi-parsed DOM representation of a template with a given title,
3275 * and its redirect destination title. Cached.
3276 */
3277 function getTemplateDom( $title ) {
3278 $cacheTitle = $title;
3279 $titleText = $title->getPrefixedDBkey();
3280
3281 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3282 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3283 $title = Title::makeTitle( $ns, $dbk );
3284 $titleText = $title->getPrefixedDBkey();
3285 }
3286 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3287 return array( $this->mTplDomCache[$titleText], $title );
3288 }
3289
3290 # Cache miss, go to the database
3291 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3292
3293 if ( $text === false ) {
3294 $this->mTplDomCache[$titleText] = false;
3295 return array( false, $title );
3296 }
3297
3298 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3299 $this->mTplDomCache[ $titleText ] = $dom;
3300
3301 if ( !$title->equals( $cacheTitle ) ) {
3302 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3303 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3304 }
3305
3306 return array( $dom, $title );
3307 }
3308
3309 /**
3310 * Fetch the unparsed text of a template and register a reference to it.
3311 */
3312 function fetchTemplateAndTitle( $title ) {
3313 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3314 $stuff = call_user_func( $templateCb, $title, $this );
3315 $text = $stuff['text'];
3316 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3317 if ( isset( $stuff['deps'] ) ) {
3318 foreach ( $stuff['deps'] as $dep ) {
3319 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3320 }
3321 }
3322 return array( $text, $finalTitle );
3323 }
3324
3325 function fetchTemplate( $title ) {
3326 $rv = $this->fetchTemplateAndTitle( $title );
3327 return $rv[0];
3328 }
3329
3330 /**
3331 * Static function to get a template
3332 * Can be overridden via ParserOptions::setTemplateCallback().
3333 */
3334 static function statelessFetchTemplate( $title, $parser=false ) {
3335 $text = $skip = false;
3336 $finalTitle = $title;
3337 $deps = array();
3338
3339 # Loop to fetch the article, with up to 1 redirect
3340 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3341 # Give extensions a chance to select the revision instead
3342 $id = false; # Assume current
3343 wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3344
3345 if ( $skip ) {
3346 $text = false;
3347 $deps[] = array(
3348 'title' => $title,
3349 'page_id' => $title->getArticleID(),
3350 'rev_id' => null );
3351 break;
3352 }
3353 $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3354 $rev_id = $rev ? $rev->getId() : 0;
3355 # If there is no current revision, there is no page
3356 if ( $id === false && !$rev ) {
3357 $linkCache = LinkCache::singleton();
3358 $linkCache->addBadLinkObj( $title );
3359 }
3360
3361 $deps[] = array(
3362 'title' => $title,
3363 'page_id' => $title->getArticleID(),
3364 'rev_id' => $rev_id );
3365
3366 if ( $rev ) {
3367 $text = $rev->getText();
3368 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3369 global $wgContLang;
3370 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3371 if ( !$message->exists() ) {
3372 $text = false;
3373 break;
3374 }
3375 $text = $message->plain();
3376 } else {
3377 break;
3378 }
3379 if ( $text === false ) {
3380 break;
3381 }
3382 # Redirect?
3383 $finalTitle = $title;
3384 $title = Title::newFromRedirect( $text );
3385 }
3386 return array(
3387 'text' => $text,
3388 'finalTitle' => $finalTitle,
3389 'deps' => $deps );
3390 }
3391
3392 /**
3393 * Transclude an interwiki link.
3394 */
3395 function interwikiTransclude( $title, $action ) {
3396 global $wgEnableScaryTranscluding;
3397
3398 if ( !$wgEnableScaryTranscluding ) {
3399 return wfMsgForContent('scarytranscludedisabled');
3400 }
3401
3402 $url = $title->getFullUrl( "action=$action" );
3403
3404 if ( strlen( $url ) > 255 ) {
3405 return wfMsgForContent( 'scarytranscludetoolong' );
3406 }
3407 return $this->fetchScaryTemplateMaybeFromCache( $url );
3408 }
3409
3410 function fetchScaryTemplateMaybeFromCache( $url ) {
3411 global $wgTranscludeCacheExpiry;
3412 $dbr = wfGetDB( DB_SLAVE );
3413 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3414 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3415 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3416 if ( $obj ) {
3417 return $obj->tc_contents;
3418 }
3419
3420 $text = Http::get( $url );
3421 if ( !$text ) {
3422 return wfMsgForContent( 'scarytranscludefailed', $url );
3423 }
3424
3425 $dbw = wfGetDB( DB_MASTER );
3426 $dbw->replace( 'transcache', array('tc_url'), array(
3427 'tc_url' => $url,
3428 'tc_time' => $dbw->timestamp( time() ),
3429 'tc_contents' => $text)
3430 );
3431 return $text;
3432 }
3433
3434
3435 /**
3436 * Triple brace replacement -- used for template arguments
3437 * @private
3438 */
3439 function argSubstitution( $piece, $frame ) {
3440 wfProfileIn( __METHOD__ );
3441
3442 $error = false;
3443 $parts = $piece['parts'];
3444 $nameWithSpaces = $frame->expand( $piece['title'] );
3445 $argName = trim( $nameWithSpaces );
3446 $object = false;
3447 $text = $frame->getArgument( $argName );
3448 if ( $text === false && $parts->getLength() > 0
3449 && (
3450 $this->ot['html']
3451 || $this->ot['pre']
3452 || ( $this->ot['wiki'] && $frame->isTemplate() )
3453 )
3454 ) {
3455 # No match in frame, use the supplied default
3456 $object = $parts->item( 0 )->getChildren();
3457 }
3458 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3459 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3460 $this->limitationWarn( 'post-expand-template-argument' );
3461 }
3462
3463 if ( $text === false && $object === false ) {
3464 # No match anywhere
3465 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3466 }
3467 if ( $error !== false ) {
3468 $text .= $error;
3469 }
3470 if ( $object !== false ) {
3471 $ret = array( 'object' => $object );
3472 } else {
3473 $ret = array( 'text' => $text );
3474 }
3475
3476 wfProfileOut( __METHOD__ );
3477 return $ret;
3478 }
3479
3480 /**
3481 * Return the text to be used for a given extension tag.
3482 * This is the ghost of strip().
3483 *
3484 * @param $params Associative array of parameters:
3485 * name PPNode for the tag name
3486 * attr PPNode for unparsed text where tag attributes are thought to be
3487 * attributes Optional associative array of parsed attributes
3488 * inner Contents of extension element
3489 * noClose Original text did not have a close tag
3490 * @param $frame PPFrame
3491 */
3492 function extensionSubstitution( $params, $frame ) {
3493 $name = $frame->expand( $params['name'] );
3494 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3495 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3496 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3497
3498 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3499 ( $this->ot['html'] || $this->ot['pre'] );
3500 if ( $isFunctionTag ) {
3501 $markerType = 'none';
3502 } else {
3503 $markerType = 'general';
3504 }
3505 if ( $this->ot['html'] || $isFunctionTag ) {
3506 $name = strtolower( $name );
3507 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3508 if ( isset( $params['attributes'] ) ) {
3509 $attributes = $attributes + $params['attributes'];
3510 }
3511
3512 if ( isset( $this->mTagHooks[$name] ) ) {
3513 # Workaround for PHP bug 35229 and similar
3514 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3515 throw new MWException( "Tag hook for $name is not callable\n" );
3516 }
3517 $output = call_user_func_array( $this->mTagHooks[$name],
3518 array( $content, $attributes, $this, $frame ) );
3519 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3520 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3521 if ( !is_callable( $callback ) ) {
3522 throw new MWException( "Tag hook for $name is not callable\n" );
3523 }
3524
3525 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3526 } else {
3527 $output = '<span class="error">Invalid tag extension name: ' .
3528 htmlspecialchars( $name ) . '</span>';
3529 }
3530
3531 if ( is_array( $output ) ) {
3532 # Extract flags to local scope (to override $markerType)
3533 $flags = $output;
3534 $output = $flags[0];
3535 unset( $flags[0] );
3536 extract( $flags );
3537 }
3538 } else {
3539 if ( is_null( $attrText ) ) {
3540 $attrText = '';
3541 }
3542 if ( isset( $params['attributes'] ) ) {
3543 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3544 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3545 htmlspecialchars( $attrValue ) . '"';
3546 }
3547 }
3548 if ( $content === null ) {
3549 $output = "<$name$attrText/>";
3550 } else {
3551 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3552 $output = "<$name$attrText>$content$close";
3553 }
3554 }
3555
3556 if ( $markerType === 'none' ) {
3557 return $output;
3558 } elseif ( $markerType === 'nowiki' ) {
3559 $this->mStripState->addNoWiki( $marker, $output );
3560 } elseif ( $markerType === 'general' ) {
3561 $this->mStripState->addGeneral( $marker, $output );
3562 } else {
3563 throw new MWException( __METHOD__.': invalid marker type' );
3564 }
3565 return $marker;
3566 }
3567
3568 /**
3569 * Increment an include size counter
3570 *
3571 * @param $type String: the type of expansion
3572 * @param $size Integer: the size of the text
3573 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3574 */
3575 function incrementIncludeSize( $type, $size ) {
3576 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3577 return false;
3578 } else {
3579 $this->mIncludeSizes[$type] += $size;
3580 return true;
3581 }
3582 }
3583
3584 /**
3585 * Increment the expensive function count
3586 *
3587 * @return Boolean: false if the limit has been exceeded
3588 */
3589 function incrementExpensiveFunctionCount() {
3590 global $wgExpensiveParserFunctionLimit;
3591 $this->mExpensiveFunctionCount++;
3592 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3593 return true;
3594 }
3595 return false;
3596 }
3597
3598 /**
3599 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3600 * Fills $this->mDoubleUnderscores, returns the modified text
3601 */
3602 function doDoubleUnderscore( $text ) {
3603 wfProfileIn( __METHOD__ );
3604
3605 # The position of __TOC__ needs to be recorded
3606 $mw = MagicWord::get( 'toc' );
3607 if ( $mw->match( $text ) ) {
3608 $this->mShowToc = true;
3609 $this->mForceTocPosition = true;
3610
3611 # Set a placeholder. At the end we'll fill it in with the TOC.
3612 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3613
3614 # Only keep the first one.
3615 $text = $mw->replace( '', $text );
3616 }
3617
3618 # Now match and remove the rest of them
3619 $mwa = MagicWord::getDoubleUnderscoreArray();
3620 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3621
3622 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3623 $this->mOutput->mNoGallery = true;
3624 }
3625 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3626 $this->mShowToc = false;
3627 }
3628 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3629 $this->addTrackingCategory( 'hidden-category-category' );
3630 }
3631 # (bug 8068) Allow control over whether robots index a page.
3632 #
3633 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3634 # is not desirable, the last one on the page should win.
3635 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3636 $this->mOutput->setIndexPolicy( 'noindex' );
3637 $this->addTrackingCategory( 'noindex-category' );
3638 }
3639 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3640 $this->mOutput->setIndexPolicy( 'index' );
3641 $this->addTrackingCategory( 'index-category' );
3642 }
3643
3644 # Cache all double underscores in the database
3645 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3646 $this->mOutput->setProperty( $key, '' );
3647 }
3648
3649 wfProfileOut( __METHOD__ );
3650 return $text;
3651 }
3652
3653 /**
3654 * Add a tracking category, getting the title from a system message,
3655 * or print a debug message if the title is invalid.
3656 *
3657 * @param $msg String: message key
3658 * @return Boolean: whether the addition was successful
3659 */
3660 protected function addTrackingCategory( $msg ) {
3661 $cat = wfMsgForContent( $msg );
3662
3663 # Allow tracking categories to be disabled by setting them to "-"
3664 if ( $cat === '-' ) {
3665 return false;
3666 }
3667
3668 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3669 if ( $containerCategory ) {
3670 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3671 return true;
3672 } else {
3673 wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3674 return false;
3675 }
3676 }
3677
3678 /**
3679 * This function accomplishes several tasks:
3680 * 1) Auto-number headings if that option is enabled
3681 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3682 * 3) Add a Table of contents on the top for users who have enabled the option
3683 * 4) Auto-anchor headings
3684 *
3685 * It loops through all headlines, collects the necessary data, then splits up the
3686 * string and re-inserts the newly formatted headlines.
3687 *
3688 * @param $text String
3689 * @param $origText String: original, untouched wikitext
3690 * @param $isMain Boolean
3691 * @private
3692 */
3693 function formatHeadings( $text, $origText, $isMain=true ) {
3694 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3695
3696 # Inhibit editsection links if requested in the page
3697 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3698 $showEditLink = 0;
3699 } else {
3700 $showEditLink = $this->mOptions->getEditSection();
3701 }
3702 if ( $showEditLink ) {
3703 $this->mOutput->setEditSectionTokens( true );
3704 }
3705
3706 # Get all headlines for numbering them and adding funky stuff like [edit]
3707 # links - this is for later, but we need the number of headlines right now
3708 $matches = array();
3709 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3710
3711 # if there are fewer than 4 headlines in the article, do not show TOC
3712 # unless it's been explicitly enabled.
3713 $enoughToc = $this->mShowToc &&
3714 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3715
3716 # Allow user to stipulate that a page should have a "new section"
3717 # link added via __NEWSECTIONLINK__
3718 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3719 $this->mOutput->setNewSection( true );
3720 }
3721
3722 # Allow user to remove the "new section"
3723 # link via __NONEWSECTIONLINK__
3724 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3725 $this->mOutput->hideNewSection( true );
3726 }
3727
3728 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3729 # override above conditions and always show TOC above first header
3730 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3731 $this->mShowToc = true;
3732 $enoughToc = true;
3733 }
3734
3735 # We need this to perform operations on the HTML
3736 $sk = $this->mOptions->getSkin( $this->mTitle );
3737
3738 # headline counter
3739 $headlineCount = 0;
3740 $numVisible = 0;
3741
3742 # Ugh .. the TOC should have neat indentation levels which can be
3743 # passed to the skin functions. These are determined here
3744 $toc = '';
3745 $full = '';
3746 $head = array();
3747 $sublevelCount = array();
3748 $levelCount = array();
3749 $level = 0;
3750 $prevlevel = 0;
3751 $toclevel = 0;
3752 $prevtoclevel = 0;
3753 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3754 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3755 $oldType = $this->mOutputType;
3756 $this->setOutputType( self::OT_WIKI );
3757 $frame = $this->getPreprocessor()->newFrame();
3758 $root = $this->preprocessToDom( $origText );
3759 $node = $root->getFirstChild();
3760 $byteOffset = 0;
3761 $tocraw = array();
3762 $refers = array();
3763
3764 foreach ( $matches[3] as $headline ) {
3765 $isTemplate = false;
3766 $titleText = false;
3767 $sectionIndex = false;
3768 $numbering = '';
3769 $markerMatches = array();
3770 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3771 $serial = $markerMatches[1];
3772 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3773 $isTemplate = ( $titleText != $baseTitleText );
3774 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3775 }
3776
3777 if ( $toclevel ) {
3778 $prevlevel = $level;
3779 }
3780 $level = $matches[1][$headlineCount];
3781
3782 if ( $level > $prevlevel ) {
3783 # Increase TOC level
3784 $toclevel++;
3785 $sublevelCount[$toclevel] = 0;
3786 if ( $toclevel<$wgMaxTocLevel ) {
3787 $prevtoclevel = $toclevel;
3788 $toc .= $sk->tocIndent();
3789 $numVisible++;
3790 }
3791 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3792 # Decrease TOC level, find level to jump to
3793
3794 for ( $i = $toclevel; $i > 0; $i-- ) {
3795 if ( $levelCount[$i] == $level ) {
3796 # Found last matching level
3797 $toclevel = $i;
3798 break;
3799 } elseif ( $levelCount[$i] < $level ) {
3800 # Found first matching level below current level
3801 $toclevel = $i + 1;
3802 break;
3803 }
3804 }
3805 if ( $i == 0 ) {
3806 $toclevel = 1;
3807 }
3808 if ( $toclevel<$wgMaxTocLevel ) {
3809 if ( $prevtoclevel < $wgMaxTocLevel ) {
3810 # Unindent only if the previous toc level was shown :p
3811 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3812 $prevtoclevel = $toclevel;
3813 } else {
3814 $toc .= $sk->tocLineEnd();
3815 }
3816 }
3817 } else {
3818 # No change in level, end TOC line
3819 if ( $toclevel<$wgMaxTocLevel ) {
3820 $toc .= $sk->tocLineEnd();
3821 }
3822 }
3823
3824 $levelCount[$toclevel] = $level;
3825
3826 # count number of headlines for each level
3827 @$sublevelCount[$toclevel]++;
3828 $dot = 0;
3829 for( $i = 1; $i <= $toclevel; $i++ ) {
3830 if ( !empty( $sublevelCount[$i] ) ) {
3831 if ( $dot ) {
3832 $numbering .= '.';
3833 }
3834 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3835 $dot = 1;
3836 }
3837 }
3838
3839 # The safe header is a version of the header text safe to use for links
3840 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3841 $safeHeadline = $this->mStripState->unstripBoth( $headline );
3842
3843 # Remove link placeholders by the link text.
3844 # <!--LINK number-->
3845 # turns into
3846 # link text with suffix
3847 $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3848
3849 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3850 $tocline = preg_replace(
3851 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3852 array( '', '<$1>' ),
3853 $safeHeadline
3854 );
3855 $tocline = trim( $tocline );
3856
3857 # For the anchor, strip out HTML-y stuff period
3858 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3859 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3860
3861 # Save headline for section edit hint before it's escaped
3862 $headlineHint = $safeHeadline;
3863
3864 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3865 # For reverse compatibility, provide an id that's
3866 # HTML4-compatible, like we used to.
3867 #
3868 # It may be worth noting, academically, that it's possible for
3869 # the legacy anchor to conflict with a non-legacy headline
3870 # anchor on the page. In this case likely the "correct" thing
3871 # would be to either drop the legacy anchors or make sure
3872 # they're numbered first. However, this would require people
3873 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3874 # manually, so let's not bother worrying about it.
3875 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3876 array( 'noninitial', 'legacy' ) );
3877 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3878
3879 if ( $legacyHeadline == $safeHeadline ) {
3880 # No reason to have both (in fact, we can't)
3881 $legacyHeadline = false;
3882 }
3883 } else {
3884 $legacyHeadline = false;
3885 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3886 'noninitial' );
3887 }
3888
3889 # HTML names must be case-insensitively unique (bug 10721).
3890 # This does not apply to Unicode characters per
3891 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3892 # FIXME: We may be changing them depending on the current locale.
3893 $arrayKey = strtolower( $safeHeadline );
3894 if ( $legacyHeadline === false ) {
3895 $legacyArrayKey = false;
3896 } else {
3897 $legacyArrayKey = strtolower( $legacyHeadline );
3898 }
3899
3900 # count how many in assoc. array so we can track dupes in anchors
3901 if ( isset( $refers[$arrayKey] ) ) {
3902 $refers[$arrayKey]++;
3903 } else {
3904 $refers[$arrayKey] = 1;
3905 }
3906 if ( isset( $refers[$legacyArrayKey] ) ) {
3907 $refers[$legacyArrayKey]++;
3908 } else {
3909 $refers[$legacyArrayKey] = 1;
3910 }
3911
3912 # Don't number the heading if it is the only one (looks silly)
3913 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
3914 # the two are different if the line contains a link
3915 $headline = $numbering . ' ' . $headline;
3916 }
3917
3918 # Create the anchor for linking from the TOC to the section
3919 $anchor = $safeHeadline;
3920 $legacyAnchor = $legacyHeadline;
3921 if ( $refers[$arrayKey] > 1 ) {
3922 $anchor .= '_' . $refers[$arrayKey];
3923 }
3924 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3925 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3926 }
3927 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3928 $toc .= $sk->tocLine( $anchor, $tocline,
3929 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3930 }
3931
3932 # Add the section to the section tree
3933 # Find the DOM node for this header
3934 while ( $node && !$isTemplate ) {
3935 if ( $node->getName() === 'h' ) {
3936 $bits = $node->splitHeading();
3937 if ( $bits['i'] == $sectionIndex ) {
3938 break;
3939 }
3940 }
3941 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3942 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3943 $node = $node->getNextSibling();
3944 }
3945 $tocraw[] = array(
3946 'toclevel' => $toclevel,
3947 'level' => $level,
3948 'line' => $tocline,
3949 'number' => $numbering,
3950 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3951 'fromtitle' => $titleText,
3952 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3953 'anchor' => $anchor,
3954 );
3955
3956 # give headline the correct <h#> tag
3957 if ( $showEditLink && $sectionIndex !== false ) {
3958 // Output edit section links as markers with styles that can be customized by skins
3959 if ( $isTemplate ) {
3960 # Put a T flag in the section identifier, to indicate to extractSections()
3961 # that sections inside <includeonly> should be counted.
3962 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3963 } else {
3964 $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
3965 }
3966 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3967 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3968 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3969 // so we don't have to worry about a user trying to input one of these markers directly.
3970 // We use a page and section attribute to stop the language converter from converting these important bits
3971 // of data, but put the headline hint inside a content block because the language converter is supposed to
3972 // be able to convert that piece of data.
3973 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
3974 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
3975 if ( isset($editlinkArgs[2]) ) {
3976 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
3977 } else {
3978 $editlink .= '/>';
3979 }
3980 } else {
3981 $editlink = '';
3982 }
3983 $head[$headlineCount] = $sk->makeHeadline( $level,
3984 $matches['attrib'][$headlineCount], $anchor, $headline,
3985 $editlink, $legacyAnchor );
3986
3987 $headlineCount++;
3988 }
3989
3990 $this->setOutputType( $oldType );
3991
3992 # Never ever show TOC if no headers
3993 if ( $numVisible < 1 ) {
3994 $enoughToc = false;
3995 }
3996
3997 if ( $enoughToc ) {
3998 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3999 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4000 }
4001 $toc = $sk->tocList( $toc, $this->mOptions->getUserLang() );
4002 $this->mOutput->setTOCHTML( $toc );
4003 }
4004
4005 if ( $isMain ) {
4006 $this->mOutput->setSections( $tocraw );
4007 }
4008
4009 # split up and insert constructed headlines
4010
4011 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4012 $i = 0;
4013
4014 foreach ( $blocks as $block ) {
4015 if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4016 # This is the [edit] link that appears for the top block of text when
4017 # section editing is enabled
4018
4019 # Disabled because it broke block formatting
4020 # For example, a bullet point in the top line
4021 # $full .= $sk->editSectionLink(0);
4022 }
4023 $full .= $block;
4024 if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4025 # Top anchor now in skin
4026 $full = $full.$toc;
4027 }
4028
4029 if ( !empty( $head[$i] ) ) {
4030 $full .= $head[$i];
4031 }
4032 $i++;
4033 }
4034 if ( $this->mForceTocPosition ) {
4035 return str_replace( '<!--MWTOC-->', $toc, $full );
4036 } else {
4037 return $full;
4038 }
4039 }
4040
4041 /**
4042 * Transform wiki markup when saving a page by doing \r\n -> \n
4043 * conversion, substitting signatures, {{subst:}} templates, etc.
4044 *
4045 * @param $text String: the text to transform
4046 * @param $title Title: the Title object for the current article
4047 * @param $user User: the User object describing the current user
4048 * @param $options ParserOptions: parsing options
4049 * @param $clearState Boolean: whether to clear the parser state first
4050 * @return String: the altered wiki markup
4051 */
4052 public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4053 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4054 $this->setUser( $user );
4055
4056 $pairs = array(
4057 "\r\n" => "\n",
4058 );
4059 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4060 if( $options->getPreSaveTransform() ) {
4061 $text = $this->pstPass2( $text, $user );
4062 }
4063 $text = $this->mStripState->unstripBoth( $text );
4064
4065 $this->setUser( null ); #Reset
4066
4067 return $text;
4068 }
4069
4070 /**
4071 * Pre-save transform helper function
4072 * @private
4073 */
4074 function pstPass2( $text, $user ) {
4075 global $wgContLang, $wgLocaltimezone;
4076
4077 # Note: This is the timestamp saved as hardcoded wikitext to
4078 # the database, we use $wgContLang here in order to give
4079 # everyone the same signature and use the default one rather
4080 # than the one selected in each user's preferences.
4081 # (see also bug 12815)
4082 $ts = $this->mOptions->getTimestamp();
4083 if ( isset( $wgLocaltimezone ) ) {
4084 $tz = $wgLocaltimezone;
4085 } else {
4086 $tz = date_default_timezone_get();
4087 }
4088
4089 $unixts = wfTimestamp( TS_UNIX, $ts );
4090 $oldtz = date_default_timezone_get();
4091 date_default_timezone_set( $tz );
4092 $ts = date( 'YmdHis', $unixts );
4093 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4094
4095 # Allow translation of timezones through wiki. date() can return
4096 # whatever crap the system uses, localised or not, so we cannot
4097 # ship premade translations.
4098 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4099 $msg = wfMessage( $key )->inContentLanguage();
4100 if ( $msg->exists() ) {
4101 $tzMsg = $msg->text();
4102 }
4103
4104 date_default_timezone_set( $oldtz );
4105
4106 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4107
4108 # Variable replacement
4109 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4110 $text = $this->replaceVariables( $text );
4111
4112 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4113 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4114
4115 # Signatures
4116 $sigText = $this->getUserSig( $user );
4117 $text = strtr( $text, array(
4118 '~~~~~' => $d,
4119 '~~~~' => "$sigText $d",
4120 '~~~' => $sigText
4121 ) );
4122
4123 # Context links: [[|name]] and [[name (context)|]]
4124 global $wgLegalTitleChars;
4125 $tc = "[$wgLegalTitleChars]";
4126 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4127
4128 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4129 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/"; # [[ns:page（context）|]]
4130 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4131 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4132
4133 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4134 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4135 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4136 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4137
4138 $t = $this->mTitle->getText();
4139 $m = array();
4140 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4141 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4142 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4143 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4144 } else {
4145 # if there's no context, don't bother duplicating the title
4146 $text = preg_replace( $p2, '[[\\1]]', $text );
4147 }
4148
4149 # Trim trailing whitespace
4150 $text = rtrim( $text );
4151
4152 return $text;
4153 }
4154
4155 /**
4156 * Fetch the user's signature text, if any, and normalize to
4157 * validated, ready-to-insert wikitext.
4158 * If you have pre-fetched the nickname or the fancySig option, you can
4159 * specify them here to save a database query.
4160 * Do not reuse this parser instance after calling getUserSig(),
4161 * as it may have changed if it's the $wgParser.
4162 *
4163 * @param $user User
4164 * @param $nickname String: nickname to use or false to use user's default nickname
4165 * @param $fancySig Boolean: whether the nicknname is the complete signature
4166 * or null to use default value
4167 * @return string
4168 */
4169 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4170 global $wgMaxSigChars;
4171
4172 $username = $user->getName();
4173
4174 # If not given, retrieve from the user object.
4175 if ( $nickname === false )
4176 $nickname = $user->getOption( 'nickname' );
4177
4178 if ( is_null( $fancySig ) ) {
4179 $fancySig = $user->getBoolOption( 'fancysig' );
4180 }
4181
4182 $nickname = $nickname == null ? $username : $nickname;
4183
4184 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4185 $nickname = $username;
4186 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4187 } elseif ( $fancySig !== false ) {
4188 # Sig. might contain markup; validate this
4189 if ( $this->validateSig( $nickname ) !== false ) {
4190 # Validated; clean up (if needed) and return it
4191 return $this->cleanSig( $nickname, true );
4192 } else {
4193 # Failed to validate; fall back to the default
4194 $nickname = $username;
4195 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4196 }
4197 }
4198
4199 # Make sure nickname doesnt get a sig in a sig
4200 $nickname = $this->cleanSigInSig( $nickname );
4201
4202 # If we're still here, make it a link to the user page
4203 $userText = wfEscapeWikiText( $username );
4204 $nickText = wfEscapeWikiText( $nickname );
4205 if ( $user->isAnon() ) {
4206 return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4207 } else {
4208 return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4209 }
4210 }
4211
4212 /**
4213 * Check that the user's signature contains no bad XML
4214 *
4215 * @param $text String
4216 * @return mixed An expanded string, or false if invalid.
4217 */
4218 function validateSig( $text ) {
4219 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4220 }
4221
4222 /**
4223 * Clean up signature text
4224 *
4225 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4226 * 2) Substitute all transclusions
4227 *
4228 * @param $text String
4229 * @param $parsing Whether we're cleaning (preferences save) or parsing
4230 * @return String: signature text
4231 */
4232 function cleanSig( $text, $parsing = false ) {
4233 if ( !$parsing ) {
4234 global $wgTitle;
4235 $this->mOptions = new ParserOptions;
4236 $this->clearState();
4237 $this->setTitle( $wgTitle );
4238 $this->setOutputType = self::OT_PREPROCESS;
4239 }
4240
4241 # Option to disable this feature
4242 if ( !$this->mOptions->getCleanSignatures() ) {
4243 return $text;
4244 }
4245
4246 # FIXME: regex doesn't respect extension tags or nowiki
4247 # => Move this logic to braceSubstitution()
4248 $substWord = MagicWord::get( 'subst' );
4249 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4250 $substText = '{{' . $substWord->getSynonym( 0 );
4251
4252 $text = preg_replace( $substRegex, $substText, $text );
4253 $text = $this->cleanSigInSig( $text );
4254 $dom = $this->preprocessToDom( $text );
4255 $frame = $this->getPreprocessor()->newFrame();
4256 $text = $frame->expand( $dom );
4257
4258 if ( !$parsing ) {
4259 $text = $this->mStripState->unstripBoth( $text );
4260 }
4261
4262 return $text;
4263 }
4264
4265 /**
4266 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4267 *
4268 * @param $text String
4269 * @return String: signature text with /~{3,5}/ removed
4270 */
4271 function cleanSigInSig( $text ) {
4272 $text = preg_replace( '/~{3,5}/', '', $text );
4273 return $text;
4274 }
4275
4276 /**
4277 * Set up some variables which are usually set up in parse()
4278 * so that an external function can call some class members with confidence
4279 */
4280 public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4281 $this->startParse( $title, $options, $outputType, $clearState );
4282 }
4283
4284 private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4285 $this->setTitle( $title );
4286 $this->mOptions = $options;
4287 $this->setOutputType( $outputType );
4288 if ( $clearState ) {
4289 $this->clearState();
4290 }
4291 }
4292
4293 /**
4294 * Wrapper for preprocess()
4295 *
4296 * @param $text String: the text to preprocess
4297 * @param $options ParserOptions: options
4298 * @param $title Title object or null to use $wgTitle
4299 * @return String
4300 */
4301 public function transformMsg( $text, $options, $title = null ) {
4302 static $executing = false;
4303
4304 # Guard against infinite recursion
4305 if ( $executing ) {
4306 return $text;
4307 }
4308 $executing = true;
4309
4310 wfProfileIn( __METHOD__ );
4311 if ( !$title ) {
4312 global $wgTitle;
4313 $title = $wgTitle;
4314 }
4315 if ( !$title ) {
4316 # It's not uncommon having a null $wgTitle in scripts. See r80898
4317 # Create a ghost title in such case
4318 $title = Title::newFromText( 'Dwimmerlaik' );
4319 }
4320 $text = $this->preprocess( $text, $title, $options );
4321
4322 $executing = false;
4323 wfProfileOut( __METHOD__ );
4324 return $text;
4325 }
4326
4327 /**
4328 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4329 * The callback should have the following form:
4330 * function myParserHook( $text, $params, $parser ) { ... }
4331 *
4332 * Transform and return $text. Use $parser for any required context, e.g. use
4333 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4334 *
4335 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4336 * @param $callback Mixed: the callback function (and object) to use for the tag
4337 * @return The old value of the mTagHooks array associated with the hook
4338 */
4339 public function setHook( $tag, $callback ) {
4340 $tag = strtolower( $tag );
4341 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4342 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4343 $this->mTagHooks[$tag] = $callback;
4344 if ( !in_array( $tag, $this->mStripList ) ) {
4345 $this->mStripList[] = $tag;
4346 }
4347
4348 return $oldVal;
4349 }
4350
4351 function setTransparentTagHook( $tag, $callback ) {
4352 $tag = strtolower( $tag );
4353 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4354 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4355 $this->mTransparentTagHooks[$tag] = $callback;
4356
4357 return $oldVal;
4358 }
4359
4360 /**
4361 * Remove all tag hooks
4362 */
4363 function clearTagHooks() {
4364 $this->mTagHooks = array();
4365 $this->mStripList = $this->mDefaultStripList;
4366 }
4367
4368 /**
4369 * Create a function, e.g. {{sum:1|2|3}}
4370 * The callback function should have the form:
4371 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4372 *
4373 * Or with SFH_OBJECT_ARGS:
4374 * function myParserFunction( $parser, $frame, $args ) { ... }
4375 *
4376 * The callback may either return the text result of the function, or an array with the text
4377 * in element 0, and a number of flags in the other elements. The names of the flags are
4378 * specified in the keys. Valid flags are:
4379 * found The text returned is valid, stop processing the template. This
4380 * is on by default.
4381 * nowiki Wiki markup in the return value should be escaped
4382 * isHTML The returned text is HTML, armour it against wikitext transformation
4383 *
4384 * @param $id String: The magic word ID
4385 * @param $callback Mixed: the callback function (and object) to use
4386 * @param $flags Integer: a combination of the following flags:
4387 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4388 *
4389 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4390 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4391 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4392 * the arguments, and to control the way they are expanded.
4393 *
4394 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4395 * arguments, for instance:
4396 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4397 *
4398 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4399 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4400 * working if/when this is changed.
4401 *
4402 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4403 * expansion.
4404 *
4405 * Please read the documentation in includes/parser/Preprocessor.php for more information
4406 * about the methods available in PPFrame and PPNode.
4407 *
4408 * @return The old callback function for this name, if any
4409 */
4410 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4411 global $wgContLang;
4412
4413 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4414 $this->mFunctionHooks[$id] = array( $callback, $flags );
4415
4416 # Add to function cache
4417 $mw = MagicWord::get( $id );
4418 if ( !$mw )
4419 throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4420
4421 $synonyms = $mw->getSynonyms();
4422 $sensitive = intval( $mw->isCaseSensitive() );
4423
4424 foreach ( $synonyms as $syn ) {
4425 # Case
4426 if ( !$sensitive ) {
4427 $syn = $wgContLang->lc( $syn );
4428 }
4429 # Add leading hash
4430 if ( !( $flags & SFH_NO_HASH ) ) {
4431 $syn = '#' . $syn;
4432 }
4433 # Remove trailing colon
4434 if ( substr( $syn, -1, 1 ) === ':' ) {
4435 $syn = substr( $syn, 0, -1 );
4436 }
4437 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4438 }
4439 return $oldVal;
4440 }
4441
4442 /**
4443 * Get all registered function hook identifiers
4444 *
4445 * @return Array
4446 */
4447 function getFunctionHooks() {
4448 return array_keys( $this->mFunctionHooks );
4449 }
4450
4451 /**
4452 * Create a tag function, e.g. <test>some stuff</test>.
4453 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4454 * Unlike parser functions, their content is not preprocessed.
4455 */
4456 function setFunctionTagHook( $tag, $callback, $flags ) {
4457 $tag = strtolower( $tag );
4458 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4459 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4460 $this->mFunctionTagHooks[$tag] : null;
4461 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4462
4463 if ( !in_array( $tag, $this->mStripList ) ) {
4464 $this->mStripList[] = $tag;
4465 }
4466
4467 return $old;
4468 }
4469
4470 /**
4471 * FIXME: update documentation. makeLinkObj() is deprecated.
4472 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4473 * Placeholders created in Skin::makeLinkObj()
4474 * Returns an array of link CSS classes, indexed by PDBK.
4475 */
4476 function replaceLinkHolders( &$text, $options = 0 ) {
4477 return $this->mLinkHolders->replace( $text );
4478 }
4479
4480 /**
4481 * Replace <!--LINK--> link placeholders with plain text of links
4482 * (not HTML-formatted).
4483 *
4484 * @param $text String
4485 * @return String
4486 */
4487 function replaceLinkHoldersText( $text ) {
4488 return $this->mLinkHolders->replaceText( $text );
4489 }
4490
4491 /**
4492 * Renders an image gallery from a text with one line per image.
4493 * text labels may be given by using |-style alternative text. E.g.
4494 * Image:one.jpg|The number "1"
4495 * Image:tree.jpg|A tree
4496 * given as text will return the HTML of a gallery with two images,
4497 * labeled 'The number "1"' and
4498 * 'A tree'.
4499 */
4500 function renderImageGallery( $text, $params ) {
4501 $ig = new ImageGallery();
4502 $ig->setContextTitle( $this->mTitle );
4503 $ig->setShowBytes( false );
4504 $ig->setShowFilename( false );
4505 $ig->setParser( $this );
4506 $ig->setHideBadImages();
4507 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4508 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4509 $ig->mRevisionId = $this->mRevisionId;
4510
4511 if ( isset( $params['showfilename'] ) ) {
4512 $ig->setShowFilename( true );
4513 } else {
4514 $ig->setShowFilename( false );
4515 }
4516 if ( isset( $params['caption'] ) ) {
4517 $caption = $params['caption'];
4518 $caption = htmlspecialchars( $caption );
4519 $caption = $this->replaceInternalLinks( $caption );
4520 $ig->setCaptionHtml( $caption );
4521 }
4522 if ( isset( $params['perrow'] ) ) {
4523 $ig->setPerRow( $params['perrow'] );
4524 }
4525 if ( isset( $params['widths'] ) ) {
4526 $ig->setWidths( $params['widths'] );
4527 }
4528 if ( isset( $params['heights'] ) ) {
4529 $ig->setHeights( $params['heights'] );
4530 }
4531
4532 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4533
4534 $lines = StringUtils::explode( "\n", $text );
4535 foreach ( $lines as $line ) {
4536 # match lines like these:
4537 # Image:someimage.jpg|This is some image
4538 $matches = array();
4539 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4540 # Skip empty lines
4541 if ( count( $matches ) == 0 ) {
4542 continue;
4543 }
4544
4545 if ( strpos( $matches[0], '%' ) !== false ) {
4546 $matches[1] = rawurldecode( $matches[1] );
4547 }
4548 $tp = Title::newFromText( $matches[1], NS_FILE );
4549 $nt =& $tp;
4550 if ( is_null( $nt ) ) {
4551 # Bogus title. Ignore these so we don't bomb out later.
4552 continue;
4553 }
4554 if ( isset( $matches[3] ) ) {
4555 $label = $matches[3];
4556 } else {
4557 $label = '';
4558 }
4559
4560 $html = $this->recursiveTagParse( trim( $label ) );
4561
4562 $ig->add( $nt, $html );
4563
4564 # Only add real images (bug #5586)
4565 if ( $nt->getNamespace() == NS_FILE ) {
4566 $this->mOutput->addImage( $nt->getDBkey() );
4567 }
4568 }
4569 return $ig->toHTML();
4570 }
4571
4572 function getImageParams( $handler ) {
4573 if ( $handler ) {
4574 $handlerClass = get_class( $handler );
4575 } else {
4576 $handlerClass = '';
4577 }
4578 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
4579 # Initialise static lists
4580 static $internalParamNames = array(
4581 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4582 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4583 'bottom', 'text-bottom' ),
4584 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4585 'upright', 'border', 'link', 'alt' ),
4586 );
4587 static $internalParamMap;
4588 if ( !$internalParamMap ) {
4589 $internalParamMap = array();
4590 foreach ( $internalParamNames as $type => $names ) {
4591 foreach ( $names as $name ) {
4592 $magicName = str_replace( '-', '_', "img_$name" );
4593 $internalParamMap[$magicName] = array( $type, $name );
4594 }
4595 }
4596 }
4597
4598 # Add handler params
4599 $paramMap = $internalParamMap;
4600 if ( $handler ) {
4601 $handlerParamMap = $handler->getParamMap();
4602 foreach ( $handlerParamMap as $magic => $paramName ) {
4603 $paramMap[$magic] = array( 'handler', $paramName );
4604 }
4605 }
4606 $this->mImageParams[$handlerClass] = $paramMap;
4607 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4608 }
4609 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4610 }
4611
4612 /**
4613 * Parse image options text and use it to make an image
4614 *
4615 * @param $title Title
4616 * @param $options String
4617 * @param $holders LinkHolderArray
4618 */
4619 function makeImage( $title, $options, $holders = false ) {
4620 # Check if the options text is of the form "options|alt text"
4621 # Options are:
4622 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4623 # * left no resizing, just left align. label is used for alt= only
4624 # * right same, but right aligned
4625 # * none same, but not aligned
4626 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4627 # * center center the image
4628 # * frame Keep original image size, no magnify-button.
4629 # * framed Same as "frame"
4630 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4631 # * upright reduce width for upright images, rounded to full __0 px
4632 # * border draw a 1px border around the image
4633 # * alt Text for HTML alt attribute (defaults to empty)
4634 # * link Set the target of the image link. Can be external, interwiki, or local
4635 # vertical-align values (no % or length right now):
4636 # * baseline
4637 # * sub
4638 # * super
4639 # * top
4640 # * text-top
4641 # * middle
4642 # * bottom
4643 # * text-bottom
4644
4645 $parts = StringUtils::explode( "|", $options );
4646 $sk = $this->mOptions->getSkin( $this->mTitle );
4647
4648 # Give extensions a chance to select the file revision for us
4649 $skip = $time = $descQuery = false;
4650 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4651
4652 if ( $skip ) {
4653 return $sk->link( $title );
4654 }
4655
4656 # Get the file
4657 $file = wfFindFile( $title, array( 'time' => $time ) );
4658 # Get parameter map
4659 $handler = $file ? $file->getHandler() : false;
4660
4661 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4662
4663 # Process the input parameters
4664 $caption = '';
4665 $params = array( 'frame' => array(), 'handler' => array(),
4666 'horizAlign' => array(), 'vertAlign' => array() );
4667 foreach ( $parts as $part ) {
4668 $part = trim( $part );
4669 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4670 $validated = false;
4671 if ( isset( $paramMap[$magicName] ) ) {
4672 list( $type, $paramName ) = $paramMap[$magicName];
4673
4674 # Special case; width and height come in one variable together
4675 if ( $type === 'handler' && $paramName === 'width' ) {
4676 $m = array();
4677 # (bug 13500) In both cases (width/height and width only),
4678 # permit trailing "px" for backward compatibility.
4679 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4680 $width = intval( $m[1] );
4681 $height = intval( $m[2] );
4682 if ( $handler->validateParam( 'width', $width ) ) {
4683 $params[$type]['width'] = $width;
4684 $validated = true;
4685 }
4686 if ( $handler->validateParam( 'height', $height ) ) {
4687 $params[$type]['height'] = $height;
4688 $validated = true;
4689 }
4690 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4691 $width = intval( $value );
4692 if ( $handler->validateParam( 'width', $width ) ) {
4693 $params[$type]['width'] = $width;
4694 $validated = true;
4695 }
4696 } # else no validation -- bug 13436
4697 } else {
4698 if ( $type === 'handler' ) {
4699 # Validate handler parameter
4700 $validated = $handler->validateParam( $paramName, $value );
4701 } else {
4702 # Validate internal parameters
4703 switch( $paramName ) {
4704 case 'manualthumb':
4705 case 'alt':
4706 # @todo Fixme: possibly check validity here for
4707 # manualthumb? downstream behavior seems odd with
4708 # missing manual thumbs.
4709 $validated = true;
4710 $value = $this->stripAltText( $value, $holders );
4711 break;
4712 case 'link':
4713 $chars = self::EXT_LINK_URL_CLASS;
4714 $prots = $this->mUrlProtocols;
4715 if ( $value === '' ) {
4716 $paramName = 'no-link';
4717 $value = true;
4718 $validated = true;
4719 } elseif ( preg_match( "/^$prots/", $value ) ) {
4720 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4721 $paramName = 'link-url';
4722 $this->mOutput->addExternalLink( $value );
4723 if ( $this->mOptions->getExternalLinkTarget() ) {
4724 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4725 }
4726 $validated = true;
4727 }
4728 } else {
4729 $linkTitle = Title::newFromText( $value );
4730 if ( $linkTitle ) {
4731 $paramName = 'link-title';
4732 $value = $linkTitle;
4733 $this->mOutput->addLink( $linkTitle );
4734 $validated = true;
4735 }
4736 }
4737 break;
4738 default:
4739 # Most other things appear to be empty or numeric...
4740 $validated = ( $value === false || is_numeric( trim( $value ) ) );
4741 }
4742 }
4743
4744 if ( $validated ) {
4745 $params[$type][$paramName] = $value;
4746 }
4747 }
4748 }
4749 if ( !$validated ) {
4750 $caption = $part;
4751 }
4752 }
4753
4754 # Process alignment parameters
4755 if ( $params['horizAlign'] ) {
4756 $params['frame']['align'] = key( $params['horizAlign'] );
4757 }
4758 if ( $params['vertAlign'] ) {
4759 $params['frame']['valign'] = key( $params['vertAlign'] );
4760 }
4761
4762 $params['frame']['caption'] = $caption;
4763
4764 # Will the image be presented in a frame, with the caption below?
4765 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4766 isset( $params['frame']['framed'] ) ||
4767 isset( $params['frame']['thumbnail'] ) ||
4768 isset( $params['frame']['manualthumb'] );
4769
4770 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
4771 # came to also set the caption, ordinary text after the image -- which
4772 # makes no sense, because that just repeats the text multiple times in
4773 # screen readers. It *also* came to set the title attribute.
4774 #
4775 # Now that we have an alt attribute, we should not set the alt text to
4776 # equal the caption: that's worse than useless, it just repeats the
4777 # text. This is the framed/thumbnail case. If there's no caption, we
4778 # use the unnamed parameter for alt text as well, just for the time be-
4779 # ing, if the unnamed param is set and the alt param is not.
4780 #
4781 # For the future, we need to figure out if we want to tweak this more,
4782 # e.g., introducing a title= parameter for the title; ignoring the un-
4783 # named parameter entirely for images without a caption; adding an ex-
4784 # plicit caption= parameter and preserving the old magic unnamed para-
4785 # meter for BC; ...
4786 if ( $imageIsFramed ) { # Framed image
4787 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4788 # No caption or alt text, add the filename as the alt text so
4789 # that screen readers at least get some description of the image
4790 $params['frame']['alt'] = $title->getText();
4791 }
4792 # Do not set $params['frame']['title'] because tooltips don't make sense
4793 # for framed images
4794 } else { # Inline image
4795 if ( !isset( $params['frame']['alt'] ) ) {
4796 # No alt text, use the "caption" for the alt text
4797 if ( $caption !== '') {
4798 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4799 } else {
4800 # No caption, fall back to using the filename for the
4801 # alt text
4802 $params['frame']['alt'] = $title->getText();
4803 }
4804 }
4805 # Use the "caption" for the tooltip text
4806 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4807 }
4808
4809 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4810
4811 # Linker does the rest
4812 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4813
4814 # Give the handler a chance to modify the parser object
4815 if ( $handler ) {
4816 $handler->parserTransformHook( $this, $file );
4817 }
4818
4819 return $ret;
4820 }
4821
4822 protected function stripAltText( $caption, $holders ) {
4823 # Strip bad stuff out of the title (tooltip). We can't just use
4824 # replaceLinkHoldersText() here, because if this function is called
4825 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4826 if ( $holders ) {
4827 $tooltip = $holders->replaceText( $caption );
4828 } else {
4829 $tooltip = $this->replaceLinkHoldersText( $caption );
4830 }
4831
4832 # make sure there are no placeholders in thumbnail attributes
4833 # that are later expanded to html- so expand them now and
4834 # remove the tags
4835 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4836 $tooltip = Sanitizer::stripAllTags( $tooltip );
4837
4838 return $tooltip;
4839 }
4840
4841 /**
4842 * Set a flag in the output object indicating that the content is dynamic and
4843 * shouldn't be cached.
4844 */
4845 function disableCache() {
4846 wfDebug( "Parser output marked as uncacheable.\n" );
4847 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4848 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4849 }
4850
4851 /**
4852 * Callback from the Sanitizer for expanding items found in HTML attribute
4853 * values, so they can be safely tested and escaped.
4854 *
4855 * @param $text String
4856 * @param $frame PPFrame
4857 * @return String
4858 * @private
4859 */
4860 function attributeStripCallback( &$text, $frame = false ) {
4861 $text = $this->replaceVariables( $text, $frame );
4862 $text = $this->mStripState->unstripBoth( $text );
4863 return $text;
4864 }
4865
4866 /**
4867 * Accessor
4868 */
4869 function getTags() {
4870 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4871 }
4872
4873 /**
4874 * Replace transparent tags in $text with the values given by the callbacks.
4875 *
4876 * Transparent tag hooks are like regular XML-style tag hooks, except they
4877 * operate late in the transformation sequence, on HTML instead of wikitext.
4878 */
4879 function replaceTransparentTags( $text ) {
4880 $matches = array();
4881 $elements = array_keys( $this->mTransparentTagHooks );
4882 $text = $this->extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
4883
4884 foreach ( $matches as $marker => $data ) {
4885 list( $element, $content, $params, $tag ) = $data;
4886 $tagName = strtolower( $element );
4887 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
4888 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
4889 } else {
4890 $output = $tag;
4891 }
4892 $this->mStripState->addGeneral( $marker, $output );
4893 }
4894 return $text;
4895 }
4896
4897 /**
4898 * Break wikitext input into sections, and either pull or replace
4899 * some particular section's text.
4900 *
4901 * External callers should use the getSection and replaceSection methods.
4902 *
4903 * @param $text String: Page wikitext
4904 * @param $section String: a section identifier string of the form:
4905 * <flag1> - <flag2> - ... - <section number>
4906 *
4907 * Currently the only recognised flag is "T", which means the target section number
4908 * was derived during a template inclusion parse, in other words this is a template
4909 * section edit link. If no flags are given, it was an ordinary section edit link.
4910 * This flag is required to avoid a section numbering mismatch when a section is
4911 * enclosed by <includeonly> (bug 6563).
4912 *
4913 * The section number 0 pulls the text before the first heading; other numbers will
4914 * pull the given section along with its lower-level subsections. If the section is
4915 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4916 *
4917 * @param $mode String: one of "get" or "replace"
4918 * @param $newText String: replacement text for section data.
4919 * @return String: for "get", the extracted section text.
4920 * for "replace", the whole page with the section replaced.
4921 */
4922 private function extractSections( $text, $section, $mode, $newText='' ) {
4923 global $wgTitle; # not generally used but removes an ugly failure mode
4924 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
4925 $outText = '';
4926 $frame = $this->getPreprocessor()->newFrame();
4927
4928 # Process section extraction flags
4929 $flags = 0;
4930 $sectionParts = explode( '-', $section );
4931 $sectionIndex = array_pop( $sectionParts );
4932 foreach ( $sectionParts as $part ) {
4933 if ( $part === 'T' ) {
4934 $flags |= self::PTD_FOR_INCLUSION;
4935 }
4936 }
4937 # Preprocess the text
4938 $root = $this->preprocessToDom( $text, $flags );
4939
4940 # <h> nodes indicate section breaks
4941 # They can only occur at the top level, so we can find them by iterating the root's children
4942 $node = $root->getFirstChild();
4943
4944 # Find the target section
4945 if ( $sectionIndex == 0 ) {
4946 # Section zero doesn't nest, level=big
4947 $targetLevel = 1000;
4948 } else {
4949 while ( $node ) {
4950 if ( $node->getName() === 'h' ) {
4951 $bits = $node->splitHeading();
4952 if ( $bits['i'] == $sectionIndex ) {
4953 $targetLevel = $bits['level'];
4954 break;
4955 }
4956 }
4957 if ( $mode === 'replace' ) {
4958 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4959 }
4960 $node = $node->getNextSibling();
4961 }
4962 }
4963
4964 if ( !$node ) {
4965 # Not found
4966 if ( $mode === 'get' ) {
4967 return $newText;
4968 } else {
4969 return $text;
4970 }
4971 }
4972
4973 # Find the end of the section, including nested sections
4974 do {
4975 if ( $node->getName() === 'h' ) {
4976 $bits = $node->splitHeading();
4977 $curLevel = $bits['level'];
4978 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4979 break;
4980 }
4981 }
4982 if ( $mode === 'get' ) {
4983 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4984 }
4985 $node = $node->getNextSibling();
4986 } while ( $node );
4987
4988 # Write out the remainder (in replace mode only)
4989 if ( $mode === 'replace' ) {
4990 # Output the replacement text
4991 # Add two newlines on -- trailing whitespace in $newText is conventionally
4992 # stripped by the editor, so we need both newlines to restore the paragraph gap
4993 # Only add trailing whitespace if there is newText
4994 if ( $newText != "" ) {
4995 $outText .= $newText . "\n\n";
4996 }
4997
4998 while ( $node ) {
4999 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5000 $node = $node->getNextSibling();
5001 }
5002 }
5003
5004 if ( is_string( $outText ) ) {
5005 # Re-insert stripped tags
5006 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5007 }
5008
5009 return $outText;
5010 }
5011
5012 /**
5013 * This function returns the text of a section, specified by a number ($section).
5014 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5015 * the first section before any such heading (section 0).
5016 *
5017 * If a section contains subsections, these are also returned.
5018 *
5019 * @param $text String: text to look in
5020 * @param $section String: section identifier
5021 * @param $deftext String: default to return if section is not found
5022 * @return string text of the requested section
5023 */
5024 public function getSection( $text, $section, $deftext='' ) {
5025 return $this->extractSections( $text, $section, "get", $deftext );
5026 }
5027
5028 /**
5029 * This function returns $oldtext after the content of the section
5030 * specified by $section has been replaced with $text.
5031 *
5032 * @param $text String: former text of the article
5033 * @param $section Numeric: section identifier
5034 * @param $text String: replacing text
5035 * #return String: modified text
5036 */
5037 public function replaceSection( $oldtext, $section, $text ) {
5038 return $this->extractSections( $oldtext, $section, "replace", $text );
5039 }
5040
5041 /**
5042 * Get the ID of the revision we are parsing
5043 *
5044 * @return Mixed: integer or null
5045 */
5046 function getRevisionId() {
5047 return $this->mRevisionId;
5048 }
5049
5050 /**
5051 * Get the revision object for $this->mRevisionId
5052 *
5053 * @return either a Revision object or null
5054 */
5055 protected function getRevisionObject() {
5056 if ( !is_null( $this->mRevisionObject ) ) {
5057 return $this->mRevisionObject;
5058 }
5059 if ( is_null( $this->mRevisionId ) ) {
5060 return null;
5061 }
5062
5063 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5064 return $this->mRevisionObject;
5065 }
5066
5067 /**
5068 * Get the timestamp associated with the current revision, adjusted for
5069 * the default server-local timestamp
5070 */
5071 function getRevisionTimestamp() {
5072 if ( is_null( $this->mRevisionTimestamp ) ) {
5073 wfProfileIn( __METHOD__ );
5074
5075 $revObject = $this->getRevisionObject();
5076 $timestamp = $revObject ? $revObject->getTimestamp() : false;
5077
5078 if( $timestamp !== false ) {
5079 global $wgContLang;
5080
5081 # The cryptic '' timezone parameter tells to use the site-default
5082 # timezone offset instead of the user settings.
5083 #
5084 # Since this value will be saved into the parser cache, served
5085 # to other users, and potentially even used inside links and such,
5086 # it needs to be consistent for all visitors.
5087 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5088 }
5089
5090 wfProfileOut( __METHOD__ );
5091 }
5092 return $this->mRevisionTimestamp;
5093 }
5094
5095 /**
5096 * Get the name of the user that edited the last revision
5097 *
5098 * @return String: user name
5099 */
5100 function getRevisionUser() {
5101 if( is_null( $this->mRevisionUser ) ) {
5102 $revObject = $this->getRevisionObject();
5103
5104 # if this template is subst: the revision id will be blank,
5105 # so just use the current user's name
5106 if( $revObject ) {
5107 $this->mRevisionUser = $revObject->getUserText();
5108 } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5109 $this->mRevisionUser = $this->getUser()->getName();
5110 }
5111 }
5112 return $this->mRevisionUser;
5113 }
5114
5115 /**
5116 * Mutator for $mDefaultSort
5117 *
5118 * @param $sort New value
5119 */
5120 public function setDefaultSort( $sort ) {
5121 $this->mDefaultSort = $sort;
5122 $this->mOutput->setProperty( 'defaultsort', $sort );
5123 }
5124
5125 /**
5126 * Accessor for $mDefaultSort
5127 * Will use the empty string if none is set.
5128 *
5129 * This value is treated as a prefix, so the
5130 * empty string is equivalent to sorting by
5131 * page name.
5132 *
5133 * @return string
5134 */
5135 public function getDefaultSort() {
5136 if ( $this->mDefaultSort !== false ) {
5137 return $this->mDefaultSort;
5138 } else {
5139 return '';
5140 }
5141 }
5142
5143 /**
5144 * Accessor for $mDefaultSort
5145 * Unlike getDefaultSort(), will return false if none is set
5146 *
5147 * @return string or false
5148 */
5149 public function getCustomDefaultSort() {
5150 return $this->mDefaultSort;
5151 }
5152
5153 /**
5154 * Try to guess the section anchor name based on a wikitext fragment
5155 * presumably extracted from a heading, for example "Header" from
5156 * "== Header ==".
5157 */
5158 public function guessSectionNameFromWikiText( $text ) {
5159 # Strip out wikitext links(they break the anchor)
5160 $text = $this->stripSectionName( $text );
5161 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5162 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5163 }
5164
5165 /**
5166 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5167 * instead. For use in redirects, since IE6 interprets Redirect: headers
5168 * as something other than UTF-8 (apparently?), resulting in breakage.
5169 *
5170 * @param $text String: The section name
5171 * @return string An anchor
5172 */
5173 public function guessLegacySectionNameFromWikiText( $text ) {
5174 # Strip out wikitext links(they break the anchor)
5175 $text = $this->stripSectionName( $text );
5176 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5177 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5178 }
5179
5180 /**
5181 * Strips a text string of wikitext for use in a section anchor
5182 *
5183 * Accepts a text string and then removes all wikitext from the
5184 * string and leaves only the resultant text (i.e. the result of
5185 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5186 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5187 * to create valid section anchors by mimicing the output of the
5188 * parser when headings are parsed.
5189 *
5190 * @param $text String: text string to be stripped of wikitext
5191 * for use in a Section anchor
5192 * @return Filtered text string
5193 */
5194 public function stripSectionName( $text ) {
5195 # Strip internal link markup
5196 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5197 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5198
5199 # Strip external link markup (FIXME: Not Tolerant to blank link text
5200 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5201 # on how many empty links there are on the page - need to figure that out.
5202 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5203
5204 # Parse wikitext quotes (italics & bold)
5205 $text = $this->doQuotes( $text );
5206
5207 # Strip HTML tags
5208 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5209 return $text;
5210 }
5211
5212 /**
5213 * strip/replaceVariables/unstrip for preprocessor regression testing
5214 */
5215 function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5216 if ( !$title instanceof Title ) {
5217 $title = Title::newFromText( $title );
5218 }
5219 $this->startParse( $title, $options, $outputType, true );
5220
5221 $text = $this->replaceVariables( $text );
5222 $text = $this->mStripState->unstripBoth( $text );
5223 $text = Sanitizer::removeHTMLtags( $text );
5224 return $text;
5225 }
5226
5227 function testPst( $text, $title, $options ) {
5228 global $wgUser;
5229 if ( !$title instanceof Title ) {
5230 $title = Title::newFromText( $title );
5231 }
5232 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5233 }
5234
5235 function testPreprocess( $text, $title, $options ) {
5236 if ( !$title instanceof Title ) {
5237 $title = Title::newFromText( $title );
5238 }
5239 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5240 }
5241
5242 /**
5243 * Call a callback function on all regions of the given text that are not
5244 * inside strip markers, and replace those regions with the return value
5245 * of the callback. For example, with input:
5246 *
5247 * aaa<MARKER>bbb
5248 *
5249 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5250 * two strings will be replaced with the value returned by the callback in
5251 * each case.
5252 */
5253 function markerSkipCallback( $s, $callback ) {
5254 $i = 0;
5255 $out = '';
5256 while ( $i < strlen( $s ) ) {
5257 $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5258 if ( $markerStart === false ) {
5259 $out .= call_user_func( $callback, substr( $s, $i ) );
5260 break;
5261 } else {
5262 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5263 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5264 if ( $markerEnd === false ) {
5265 $out .= substr( $s, $markerStart );
5266 break;
5267 } else {
5268 $markerEnd += strlen( self::MARKER_SUFFIX );
5269 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5270 $i = $markerEnd;
5271 }
5272 }
5273 }
5274 return $out;
5275 }
5276
5277 /**
5278 * Save the parser state required to convert the given half-parsed text to
5279 * HTML. "Half-parsed" in this context means the output of
5280 * recursiveTagParse() or internalParse(). This output has strip markers
5281 * from replaceVariables (extensionSubstitution() etc.), and link
5282 * placeholders from replaceLinkHolders().
5283 *
5284 * Returns an array which can be serialized and stored persistently. This
5285 * array can later be loaded into another parser instance with
5286 * unserializeHalfParsedText(). The text can then be safely incorporated into
5287 * the return value of a parser hook.
5288 */
5289 function serializeHalfParsedText( $text ) {
5290 wfProfileIn( __METHOD__ );
5291 $data = array(
5292 'text' => $text,
5293 'version' => self::HALF_PARSED_VERSION,
5294 'stripState' => $this->mStripState->getSubState( $text ),
5295 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5296 );
5297 wfProfileOut( __METHOD__ );
5298 return $data;
5299 }
5300
5301 /**
5302 * Load the parser state given in the $data array, which is assumed to
5303 * have been generated by serializeHalfParsedText(). The text contents is
5304 * extracted from the array, and its markers are transformed into markers
5305 * appropriate for the current Parser instance. This transformed text is
5306 * returned, and can be safely included in the return value of a parser
5307 * hook.
5308 *
5309 * If the $data array has been stored persistently, the caller should first
5310 * check whether it is still valid, by calling isValidHalfParsedText().
5311 *
5312 * @param $data Serialized data
5313 * @return String
5314 */
5315 function unserializeHalfParsedText( $data ) {
5316 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5317 throw new MWException( __METHOD__.': invalid version' );
5318 }
5319
5320 # First, extract the strip state.
5321 $texts = array( $data['text'] );
5322 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5323
5324 # Now renumber links
5325 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5326
5327 # Should be good to go.
5328 return $texts[0];
5329 }
5330
5331 /**
5332 * Returns true if the given array, presumed to be generated by
5333 * serializeHalfParsedText(), is compatible with the current version of the
5334 * parser.
5335 *
5336 * @param $data Array.
5337 */
5338 function isValidHalfParsedText( $data ) {
5339 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5340 }
5341 }
Wiklou, le Wiki du Wiklou