3 * @defgroup Parser Parser
7 * File for Parser and related classes
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).
18 * There are five main entry points into the Parser class:
20 * produces HTML output
22 * produces altered wiki markup.
24 * removes HTML comments and expands templates
25 * cleanSig() / cleanSigInSig()
26 * Cleans a signature before saving it to preferences
28 * Return the content of a section from an article for section editing
30 * Replaces a section by number inside an article
32 * Removes <noinclude> sections, and <includeonly> tags.
37 * NOT $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
40 * $wgUseDynamicDates*, $wgInterwikiMagic*,
41 * $wgNamespacesWithSubpages, $wgAllowExternalImages*,
42 * $wgLocaltimezone, $wgAllowSpecialInclusion*,
45 * * only within ParserOptions
52 * Update this version number when the ParserOutput format
53 * changes in an incompatible way, so the parser cache
54 * can automatically discard old data.
56 const VERSION
= '1.6.4';
59 * Update this version number when the output of serialiseHalfParsedText()
60 * changes in an incompatible way
62 const HALF_PARSED_VERSION
= 2;
64 # Flags for Parser::setFunctionHook
65 # Also available as global constants from Defines.php
66 const SFH_NO_HASH
= 1;
67 const SFH_OBJECT_ARGS
= 2;
69 # Constants needed for external link processing
70 # Everything except bracket, space, or control characters
71 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
72 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
73 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
74 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
75 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
77 # State constants for the definition list colon extraction
78 const COLON_STATE_TEXT
= 0;
79 const COLON_STATE_TAG
= 1;
80 const COLON_STATE_TAGSTART
= 2;
81 const COLON_STATE_CLOSETAG
= 3;
82 const COLON_STATE_TAGSLASH
= 4;
83 const COLON_STATE_COMMENT
= 5;
84 const COLON_STATE_COMMENTDASH
= 6;
85 const COLON_STATE_COMMENTDASHDASH
= 7;
87 # Flags for preprocessToDom
88 const PTD_FOR_INCLUSION
= 1;
90 # Allowed values for $this->mOutputType
91 # Parameter to startExternalParse().
92 const OT_HTML
= 1; # like parse()
93 const OT_WIKI
= 2; # like preSaveTransform()
94 const OT_PREPROCESS
= 3; # like preprocess()
96 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
98 # Marker Suffix needs to be accessible staticly.
99 const MARKER_SUFFIX
= "-QINU\x7f";
102 var $mTagHooks = array();
103 var $mTransparentTagHooks = array();
104 var $mFunctionHooks = array();
105 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
106 var $mFunctionTagHooks = array();
107 var $mStripList = array();
108 var $mDefaultStripList = array();
109 var $mVarCache = array();
110 var $mImageParams = array();
111 var $mImageParamsMagicArray = array();
112 var $mMarkerIndex = 0;
113 var $mFirstCall = true;
115 # Initialised by initialiseVariables()
118 * @var MagicWordArray
123 * @var MagicWordArray
126 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
128 # Cleared with clearState():
133 var $mAutonumber, $mDTopen;
140 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
142 * @var LinkHolderArray
147 var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
148 var $mTplExpandCache; # empty-frame expansion cache
149 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
150 var $mExpensiveFunctionCount; # number of expensive parser function calls
151 var $mShowToc, $mForceTocPosition;
156 var $mUser; # User object; only used when doing pre-save transform
159 # These are variables reset at least once per parse regardless of $clearState
169 var $mTitle; # Title context, used for self-link rendering and similar things
170 var $mOutputType; # Output type, one of the OT_xxx constants
171 var $ot; # Shortcut alias, see setOutputType()
172 var $mRevisionObject; # The revision object of the specified revision ID
173 var $mRevisionId; # ID to display in {{REVISIONID}} tags
174 var $mRevisionTimestamp; # The timestamp of the specified revision ID
175 var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
176 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
188 public function __construct( $conf = array() ) {
189 $this->mConf
= $conf;
190 $this->mUrlProtocols
= wfUrlProtocols();
191 $this->mExtLinkBracketedRegex
= '/\[((' . wfUrlProtocols() . ')'.
192 self
::EXT_LINK_URL_CLASS
.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
193 if ( isset( $conf['preprocessorClass'] ) ) {
194 $this->mPreprocessorClass
= $conf['preprocessorClass'];
195 } elseif ( defined( 'MW_COMPILED' ) ) {
196 # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
197 $this->mPreprocessorClass
= 'Preprocessor_Hash';
198 } elseif ( extension_loaded( 'domxml' ) ) {
199 # PECL extension that conflicts with the core DOM extension (bug 13770)
200 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
201 $this->mPreprocessorClass
= 'Preprocessor_Hash';
202 } elseif ( extension_loaded( 'dom' ) ) {
203 $this->mPreprocessorClass
= 'Preprocessor_DOM';
205 $this->mPreprocessorClass
= 'Preprocessor_Hash';
207 wfDebug( __CLASS__
. ": using preprocessor: {$this->mPreprocessorClass}\n" );
211 * Reduce memory usage to reduce the impact of circular references
213 function __destruct() {
214 if ( isset( $this->mLinkHolders
) ) {
215 unset( $this->mLinkHolders
);
217 foreach ( $this as $name => $value ) {
218 unset( $this->$name );
223 * Do various kinds of initialisation on the first call of the parser
225 function firstCallInit() {
226 if ( !$this->mFirstCall
) {
229 $this->mFirstCall
= false;
231 wfProfileIn( __METHOD__
);
233 CoreParserFunctions
::register( $this );
234 CoreTagHooks
::register( $this );
235 $this->initialiseVariables();
237 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
238 wfProfileOut( __METHOD__
);
246 function clearState() {
247 wfProfileIn( __METHOD__
);
248 if ( $this->mFirstCall
) {
249 $this->firstCallInit();
251 $this->mOutput
= new ParserOutput
;
252 $this->mOptions
->registerWatcher( array( $this->mOutput
, 'recordOption' ) );
253 $this->mAutonumber
= 0;
254 $this->mLastSection
= '';
255 $this->mDTopen
= false;
256 $this->mIncludeCount
= array();
257 $this->mArgStack
= false;
258 $this->mInPre
= false;
259 $this->mLinkHolders
= new LinkHolderArray( $this );
261 $this->mRevisionObject
= $this->mRevisionTimestamp
=
262 $this->mRevisionId
= $this->mRevisionUser
= null;
263 $this->mVarCache
= array();
267 * Prefix for temporary replacement strings for the multipass parser.
268 * \x07 should never appear in input as it's disallowed in XML.
269 * Using it at the front also gives us a little extra robustness
270 * since it shouldn't match when butted up against identifier-like
273 * Must not consist of all title characters, or else it will change
274 * the behaviour of <nowiki> in a link.
276 $this->mUniqPrefix
= "\x7fUNIQ" . self
::getRandomString();
277 $this->mStripState
= new StripState( $this->mUniqPrefix
);
280 # Clear these on every parse, bug 4549
281 $this->mTplExpandCache
= $this->mTplRedirCache
= $this->mTplDomCache
= array();
283 $this->mShowToc
= true;
284 $this->mForceTocPosition
= false;
285 $this->mIncludeSizes
= array(
289 $this->mPPNodeCount
= 0;
290 $this->mDefaultSort
= false;
291 $this->mHeadings
= array();
292 $this->mDoubleUnderscores
= array();
293 $this->mExpensiveFunctionCount
= 0;
296 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
297 $this->mPreprocessor
= null;
300 wfRunHooks( 'ParserClearState', array( &$this ) );
301 wfProfileOut( __METHOD__
);
305 * Convert wikitext to HTML
306 * Do not call this function recursively.
308 * @param $text String: text we want to parse
309 * @param $title Title object
310 * @param $options ParserOptions
311 * @param $linestart boolean
312 * @param $clearState boolean
313 * @param $revid Int: number to pass in {{REVISIONID}}
314 * @return ParserOutput a ParserOutput
316 public function parse( $text, Title
$title, ParserOptions
$options, $linestart = true, $clearState = true, $revid = null ) {
318 * First pass--just handle <nowiki> sections, pass the rest off
319 * to internalParse() which does all the real work.
322 global $wgUseTidy, $wgAlwaysUseTidy, $wgDisableLangConversion, $wgDisableTitleConversion;
323 $fname = __METHOD__
.'-' . wfGetCaller();
324 wfProfileIn( __METHOD__
);
325 wfProfileIn( $fname );
327 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
329 # Remove the strip marker tag prefix from the input, if present.
331 $text = str_replace( $this->mUniqPrefix
, '', $text );
334 $oldRevisionId = $this->mRevisionId
;
335 $oldRevisionObject = $this->mRevisionObject
;
336 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
337 $oldRevisionUser = $this->mRevisionUser
;
338 if ( $revid !== null ) {
339 $this->mRevisionId
= $revid;
340 $this->mRevisionObject
= null;
341 $this->mRevisionTimestamp
= null;
342 $this->mRevisionUser
= null;
345 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
347 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
348 $text = $this->internalParse( $text );
350 $text = $this->mStripState
->unstripGeneral( $text );
352 # Clean up special characters, only run once, next-to-last before doBlockLevels
354 # french spaces, last one Guillemet-left
355 # only if there is something before the space
356 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1 ',
357 # french spaces, Guillemet-right
358 '/(\\302\\253) /' => '\\1 ',
359 '/ (!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
361 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
363 $text = $this->doBlockLevels( $text, $linestart );
365 $this->replaceLinkHolders( $text );
368 * The input doesn't get language converted if
370 * b) Content isn't converted
371 * c) It's a conversion table
372 * d) it is an interface message (which is in the user language)
374 if ( !( $wgDisableLangConversion
375 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
376 ||
$this->mTitle
->isConversionTable() ) )
378 # Run convert unconditionally in 1.18-compatible mode
379 global $wgBug34832TransitionalRollback;
380 if ( $wgBug34832TransitionalRollback ||
!$this->mOptions
->getInterfaceMessage() ) {
381 # The position of the convert() call should not be changed. it
382 # assumes that the links are all replaced and the only thing left
383 # is the <nowiki> mark.
384 $text = $this->getConverterLanguage()->convert( $text );
389 * A converted title will be provided in the output object if title and
390 * content conversion are enabled, the article text does not contain
391 * a conversion-suppressing double-underscore tag, and no
392 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
393 * automatic link conversion.
395 if ( !( $wgDisableLangConversion
396 ||
$wgDisableTitleConversion
397 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
398 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
399 ||
$this->mOutput
->getDisplayTitle() !== false ) )
401 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
402 if ( $convruletitle ) {
403 $this->mOutput
->setTitleText( $convruletitle );
405 $titleText = $this->getConverterLanguage()->convertTitle( $title );
406 $this->mOutput
->setTitleText( $titleText );
410 $text = $this->mStripState
->unstripNoWiki( $text );
412 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
414 $text = $this->replaceTransparentTags( $text );
415 $text = $this->mStripState
->unstripGeneral( $text );
417 $text = Sanitizer
::normalizeCharReferences( $text );
419 if ( ( $wgUseTidy && $this->mOptions
->getTidy() ) ||
$wgAlwaysUseTidy ) {
420 $text = MWTidy
::tidy( $text );
422 # attempt to sanitize at least some nesting problems
423 # (bug #2702 and quite a few others)
425 # ''Something [http://www.cool.com cool''] -->
426 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
427 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
428 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
429 # fix up an anchor inside another anchor, only
430 # at least for a single single nested link (bug 3695)
431 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
432 '\\1\\2</a>\\3</a>\\1\\4</a>',
433 # fix div inside inline elements- doBlockLevels won't wrap a line which
434 # contains a div, so fix it up here; replace
435 # div with escaped text
436 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
437 '\\1\\3<div\\5>\\6</div>\\8\\9',
438 # remove empty italic or bold tag pairs, some
439 # introduced by rules above
440 '/<([bi])><\/\\1>/' => '',
443 $text = preg_replace(
444 array_keys( $tidyregs ),
445 array_values( $tidyregs ),
448 global $wgExpensiveParserFunctionLimit;
449 if ( $this->mExpensiveFunctionCount
> $wgExpensiveParserFunctionLimit ) {
450 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount
, $wgExpensiveParserFunctionLimit );
453 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
455 # Information on include size limits, for the benefit of users who try to skirt them
456 if ( $this->mOptions
->getEnableLimitReport() ) {
457 $max = $this->mOptions
->getMaxIncludeSize();
458 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
460 "NewPP limit report\n" .
461 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
462 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
463 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
465 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
466 $text .= "\n<!-- \n$limitReport-->\n";
468 $this->mOutput
->setText( $text );
470 $this->mRevisionId
= $oldRevisionId;
471 $this->mRevisionObject
= $oldRevisionObject;
472 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
473 $this->mRevisionUser
= $oldRevisionUser;
474 wfProfileOut( $fname );
475 wfProfileOut( __METHOD__
);
477 return $this->mOutput
;
481 * Recursive parser entry point that can be called from an extension tag
484 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
486 * @param $text String: text extension wants to have parsed
487 * @param $frame PPFrame: The frame to use for expanding any template variables
491 function recursiveTagParse( $text, $frame=false ) {
492 wfProfileIn( __METHOD__
);
493 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
494 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
495 $text = $this->internalParse( $text, false, $frame );
496 wfProfileOut( __METHOD__
);
501 * Expand templates and variables in the text, producing valid, static wikitext.
502 * Also removes comments.
503 * @return mixed|string
505 function preprocess( $text, Title
$title, ParserOptions
$options, $revid = null ) {
506 wfProfileIn( __METHOD__
);
507 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
508 if ( $revid !== null ) {
509 $this->mRevisionId
= $revid;
511 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
512 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
513 $text = $this->replaceVariables( $text );
514 $text = $this->mStripState
->unstripBoth( $text );
515 wfProfileOut( __METHOD__
);
520 * Recursive parser entry point that can be called from an extension tag
523 * @param $text String: text to be expanded
524 * @param $frame PPFrame: The frame to use for expanding any template variables
528 public function recursivePreprocess( $text, $frame = false ) {
529 wfProfileIn( __METHOD__
);
530 $text = $this->replaceVariables( $text, $frame );
531 $text = $this->mStripState
->unstripBoth( $text );
532 wfProfileOut( __METHOD__
);
537 * Process the wikitext for the ?preload= feature. (bug 5210)
539 * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
540 * comments, templates, arguments, tags hooks and parser functions are untouched.
542 * @param $text String
543 * @param $title Title
544 * @param $options ParserOptions
547 public function getPreloadText( $text, Title
$title, ParserOptions
$options ) {
548 # Parser (re)initialisation
549 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
551 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
552 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
553 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
554 $text = $this->mStripState
->unstripBoth( $text );
559 * Get a random string
563 static public function getRandomString() {
564 return wfRandomString( 16 );
568 * Set the current user.
569 * Should only be used when doing pre-save transform.
571 * @param $user Mixed: User object or null (to reset)
573 function setUser( $user ) {
574 $this->mUser
= $user;
578 * Accessor for mUniqPrefix.
582 public function uniqPrefix() {
583 if ( !isset( $this->mUniqPrefix
) ) {
584 # @todo FIXME: This is probably *horribly wrong*
585 # LanguageConverter seems to want $wgParser's uniqPrefix, however
586 # if this is called for a parser cache hit, the parser may not
587 # have ever been initialized in the first place.
588 # Not really sure what the heck is supposed to be going on here.
590 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
592 return $this->mUniqPrefix
;
596 * Set the context title
600 function setTitle( $t ) {
601 if ( !$t ||
$t instanceof FakeTitle
) {
602 $t = Title
::newFromText( 'NO TITLE' );
605 if ( strval( $t->getFragment() ) !== '' ) {
606 # Strip the fragment to avoid various odd effects
607 $this->mTitle
= clone $t;
608 $this->mTitle
->setFragment( '' );
615 * Accessor for the Title object
617 * @return Title object
619 function getTitle() {
620 return $this->mTitle
;
624 * Accessor/mutator for the Title object
626 * @param $x Title object or null to just get the current one
627 * @return Title object
629 function Title( $x = null ) {
630 return wfSetVar( $this->mTitle
, $x );
634 * Set the output type
636 * @param $ot Integer: new value
638 function setOutputType( $ot ) {
639 $this->mOutputType
= $ot;
642 'html' => $ot == self
::OT_HTML
,
643 'wiki' => $ot == self
::OT_WIKI
,
644 'pre' => $ot == self
::OT_PREPROCESS
,
645 'plain' => $ot == self
::OT_PLAIN
,
650 * Accessor/mutator for the output type
652 * @param $x int|null New value or null to just get the current one
655 function OutputType( $x = null ) {
656 return wfSetVar( $this->mOutputType
, $x );
660 * Get the ParserOutput object
662 * @return ParserOutput object
664 function getOutput() {
665 return $this->mOutput
;
669 * Get the ParserOptions object
671 * @return ParserOptions object
673 function getOptions() {
674 return $this->mOptions
;
678 * Accessor/mutator for the ParserOptions object
680 * @param $x ParserOptions New value or null to just get the current one
681 * @return ParserOptions Current ParserOptions object
683 function Options( $x = null ) {
684 return wfSetVar( $this->mOptions
, $x );
690 function nextLinkID() {
691 return $this->mLinkID++
;
697 function setLinkID( $id ) {
698 $this->mLinkID
= $id;
702 * Get a language object for use in parser functions such as {{FORMATNUM:}}
705 function getFunctionLang() {
706 return $this->getTargetLanguage();
710 * Get the target language for the content being parsed. This is usually the
711 * language that the content is in.
713 function getTargetLanguage() {
714 $target = $this->mOptions
->getTargetLanguage();
715 if ( $target !== null ) {
717 } elseif( $this->mOptions
->getInterfaceMessage() ) {
718 return $this->mOptions
->getUserLangObj();
719 } elseif( is_null( $this->mTitle
) ) {
720 throw new MWException( __METHOD__
.': $this->mTitle is null' );
722 return $this->mTitle
->getPageLanguage();
726 * Get the language object for language conversion
728 function getConverterLanguage() {
729 global $wgBug34832TransitionalRollback, $wgContLang;
730 if ( $wgBug34832TransitionalRollback ) {
733 return $this->getTargetLanguage();
738 * Get a User object either from $this->mUser, if set, or from the
739 * ParserOptions object otherwise
741 * @return User object
744 if ( !is_null( $this->mUser
) ) {
747 return $this->mOptions
->getUser();
751 * Get a preprocessor object
753 * @return Preprocessor instance
755 function getPreprocessor() {
756 if ( !isset( $this->mPreprocessor
) ) {
757 $class = $this->mPreprocessorClass
;
758 $this->mPreprocessor
= new $class( $this );
760 return $this->mPreprocessor
;
764 * Replaces all occurrences of HTML-style comments and the given tags
765 * in the text with a random marker and returns the next text. The output
766 * parameter $matches will be an associative array filled with data in
768 * 'UNIQ-xxxxx' => array(
771 * array( 'param' => 'x' ),
772 * '<element param="x">tag content</element>' ) )
774 * @param $elements array list of element names. Comments are always extracted.
775 * @param $text string Source text string.
776 * @param $matches array Out parameter, Array: extracted tags
777 * @param $uniq_prefix string
778 * @return String: stripped text
780 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
785 $taglist = implode( '|', $elements );
786 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
788 while ( $text != '' ) {
789 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
791 if ( count( $p ) < 5 ) {
794 if ( count( $p ) > 5 ) {
808 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
809 $stripped .= $marker;
811 if ( $close === '/>' ) {
812 # Empty element tag, <tag />
817 if ( $element === '!--' ) {
820 $end = "/(<\\/$element\\s*>)/i";
822 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
824 if ( count( $q ) < 3 ) {
825 # No end tag -- let it run out to the end of the text.
834 $matches[$marker] = array( $element,
836 Sanitizer
::decodeTagAttributes( $attributes ),
837 "<$element$attributes$close$content$tail" );
843 * Get a list of strippable XML-like elements
847 function getStripList() {
848 return $this->mStripList
;
852 * Add an item to the strip state
853 * Returns the unique tag which must be inserted into the stripped text
854 * The tag will be replaced with the original text in unstrip()
856 * @param $text string
860 function insertStripItem( $text ) {
861 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
862 $this->mMarkerIndex++
;
863 $this->mStripState
->addGeneral( $rnd, $text );
868 * parse the wiki syntax used to render tables
873 function doTableStuff( $text ) {
874 wfProfileIn( __METHOD__
);
876 $lines = StringUtils
::explode( "\n", $text );
878 $td_history = array(); # Is currently a td tag open?
879 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
880 $tr_history = array(); # Is currently a tr tag open?
881 $tr_attributes = array(); # history of tr attributes
882 $has_opened_tr = array(); # Did this table open a <tr> element?
883 $indent_level = 0; # indent level of the table
885 foreach ( $lines as $outLine ) {
886 $line = trim( $outLine );
888 if ( $line === '' ) { # empty line, go to next line
889 $out .= $outLine."\n";
893 $first_character = $line[0];
896 if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
897 # First check if we are starting a new table
898 $indent_level = strlen( $matches[1] );
900 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
901 $attributes = Sanitizer
::fixTagAttributes( $attributes , 'table' );
903 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
904 array_push( $td_history , false );
905 array_push( $last_tag_history , '' );
906 array_push( $tr_history , false );
907 array_push( $tr_attributes , '' );
908 array_push( $has_opened_tr , false );
909 } elseif ( count( $td_history ) == 0 ) {
910 # Don't do any of the following
911 $out .= $outLine."\n";
913 } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
914 # We are ending a table
915 $line = '</table>' . substr( $line , 2 );
916 $last_tag = array_pop( $last_tag_history );
918 if ( !array_pop( $has_opened_tr ) ) {
919 $line = "<tr><td></td></tr>{$line}";
922 if ( array_pop( $tr_history ) ) {
923 $line = "</tr>{$line}";
926 if ( array_pop( $td_history ) ) {
927 $line = "</{$last_tag}>{$line}";
929 array_pop( $tr_attributes );
930 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
931 } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
932 # Now we have a table row
933 $line = preg_replace( '#^\|-+#', '', $line );
935 # Whats after the tag is now only attributes
936 $attributes = $this->mStripState
->unstripBoth( $line );
937 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
938 array_pop( $tr_attributes );
939 array_push( $tr_attributes, $attributes );
942 $last_tag = array_pop( $last_tag_history );
943 array_pop( $has_opened_tr );
944 array_push( $has_opened_tr , true );
946 if ( array_pop( $tr_history ) ) {
950 if ( array_pop( $td_history ) ) {
951 $line = "</{$last_tag}>{$line}";
955 array_push( $tr_history , false );
956 array_push( $td_history , false );
957 array_push( $last_tag_history , '' );
958 } elseif ( $first_character === '|' ||
$first_character === '!' ||
substr( $line , 0 , 2 ) === '|+' ) {
959 # This might be cell elements, td, th or captions
960 if ( substr( $line , 0 , 2 ) === '|+' ) {
961 $first_character = '+';
962 $line = substr( $line , 1 );
965 $line = substr( $line , 1 );
967 if ( $first_character === '!' ) {
968 $line = str_replace( '!!' , '||' , $line );
971 # Split up multiple cells on the same line.
972 # FIXME : This can result in improper nesting of tags processed
973 # by earlier parser steps, but should avoid splitting up eg
974 # attribute values containing literal "||".
975 $cells = StringUtils
::explodeMarkup( '||' , $line );
979 # Loop through each table cell
980 foreach ( $cells as $cell ) {
982 if ( $first_character !== '+' ) {
983 $tr_after = array_pop( $tr_attributes );
984 if ( !array_pop( $tr_history ) ) {
985 $previous = "<tr{$tr_after}>\n";
987 array_push( $tr_history , true );
988 array_push( $tr_attributes , '' );
989 array_pop( $has_opened_tr );
990 array_push( $has_opened_tr , true );
993 $last_tag = array_pop( $last_tag_history );
995 if ( array_pop( $td_history ) ) {
996 $previous = "</{$last_tag}>\n{$previous}";
999 if ( $first_character === '|' ) {
1001 } elseif ( $first_character === '!' ) {
1003 } elseif ( $first_character === '+' ) {
1004 $last_tag = 'caption';
1009 array_push( $last_tag_history , $last_tag );
1011 # A cell could contain both parameters and data
1012 $cell_data = explode( '|' , $cell , 2 );
1014 # Bug 553: Note that a '|' inside an invalid link should not
1015 # be mistaken as delimiting cell parameters
1016 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1017 $cell = "{$previous}<{$last_tag}>{$cell}";
1018 } elseif ( count( $cell_data ) == 1 ) {
1019 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1021 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1022 $attributes = Sanitizer
::fixTagAttributes( $attributes , $last_tag );
1023 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1027 array_push( $td_history , true );
1030 $out .= $outLine . "\n";
1033 # Closing open td, tr && table
1034 while ( count( $td_history ) > 0 ) {
1035 if ( array_pop( $td_history ) ) {
1038 if ( array_pop( $tr_history ) ) {
1041 if ( !array_pop( $has_opened_tr ) ) {
1042 $out .= "<tr><td></td></tr>\n" ;
1045 $out .= "</table>\n";
1048 # Remove trailing line-ending (b/c)
1049 if ( substr( $out, -1 ) === "\n" ) {
1050 $out = substr( $out, 0, -1 );
1053 # special case: don't return empty table
1054 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1058 wfProfileOut( __METHOD__
);
1064 * Helper function for parse() that transforms wiki markup into
1065 * HTML. Only called for $mOutputType == self::OT_HTML.
1069 * @param $text string
1070 * @param $isMain bool
1071 * @param $frame bool
1075 function internalParse( $text, $isMain = true, $frame = false ) {
1076 wfProfileIn( __METHOD__
);
1080 # Hook to suspend the parser in this state
1081 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState
) ) ) {
1082 wfProfileOut( __METHOD__
);
1086 # if $frame is provided, then use $frame for replacing any variables
1088 # use frame depth to infer how include/noinclude tags should be handled
1089 # depth=0 means this is the top-level document; otherwise it's an included document
1090 if ( !$frame->depth
) {
1093 $flag = Parser
::PTD_FOR_INCLUSION
;
1095 $dom = $this->preprocessToDom( $text, $flag );
1096 $text = $frame->expand( $dom );
1098 # if $frame is not provided, then use old-style replaceVariables
1099 $text = $this->replaceVariables( $text );
1102 $text = Sanitizer
::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks
) );
1103 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState
) );
1105 # Tables need to come after variable replacement for things to work
1106 # properly; putting them before other transformations should keep
1107 # exciting things like link expansions from showing up in surprising
1109 $text = $this->doTableStuff( $text );
1111 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1113 $text = $this->doDoubleUnderscore( $text );
1115 $text = $this->doHeadings( $text );
1116 if ( $this->mOptions
->getUseDynamicDates() ) {
1117 $df = DateFormatter
::getInstance();
1118 $text = $df->reformat( $this->mOptions
->getDateFormat(), $text );
1120 $text = $this->replaceInternalLinks( $text );
1121 $text = $this->doAllQuotes( $text );
1122 $text = $this->replaceExternalLinks( $text );
1124 # replaceInternalLinks may sometimes leave behind
1125 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1126 $text = str_replace( $this->mUniqPrefix
.'NOPARSE', '', $text );
1128 $text = $this->doMagicLinks( $text );
1129 $text = $this->formatHeadings( $text, $origText, $isMain );
1131 wfProfileOut( __METHOD__
);
1136 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1137 * magic external links.
1142 * @param $text string
1146 function doMagicLinks( $text ) {
1147 wfProfileIn( __METHOD__
);
1148 $prots = wfUrlProtocolsWithoutProtRel();
1149 $urlChar = self
::EXT_LINK_URL_CLASS
;
1150 $text = preg_replace_callback(
1152 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1153 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1154 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1155 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1156 ISBN\s+(\b # m[5]: ISBN, capture number
1157 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1158 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1159 [0-9Xx] # check digit
1161 )!xu', array( &$this, 'magicLinkCallback' ), $text );
1162 wfProfileOut( __METHOD__
);
1167 * @throws MWException
1169 * @return HTML|string
1171 function magicLinkCallback( $m ) {
1172 if ( isset( $m[1] ) && $m[1] !== '' ) {
1175 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1178 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1179 # Free external link
1180 return $this->makeFreeExternalLink( $m[0] );
1181 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1183 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1186 $CssClass = 'mw-magiclink-rfc';
1188 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1190 $urlmsg = 'pubmedurl';
1191 $CssClass = 'mw-magiclink-pmid';
1194 throw new MWException( __METHOD__
.': unrecognised match type "' .
1195 substr( $m[0], 0, 20 ) . '"' );
1197 $url = wfMsgForContent( $urlmsg, $id );
1198 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1199 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1202 $num = strtr( $isbn, array(
1207 $titleObj = SpecialPage
::getTitleFor( 'Booksources', $num );
1209 htmlspecialchars( $titleObj->getLocalUrl() ) .
1210 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1217 * Make a free external link, given a user-supplied URL
1219 * @param $url string
1221 * @return string HTML
1224 function makeFreeExternalLink( $url ) {
1225 wfProfileIn( __METHOD__
);
1229 # The characters '<' and '>' (which were escaped by
1230 # removeHTMLtags()) should not be included in
1231 # URLs, per RFC 2396.
1233 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1234 $trail = substr( $url, $m2[0][1] ) . $trail;
1235 $url = substr( $url, 0, $m2[0][1] );
1238 # Move trailing punctuation to $trail
1240 # If there is no left bracket, then consider right brackets fair game too
1241 if ( strpos( $url, '(' ) === false ) {
1245 $numSepChars = strspn( strrev( $url ), $sep );
1246 if ( $numSepChars ) {
1247 $trail = substr( $url, -$numSepChars ) . $trail;
1248 $url = substr( $url, 0, -$numSepChars );
1251 $url = Sanitizer
::cleanUrl( $url );
1253 # Is this an external image?
1254 $text = $this->maybeMakeExternalImage( $url );
1255 if ( $text === false ) {
1256 # Not an image, make a link
1257 $text = Linker
::makeExternalLink( $url,
1258 $this->getConverterLanguage()->markNoConversion($url), true, 'free',
1259 $this->getExternalLinkAttribs( $url ) );
1260 # Register it in the output object...
1261 # Replace unnecessary URL escape codes with their equivalent characters
1262 $pasteurized = self
::replaceUnusualEscapes( $url );
1263 $this->mOutput
->addExternalLink( $pasteurized );
1265 wfProfileOut( __METHOD__
);
1266 return $text . $trail;
1271 * Parse headers and return html
1275 * @param $text string
1279 function doHeadings( $text ) {
1280 wfProfileIn( __METHOD__
);
1281 for ( $i = 6; $i >= 1; --$i ) {
1282 $h = str_repeat( '=', $i );
1283 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1284 "<h$i>\\1</h$i>", $text );
1286 wfProfileOut( __METHOD__
);
1291 * Replace single quotes with HTML markup
1294 * @param $text string
1296 * @return string the altered text
1298 function doAllQuotes( $text ) {
1299 wfProfileIn( __METHOD__
);
1301 $lines = StringUtils
::explode( "\n", $text );
1302 foreach ( $lines as $line ) {
1303 $outtext .= $this->doQuotes( $line ) . "\n";
1305 $outtext = substr( $outtext, 0,-1 );
1306 wfProfileOut( __METHOD__
);
1311 * Helper function for doAllQuotes()
1313 * @param $text string
1317 public function doQuotes( $text ) {
1318 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1319 if ( count( $arr ) == 1 ) {
1322 # First, do some preliminary work. This may shift some apostrophes from
1323 # being mark-up to being text. It also counts the number of occurrences
1324 # of bold and italics mark-ups.
1327 for ( $i = 0; $i < count( $arr ); $i++
) {
1328 if ( ( $i %
2 ) == 1 ) {
1329 # If there are ever four apostrophes, assume the first is supposed to
1330 # be text, and the remaining three constitute mark-up for bold text.
1331 if ( strlen( $arr[$i] ) == 4 ) {
1334 } elseif ( strlen( $arr[$i] ) > 5 ) {
1335 # If there are more than 5 apostrophes in a row, assume they're all
1336 # text except for the last 5.
1337 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1340 # Count the number of occurrences of bold and italics mark-ups.
1341 # We are not counting sequences of five apostrophes.
1342 if ( strlen( $arr[$i] ) == 2 ) {
1344 } elseif ( strlen( $arr[$i] ) == 3 ) {
1346 } elseif ( strlen( $arr[$i] ) == 5 ) {
1353 # If there is an odd number of both bold and italics, it is likely
1354 # that one of the bold ones was meant to be an apostrophe followed
1355 # by italics. Which one we cannot know for certain, but it is more
1356 # likely to be one that has a single-letter word before it.
1357 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1359 $firstsingleletterword = -1;
1360 $firstmultiletterword = -1;
1362 foreach ( $arr as $r ) {
1363 if ( ( $i %
2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1364 $x1 = substr( $arr[$i-1], -1 );
1365 $x2 = substr( $arr[$i-1], -2, 1 );
1366 if ( $x1 === ' ' ) {
1367 if ( $firstspace == -1 ) {
1370 } elseif ( $x2 === ' ') {
1371 if ( $firstsingleletterword == -1 ) {
1372 $firstsingleletterword = $i;
1375 if ( $firstmultiletterword == -1 ) {
1376 $firstmultiletterword = $i;
1383 # If there is a single-letter word, use it!
1384 if ( $firstsingleletterword > -1 ) {
1385 $arr[$firstsingleletterword] = "''";
1386 $arr[$firstsingleletterword-1] .= "'";
1387 } elseif ( $firstmultiletterword > -1 ) {
1388 # If not, but there's a multi-letter word, use that one.
1389 $arr[$firstmultiletterword] = "''";
1390 $arr[$firstmultiletterword-1] .= "'";
1391 } elseif ( $firstspace > -1 ) {
1392 # ... otherwise use the first one that has neither.
1393 # (notice that it is possible for all three to be -1 if, for example,
1394 # there is only one pentuple-apostrophe in the line)
1395 $arr[$firstspace] = "''";
1396 $arr[$firstspace-1] .= "'";
1400 # Now let's actually convert our apostrophic mush to HTML!
1405 foreach ( $arr as $r ) {
1406 if ( ( $i %
2 ) == 0 ) {
1407 if ( $state === 'both' ) {
1413 if ( strlen( $r ) == 2 ) {
1414 if ( $state === 'i' ) {
1415 $output .= '</i>'; $state = '';
1416 } elseif ( $state === 'bi' ) {
1417 $output .= '</i>'; $state = 'b';
1418 } elseif ( $state === 'ib' ) {
1419 $output .= '</b></i><b>'; $state = 'b';
1420 } elseif ( $state === 'both' ) {
1421 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1422 } else { # $state can be 'b' or ''
1423 $output .= '<i>'; $state .= 'i';
1425 } elseif ( strlen( $r ) == 3 ) {
1426 if ( $state === 'b' ) {
1427 $output .= '</b>'; $state = '';
1428 } elseif ( $state === 'bi' ) {
1429 $output .= '</i></b><i>'; $state = 'i';
1430 } elseif ( $state === 'ib' ) {
1431 $output .= '</b>'; $state = 'i';
1432 } elseif ( $state === 'both' ) {
1433 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1434 } else { # $state can be 'i' or ''
1435 $output .= '<b>'; $state .= 'b';
1437 } elseif ( strlen( $r ) == 5 ) {
1438 if ( $state === 'b' ) {
1439 $output .= '</b><i>'; $state = 'i';
1440 } elseif ( $state === 'i' ) {
1441 $output .= '</i><b>'; $state = 'b';
1442 } elseif ( $state === 'bi' ) {
1443 $output .= '</i></b>'; $state = '';
1444 } elseif ( $state === 'ib' ) {
1445 $output .= '</b></i>'; $state = '';
1446 } elseif ( $state === 'both' ) {
1447 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1448 } else { # ($state == '')
1449 $buffer = ''; $state = 'both';
1455 # Now close all remaining tags. Notice that the order is important.
1456 if ( $state === 'b' ||
$state === 'ib' ) {
1459 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1462 if ( $state === 'bi' ) {
1465 # There might be lonely ''''', so make sure we have a buffer
1466 if ( $state === 'both' && $buffer ) {
1467 $output .= '<b><i>'.$buffer.'</i></b>';
1474 * Replace external links (REL)
1476 * Note: this is all very hackish and the order of execution matters a lot.
1477 * Make sure to run maintenance/parserTests.php if you change this code.
1481 * @param $text string
1485 function replaceExternalLinks( $text ) {
1486 wfProfileIn( __METHOD__
);
1488 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1489 $s = array_shift( $bits );
1492 while ( $i<count( $bits ) ) {
1494 $protocol = $bits[$i++
];
1495 $text = $bits[$i++
];
1496 $trail = $bits[$i++
];
1498 # The characters '<' and '>' (which were escaped by
1499 # removeHTMLtags()) should not be included in
1500 # URLs, per RFC 2396.
1502 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1503 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1504 $url = substr( $url, 0, $m2[0][1] );
1507 # If the link text is an image URL, replace it with an <img> tag
1508 # This happened by accident in the original parser, but some people used it extensively
1509 $img = $this->maybeMakeExternalImage( $text );
1510 if ( $img !== false ) {
1516 # Set linktype for CSS - if URL==text, link is essentially free
1517 $linktype = ( $text === $url ) ?
'free' : 'text';
1519 # No link text, e.g. [http://domain.tld/some.link]
1520 if ( $text == '' ) {
1522 $langObj = $this->getTargetLanguage();
1523 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1524 $linktype = 'autonumber';
1526 # Have link text, e.g. [http://domain.tld/some.link text]s
1528 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1531 $text = $this->getConverterLanguage()->markNoConversion( $text );
1533 $url = Sanitizer
::cleanUrl( $url );
1535 # Use the encoded URL
1536 # This means that users can paste URLs directly into the text
1537 # Funny characters like ö aren't valid in URLs anyway
1538 # This was changed in August 2004
1539 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
1540 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1542 # Register link in the output object.
1543 # Replace unnecessary URL escape codes with the referenced character
1544 # This prevents spammers from hiding links from the filters
1545 $pasteurized = self
::replaceUnusualEscapes( $url );
1546 $this->mOutput
->addExternalLink( $pasteurized );
1549 wfProfileOut( __METHOD__
);
1554 * Get an associative array of additional HTML attributes appropriate for a
1555 * particular external link. This currently may include rel => nofollow
1556 * (depending on configuration, namespace, and the URL's domain) and/or a
1557 * target attribute (depending on configuration).
1559 * @param $url String|bool optional URL, to extract the domain from for rel =>
1560 * nofollow if appropriate
1561 * @return Array associative array of HTML attributes
1563 function getExternalLinkAttribs( $url = false ) {
1565 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1566 $ns = $this->mTitle
->getNamespace();
1567 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1568 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1570 $attribs['rel'] = 'nofollow';
1572 if ( $this->mOptions
->getExternalLinkTarget() ) {
1573 $attribs['target'] = $this->mOptions
->getExternalLinkTarget();
1579 * Replace unusual URL escape codes with their equivalent characters
1581 * @param $url String
1584 * @todo This can merge genuinely required bits in the path or query string,
1585 * breaking legit URLs. A proper fix would treat the various parts of
1586 * the URL differently; as a workaround, just use the output for
1587 * statistical records, not for actual linking/output.
1589 static function replaceUnusualEscapes( $url ) {
1590 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1591 array( __CLASS__
, 'replaceUnusualEscapesCallback' ), $url );
1595 * Callback function used in replaceUnusualEscapes().
1596 * Replaces unusual URL escape codes with their equivalent character
1598 * @param $matches array
1602 private static function replaceUnusualEscapesCallback( $matches ) {
1603 $char = urldecode( $matches[0] );
1604 $ord = ord( $char );
1605 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1606 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1607 # No, shouldn't be escaped
1610 # Yes, leave it escaped
1616 * make an image if it's allowed, either through the global
1617 * option, through the exception, or through the on-wiki whitelist
1620 * $param $url string
1624 function maybeMakeExternalImage( $url ) {
1625 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
1626 $imagesexception = !empty( $imagesfrom );
1628 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1629 if ( $imagesexception && is_array( $imagesfrom ) ) {
1630 $imagematch = false;
1631 foreach ( $imagesfrom as $match ) {
1632 if ( strpos( $url, $match ) === 0 ) {
1637 } elseif ( $imagesexception ) {
1638 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1640 $imagematch = false;
1642 if ( $this->mOptions
->getAllowExternalImages()
1643 ||
( $imagesexception && $imagematch ) ) {
1644 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1646 $text = Linker
::makeExternalImage( $url );
1649 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
1650 && preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1651 $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1652 foreach ( $whitelist as $entry ) {
1653 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1654 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
1657 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1658 # Image matches a whitelist entry
1659 $text = Linker
::makeExternalImage( $url );
1668 * Process [[ ]] wikilinks
1672 * @return String: processed text
1676 function replaceInternalLinks( $s ) {
1677 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
1682 * Process [[ ]] wikilinks (RIL)
1683 * @return LinkHolderArray
1687 function replaceInternalLinks2( &$s ) {
1688 wfProfileIn( __METHOD__
);
1690 wfProfileIn( __METHOD__
.'-setup' );
1691 static $tc = FALSE, $e1, $e1_img;
1692 # the % is needed to support urlencoded titles as well
1694 $tc = Title
::legalChars() . '#%';
1695 # Match a link having the form [[namespace:link|alternate]]trail
1696 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1697 # Match cases where there is no "]]", which might still be images
1698 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1701 $holders = new LinkHolderArray( $this );
1703 # split the entire text string on occurrences of [[
1704 $a = StringUtils
::explode( '[[', ' ' . $s );
1705 # get the first element (all text up to first [[), and remove the space we added
1708 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1709 $s = substr( $s, 1 );
1711 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1713 if ( $useLinkPrefixExtension ) {
1714 # Match the end of a line for a word that's not followed by whitespace,
1715 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1716 $e2 = wfMsgForContent( 'linkprefix' );
1719 if ( is_null( $this->mTitle
) ) {
1720 wfProfileOut( __METHOD__
.'-setup' );
1721 wfProfileOut( __METHOD__
);
1722 throw new MWException( __METHOD__
.": \$this->mTitle is null\n" );
1724 $nottalk = !$this->mTitle
->isTalkPage();
1726 if ( $useLinkPrefixExtension ) {
1728 if ( preg_match( $e2, $s, $m ) ) {
1729 $first_prefix = $m[2];
1731 $first_prefix = false;
1737 if ( $this->getConverterLanguage()->hasVariants() ) {
1738 $selflink = $this->getConverterLanguage()->autoConvertToAllVariants(
1739 $this->mTitle
->getPrefixedText() );
1741 $selflink = array( $this->mTitle
->getPrefixedText() );
1743 $useSubpages = $this->areSubpagesAllowed();
1744 wfProfileOut( __METHOD__
.'-setup' );
1746 # Loop for each link
1747 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1748 # Check for excessive memory usage
1749 if ( $holders->isBig() ) {
1751 # Do the existence check, replace the link holders and clear the array
1752 $holders->replace( $s );
1756 if ( $useLinkPrefixExtension ) {
1757 wfProfileIn( __METHOD__
.'-prefixhandling' );
1758 if ( preg_match( $e2, $s, $m ) ) {
1765 if ( $first_prefix ) {
1766 $prefix = $first_prefix;
1767 $first_prefix = false;
1769 wfProfileOut( __METHOD__
.'-prefixhandling' );
1772 $might_be_img = false;
1774 wfProfileIn( __METHOD__
."-e1" );
1775 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1777 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1778 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1779 # the real problem is with the $e1 regex
1782 # Still some problems for cases where the ] is meant to be outside punctuation,
1783 # and no image is in sight. See bug 2095.
1785 if ( $text !== '' &&
1786 substr( $m[3], 0, 1 ) === ']' &&
1787 strpos( $text, '[' ) !== false
1790 $text .= ']'; # so that replaceExternalLinks($text) works later
1791 $m[3] = substr( $m[3], 1 );
1793 # fix up urlencoded title texts
1794 if ( strpos( $m[1], '%' ) !== false ) {
1795 # Should anchors '#' also be rejected?
1796 $m[1] = str_replace( array('<', '>'), array('<', '>'), rawurldecode( $m[1] ) );
1799 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1800 $might_be_img = true;
1802 if ( strpos( $m[1], '%' ) !== false ) {
1803 $m[1] = rawurldecode( $m[1] );
1806 } else { # Invalid form; output directly
1807 $s .= $prefix . '[[' . $line ;
1808 wfProfileOut( __METHOD__
."-e1" );
1811 wfProfileOut( __METHOD__
."-e1" );
1812 wfProfileIn( __METHOD__
."-misc" );
1814 # Don't allow internal links to pages containing
1815 # PROTO: where PROTO is a valid URL protocol; these
1816 # should be external links.
1817 if ( preg_match( '/^(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1818 $s .= $prefix . '[[' . $line ;
1819 wfProfileOut( __METHOD__
."-misc" );
1823 # Make subpage if necessary
1824 if ( $useSubpages ) {
1825 $link = $this->maybeDoSubpageLink( $m[1], $text );
1830 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1832 # Strip off leading ':'
1833 $link = substr( $link, 1 );
1836 wfProfileOut( __METHOD__
."-misc" );
1837 wfProfileIn( __METHOD__
."-title" );
1838 $nt = Title
::newFromText( $this->mStripState
->unstripNoWiki( $link ) );
1839 if ( $nt === null ) {
1840 $s .= $prefix . '[[' . $line;
1841 wfProfileOut( __METHOD__
."-title" );
1845 $ns = $nt->getNamespace();
1846 $iw = $nt->getInterWiki();
1847 wfProfileOut( __METHOD__
."-title" );
1849 if ( $might_be_img ) { # if this is actually an invalid link
1850 wfProfileIn( __METHOD__
."-might_be_img" );
1851 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
1854 # look at the next 'line' to see if we can close it there
1856 $next_line = $a->current();
1857 if ( $next_line === false ||
$next_line === null ) {
1860 $m = explode( ']]', $next_line, 3 );
1861 if ( count( $m ) == 3 ) {
1862 # the first ]] closes the inner link, the second the image
1864 $text .= "[[{$m[0]}]]{$m[1]}";
1867 } elseif ( count( $m ) == 2 ) {
1868 # if there's exactly one ]] that's fine, we'll keep looking
1869 $text .= "[[{$m[0]}]]{$m[1]}";
1871 # if $next_line is invalid too, we need look no further
1872 $text .= '[[' . $next_line;
1877 # we couldn't find the end of this imageLink, so output it raw
1878 # but don't ignore what might be perfectly normal links in the text we've examined
1879 $holders->merge( $this->replaceInternalLinks2( $text ) );
1880 $s .= "{$prefix}[[$link|$text";
1881 # note: no $trail, because without an end, there *is* no trail
1882 wfProfileOut( __METHOD__
."-might_be_img" );
1885 } else { # it's not an image, so output it raw
1886 $s .= "{$prefix}[[$link|$text";
1887 # note: no $trail, because without an end, there *is* no trail
1888 wfProfileOut( __METHOD__
."-might_be_img" );
1891 wfProfileOut( __METHOD__
."-might_be_img" );
1894 $wasblank = ( $text == '' );
1898 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1899 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1900 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1901 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1902 $text = $this->doQuotes( $text );
1905 # Link not escaped by : , create the various objects
1908 wfProfileIn( __METHOD__
."-interwiki" );
1909 if ( $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && Language
::fetchLanguageName( $iw, null, 'mw' ) ) {
1910 $this->mOutput
->addLanguageLink( $nt->getFullText() );
1911 $s = rtrim( $s . $prefix );
1912 $s .= trim( $trail, "\n" ) == '' ?
'': $prefix . $trail;
1913 wfProfileOut( __METHOD__
."-interwiki" );
1916 wfProfileOut( __METHOD__
."-interwiki" );
1918 if ( $ns == NS_FILE
) {
1919 wfProfileIn( __METHOD__
."-image" );
1920 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
1922 # if no parameters were passed, $text
1923 # becomes something like "File:Foo.png",
1924 # which we don't want to pass on to the
1928 # recursively parse links inside the image caption
1929 # actually, this will parse them in any other parameters, too,
1930 # but it might be hard to fix that, and it doesn't matter ATM
1931 $text = $this->replaceExternalLinks( $text );
1932 $holders->merge( $this->replaceInternalLinks2( $text ) );
1934 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1935 $s .= $prefix . $this->armorLinks(
1936 $this->makeImage( $nt, $text, $holders ) ) . $trail;
1938 $s .= $prefix . $trail;
1940 wfProfileOut( __METHOD__
."-image" );
1944 if ( $ns == NS_CATEGORY
) {
1945 wfProfileIn( __METHOD__
."-category" );
1946 $s = rtrim( $s . "\n" ); # bug 87
1949 $sortkey = $this->getDefaultSort();
1953 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
1954 $sortkey = str_replace( "\n", '', $sortkey );
1955 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
1956 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
1959 * Strip the whitespace Category links produce, see bug 87
1960 * @todo We might want to use trim($tmp, "\n") here.
1962 $s .= trim( $prefix . $trail, "\n" ) == '' ?
'' : $prefix . $trail;
1964 wfProfileOut( __METHOD__
."-category" );
1969 # Self-link checking
1970 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL
) {
1971 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1972 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
1977 # NS_MEDIA is a pseudo-namespace for linking directly to a file
1978 # @todo FIXME: Should do batch file existence checks, see comment below
1979 if ( $ns == NS_MEDIA
) {
1980 wfProfileIn( __METHOD__
."-media" );
1981 # Give extensions a chance to select the file revision for us
1984 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1985 array( $this, $nt, &$options, &$descQuery ) );
1986 # Fetch and register the file (file title may be different via hooks)
1987 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
1988 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1989 $s .= $prefix . $this->armorLinks(
1990 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1991 wfProfileOut( __METHOD__
."-media" );
1995 wfProfileIn( __METHOD__
."-always_known" );
1996 # Some titles, such as valid special pages or files in foreign repos, should
1997 # be shown as bluelinks even though they're not included in the page table
1999 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2000 # batch file existence checks for NS_FILE and NS_MEDIA
2001 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2002 $this->mOutput
->addLink( $nt );
2003 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2005 # Links will be added to the output link list after checking
2006 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2008 wfProfileOut( __METHOD__
."-always_known" );
2010 wfProfileOut( __METHOD__
);
2015 * Render a forced-blue link inline; protect against double expansion of
2016 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2017 * Since this little disaster has to split off the trail text to avoid
2018 * breaking URLs in the following text without breaking trails on the
2019 * wiki links, it's been made into a horrible function.
2022 * @param $text String
2023 * @param $query Array or String
2024 * @param $trail String
2025 * @param $prefix String
2026 * @return String: HTML-wikitext mix oh yuck
2028 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2029 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2031 if ( is_string( $query ) ) {
2032 $query = wfCgiToArray( $query );
2034 if ( $text == '' ) {
2035 $text = htmlspecialchars( $nt->getPrefixedText() );
2038 $link = Linker
::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2040 return $this->armorLinks( $link ) . $trail;
2044 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2045 * going to go through further parsing steps before inline URL expansion.
2047 * Not needed quite as much as it used to be since free links are a bit
2048 * more sensible these days. But bracketed links are still an issue.
2050 * @param $text String: more-or-less HTML
2051 * @return String: less-or-more HTML with NOPARSE bits
2053 function armorLinks( $text ) {
2054 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2055 "{$this->mUniqPrefix}NOPARSE$1", $text );
2059 * Return true if subpage links should be expanded on this page.
2062 function areSubpagesAllowed() {
2063 # Some namespaces don't allow subpages
2064 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2068 * Handle link to subpage if necessary
2070 * @param $target String: the source of the link
2071 * @param &$text String: the link text, modified as necessary
2072 * @return string the full name of the link
2075 function maybeDoSubpageLink( $target, &$text ) {
2076 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2080 * Used by doBlockLevels()
2085 function closeParagraph() {
2087 if ( $this->mLastSection
!= '' ) {
2088 $result = '</' . $this->mLastSection
. ">\n";
2090 $this->mInPre
= false;
2091 $this->mLastSection
= '';
2096 * getCommon() returns the length of the longest common substring
2097 * of both arguments, starting at the beginning of both.
2100 * @param $st1 string
2101 * @param $st2 string
2105 function getCommon( $st1, $st2 ) {
2106 $fl = strlen( $st1 );
2107 $shorter = strlen( $st2 );
2108 if ( $fl < $shorter ) {
2112 for ( $i = 0; $i < $shorter; ++
$i ) {
2113 if ( $st1[$i] != $st2[$i] ) {
2121 * These next three functions open, continue, and close the list
2122 * element appropriate to the prefix character passed into them.
2125 * @param $char string
2129 function openList( $char ) {
2130 $result = $this->closeParagraph();
2132 if ( '*' === $char ) {
2133 $result .= '<ul><li>';
2134 } elseif ( '#' === $char ) {
2135 $result .= '<ol><li>';
2136 } elseif ( ':' === $char ) {
2137 $result .= '<dl><dd>';
2138 } elseif ( ';' === $char ) {
2139 $result .= '<dl><dt>';
2140 $this->mDTopen
= true;
2142 $result = '<!-- ERR 1 -->';
2150 * @param $char String
2155 function nextItem( $char ) {
2156 if ( '*' === $char ||
'#' === $char ) {
2158 } elseif ( ':' === $char ||
';' === $char ) {
2160 if ( $this->mDTopen
) {
2163 if ( ';' === $char ) {
2164 $this->mDTopen
= true;
2165 return $close . '<dt>';
2167 $this->mDTopen
= false;
2168 return $close . '<dd>';
2171 return '<!-- ERR 2 -->';
2176 * @param $char String
2181 function closeList( $char ) {
2182 if ( '*' === $char ) {
2183 $text = '</li></ul>';
2184 } elseif ( '#' === $char ) {
2185 $text = '</li></ol>';
2186 } elseif ( ':' === $char ) {
2187 if ( $this->mDTopen
) {
2188 $this->mDTopen
= false;
2189 $text = '</dt></dl>';
2191 $text = '</dd></dl>';
2194 return '<!-- ERR 3 -->';
2201 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2203 * @param $text String
2204 * @param $linestart Boolean: whether or not this is at the start of a line.
2206 * @return string the lists rendered as HTML
2208 function doBlockLevels( $text, $linestart ) {
2209 wfProfileIn( __METHOD__
);
2211 # Parsing through the text line by line. The main thing
2212 # happening here is handling of block-level elements p, pre,
2213 # and making lists from lines starting with * # : etc.
2215 $textLines = StringUtils
::explode( "\n", $text );
2217 $lastPrefix = $output = '';
2218 $this->mDTopen
= $inBlockElem = false;
2220 $paragraphStack = false;
2222 foreach ( $textLines as $oLine ) {
2224 if ( !$linestart ) {
2234 $lastPrefixLength = strlen( $lastPrefix );
2235 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2236 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2237 # If not in a <pre> element, scan for and figure out what prefixes are there.
2238 if ( !$this->mInPre
) {
2239 # Multiple prefixes may abut each other for nested lists.
2240 $prefixLength = strspn( $oLine, '*#:;' );
2241 $prefix = substr( $oLine, 0, $prefixLength );
2244 # ; and : are both from definition-lists, so they're equivalent
2245 # for the purposes of determining whether or not we need to open/close
2247 $prefix2 = str_replace( ';', ':', $prefix );
2248 $t = substr( $oLine, $prefixLength );
2249 $this->mInPre
= (bool)$preOpenMatch;
2251 # Don't interpret any other prefixes in preformatted text
2253 $prefix = $prefix2 = '';
2258 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2259 # Same as the last item, so no need to deal with nesting or opening stuff
2260 $output .= $this->nextItem( substr( $prefix, -1 ) );
2261 $paragraphStack = false;
2263 if ( substr( $prefix, -1 ) === ';') {
2264 # The one nasty exception: definition lists work like this:
2265 # ; title : definition text
2266 # So we check for : in the remainder text to split up the
2267 # title and definition, without b0rking links.
2269 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2271 $output .= $term . $this->nextItem( ':' );
2274 } elseif ( $prefixLength ||
$lastPrefixLength ) {
2275 # We need to open or close prefixes, or both.
2277 # Either open or close a level...
2278 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2279 $paragraphStack = false;
2281 # Close all the prefixes which aren't shared.
2282 while ( $commonPrefixLength < $lastPrefixLength ) {
2283 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2284 --$lastPrefixLength;
2287 # Continue the current prefix if appropriate.
2288 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2289 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2292 # Open prefixes where appropriate.
2293 while ( $prefixLength > $commonPrefixLength ) {
2294 $char = substr( $prefix, $commonPrefixLength, 1 );
2295 $output .= $this->openList( $char );
2297 if ( ';' === $char ) {
2298 # @todo FIXME: This is dupe of code above
2299 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2301 $output .= $term . $this->nextItem( ':' );
2304 ++
$commonPrefixLength;
2306 $lastPrefix = $prefix2;
2309 # If we have no prefixes, go to paragraph mode.
2310 if ( 0 == $prefixLength ) {
2311 wfProfileIn( __METHOD__
."-paragraph" );
2312 # No prefix (not in list)--go to paragraph mode
2313 # XXX: use a stack for nestable elements like span, table and div
2314 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2315 $closematch = preg_match(
2316 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2317 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix
.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2318 if ( $openmatch or $closematch ) {
2319 $paragraphStack = false;
2320 # TODO bug 5718: paragraph closed
2321 $output .= $this->closeParagraph();
2322 if ( $preOpenMatch and !$preCloseMatch ) {
2323 $this->mInPre
= true;
2325 $inBlockElem = !$closematch;
2326 } elseif ( !$inBlockElem && !$this->mInPre
) {
2327 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection
=== 'pre' ||
trim( $t ) != '' ) ) {
2329 if ( $this->mLastSection
!== 'pre' ) {
2330 $paragraphStack = false;
2331 $output .= $this->closeParagraph().'<pre>';
2332 $this->mLastSection
= 'pre';
2334 $t = substr( $t, 1 );
2337 if ( trim( $t ) === '' ) {
2338 if ( $paragraphStack ) {
2339 $output .= $paragraphStack.'<br />';
2340 $paragraphStack = false;
2341 $this->mLastSection
= 'p';
2343 if ( $this->mLastSection
!== 'p' ) {
2344 $output .= $this->closeParagraph();
2345 $this->mLastSection
= '';
2346 $paragraphStack = '<p>';
2348 $paragraphStack = '</p><p>';
2352 if ( $paragraphStack ) {
2353 $output .= $paragraphStack;
2354 $paragraphStack = false;
2355 $this->mLastSection
= 'p';
2356 } elseif ( $this->mLastSection
!== 'p' ) {
2357 $output .= $this->closeParagraph().'<p>';
2358 $this->mLastSection
= 'p';
2363 wfProfileOut( __METHOD__
."-paragraph" );
2365 # somewhere above we forget to get out of pre block (bug 785)
2366 if ( $preCloseMatch && $this->mInPre
) {
2367 $this->mInPre
= false;
2369 if ( $paragraphStack === false ) {
2373 while ( $prefixLength ) {
2374 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2377 if ( $this->mLastSection
!= '' ) {
2378 $output .= '</' . $this->mLastSection
. '>';
2379 $this->mLastSection
= '';
2382 wfProfileOut( __METHOD__
);
2387 * Split up a string on ':', ignoring any occurrences inside tags
2388 * to prevent illegal overlapping.
2390 * @param $str String the string to split
2391 * @param &$before String set to everything before the ':'
2392 * @param &$after String set to everything after the ':'
2393 * @return String the position of the ':', or false if none found
2395 function findColonNoLinks( $str, &$before, &$after ) {
2396 wfProfileIn( __METHOD__
);
2398 $pos = strpos( $str, ':' );
2399 if ( $pos === false ) {
2401 wfProfileOut( __METHOD__
);
2405 $lt = strpos( $str, '<' );
2406 if ( $lt === false ||
$lt > $pos ) {
2407 # Easy; no tag nesting to worry about
2408 $before = substr( $str, 0, $pos );
2409 $after = substr( $str, $pos+
1 );
2410 wfProfileOut( __METHOD__
);
2414 # Ugly state machine to walk through avoiding tags.
2415 $state = self
::COLON_STATE_TEXT
;
2417 $len = strlen( $str );
2418 for( $i = 0; $i < $len; $i++
) {
2422 # (Using the number is a performance hack for common cases)
2423 case 0: # self::COLON_STATE_TEXT:
2426 # Could be either a <start> tag or an </end> tag
2427 $state = self
::COLON_STATE_TAGSTART
;
2430 if ( $stack == 0 ) {
2432 $before = substr( $str, 0, $i );
2433 $after = substr( $str, $i +
1 );
2434 wfProfileOut( __METHOD__
);
2437 # Embedded in a tag; don't break it.
2440 # Skip ahead looking for something interesting
2441 $colon = strpos( $str, ':', $i );
2442 if ( $colon === false ) {
2443 # Nothing else interesting
2444 wfProfileOut( __METHOD__
);
2447 $lt = strpos( $str, '<', $i );
2448 if ( $stack === 0 ) {
2449 if ( $lt === false ||
$colon < $lt ) {
2451 $before = substr( $str, 0, $colon );
2452 $after = substr( $str, $colon +
1 );
2453 wfProfileOut( __METHOD__
);
2457 if ( $lt === false ) {
2458 # Nothing else interesting to find; abort!
2459 # We're nested, but there's no close tags left. Abort!
2462 # Skip ahead to next tag start
2464 $state = self
::COLON_STATE_TAGSTART
;
2467 case 1: # self::COLON_STATE_TAG:
2472 $state = self
::COLON_STATE_TEXT
;
2475 # Slash may be followed by >?
2476 $state = self
::COLON_STATE_TAGSLASH
;
2482 case 2: # self::COLON_STATE_TAGSTART:
2485 $state = self
::COLON_STATE_CLOSETAG
;
2488 $state = self
::COLON_STATE_COMMENT
;
2491 # Illegal early close? This shouldn't happen D:
2492 $state = self
::COLON_STATE_TEXT
;
2495 $state = self
::COLON_STATE_TAG
;
2498 case 3: # self::COLON_STATE_CLOSETAG:
2503 wfDebug( __METHOD__
.": Invalid input; too many close tags\n" );
2504 wfProfileOut( __METHOD__
);
2507 $state = self
::COLON_STATE_TEXT
;
2510 case self
::COLON_STATE_TAGSLASH
:
2512 # Yes, a self-closed tag <blah/>
2513 $state = self
::COLON_STATE_TEXT
;
2515 # Probably we're jumping the gun, and this is an attribute
2516 $state = self
::COLON_STATE_TAG
;
2519 case 5: # self::COLON_STATE_COMMENT:
2521 $state = self
::COLON_STATE_COMMENTDASH
;
2524 case self
::COLON_STATE_COMMENTDASH
:
2526 $state = self
::COLON_STATE_COMMENTDASHDASH
;
2528 $state = self
::COLON_STATE_COMMENT
;
2531 case self
::COLON_STATE_COMMENTDASHDASH
:
2533 $state = self
::COLON_STATE_TEXT
;
2535 $state = self
::COLON_STATE_COMMENT
;
2539 throw new MWException( "State machine error in " . __METHOD__
);
2543 wfDebug( __METHOD__
.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2544 wfProfileOut( __METHOD__
);
2547 wfProfileOut( __METHOD__
);
2552 * Return value of a magic variable (like PAGENAME)
2556 * @param $index integer
2557 * @param $frame PPFrame
2561 function getVariableValue( $index, $frame = false ) {
2562 global $wgContLang, $wgSitename, $wgServer;
2563 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2565 if ( is_null( $this->mTitle
) ) {
2566 // If no title set, bad things are going to happen
2567 // later. Title should always be set since this
2568 // should only be called in the middle of a parse
2569 // operation (but the unit-tests do funky stuff)
2570 throw new MWException( __METHOD__
. ' Should only be '
2571 . ' called while parsing (no title set)' );
2575 * Some of these require message or data lookups and can be
2576 * expensive to check many times.
2578 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache
) ) ) {
2579 if ( isset( $this->mVarCache
[$index] ) ) {
2580 return $this->mVarCache
[$index];
2584 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2585 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2588 global $wgLocaltimezone;
2589 if ( isset( $wgLocaltimezone ) ) {
2590 $oldtz = date_default_timezone_get();
2591 date_default_timezone_set( $wgLocaltimezone );
2594 $localTimestamp = date( 'YmdHis', $ts );
2595 $localMonth = date( 'm', $ts );
2596 $localMonth1 = date( 'n', $ts );
2597 $localMonthName = date( 'n', $ts );
2598 $localDay = date( 'j', $ts );
2599 $localDay2 = date( 'd', $ts );
2600 $localDayOfWeek = date( 'w', $ts );
2601 $localWeek = date( 'W', $ts );
2602 $localYear = date( 'Y', $ts );
2603 $localHour = date( 'H', $ts );
2604 if ( isset( $wgLocaltimezone ) ) {
2605 date_default_timezone_set( $oldtz );
2608 $pageLang = $this->getFunctionLang();
2611 case 'currentmonth':
2612 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2614 case 'currentmonth1':
2615 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2617 case 'currentmonthname':
2618 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2620 case 'currentmonthnamegen':
2621 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2623 case 'currentmonthabbrev':
2624 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2627 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2630 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2633 $value = $pageLang->formatNum( $localMonth );
2636 $value = $pageLang->formatNum( $localMonth1 );
2638 case 'localmonthname':
2639 $value = $pageLang->getMonthName( $localMonthName );
2641 case 'localmonthnamegen':
2642 $value = $pageLang->getMonthNameGen( $localMonthName );
2644 case 'localmonthabbrev':
2645 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2648 $value = $pageLang->formatNum( $localDay );
2651 $value = $pageLang->formatNum( $localDay2 );
2654 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2657 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2659 case 'fullpagename':
2660 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2662 case 'fullpagenamee':
2663 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2666 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2668 case 'subpagenamee':
2669 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2671 case 'basepagename':
2672 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2674 case 'basepagenamee':
2675 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle
->getBaseText() ) ) );
2677 case 'talkpagename':
2678 if ( $this->mTitle
->canTalk() ) {
2679 $talkPage = $this->mTitle
->getTalkPage();
2680 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2685 case 'talkpagenamee':
2686 if ( $this->mTitle
->canTalk() ) {
2687 $talkPage = $this->mTitle
->getTalkPage();
2688 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2693 case 'subjectpagename':
2694 $subjPage = $this->mTitle
->getSubjectPage();
2695 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2697 case 'subjectpagenamee':
2698 $subjPage = $this->mTitle
->getSubjectPage();
2699 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2702 # Let the edit saving system know we should parse the page
2703 # *after* a revision ID has been assigned.
2704 $this->mOutput
->setFlag( 'vary-revision' );
2705 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision...\n" );
2706 $value = $this->mRevisionId
;
2709 # Let the edit saving system know we should parse the page
2710 # *after* a revision ID has been assigned. This is for null edits.
2711 $this->mOutput
->setFlag( 'vary-revision' );
2712 wfDebug( __METHOD__
. ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2713 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2715 case 'revisionday2':
2716 # Let the edit saving system know we should parse the page
2717 # *after* a revision ID has been assigned. This is for null edits.
2718 $this->mOutput
->setFlag( 'vary-revision' );
2719 wfDebug( __METHOD__
. ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2720 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2722 case 'revisionmonth':
2723 # Let the edit saving system know we should parse the page
2724 # *after* a revision ID has been assigned. This is for null edits.
2725 $this->mOutput
->setFlag( 'vary-revision' );
2726 wfDebug( __METHOD__
. ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2727 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2729 case 'revisionmonth1':
2730 # Let the edit saving system know we should parse the page
2731 # *after* a revision ID has been assigned. This is for null edits.
2732 $this->mOutput
->setFlag( 'vary-revision' );
2733 wfDebug( __METHOD__
. ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2734 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2736 case 'revisionyear':
2737 # Let the edit saving system know we should parse the page
2738 # *after* a revision ID has been assigned. This is for null edits.
2739 $this->mOutput
->setFlag( 'vary-revision' );
2740 wfDebug( __METHOD__
. ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2741 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2743 case 'revisiontimestamp':
2744 # Let the edit saving system know we should parse the page
2745 # *after* a revision ID has been assigned. This is for null edits.
2746 $this->mOutput
->setFlag( 'vary-revision' );
2747 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2748 $value = $this->getRevisionTimestamp();
2750 case 'revisionuser':
2751 # Let the edit saving system know we should parse the page
2752 # *after* a revision ID has been assigned. This is for null edits.
2753 $this->mOutput
->setFlag( 'vary-revision' );
2754 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2755 $value = $this->getRevisionUser();
2758 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2761 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2764 $value = $this->mTitle
->canTalk() ?
str_replace( '_',' ',$this->mTitle
->getTalkNsText() ) : '';
2767 $value = $this->mTitle
->canTalk() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2769 case 'subjectspace':
2770 $value = $this->mTitle
->getSubjectNsText();
2772 case 'subjectspacee':
2773 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2775 case 'currentdayname':
2776 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) +
1 );
2779 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2782 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2785 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2788 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2789 # int to remove the padding
2790 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2793 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2795 case 'localdayname':
2796 $value = $pageLang->getWeekdayName( $localDayOfWeek +
1 );
2799 $value = $pageLang->formatNum( $localYear, true );
2802 $value = $pageLang->time( $localTimestamp, false, false );
2805 $value = $pageLang->formatNum( $localHour, true );
2808 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2809 # int to remove the padding
2810 $value = $pageLang->formatNum( (int)$localWeek );
2813 $value = $pageLang->formatNum( $localDayOfWeek );
2815 case 'numberofarticles':
2816 $value = $pageLang->formatNum( SiteStats
::articles() );
2818 case 'numberoffiles':
2819 $value = $pageLang->formatNum( SiteStats
::images() );
2821 case 'numberofusers':
2822 $value = $pageLang->formatNum( SiteStats
::users() );
2824 case 'numberofactiveusers':
2825 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2827 case 'numberofpages':
2828 $value = $pageLang->formatNum( SiteStats
::pages() );
2830 case 'numberofadmins':
2831 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2833 case 'numberofedits':
2834 $value = $pageLang->formatNum( SiteStats
::edits() );
2836 case 'numberofviews':
2837 $value = $pageLang->formatNum( SiteStats
::views() );
2839 case 'currenttimestamp':
2840 $value = wfTimestamp( TS_MW
, $ts );
2842 case 'localtimestamp':
2843 $value = $localTimestamp;
2845 case 'currentversion':
2846 $value = SpecialVersion
::getVersion();
2849 return $wgArticlePath;
2855 $serverParts = wfParseUrl( $wgServer );
2856 return $serverParts && isset( $serverParts['host'] ) ?
$serverParts['host'] : $wgServer;
2858 return $wgScriptPath;
2860 return $wgStylePath;
2861 case 'directionmark':
2862 return $pageLang->getDirMark();
2863 case 'contentlanguage':
2864 global $wgLanguageCode;
2865 return $wgLanguageCode;
2868 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache
, &$index, &$ret, &$frame ) ) ) {
2876 $this->mVarCache
[$index] = $value;
2883 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2887 function initialiseVariables() {
2888 wfProfileIn( __METHOD__
);
2889 $variableIDs = MagicWord
::getVariableIDs();
2890 $substIDs = MagicWord
::getSubstIDs();
2892 $this->mVariables
= new MagicWordArray( $variableIDs );
2893 $this->mSubstWords
= new MagicWordArray( $substIDs );
2894 wfProfileOut( __METHOD__
);
2898 * Preprocess some wikitext and return the document tree.
2899 * This is the ghost of replace_variables().
2901 * @param $text String: The text to parse
2902 * @param $flags Integer: bitwise combination of:
2903 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2904 * included. Default is to assume a direct page view.
2906 * The generated DOM tree must depend only on the input text and the flags.
2907 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2909 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2910 * change in the DOM tree for a given text, must be passed through the section identifier
2911 * in the section edit link and thus back to extractSections().
2913 * The output of this function is currently only cached in process memory, but a persistent
2914 * cache may be implemented at a later date which takes further advantage of these strict
2915 * dependency requirements.
2921 function preprocessToDom( $text, $flags = 0 ) {
2922 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2927 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2933 public static function splitWhitespace( $s ) {
2934 $ltrimmed = ltrim( $s );
2935 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2936 $trimmed = rtrim( $ltrimmed );
2937 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2939 $w2 = substr( $ltrimmed, -$diff );
2943 return array( $w1, $trimmed, $w2 );
2947 * Replace magic variables, templates, and template arguments
2948 * with the appropriate text. Templates are substituted recursively,
2949 * taking care to avoid infinite loops.
2951 * Note that the substitution depends on value of $mOutputType:
2952 * self::OT_WIKI: only {{subst:}} templates
2953 * self::OT_PREPROCESS: templates but not extension tags
2954 * self::OT_HTML: all templates and extension tags
2956 * @param $text String the text to transform
2957 * @param $frame PPFrame Object describing the arguments passed to the template.
2958 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2959 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2960 * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
2965 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2966 # Is there any text? Also, Prevent too big inclusions!
2967 if ( strlen( $text ) < 1 ||
strlen( $text ) > $this->mOptions
->getMaxIncludeSize() ) {
2970 wfProfileIn( __METHOD__
);
2972 if ( $frame === false ) {
2973 $frame = $this->getPreprocessor()->newFrame();
2974 } elseif ( !( $frame instanceof PPFrame
) ) {
2975 wfDebug( __METHOD__
." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2976 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2979 $dom = $this->preprocessToDom( $text );
2980 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
2981 $text = $frame->expand( $dom, $flags );
2983 wfProfileOut( __METHOD__
);
2988 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2990 * @param $args array
2994 static function createAssocArgs( $args ) {
2995 $assocArgs = array();
2997 foreach ( $args as $arg ) {
2998 $eqpos = strpos( $arg, '=' );
2999 if ( $eqpos === false ) {
3000 $assocArgs[$index++
] = $arg;
3002 $name = trim( substr( $arg, 0, $eqpos ) );
3003 $value = trim( substr( $arg, $eqpos+
1 ) );
3004 if ( $value === false ) {
3007 if ( $name !== false ) {
3008 $assocArgs[$name] = $value;
3017 * Warn the user when a parser limitation is reached
3018 * Will warn at most once the user per limitation type
3020 * @param $limitationType String: should be one of:
3021 * 'expensive-parserfunction' (corresponding messages:
3022 * 'expensive-parserfunction-warning',
3023 * 'expensive-parserfunction-category')
3024 * 'post-expand-template-argument' (corresponding messages:
3025 * 'post-expand-template-argument-warning',
3026 * 'post-expand-template-argument-category')
3027 * 'post-expand-template-inclusion' (corresponding messages:
3028 * 'post-expand-template-inclusion-warning',
3029 * 'post-expand-template-inclusion-category')
3030 * @param $current int|null Current value
3031 * @param $max int|null Maximum allowed, when an explicit limit has been
3032 * exceeded, provide the values (optional)
3034 function limitationWarn( $limitationType, $current = null, $max = null) {
3035 # does no harm if $current and $max are present but are unnecessary for the message
3036 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3037 $this->mOutput
->addWarning( $warning );
3038 $this->addTrackingCategory( "$limitationType-category" );
3042 * Return the text of a template, after recursively
3043 * replacing any variables or templates within the template.
3045 * @param $piece Array: the parts of the template
3046 * $piece['title']: the title, i.e. the part before the |
3047 * $piece['parts']: the parameter array
3048 * $piece['lineStart']: whether the brace was at the start of a line
3049 * @param $frame PPFrame The current frame, contains template arguments
3050 * @return String: the text of the template
3053 function braceSubstitution( $piece, $frame ) {
3054 global $wgNonincludableNamespaces, $wgContLang;
3055 wfProfileIn( __METHOD__
);
3056 wfProfileIn( __METHOD__
.'-setup' );
3059 $found = false; # $text has been filled
3060 $nowiki = false; # wiki markup in $text should be escaped
3061 $isHTML = false; # $text is HTML, armour it against wikitext transformation
3062 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3063 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
3064 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
3066 # Title object, where $text came from
3069 # $part1 is the bit before the first |, and must contain only title characters.
3070 # Various prefixes will be stripped from it later.
3071 $titleWithSpaces = $frame->expand( $piece['title'] );
3072 $part1 = trim( $titleWithSpaces );
3075 # Original title text preserved for various purposes
3076 $originalTitle = $part1;
3078 # $args is a list of argument nodes, starting from index 0, not including $part1
3079 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3080 $args = ( null == $piece['parts'] ) ?
array() : $piece['parts'];
3081 wfProfileOut( __METHOD__
.'-setup' );
3083 $titleProfileIn = null; // profile templates
3086 wfProfileIn( __METHOD__
.'-modifiers' );
3089 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3091 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3092 # Decide whether to expand template or keep wikitext as-is.
3093 if ( $this->ot
['wiki'] ) {
3094 if ( $substMatch === false ) {
3095 $literal = true; # literal when in PST with no prefix
3097 $literal = false; # expand when in PST with subst: or safesubst:
3100 if ( $substMatch == 'subst' ) {
3101 $literal = true; # literal when not in PST with plain subst:
3103 $literal = false; # expand when not in PST with safesubst: or no prefix
3107 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3114 if ( !$found && $args->getLength() == 0 ) {
3115 $id = $this->mVariables
->matchStartToEnd( $part1 );
3116 if ( $id !== false ) {
3117 $text = $this->getVariableValue( $id, $frame );
3118 if ( MagicWord
::getCacheTTL( $id ) > -1 ) {
3119 $this->mOutput
->updateCacheExpiry( MagicWord
::getCacheTTL( $id ) );
3125 # MSG, MSGNW and RAW
3128 $mwMsgnw = MagicWord
::get( 'msgnw' );
3129 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3132 # Remove obsolete MSG:
3133 $mwMsg = MagicWord
::get( 'msg' );
3134 $mwMsg->matchStartAndRemove( $part1 );
3138 $mwRaw = MagicWord
::get( 'raw' );
3139 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3140 $forceRawInterwiki = true;
3143 wfProfileOut( __METHOD__
.'-modifiers' );
3147 wfProfileIn( __METHOD__
. '-pfunc' );
3149 $colonPos = strpos( $part1, ':' );
3150 if ( $colonPos !== false ) {
3151 # Case sensitive functions
3152 $function = substr( $part1, 0, $colonPos );
3153 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3154 $function = $this->mFunctionSynonyms
[1][$function];
3156 # Case insensitive functions
3157 $function = $wgContLang->lc( $function );
3158 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3159 $function = $this->mFunctionSynonyms
[0][$function];
3165 wfProfileIn( __METHOD__
. '-pfunc-' . $function );
3166 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3167 $initialArgs = array( &$this );
3168 $funcArgs = array( trim( substr( $part1, $colonPos +
1 ) ) );
3169 if ( $flags & SFH_OBJECT_ARGS
) {
3170 # Add a frame parameter, and pass the arguments as an array
3171 $allArgs = $initialArgs;
3172 $allArgs[] = $frame;
3173 for ( $i = 0; $i < $args->getLength(); $i++
) {
3174 $funcArgs[] = $args->item( $i );
3176 $allArgs[] = $funcArgs;
3178 # Convert arguments to plain text
3179 for ( $i = 0; $i < $args->getLength(); $i++
) {
3180 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3182 $allArgs = array_merge( $initialArgs, $funcArgs );
3185 # Workaround for PHP bug 35229 and similar
3186 if ( !is_callable( $callback ) ) {
3187 wfProfileOut( __METHOD__
. '-pfunc-' . $function );
3188 wfProfileOut( __METHOD__
. '-pfunc' );
3189 wfProfileOut( __METHOD__
);
3190 throw new MWException( "Tag hook for $function is not callable\n" );
3192 $result = call_user_func_array( $callback, $allArgs );
3195 $preprocessFlags = 0;
3197 if ( is_array( $result ) ) {
3198 if ( isset( $result[0] ) ) {
3200 unset( $result[0] );
3203 # Extract flags into the local scope
3204 # This allows callers to set flags such as nowiki, found, etc.
3210 $text = $this->preprocessToDom( $text, $preprocessFlags );
3213 wfProfileOut( __METHOD__
. '-pfunc-' . $function );
3216 wfProfileOut( __METHOD__
. '-pfunc' );
3219 # Finish mangling title and then check for loops.
3220 # Set $title to a Title object and $titleText to the PDBK
3223 # Split the title into page and subpage
3225 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3226 if ( $subpage !== '' ) {
3227 $ns = $this->mTitle
->getNamespace();
3229 $title = Title
::newFromText( $part1, $ns );
3231 $titleText = $title->getPrefixedText();
3232 # Check for language variants if the template is not found
3233 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3234 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3236 # Do recursion depth check
3237 $limit = $this->mOptions
->getMaxTemplateDepth();
3238 if ( $frame->depth
>= $limit ) {
3240 $text = '<span class="error">'
3241 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3247 # Load from database
3248 if ( !$found && $title ) {
3249 $titleProfileIn = __METHOD__
. "-title-" . $title->getDBKey();
3250 wfProfileIn( $titleProfileIn ); // template in
3251 wfProfileIn( __METHOD__
. '-loadtpl' );
3252 if ( !$title->isExternal() ) {
3253 if ( $title->isSpecialPage()
3254 && $this->mOptions
->getAllowSpecialInclusion()
3255 && $this->ot
['html'] )
3257 // Pass the template arguments as URL parameters.
3258 // "uselang" will have no effect since the Language object
3259 // is forced to the one defined in ParserOptions.
3260 $pageArgs = array();
3261 for ( $i = 0; $i < $args->getLength(); $i++
) {
3262 $bits = $args->item( $i )->splitArg();
3263 if ( strval( $bits['index'] ) === '' ) {
3264 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3265 $value = trim( $frame->expand( $bits['value'] ) );
3266 $pageArgs[$name] = $value;
3270 // Create a new context to execute the special page
3271 $context = new RequestContext
;
3272 $context->setTitle( $title );
3273 $context->setRequest( new FauxRequest( $pageArgs ) );
3274 $context->setUser( $this->getUser() );
3275 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3276 $ret = SpecialPageFactory
::capturePath( $title, $context );
3278 $text = $context->getOutput()->getHTML();
3279 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3282 $this->disableCache();
3284 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3285 $found = false; # access denied
3286 wfDebug( __METHOD__
.": template inclusion denied for " . $title->getPrefixedDBkey() );
3288 list( $text, $title ) = $this->getTemplateDom( $title );
3289 if ( $text !== false ) {
3295 # If the title is valid but undisplayable, make a link to it
3296 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3297 $text = "[[:$titleText]]";
3300 } elseif ( $title->isTrans() ) {
3301 # Interwiki transclusion
3302 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3303 $text = $this->interwikiTransclude( $title, 'render' );
3306 $text = $this->interwikiTransclude( $title, 'raw' );
3307 # Preprocess it like a template
3308 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3314 # Do infinite loop check
3315 # This has to be done after redirect resolution to avoid infinite loops via redirects
3316 if ( !$frame->loopCheck( $title ) ) {
3318 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3319 wfDebug( __METHOD__
.": template loop broken at '$titleText'\n" );
3321 wfProfileOut( __METHOD__
. '-loadtpl' );
3324 # If we haven't found text to substitute by now, we're done
3325 # Recover the source wikitext and return it
3327 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3328 if ( $titleProfileIn ) {
3329 wfProfileOut( $titleProfileIn ); // template out
3331 wfProfileOut( __METHOD__
);
3332 return array( 'object' => $text );
3335 # Expand DOM-style return values in a child frame
3336 if ( $isChildObj ) {
3337 # Clean up argument array
3338 $newFrame = $frame->newChild( $args, $title );
3341 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3342 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3343 # Expansion is eligible for the empty-frame cache
3344 if ( isset( $this->mTplExpandCache
[$titleText] ) ) {
3345 $text = $this->mTplExpandCache
[$titleText];
3347 $text = $newFrame->expand( $text );
3348 $this->mTplExpandCache
[$titleText] = $text;
3351 # Uncached expansion
3352 $text = $newFrame->expand( $text );
3355 if ( $isLocalObj && $nowiki ) {
3356 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3357 $isLocalObj = false;
3360 if ( $titleProfileIn ) {
3361 wfProfileOut( $titleProfileIn ); // template out
3364 # Replace raw HTML by a placeholder
3365 # Add a blank line preceding, to prevent it from mucking up
3366 # immediately preceding headings
3368 $text = "\n\n" . $this->insertStripItem( $text );
3369 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3370 # Escape nowiki-style return values
3371 $text = wfEscapeWikiText( $text );
3372 } elseif ( is_string( $text )
3373 && !$piece['lineStart']
3374 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3376 # Bug 529: if the template begins with a table or block-level
3377 # element, it should be treated as beginning a new line.
3378 # This behaviour is somewhat controversial.
3379 $text = "\n" . $text;
3382 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3383 # Error, oversize inclusion
3384 if ( $titleText !== false ) {
3385 # Make a working, properly escaped link if possible (bug 23588)
3386 $text = "[[:$titleText]]";
3388 # This will probably not be a working link, but at least it may
3389 # provide some hint of where the problem is
3390 preg_replace( '/^:/', '', $originalTitle );
3391 $text = "[[:$originalTitle]]";
3393 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3394 $this->limitationWarn( 'post-expand-template-inclusion' );
3397 if ( $isLocalObj ) {
3398 $ret = array( 'object' => $text );
3400 $ret = array( 'text' => $text );
3403 wfProfileOut( __METHOD__
);
3408 * Get the semi-parsed DOM representation of a template with a given title,
3409 * and its redirect destination title. Cached.
3411 * @param $title Title
3415 function getTemplateDom( $title ) {
3416 $cacheTitle = $title;
3417 $titleText = $title->getPrefixedDBkey();
3419 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3420 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3421 $title = Title
::makeTitle( $ns, $dbk );
3422 $titleText = $title->getPrefixedDBkey();
3424 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3425 return array( $this->mTplDomCache
[$titleText], $title );
3428 # Cache miss, go to the database
3429 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3431 if ( $text === false ) {
3432 $this->mTplDomCache
[$titleText] = false;
3433 return array( false, $title );
3436 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3437 $this->mTplDomCache
[ $titleText ] = $dom;
3439 if ( !$title->equals( $cacheTitle ) ) {
3440 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3441 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3444 return array( $dom, $title );
3448 * Fetch the unparsed text of a template and register a reference to it.
3449 * @param Title $title
3450 * @return Array ( string or false, Title )
3452 function fetchTemplateAndTitle( $title ) {
3453 $templateCb = $this->mOptions
->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3454 $stuff = call_user_func( $templateCb, $title, $this );
3455 $text = $stuff['text'];
3456 $finalTitle = isset( $stuff['finalTitle'] ) ?
$stuff['finalTitle'] : $title;
3457 if ( isset( $stuff['deps'] ) ) {
3458 foreach ( $stuff['deps'] as $dep ) {
3459 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3462 return array( $text, $finalTitle );
3466 * Fetch the unparsed text of a template and register a reference to it.
3467 * @param Title $title
3468 * @return mixed string or false
3470 function fetchTemplate( $title ) {
3471 $rv = $this->fetchTemplateAndTitle( $title );
3476 * Static function to get a template
3477 * Can be overridden via ParserOptions::setTemplateCallback().
3479 * @parma $title Title
3480 * @param $parser Parser
3484 static function statelessFetchTemplate( $title, $parser = false ) {
3485 $text = $skip = false;
3486 $finalTitle = $title;
3489 # Loop to fetch the article, with up to 1 redirect
3490 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3491 # Give extensions a chance to select the revision instead
3492 $id = false; # Assume current
3493 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3494 array( $parser, $title, &$skip, &$id ) );
3500 'page_id' => $title->getArticleID(),
3507 ? Revision
::newFromId( $id )
3508 : Revision
::newFromTitle( $title );
3509 $rev_id = $rev ?
$rev->getId() : 0;
3510 # If there is no current revision, there is no page
3511 if ( $id === false && !$rev ) {
3512 $linkCache = LinkCache
::singleton();
3513 $linkCache->addBadLinkObj( $title );
3518 'page_id' => $title->getArticleID(),
3519 'rev_id' => $rev_id );
3520 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3521 # We fetched a rev from a different title; register it too...
3523 'title' => $rev->getTitle(),
3524 'page_id' => $rev->getPage(),
3525 'rev_id' => $rev_id );
3529 $text = $rev->getText();
3530 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3532 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3533 if ( !$message->exists() ) {
3537 $text = $message->plain();
3541 if ( $text === false ) {
3545 $finalTitle = $title;
3546 $title = Title
::newFromRedirect( $text );
3550 'finalTitle' => $finalTitle,
3555 * Fetch a file and its title and register a reference to it.
3556 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3557 * @param Title $title
3558 * @param Array $options Array of options to RepoGroup::findFile
3561 function fetchFile( $title, $options = array() ) {
3562 $res = $this->fetchFileAndTitle( $title, $options );
3567 * Fetch a file and its title and register a reference to it.
3568 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3569 * @param Title $title
3570 * @param Array $options Array of options to RepoGroup::findFile
3571 * @return Array ( File or false, Title of file )
3573 function fetchFileAndTitle( $title, $options = array() ) {
3574 if ( isset( $options['broken'] ) ) {
3575 $file = false; // broken thumbnail forced by hook
3576 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3577 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3578 } else { // get by (name,timestamp)
3579 $file = wfFindFile( $title, $options );
3581 $time = $file ?
$file->getTimestamp() : false;
3582 $sha1 = $file ?
$file->getSha1() : false;
3583 # Register the file as a dependency...
3584 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3585 if ( $file && !$title->equals( $file->getTitle() ) ) {
3586 # Update fetched file title
3587 $title = $file->getTitle();
3588 if ( is_null( $file->getRedirectedTitle() ) ) {
3589 # This file was not a redirect, but the title does not match.
3590 # Register under the new name because otherwise the link will
3592 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3595 return array( $file, $title );
3599 * Transclude an interwiki link.
3601 * @param $title Title
3606 function interwikiTransclude( $title, $action ) {
3607 global $wgEnableScaryTranscluding;
3609 if ( !$wgEnableScaryTranscluding ) {
3610 return wfMsgForContent('scarytranscludedisabled');
3613 $url = $title->getFullUrl( "action=$action" );
3615 if ( strlen( $url ) > 255 ) {
3616 return wfMsgForContent( 'scarytranscludetoolong' );
3618 return $this->fetchScaryTemplateMaybeFromCache( $url );
3622 * @param $url string
3623 * @return Mixed|String
3625 function fetchScaryTemplateMaybeFromCache( $url ) {
3626 global $wgTranscludeCacheExpiry;
3627 $dbr = wfGetDB( DB_SLAVE
);
3628 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3629 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3630 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3632 return $obj->tc_contents
;
3635 $text = Http
::get( $url );
3637 return wfMsgForContent( 'scarytranscludefailed', $url );
3640 $dbw = wfGetDB( DB_MASTER
);
3641 $dbw->replace( 'transcache', array('tc_url'), array(
3643 'tc_time' => $dbw->timestamp( time() ),
3644 'tc_contents' => $text)
3650 * Triple brace replacement -- used for template arguments
3653 * @param $peice array
3654 * @param $frame PPFrame
3658 function argSubstitution( $piece, $frame ) {
3659 wfProfileIn( __METHOD__
);
3662 $parts = $piece['parts'];
3663 $nameWithSpaces = $frame->expand( $piece['title'] );
3664 $argName = trim( $nameWithSpaces );
3666 $text = $frame->getArgument( $argName );
3667 if ( $text === false && $parts->getLength() > 0
3671 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3674 # No match in frame, use the supplied default
3675 $object = $parts->item( 0 )->getChildren();
3677 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3678 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3679 $this->limitationWarn( 'post-expand-template-argument' );
3682 if ( $text === false && $object === false ) {
3684 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3686 if ( $error !== false ) {
3689 if ( $object !== false ) {
3690 $ret = array( 'object' => $object );
3692 $ret = array( 'text' => $text );
3695 wfProfileOut( __METHOD__
);
3700 * Return the text to be used for a given extension tag.
3701 * This is the ghost of strip().
3703 * @param $params array Associative array of parameters:
3704 * name PPNode for the tag name
3705 * attr PPNode for unparsed text where tag attributes are thought to be
3706 * attributes Optional associative array of parsed attributes
3707 * inner Contents of extension element
3708 * noClose Original text did not have a close tag
3709 * @param $frame PPFrame
3713 function extensionSubstitution( $params, $frame ) {
3714 $name = $frame->expand( $params['name'] );
3715 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3716 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3717 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3719 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower($name)] ) &&
3720 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3721 if ( $isFunctionTag ) {
3722 $markerType = 'none';
3724 $markerType = 'general';
3726 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3727 $name = strtolower( $name );
3728 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3729 if ( isset( $params['attributes'] ) ) {
3730 $attributes = $attributes +
$params['attributes'];
3733 if ( isset( $this->mTagHooks
[$name] ) ) {
3734 # Workaround for PHP bug 35229 and similar
3735 if ( !is_callable( $this->mTagHooks
[$name] ) ) {
3736 throw new MWException( "Tag hook for $name is not callable\n" );
3738 $output = call_user_func_array( $this->mTagHooks
[$name],
3739 array( $content, $attributes, $this, $frame ) );
3740 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
3741 list( $callback, $flags ) = $this->mFunctionTagHooks
[$name];
3742 if ( !is_callable( $callback ) ) {
3743 throw new MWException( "Tag hook for $name is not callable\n" );
3746 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3748 $output = '<span class="error">Invalid tag extension name: ' .
3749 htmlspecialchars( $name ) . '</span>';
3752 if ( is_array( $output ) ) {
3753 # Extract flags to local scope (to override $markerType)
3755 $output = $flags[0];
3760 if ( is_null( $attrText ) ) {
3763 if ( isset( $params['attributes'] ) ) {
3764 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3765 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3766 htmlspecialchars( $attrValue ) . '"';
3769 if ( $content === null ) {
3770 $output = "<$name$attrText/>";
3772 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
3773 $output = "<$name$attrText>$content$close";
3777 if ( $markerType === 'none' ) {
3779 } elseif ( $markerType === 'nowiki' ) {
3780 $this->mStripState
->addNoWiki( $marker, $output );
3781 } elseif ( $markerType === 'general' ) {
3782 $this->mStripState
->addGeneral( $marker, $output );
3784 throw new MWException( __METHOD__
.': invalid marker type' );
3790 * Increment an include size counter
3792 * @param $type String: the type of expansion
3793 * @param $size Integer: the size of the text
3794 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3796 function incrementIncludeSize( $type, $size ) {
3797 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
3800 $this->mIncludeSizes
[$type] +
= $size;
3806 * Increment the expensive function count
3808 * @return Boolean: false if the limit has been exceeded
3810 function incrementExpensiveFunctionCount() {
3811 global $wgExpensiveParserFunctionLimit;
3812 $this->mExpensiveFunctionCount++
;
3813 if ( $this->mExpensiveFunctionCount
<= $wgExpensiveParserFunctionLimit ) {
3820 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3821 * Fills $this->mDoubleUnderscores, returns the modified text
3823 * @param $text string
3827 function doDoubleUnderscore( $text ) {
3828 wfProfileIn( __METHOD__
);
3830 # The position of __TOC__ needs to be recorded
3831 $mw = MagicWord
::get( 'toc' );
3832 if ( $mw->match( $text ) ) {
3833 $this->mShowToc
= true;
3834 $this->mForceTocPosition
= true;
3836 # Set a placeholder. At the end we'll fill it in with the TOC.
3837 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3839 # Only keep the first one.
3840 $text = $mw->replace( '', $text );
3843 # Now match and remove the rest of them
3844 $mwa = MagicWord
::getDoubleUnderscoreArray();
3845 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
3847 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
3848 $this->mOutput
->mNoGallery
= true;
3850 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
3851 $this->mShowToc
= false;
3853 if ( isset( $this->mDoubleUnderscores
['hiddencat'] ) && $this->mTitle
->getNamespace() == NS_CATEGORY
) {
3854 $this->addTrackingCategory( 'hidden-category-category' );
3856 # (bug 8068) Allow control over whether robots index a page.
3858 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3859 # is not desirable, the last one on the page should win.
3860 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
3861 $this->mOutput
->setIndexPolicy( 'noindex' );
3862 $this->addTrackingCategory( 'noindex-category' );
3864 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
3865 $this->mOutput
->setIndexPolicy( 'index' );
3866 $this->addTrackingCategory( 'index-category' );
3869 # Cache all double underscores in the database
3870 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
3871 $this->mOutput
->setProperty( $key, '' );
3874 wfProfileOut( __METHOD__
);
3879 * Add a tracking category, getting the title from a system message,
3880 * or print a debug message if the title is invalid.
3882 * @param $msg String: message key
3883 * @return Boolean: whether the addition was successful
3885 public function addTrackingCategory( $msg ) {
3886 if ( $this->mTitle
->getNamespace() === NS_SPECIAL
) {
3887 wfDebug( __METHOD__
.": Not adding tracking category $msg to special page!\n" );
3890 // Important to parse with correct title (bug 31469)
3891 $cat = wfMessage( $msg )
3892 ->title( $this->getTitle() )
3893 ->inContentLanguage()
3896 # Allow tracking categories to be disabled by setting them to "-"
3897 if ( $cat === '-' ) {
3901 $containerCategory = Title
::makeTitleSafe( NS_CATEGORY
, $cat );
3902 if ( $containerCategory ) {
3903 $this->mOutput
->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3906 wfDebug( __METHOD__
.": [[MediaWiki:$msg]] is not a valid title!\n" );
3912 * This function accomplishes several tasks:
3913 * 1) Auto-number headings if that option is enabled
3914 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3915 * 3) Add a Table of contents on the top for users who have enabled the option
3916 * 4) Auto-anchor headings
3918 * It loops through all headlines, collects the necessary data, then splits up the
3919 * string and re-inserts the newly formatted headlines.
3921 * @param $text String
3922 * @param $origText String: original, untouched wikitext
3923 * @param $isMain Boolean
3924 * @return mixed|string
3927 function formatHeadings( $text, $origText, $isMain=true ) {
3928 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3930 # Inhibit editsection links if requested in the page
3931 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
3932 $maybeShowEditLink = $showEditLink = false;
3934 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3935 $showEditLink = $this->mOptions
->getEditSection();
3937 if ( $showEditLink ) {
3938 $this->mOutput
->setEditSectionTokens( true );
3941 # Get all headlines for numbering them and adding funky stuff like [edit]
3942 # links - this is for later, but we need the number of headlines right now
3944 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3946 # if there are fewer than 4 headlines in the article, do not show TOC
3947 # unless it's been explicitly enabled.
3948 $enoughToc = $this->mShowToc
&&
3949 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
3951 # Allow user to stipulate that a page should have a "new section"
3952 # link added via __NEWSECTIONLINK__
3953 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
3954 $this->mOutput
->setNewSection( true );
3957 # Allow user to remove the "new section"
3958 # link via __NONEWSECTIONLINK__
3959 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
3960 $this->mOutput
->hideNewSection( true );
3963 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3964 # override above conditions and always show TOC above first header
3965 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
3966 $this->mShowToc
= true;
3974 # Ugh .. the TOC should have neat indentation levels which can be
3975 # passed to the skin functions. These are determined here
3979 $sublevelCount = array();
3980 $levelCount = array();
3985 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self
::MARKER_SUFFIX
;
3986 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
3987 $oldType = $this->mOutputType
;
3988 $this->setOutputType( self
::OT_WIKI
);
3989 $frame = $this->getPreprocessor()->newFrame();
3990 $root = $this->preprocessToDom( $origText );
3991 $node = $root->getFirstChild();
3996 foreach ( $matches[3] as $headline ) {
3997 $isTemplate = false;
3999 $sectionIndex = false;
4001 $markerMatches = array();
4002 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
4003 $serial = $markerMatches[1];
4004 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4005 $isTemplate = ( $titleText != $baseTitleText );
4006 $headline = preg_replace( "/^$markerRegex/", "", $headline );
4010 $prevlevel = $level;
4012 $level = $matches[1][$headlineCount];
4014 if ( $level > $prevlevel ) {
4015 # Increase TOC level
4017 $sublevelCount[$toclevel] = 0;
4018 if ( $toclevel<$wgMaxTocLevel ) {
4019 $prevtoclevel = $toclevel;
4020 $toc .= Linker
::tocIndent();
4023 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4024 # Decrease TOC level, find level to jump to
4026 for ( $i = $toclevel; $i > 0; $i-- ) {
4027 if ( $levelCount[$i] == $level ) {
4028 # Found last matching level
4031 } elseif ( $levelCount[$i] < $level ) {
4032 # Found first matching level below current level
4040 if ( $toclevel<$wgMaxTocLevel ) {
4041 if ( $prevtoclevel < $wgMaxTocLevel ) {
4042 # Unindent only if the previous toc level was shown :p
4043 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4044 $prevtoclevel = $toclevel;
4046 $toc .= Linker
::tocLineEnd();
4050 # No change in level, end TOC line
4051 if ( $toclevel<$wgMaxTocLevel ) {
4052 $toc .= Linker
::tocLineEnd();
4056 $levelCount[$toclevel] = $level;
4058 # count number of headlines for each level
4059 @$sublevelCount[$toclevel]++
;
4061 for( $i = 1; $i <= $toclevel; $i++
) {
4062 if ( !empty( $sublevelCount[$i] ) ) {
4066 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4071 # The safe header is a version of the header text safe to use for links
4073 # Remove link placeholders by the link text.
4074 # <!--LINK number-->
4076 # link text with suffix
4077 # Do this before unstrip since link text can contain strip markers
4078 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4080 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4081 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4083 # Strip out HTML (first regex removes any tag not allowed)
4084 # Allowed tags are <sup> and <sub> (bug 8393), <i> (bug 26375) and <b> (r105284)
4085 # We strip any parameter from accepted tags (second regex)
4086 $tocline = preg_replace(
4087 array( '#<(?!/?(sup|sub|i|b)(?: [^>]*)?>).*?'.'>#', '#<(/?(sup|sub|i|b))(?: .*?)?'.'>#' ),
4088 array( '', '<$1>' ),
4091 $tocline = trim( $tocline );
4093 # For the anchor, strip out HTML-y stuff period
4094 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4095 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4097 # Save headline for section edit hint before it's escaped
4098 $headlineHint = $safeHeadline;
4100 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4101 # For reverse compatibility, provide an id that's
4102 # HTML4-compatible, like we used to.
4104 # It may be worth noting, academically, that it's possible for
4105 # the legacy anchor to conflict with a non-legacy headline
4106 # anchor on the page. In this case likely the "correct" thing
4107 # would be to either drop the legacy anchors or make sure
4108 # they're numbered first. However, this would require people
4109 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4110 # manually, so let's not bother worrying about it.
4111 $legacyHeadline = Sanitizer
::escapeId( $safeHeadline,
4112 array( 'noninitial', 'legacy' ) );
4113 $safeHeadline = Sanitizer
::escapeId( $safeHeadline );
4115 if ( $legacyHeadline == $safeHeadline ) {
4116 # No reason to have both (in fact, we can't)
4117 $legacyHeadline = false;
4120 $legacyHeadline = false;
4121 $safeHeadline = Sanitizer
::escapeId( $safeHeadline,
4125 # HTML names must be case-insensitively unique (bug 10721).
4126 # This does not apply to Unicode characters per
4127 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4128 # @todo FIXME: We may be changing them depending on the current locale.
4129 $arrayKey = strtolower( $safeHeadline );
4130 if ( $legacyHeadline === false ) {
4131 $legacyArrayKey = false;
4133 $legacyArrayKey = strtolower( $legacyHeadline );
4136 # count how many in assoc. array so we can track dupes in anchors
4137 if ( isset( $refers[$arrayKey] ) ) {
4138 $refers[$arrayKey]++
;
4140 $refers[$arrayKey] = 1;
4142 if ( isset( $refers[$legacyArrayKey] ) ) {
4143 $refers[$legacyArrayKey]++
;
4145 $refers[$legacyArrayKey] = 1;
4148 # Don't number the heading if it is the only one (looks silly)
4149 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4150 # the two are different if the line contains a link
4151 $headline = $numbering . ' ' . $headline;
4154 # Create the anchor for linking from the TOC to the section
4155 $anchor = $safeHeadline;
4156 $legacyAnchor = $legacyHeadline;
4157 if ( $refers[$arrayKey] > 1 ) {
4158 $anchor .= '_' . $refers[$arrayKey];
4160 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4161 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4163 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) ||
$toclevel < $wgMaxTocLevel ) ) {
4164 $toc .= Linker
::tocLine( $anchor, $tocline,
4165 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4168 # Add the section to the section tree
4169 # Find the DOM node for this header
4170 while ( $node && !$isTemplate ) {
4171 if ( $node->getName() === 'h' ) {
4172 $bits = $node->splitHeading();
4173 if ( $bits['i'] == $sectionIndex ) {
4177 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4178 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4179 $node = $node->getNextSibling();
4182 'toclevel' => $toclevel,
4185 'number' => $numbering,
4186 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4187 'fromtitle' => $titleText,
4188 'byteoffset' => ( $isTemplate ?
null : $byteOffset ),
4189 'anchor' => $anchor,
4192 # give headline the correct <h#> tag
4193 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4194 // Output edit section links as markers with styles that can be customized by skins
4195 if ( $isTemplate ) {
4196 # Put a T flag in the section identifier, to indicate to extractSections()
4197 # that sections inside <includeonly> should be counted.
4198 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4200 $editlinkArgs = array( $this->mTitle
->getPrefixedText(), $sectionIndex, $headlineHint );
4202 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4203 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4204 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4205 // so we don't have to worry about a user trying to input one of these markers directly.
4206 // We use a page and section attribute to stop the language converter from converting these important bits
4207 // of data, but put the headline hint inside a content block because the language converter is supposed to
4208 // be able to convert that piece of data.
4209 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4210 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4211 if ( isset($editlinkArgs[2]) ) {
4212 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4219 $head[$headlineCount] = Linker
::makeHeadline( $level,
4220 $matches['attrib'][$headlineCount], $anchor, $headline,
4221 $editlink, $legacyAnchor );
4226 $this->setOutputType( $oldType );
4228 # Never ever show TOC if no headers
4229 if ( $numVisible < 1 ) {
4234 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4235 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4237 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4238 $this->mOutput
->setTOCHTML( $toc );
4242 $this->mOutput
->setSections( $tocraw );
4245 # split up and insert constructed headlines
4246 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4249 // build an array of document sections
4250 $sections = array();
4251 foreach ( $blocks as $block ) {
4252 // $head is zero-based, sections aren't.
4253 if ( empty( $head[$i - 1] ) ) {
4254 $sections[$i] = $block;
4256 $sections[$i] = $head[$i - 1] . $block;
4260 * Send a hook, one per section.
4261 * The idea here is to be able to make section-level DIVs, but to do so in a
4262 * lower-impact, more correct way than r50769
4265 * $section : the section number
4266 * &$sectionContent : ref to the content of the section
4267 * $showEditLinks : boolean describing whether this section has an edit link
4269 wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4274 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4275 // append the TOC at the beginning
4276 // Top anchor now in skin
4277 $sections[0] = $sections[0] . $toc . "\n";
4280 $full .= join( '', $sections );
4282 if ( $this->mForceTocPosition
) {
4283 return str_replace( '<!--MWTOC-->', $toc, $full );
4290 * Transform wiki markup when saving a page by doing \r\n -> \n
4291 * conversion, substitting signatures, {{subst:}} templates, etc.
4293 * @param $text String: the text to transform
4294 * @param $title Title: the Title object for the current article
4295 * @param $user User: the User object describing the current user
4296 * @param $options ParserOptions: parsing options
4297 * @param $clearState Boolean: whether to clear the parser state first
4298 * @return String: the altered wiki markup
4300 public function preSaveTransform( $text, Title
$title, User
$user, ParserOptions
$options, $clearState = true ) {
4301 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4302 $this->setUser( $user );
4307 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4308 if( $options->getPreSaveTransform() ) {
4309 $text = $this->pstPass2( $text, $user );
4311 $text = $this->mStripState
->unstripBoth( $text );
4313 $this->setUser( null ); #Reset
4319 * Pre-save transform helper function
4322 * @param $text string
4327 function pstPass2( $text, $user ) {
4328 global $wgContLang, $wgLocaltimezone;
4330 # Note: This is the timestamp saved as hardcoded wikitext to
4331 # the database, we use $wgContLang here in order to give
4332 # everyone the same signature and use the default one rather
4333 # than the one selected in each user's preferences.
4334 # (see also bug 12815)
4335 $ts = $this->mOptions
->getTimestamp();
4336 if ( isset( $wgLocaltimezone ) ) {
4337 $tz = $wgLocaltimezone;
4339 $tz = date_default_timezone_get();
4342 $unixts = wfTimestamp( TS_UNIX
, $ts );
4343 $oldtz = date_default_timezone_get();
4344 date_default_timezone_set( $tz );
4345 $ts = date( 'YmdHis', $unixts );
4346 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4348 # Allow translation of timezones through wiki. date() can return
4349 # whatever crap the system uses, localised or not, so we cannot
4350 # ship premade translations.
4351 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4352 $msg = wfMessage( $key )->inContentLanguage();
4353 if ( $msg->exists() ) {
4354 $tzMsg = $msg->text();
4357 date_default_timezone_set( $oldtz );
4359 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4361 # Variable replacement
4362 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4363 $text = $this->replaceVariables( $text );
4365 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4366 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4369 $sigText = $this->getUserSig( $user );
4370 $text = strtr( $text, array(
4372 '~~~~' => "$sigText $d",
4376 # Context links: [[|name]] and [[name (context)|]]
4377 global $wgLegalTitleChars;
4378 $tc = "[$wgLegalTitleChars]";
4379 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4381 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4382 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/"; # [[ns:page(context)|]]
4383 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4384 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4386 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4387 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4388 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4389 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4391 $t = $this->mTitle
->getText();
4393 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4394 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4395 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4396 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4398 # if there's no context, don't bother duplicating the title
4399 $text = preg_replace( $p2, '[[\\1]]', $text );
4402 # Trim trailing whitespace
4403 $text = rtrim( $text );
4409 * Fetch the user's signature text, if any, and normalize to
4410 * validated, ready-to-insert wikitext.
4411 * If you have pre-fetched the nickname or the fancySig option, you can
4412 * specify them here to save a database query.
4413 * Do not reuse this parser instance after calling getUserSig(),
4414 * as it may have changed if it's the $wgParser.
4417 * @param $nickname String|bool nickname to use or false to use user's default nickname
4418 * @param $fancySig Boolean|null whether the nicknname is the complete signature
4419 * or null to use default value
4422 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4423 global $wgMaxSigChars;
4425 $username = $user->getName();
4427 # If not given, retrieve from the user object.
4428 if ( $nickname === false )
4429 $nickname = $user->getOption( 'nickname' );
4431 if ( is_null( $fancySig ) ) {
4432 $fancySig = $user->getBoolOption( 'fancysig' );
4435 $nickname = $nickname == null ?
$username : $nickname;
4437 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4438 $nickname = $username;
4439 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4440 } elseif ( $fancySig !== false ) {
4441 # Sig. might contain markup; validate this
4442 if ( $this->validateSig( $nickname ) !== false ) {
4443 # Validated; clean up (if needed) and return it
4444 return $this->cleanSig( $nickname, true );
4446 # Failed to validate; fall back to the default
4447 $nickname = $username;
4448 wfDebug( __METHOD__
.": $username has bad XML tags in signature.\n" );
4452 # Make sure nickname doesnt get a sig in a sig
4453 $nickname = self
::cleanSigInSig( $nickname );
4455 # If we're still here, make it a link to the user page
4456 $userText = wfEscapeWikiText( $username );
4457 $nickText = wfEscapeWikiText( $nickname );
4458 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4460 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4464 * Check that the user's signature contains no bad XML
4466 * @param $text String
4467 * @return mixed An expanded string, or false if invalid.
4469 function validateSig( $text ) {
4470 return( Xml
::isWellFormedXmlFragment( $text ) ?
$text : false );
4474 * Clean up signature text
4476 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4477 * 2) Substitute all transclusions
4479 * @param $text String
4480 * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4481 * @return String: signature text
4483 public function cleanSig( $text, $parsing = false ) {
4486 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4489 # Option to disable this feature
4490 if ( !$this->mOptions
->getCleanSignatures() ) {
4494 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4495 # => Move this logic to braceSubstitution()
4496 $substWord = MagicWord
::get( 'subst' );
4497 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4498 $substText = '{{' . $substWord->getSynonym( 0 );
4500 $text = preg_replace( $substRegex, $substText, $text );
4501 $text = self
::cleanSigInSig( $text );
4502 $dom = $this->preprocessToDom( $text );
4503 $frame = $this->getPreprocessor()->newFrame();
4504 $text = $frame->expand( $dom );
4507 $text = $this->mStripState
->unstripBoth( $text );
4514 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4516 * @param $text String
4517 * @return String: signature text with /~{3,5}/ removed
4519 public static function cleanSigInSig( $text ) {
4520 $text = preg_replace( '/~{3,5}/', '', $text );
4525 * Set up some variables which are usually set up in parse()
4526 * so that an external function can call some class members with confidence
4528 * @param $title Title|null
4529 * @param $options ParserOptions
4530 * @param $outputType
4531 * @param $clearState bool
4533 public function startExternalParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4534 $this->startParse( $title, $options, $outputType, $clearState );
4538 * @param $title Title|null
4539 * @param $options ParserOptions
4540 * @param $outputType
4541 * @param $clearState bool
4543 private function startParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4544 $this->setTitle( $title );
4545 $this->mOptions
= $options;
4546 $this->setOutputType( $outputType );
4547 if ( $clearState ) {
4548 $this->clearState();
4553 * Wrapper for preprocess()
4555 * @param $text String: the text to preprocess
4556 * @param $options ParserOptions: options
4557 * @param $title Title object or null to use $wgTitle
4560 public function transformMsg( $text, $options, $title = null ) {
4561 static $executing = false;
4563 # Guard against infinite recursion
4569 wfProfileIn( __METHOD__
);
4575 # It's not uncommon having a null $wgTitle in scripts. See r80898
4576 # Create a ghost title in such case
4577 $title = Title
::newFromText( 'Dwimmerlaik' );
4579 $text = $this->preprocess( $text, $title, $options );
4582 wfProfileOut( __METHOD__
);
4587 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4588 * The callback should have the following form:
4589 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4591 * Transform and return $text. Use $parser for any required context, e.g. use
4592 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4594 * Hooks may return extended information by returning an array, of which the
4595 * first numbered element (index 0) must be the return string, and all other
4596 * entries are extracted into local variables within an internal function
4597 * in the Parser class.
4599 * This interface (introduced r61913) appears to be undocumented, but
4600 * 'markerName' is used by some core tag hooks to override which strip
4601 * array their results are placed in. **Use great caution if attempting
4602 * this interface, as it is not documented and injudicious use could smash
4603 * private variables.**
4605 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4606 * @param $callback Mixed: the callback function (and object) to use for the tag
4607 * @return Mixed|null The old value of the mTagHooks array associated with the hook
4609 public function setHook( $tag, $callback ) {
4610 $tag = strtolower( $tag );
4611 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4612 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4614 $oldVal = isset( $this->mTagHooks
[$tag] ) ?
$this->mTagHooks
[$tag] : null;
4615 $this->mTagHooks
[$tag] = $callback;
4616 if ( !in_array( $tag, $this->mStripList
) ) {
4617 $this->mStripList
[] = $tag;
4624 * As setHook(), but letting the contents be parsed.
4626 * Transparent tag hooks are like regular XML-style tag hooks, except they
4627 * operate late in the transformation sequence, on HTML instead of wikitext.
4629 * This is probably obsoleted by things dealing with parser frames?
4630 * The only extension currently using it is geoserver.
4633 * @todo better document or deprecate this
4635 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4636 * @param $callback Mixed: the callback function (and object) to use for the tag
4637 * @return Mixed|null The old value of the mTagHooks array associated with the hook
4639 function setTransparentTagHook( $tag, $callback ) {
4640 $tag = strtolower( $tag );
4641 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4642 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4644 $oldVal = isset( $this->mTransparentTagHooks
[$tag] ) ?
$this->mTransparentTagHooks
[$tag] : null;
4645 $this->mTransparentTagHooks
[$tag] = $callback;
4651 * Remove all tag hooks
4653 function clearTagHooks() {
4654 $this->mTagHooks
= array();
4655 $this->mFunctionTagHooks
= array();
4656 $this->mStripList
= $this->mDefaultStripList
;
4660 * Create a function, e.g. {{sum:1|2|3}}
4661 * The callback function should have the form:
4662 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4664 * Or with SFH_OBJECT_ARGS:
4665 * function myParserFunction( $parser, $frame, $args ) { ... }
4667 * The callback may either return the text result of the function, or an array with the text
4668 * in element 0, and a number of flags in the other elements. The names of the flags are
4669 * specified in the keys. Valid flags are:
4670 * found The text returned is valid, stop processing the template. This
4672 * nowiki Wiki markup in the return value should be escaped
4673 * isHTML The returned text is HTML, armour it against wikitext transformation
4675 * @param $id String: The magic word ID
4676 * @param $callback Mixed: the callback function (and object) to use
4677 * @param $flags Integer: a combination of the following flags:
4678 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4680 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4681 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4682 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4683 * the arguments, and to control the way they are expanded.
4685 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4686 * arguments, for instance:
4687 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4689 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4690 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4691 * working if/when this is changed.
4693 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4696 * Please read the documentation in includes/parser/Preprocessor.php for more information
4697 * about the methods available in PPFrame and PPNode.
4699 * @return string|callback The old callback function for this name, if any
4701 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4704 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4705 $this->mFunctionHooks
[$id] = array( $callback, $flags );
4707 # Add to function cache
4708 $mw = MagicWord
::get( $id );
4710 throw new MWException( __METHOD__
.'() expecting a magic word identifier.' );
4712 $synonyms = $mw->getSynonyms();
4713 $sensitive = intval( $mw->isCaseSensitive() );
4715 foreach ( $synonyms as $syn ) {
4717 if ( !$sensitive ) {
4718 $syn = $wgContLang->lc( $syn );
4721 if ( !( $flags & SFH_NO_HASH
) ) {
4724 # Remove trailing colon
4725 if ( substr( $syn, -1, 1 ) === ':' ) {
4726 $syn = substr( $syn, 0, -1 );
4728 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4734 * Get all registered function hook identifiers
4738 function getFunctionHooks() {
4739 return array_keys( $this->mFunctionHooks
);
4743 * Create a tag function, e.g. <test>some stuff</test>.
4744 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4745 * Unlike parser functions, their content is not preprocessed.
4748 function setFunctionTagHook( $tag, $callback, $flags ) {
4749 $tag = strtolower( $tag );
4750 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4751 $old = isset( $this->mFunctionTagHooks
[$tag] ) ?
4752 $this->mFunctionTagHooks
[$tag] : null;
4753 $this->mFunctionTagHooks
[$tag] = array( $callback, $flags );
4755 if ( !in_array( $tag, $this->mStripList
) ) {
4756 $this->mStripList
[] = $tag;
4763 * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4764 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4765 * Placeholders created in Skin::makeLinkObj()
4767 * @param $text string
4768 * @param $options int
4770 * @return array of link CSS classes, indexed by PDBK.
4772 function replaceLinkHolders( &$text, $options = 0 ) {
4773 return $this->mLinkHolders
->replace( $text );
4777 * Replace <!--LINK--> link placeholders with plain text of links
4778 * (not HTML-formatted).
4780 * @param $text String
4783 function replaceLinkHoldersText( $text ) {
4784 return $this->mLinkHolders
->replaceText( $text );
4788 * Renders an image gallery from a text with one line per image.
4789 * text labels may be given by using |-style alternative text. E.g.
4790 * Image:one.jpg|The number "1"
4791 * Image:tree.jpg|A tree
4792 * given as text will return the HTML of a gallery with two images,
4793 * labeled 'The number "1"' and
4796 * @param string $text
4797 * @param array $params
4798 * @return string HTML
4800 function renderImageGallery( $text, $params ) {
4801 $ig = new ImageGallery();
4802 $ig->setContextTitle( $this->mTitle
);
4803 $ig->setShowBytes( false );
4804 $ig->setShowFilename( false );
4805 $ig->setParser( $this );
4806 $ig->setHideBadImages();
4807 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'table' ) );
4809 if ( isset( $params['showfilename'] ) ) {
4810 $ig->setShowFilename( true );
4812 $ig->setShowFilename( false );
4814 if ( isset( $params['caption'] ) ) {
4815 $caption = $params['caption'];
4816 $caption = htmlspecialchars( $caption );
4817 $caption = $this->replaceInternalLinks( $caption );
4818 $ig->setCaptionHtml( $caption );
4820 if ( isset( $params['perrow'] ) ) {
4821 $ig->setPerRow( $params['perrow'] );
4823 if ( isset( $params['widths'] ) ) {
4824 $ig->setWidths( $params['widths'] );
4826 if ( isset( $params['heights'] ) ) {
4827 $ig->setHeights( $params['heights'] );
4830 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4832 $lines = StringUtils
::explode( "\n", $text );
4833 foreach ( $lines as $line ) {
4834 # match lines like these:
4835 # Image:someimage.jpg|This is some image
4837 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4839 if ( count( $matches ) == 0 ) {
4843 if ( strpos( $matches[0], '%' ) !== false ) {
4844 $matches[1] = rawurldecode( $matches[1] );
4846 $title = Title
::newFromText( $matches[1], NS_FILE
);
4847 if ( is_null( $title ) ) {
4848 # Bogus title. Ignore these so we don't bomb out later.
4854 if ( isset( $matches[3] ) ) {
4855 // look for an |alt= definition while trying not to break existing
4856 // captions with multiple pipes (|) in it, until a more sensible grammar
4857 // is defined for images in galleries
4859 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4860 $altmatches = StringUtils
::explode('|', $matches[3]);
4861 $magicWordAlt = MagicWord
::get( 'img_alt' );
4863 foreach ( $altmatches as $altmatch ) {
4864 $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4866 $alt = $this->stripAltText( $match, false );
4869 // concatenate all other pipes
4870 $label .= '|' . $altmatch;
4873 // remove the first pipe
4874 $label = substr( $label, 1 );
4877 $ig->add( $title, $label, $alt );
4879 return $ig->toHTML();
4886 function getImageParams( $handler ) {
4888 $handlerClass = get_class( $handler );
4892 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
4893 # Initialise static lists
4894 static $internalParamNames = array(
4895 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4896 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4897 'bottom', 'text-bottom' ),
4898 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4899 'upright', 'border', 'link', 'alt' ),
4901 static $internalParamMap;
4902 if ( !$internalParamMap ) {
4903 $internalParamMap = array();
4904 foreach ( $internalParamNames as $type => $names ) {
4905 foreach ( $names as $name ) {
4906 $magicName = str_replace( '-', '_', "img_$name" );
4907 $internalParamMap[$magicName] = array( $type, $name );
4912 # Add handler params
4913 $paramMap = $internalParamMap;
4915 $handlerParamMap = $handler->getParamMap();
4916 foreach ( $handlerParamMap as $magic => $paramName ) {
4917 $paramMap[$magic] = array( 'handler', $paramName );
4920 $this->mImageParams
[$handlerClass] = $paramMap;
4921 $this->mImageParamsMagicArray
[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4923 return array( $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] );
4927 * Parse image options text and use it to make an image
4929 * @param $title Title
4930 * @param $options String
4931 * @param $holders LinkHolderArray|bool
4932 * @return string HTML
4934 function makeImage( $title, $options, $holders = false ) {
4935 # Check if the options text is of the form "options|alt text"
4937 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4938 # * left no resizing, just left align. label is used for alt= only
4939 # * right same, but right aligned
4940 # * none same, but not aligned
4941 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4942 # * center center the image
4943 # * frame Keep original image size, no magnify-button.
4944 # * framed Same as "frame"
4945 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4946 # * upright reduce width for upright images, rounded to full __0 px
4947 # * border draw a 1px border around the image
4948 # * alt Text for HTML alt attribute (defaults to empty)
4949 # * link Set the target of the image link. Can be external, interwiki, or local
4950 # vertical-align values (no % or length right now):
4960 $parts = StringUtils
::explode( "|", $options );
4962 # Give extensions a chance to select the file revision for us
4965 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4966 array( $this, $title, &$options, &$descQuery ) );
4967 # Fetch and register the file (file title may be different via hooks)
4968 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
4971 $handler = $file ?
$file->getHandler() : false;
4973 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4976 $this->addTrackingCategory( 'broken-file-category' );
4979 # Process the input parameters
4981 $params = array( 'frame' => array(), 'handler' => array(),
4982 'horizAlign' => array(), 'vertAlign' => array() );
4983 foreach ( $parts as $part ) {
4984 $part = trim( $part );
4985 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4987 if ( isset( $paramMap[$magicName] ) ) {
4988 list( $type, $paramName ) = $paramMap[$magicName];
4990 # Special case; width and height come in one variable together
4991 if ( $type === 'handler' && $paramName === 'width' ) {
4993 # (bug 13500) In both cases (width/height and width only),
4994 # permit trailing "px" for backward compatibility.
4995 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4996 $width = intval( $m[1] );
4997 $height = intval( $m[2] );
4998 if ( $handler->validateParam( 'width', $width ) ) {
4999 $params[$type]['width'] = $width;
5002 if ( $handler->validateParam( 'height', $height ) ) {
5003 $params[$type]['height'] = $height;
5006 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5007 $width = intval( $value );
5008 if ( $handler->validateParam( 'width', $width ) ) {
5009 $params[$type]['width'] = $width;
5012 } # else no validation -- bug 13436
5014 if ( $type === 'handler' ) {
5015 # Validate handler parameter
5016 $validated = $handler->validateParam( $paramName, $value );
5018 # Validate internal parameters
5019 switch( $paramName ) {
5022 # @todo FIXME: Possibly check validity here for
5023 # manualthumb? downstream behavior seems odd with
5024 # missing manual thumbs.
5026 $value = $this->stripAltText( $value, $holders );
5029 $chars = self
::EXT_LINK_URL_CLASS
;
5030 $prots = $this->mUrlProtocols
;
5031 if ( $value === '' ) {
5032 $paramName = 'no-link';
5035 } elseif ( preg_match( "/^$prots/", $value ) ) {
5036 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
5037 $paramName = 'link-url';
5038 $this->mOutput
->addExternalLink( $value );
5039 if ( $this->mOptions
->getExternalLinkTarget() ) {
5040 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5045 $linkTitle = Title
::newFromText( $value );
5047 $paramName = 'link-title';
5048 $value = $linkTitle;
5049 $this->mOutput
->addLink( $linkTitle );
5055 # Most other things appear to be empty or numeric...
5056 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5061 $params[$type][$paramName] = $value;
5065 if ( !$validated ) {
5070 # Process alignment parameters
5071 if ( $params['horizAlign'] ) {
5072 $params['frame']['align'] = key( $params['horizAlign'] );
5074 if ( $params['vertAlign'] ) {
5075 $params['frame']['valign'] = key( $params['vertAlign'] );
5078 $params['frame']['caption'] = $caption;
5080 # Will the image be presented in a frame, with the caption below?
5081 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5082 isset( $params['frame']['framed'] ) ||
5083 isset( $params['frame']['thumbnail'] ) ||
5084 isset( $params['frame']['manualthumb'] );
5086 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5087 # came to also set the caption, ordinary text after the image -- which
5088 # makes no sense, because that just repeats the text multiple times in
5089 # screen readers. It *also* came to set the title attribute.
5091 # Now that we have an alt attribute, we should not set the alt text to
5092 # equal the caption: that's worse than useless, it just repeats the
5093 # text. This is the framed/thumbnail case. If there's no caption, we
5094 # use the unnamed parameter for alt text as well, just for the time be-
5095 # ing, if the unnamed param is set and the alt param is not.
5097 # For the future, we need to figure out if we want to tweak this more,
5098 # e.g., introducing a title= parameter for the title; ignoring the un-
5099 # named parameter entirely for images without a caption; adding an ex-
5100 # plicit caption= parameter and preserving the old magic unnamed para-
5102 if ( $imageIsFramed ) { # Framed image
5103 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5104 # No caption or alt text, add the filename as the alt text so
5105 # that screen readers at least get some description of the image
5106 $params['frame']['alt'] = $title->getText();
5108 # Do not set $params['frame']['title'] because tooltips don't make sense
5110 } else { # Inline image
5111 if ( !isset( $params['frame']['alt'] ) ) {
5112 # No alt text, use the "caption" for the alt text
5113 if ( $caption !== '') {
5114 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5116 # No caption, fall back to using the filename for the
5118 $params['frame']['alt'] = $title->getText();
5121 # Use the "caption" for the tooltip text
5122 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5125 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5127 # Linker does the rest
5128 $time = isset( $options['time'] ) ?
$options['time'] : false;
5129 $ret = Linker
::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5130 $time, $descQuery, $this->mOptions
->getThumbSize() );
5132 # Give the handler a chance to modify the parser object
5134 $handler->parserTransformHook( $this, $file );
5142 * @param $holders LinkHolderArray
5143 * @return mixed|String
5145 protected function stripAltText( $caption, $holders ) {
5146 # Strip bad stuff out of the title (tooltip). We can't just use
5147 # replaceLinkHoldersText() here, because if this function is called
5148 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5150 $tooltip = $holders->replaceText( $caption );
5152 $tooltip = $this->replaceLinkHoldersText( $caption );
5155 # make sure there are no placeholders in thumbnail attributes
5156 # that are later expanded to html- so expand them now and
5158 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5159 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5165 * Set a flag in the output object indicating that the content is dynamic and
5166 * shouldn't be cached.
5168 function disableCache() {
5169 wfDebug( "Parser output marked as uncacheable.\n" );
5170 if ( !$this->mOutput
) {
5171 throw new MWException( __METHOD__
.
5172 " can only be called when actually parsing something" );
5174 $this->mOutput
->setCacheTime( -1 ); // old style, for compatibility
5175 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5179 * Callback from the Sanitizer for expanding items found in HTML attribute
5180 * values, so they can be safely tested and escaped.
5182 * @param $text String
5183 * @param $frame PPFrame
5186 function attributeStripCallback( &$text, $frame = false ) {
5187 $text = $this->replaceVariables( $text, $frame );
5188 $text = $this->mStripState
->unstripBoth( $text );
5197 function getTags() {
5198 return array_merge( array_keys( $this->mTransparentTagHooks
), array_keys( $this->mTagHooks
), array_keys( $this->mFunctionTagHooks
) );
5202 * Replace transparent tags in $text with the values given by the callbacks.
5204 * Transparent tag hooks are like regular XML-style tag hooks, except they
5205 * operate late in the transformation sequence, on HTML instead of wikitext.
5207 * @param $text string
5211 function replaceTransparentTags( $text ) {
5213 $elements = array_keys( $this->mTransparentTagHooks
);
5214 $text = self
::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix
);
5215 $replacements = array();
5217 foreach ( $matches as $marker => $data ) {
5218 list( $element, $content, $params, $tag ) = $data;
5219 $tagName = strtolower( $element );
5220 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5221 $output = call_user_func_array( $this->mTransparentTagHooks
[$tagName], array( $content, $params, $this ) );
5225 $replacements[$marker] = $output;
5227 return strtr( $text, $replacements );
5231 * Break wikitext input into sections, and either pull or replace
5232 * some particular section's text.
5234 * External callers should use the getSection and replaceSection methods.
5236 * @param $text String: Page wikitext
5237 * @param $section String: a section identifier string of the form:
5238 * <flag1> - <flag2> - ... - <section number>
5240 * Currently the only recognised flag is "T", which means the target section number
5241 * was derived during a template inclusion parse, in other words this is a template
5242 * section edit link. If no flags are given, it was an ordinary section edit link.
5243 * This flag is required to avoid a section numbering mismatch when a section is
5244 * enclosed by <includeonly> (bug 6563).
5246 * The section number 0 pulls the text before the first heading; other numbers will
5247 * pull the given section along with its lower-level subsections. If the section is
5248 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5250 * Section 0 is always considered to exist, even if it only contains the empty
5251 * string. If $text is the empty string and section 0 is replaced, $newText is
5254 * @param $mode String: one of "get" or "replace"
5255 * @param $newText String: replacement text for section data.
5256 * @return String: for "get", the extracted section text.
5257 * for "replace", the whole page with the section replaced.
5259 private function extractSections( $text, $section, $mode, $newText='' ) {
5260 global $wgTitle; # not generally used but removes an ugly failure mode
5261 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5263 $frame = $this->getPreprocessor()->newFrame();
5265 # Process section extraction flags
5267 $sectionParts = explode( '-', $section );
5268 $sectionIndex = array_pop( $sectionParts );
5269 foreach ( $sectionParts as $part ) {
5270 if ( $part === 'T' ) {
5271 $flags |
= self
::PTD_FOR_INCLUSION
;
5275 # Check for empty input
5276 if ( strval( $text ) === '' ) {
5277 # Only sections 0 and T-0 exist in an empty document
5278 if ( $sectionIndex == 0 ) {
5279 if ( $mode === 'get' ) {
5285 if ( $mode === 'get' ) {
5293 # Preprocess the text
5294 $root = $this->preprocessToDom( $text, $flags );
5296 # <h> nodes indicate section breaks
5297 # They can only occur at the top level, so we can find them by iterating the root's children
5298 $node = $root->getFirstChild();
5300 # Find the target section
5301 if ( $sectionIndex == 0 ) {
5302 # Section zero doesn't nest, level=big
5303 $targetLevel = 1000;
5306 if ( $node->getName() === 'h' ) {
5307 $bits = $node->splitHeading();
5308 if ( $bits['i'] == $sectionIndex ) {
5309 $targetLevel = $bits['level'];
5313 if ( $mode === 'replace' ) {
5314 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5316 $node = $node->getNextSibling();
5322 if ( $mode === 'get' ) {
5329 # Find the end of the section, including nested sections
5331 if ( $node->getName() === 'h' ) {
5332 $bits = $node->splitHeading();
5333 $curLevel = $bits['level'];
5334 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5338 if ( $mode === 'get' ) {
5339 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5341 $node = $node->getNextSibling();
5344 # Write out the remainder (in replace mode only)
5345 if ( $mode === 'replace' ) {
5346 # Output the replacement text
5347 # Add two newlines on -- trailing whitespace in $newText is conventionally
5348 # stripped by the editor, so we need both newlines to restore the paragraph gap
5349 # Only add trailing whitespace if there is newText
5350 if ( $newText != "" ) {
5351 $outText .= $newText . "\n\n";
5355 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5356 $node = $node->getNextSibling();
5360 if ( is_string( $outText ) ) {
5361 # Re-insert stripped tags
5362 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5369 * This function returns the text of a section, specified by a number ($section).
5370 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5371 * the first section before any such heading (section 0).
5373 * If a section contains subsections, these are also returned.
5375 * @param $text String: text to look in
5376 * @param $section String: section identifier
5377 * @param $deftext String: default to return if section is not found
5378 * @return string text of the requested section
5380 public function getSection( $text, $section, $deftext='' ) {
5381 return $this->extractSections( $text, $section, "get", $deftext );
5385 * This function returns $oldtext after the content of the section
5386 * specified by $section has been replaced with $text. If the target
5387 * section does not exist, $oldtext is returned unchanged.
5389 * @param $oldtext String: former text of the article
5390 * @param $section int section identifier
5391 * @param $text String: replacing text
5392 * @return String: modified text
5394 public function replaceSection( $oldtext, $section, $text ) {
5395 return $this->extractSections( $oldtext, $section, "replace", $text );
5399 * Get the ID of the revision we are parsing
5401 * @return Mixed: integer or null
5403 function getRevisionId() {
5404 return $this->mRevisionId
;
5408 * Get the revision object for $this->mRevisionId
5410 * @return Revision|null either a Revision object or null
5412 protected function getRevisionObject() {
5413 if ( !is_null( $this->mRevisionObject
) ) {
5414 return $this->mRevisionObject
;
5416 if ( is_null( $this->mRevisionId
) ) {
5420 $this->mRevisionObject
= Revision
::newFromId( $this->mRevisionId
);
5421 return $this->mRevisionObject
;
5425 * Get the timestamp associated with the current revision, adjusted for
5426 * the default server-local timestamp
5428 function getRevisionTimestamp() {
5429 if ( is_null( $this->mRevisionTimestamp
) ) {
5430 wfProfileIn( __METHOD__
);
5434 $revObject = $this->getRevisionObject();
5435 $timestamp = $revObject ?
$revObject->getTimestamp() : wfTimestampNow();
5437 # The cryptic '' timezone parameter tells to use the site-default
5438 # timezone offset instead of the user settings.
5440 # Since this value will be saved into the parser cache, served
5441 # to other users, and potentially even used inside links and such,
5442 # it needs to be consistent for all visitors.
5443 $this->mRevisionTimestamp
= $wgContLang->userAdjust( $timestamp, '' );
5445 wfProfileOut( __METHOD__
);
5447 return $this->mRevisionTimestamp
;
5451 * Get the name of the user that edited the last revision
5453 * @return String: user name
5455 function getRevisionUser() {
5456 if( is_null( $this->mRevisionUser
) ) {
5457 $revObject = $this->getRevisionObject();
5459 # if this template is subst: the revision id will be blank,
5460 # so just use the current user's name
5462 $this->mRevisionUser
= $revObject->getUserText();
5463 } elseif( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5464 $this->mRevisionUser
= $this->getUser()->getName();
5467 return $this->mRevisionUser
;
5471 * Mutator for $mDefaultSort
5473 * @param $sort string New value
5475 public function setDefaultSort( $sort ) {
5476 $this->mDefaultSort
= $sort;
5477 $this->mOutput
->setProperty( 'defaultsort', $sort );
5481 * Accessor for $mDefaultSort
5482 * Will use the empty string if none is set.
5484 * This value is treated as a prefix, so the
5485 * empty string is equivalent to sorting by
5490 public function getDefaultSort() {
5491 if ( $this->mDefaultSort
!== false ) {
5492 return $this->mDefaultSort
;
5499 * Accessor for $mDefaultSort
5500 * Unlike getDefaultSort(), will return false if none is set
5502 * @return string or false
5504 public function getCustomDefaultSort() {
5505 return $this->mDefaultSort
;
5509 * Try to guess the section anchor name based on a wikitext fragment
5510 * presumably extracted from a heading, for example "Header" from
5513 * @param $text string
5517 public function guessSectionNameFromWikiText( $text ) {
5518 # Strip out wikitext links(they break the anchor)
5519 $text = $this->stripSectionName( $text );
5520 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5521 return '#' . Sanitizer
::escapeId( $text, 'noninitial' );
5525 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5526 * instead. For use in redirects, since IE6 interprets Redirect: headers
5527 * as something other than UTF-8 (apparently?), resulting in breakage.
5529 * @param $text String: The section name
5530 * @return string An anchor
5532 public function guessLegacySectionNameFromWikiText( $text ) {
5533 # Strip out wikitext links(they break the anchor)
5534 $text = $this->stripSectionName( $text );
5535 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5536 return '#' . Sanitizer
::escapeId( $text, array( 'noninitial', 'legacy' ) );
5540 * Strips a text string of wikitext for use in a section anchor
5542 * Accepts a text string and then removes all wikitext from the
5543 * string and leaves only the resultant text (i.e. the result of
5544 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5545 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5546 * to create valid section anchors by mimicing the output of the
5547 * parser when headings are parsed.
5549 * @param $text String: text string to be stripped of wikitext
5550 * for use in a Section anchor
5551 * @return string Filtered text string
5553 public function stripSectionName( $text ) {
5554 # Strip internal link markup
5555 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5556 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5558 # Strip external link markup
5559 # @todo FIXME: Not tolerant to blank link text
5560 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5561 # on how many empty links there are on the page - need to figure that out.
5562 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5564 # Parse wikitext quotes (italics & bold)
5565 $text = $this->doQuotes( $text );
5568 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
5573 * strip/replaceVariables/unstrip for preprocessor regression testing
5575 * @param $text string
5576 * @param $title Title
5577 * @param $options ParserOptions
5578 * @param $outputType int
5582 function testSrvus( $text, Title
$title, ParserOptions
$options, $outputType = self
::OT_HTML
) {
5583 $this->startParse( $title, $options, $outputType, true );
5585 $text = $this->replaceVariables( $text );
5586 $text = $this->mStripState
->unstripBoth( $text );
5587 $text = Sanitizer
::removeHTMLtags( $text );
5592 * @param $text string
5593 * @param $title Title
5594 * @param $options ParserOptions
5597 function testPst( $text, Title
$title, ParserOptions
$options ) {
5598 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5603 * @param $title Title
5604 * @param $options ParserOptions
5607 function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
5608 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
5612 * Call a callback function on all regions of the given text that are not
5613 * inside strip markers, and replace those regions with the return value
5614 * of the callback. For example, with input:
5618 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5619 * two strings will be replaced with the value returned by the callback in
5627 function markerSkipCallback( $s, $callback ) {
5630 while ( $i < strlen( $s ) ) {
5631 $markerStart = strpos( $s, $this->mUniqPrefix
, $i );
5632 if ( $markerStart === false ) {
5633 $out .= call_user_func( $callback, substr( $s, $i ) );
5636 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5637 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
5638 if ( $markerEnd === false ) {
5639 $out .= substr( $s, $markerStart );
5642 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
5643 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5652 * Remove any strip markers found in the given text.
5654 * @param $text Input string
5657 function killMarkers( $text ) {
5658 return $this->mStripState
->killMarkers( $text );
5662 * Save the parser state required to convert the given half-parsed text to
5663 * HTML. "Half-parsed" in this context means the output of
5664 * recursiveTagParse() or internalParse(). This output has strip markers
5665 * from replaceVariables (extensionSubstitution() etc.), and link
5666 * placeholders from replaceLinkHolders().
5668 * Returns an array which can be serialized and stored persistently. This
5669 * array can later be loaded into another parser instance with
5670 * unserializeHalfParsedText(). The text can then be safely incorporated into
5671 * the return value of a parser hook.
5673 * @param $text string
5677 function serializeHalfParsedText( $text ) {
5678 wfProfileIn( __METHOD__
);
5681 'version' => self
::HALF_PARSED_VERSION
,
5682 'stripState' => $this->mStripState
->getSubState( $text ),
5683 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
5685 wfProfileOut( __METHOD__
);
5690 * Load the parser state given in the $data array, which is assumed to
5691 * have been generated by serializeHalfParsedText(). The text contents is
5692 * extracted from the array, and its markers are transformed into markers
5693 * appropriate for the current Parser instance. This transformed text is
5694 * returned, and can be safely included in the return value of a parser
5697 * If the $data array has been stored persistently, the caller should first
5698 * check whether it is still valid, by calling isValidHalfParsedText().
5700 * @param $data array Serialized data
5703 function unserializeHalfParsedText( $data ) {
5704 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
5705 throw new MWException( __METHOD__
.': invalid version' );
5708 # First, extract the strip state.
5709 $texts = array( $data['text'] );
5710 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
5712 # Now renumber links
5713 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
5715 # Should be good to go.
5720 * Returns true if the given array, presumed to be generated by
5721 * serializeHalfParsedText(), is compatible with the current version of the
5724 * @param $data Array
5728 function isValidHalfParsedText( $data ) {
5729 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;