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.
35 * objects: $wgLang, $wgContLang
37 * NOT $wgArticle, $wgUser or $wgTitle. 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
= '1.6.5';
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 const EXT_LINK_URL_CLASS
= '(?:[^\]\[<>"\\x00-\\x20\\x7F]|(?:\[\]))';
72 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
73 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
75 # State constants for the definition list colon extraction
76 const COLON_STATE_TEXT
= 0;
77 const COLON_STATE_TAG
= 1;
78 const COLON_STATE_TAGSTART
= 2;
79 const COLON_STATE_CLOSETAG
= 3;
80 const COLON_STATE_TAGSLASH
= 4;
81 const COLON_STATE_COMMENT
= 5;
82 const COLON_STATE_COMMENTDASH
= 6;
83 const COLON_STATE_COMMENTDASHDASH
= 7;
85 # Flags for preprocessToDom
86 const PTD_FOR_INCLUSION
= 1;
88 # Allowed values for $this->mOutputType
89 # Parameter to startExternalParse().
90 const OT_HTML
= 1; # like parse()
91 const OT_WIKI
= 2; # like preSaveTransform()
92 const OT_PREPROCESS
= 3; # like preprocess()
94 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
96 # Marker Suffix needs to be accessible staticly.
97 const MARKER_SUFFIX
= "-QINU\x7f";
100 var $mTagHooks = array();
101 var $mTransparentTagHooks = array();
102 var $mFunctionHooks = array();
103 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
104 var $mFunctionTagHooks = array();
105 var $mStripList = array();
106 var $mDefaultStripList = array();
107 var $mVarCache = array();
108 var $mImageParams = array();
109 var $mImageParamsMagicArray = array();
110 var $mMarkerIndex = 0;
111 var $mFirstCall = true;
112 var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
113 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
115 # Cleared with clearState():
120 var $mAutonumber, $mDTopen;
127 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
128 var $mLinkHolders, $mLinkID;
129 var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
130 var $mTplExpandCache; # empty-frame expansion cache
131 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
132 var $mExpensiveFunctionCount; # number of expensive parser function calls
137 var $mUser; # User object; only used when doing pre-save transform
140 # These are variables reset at least once per parse regardless of $clearState
150 var $mTitle; # Title context, used for self-link rendering and similar things
151 var $mOutputType; # Output type, one of the OT_xxx constants
152 var $ot; # Shortcut alias, see setOutputType()
153 var $mRevisionObject; # The revision object of the specified revision ID
154 var $mRevisionId; # ID to display in {{REVISIONID}} tags
155 var $mRevisionTimestamp; # The timestamp of the specified revision ID
156 var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
157 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
162 public function __construct( $conf = array() ) {
163 $this->mConf
= $conf;
164 $this->mUrlProtocols
= wfUrlProtocols();
165 $this->mExtLinkBracketedRegex
= '/\[(\b(' . wfUrlProtocols() . ')'.
166 '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
167 if ( isset( $conf['preprocessorClass'] ) ) {
168 $this->mPreprocessorClass
= $conf['preprocessorClass'];
169 } elseif ( extension_loaded( 'domxml' ) ) {
170 # PECL extension that conflicts with the core DOM extension (bug 13770)
171 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
172 $this->mPreprocessorClass
= 'Preprocessor_Hash';
173 } elseif ( extension_loaded( 'dom' ) ) {
174 $this->mPreprocessorClass
= 'Preprocessor_DOM';
176 $this->mPreprocessorClass
= 'Preprocessor_Hash';
181 * Reduce memory usage to reduce the impact of circular references
183 function __destruct() {
184 if ( isset( $this->mLinkHolders
) ) {
185 $this->mLinkHolders
->__destruct();
187 foreach ( $this as $name => $value ) {
188 unset( $this->$name );
193 * Do various kinds of initialisation on the first call of the parser
195 function firstCallInit() {
196 if ( !$this->mFirstCall
) {
199 $this->mFirstCall
= false;
201 wfProfileIn( __METHOD__
);
203 CoreParserFunctions
::register( $this );
204 CoreTagHooks
::register( $this );
205 $this->initialiseVariables();
207 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
208 wfProfileOut( __METHOD__
);
216 function clearState() {
217 wfProfileIn( __METHOD__
);
218 if ( $this->mFirstCall
) {
219 $this->firstCallInit();
221 $this->mOutput
= new ParserOutput
;
222 $this->mOptions
->registerWatcher( array( $this->mOutput
, 'recordOption' ) );
223 $this->mAutonumber
= 0;
224 $this->mLastSection
= '';
225 $this->mDTopen
= false;
226 $this->mIncludeCount
= array();
227 $this->mArgStack
= false;
228 $this->mInPre
= false;
229 $this->mLinkHolders
= new LinkHolderArray( $this );
231 $this->mRevisionObject
= $this->mRevisionTimestamp
=
232 $this->mRevisionId
= $this->mRevisionUser
= null;
233 $this->mVarCache
= array();
237 * Prefix for temporary replacement strings for the multipass parser.
238 * \x07 should never appear in input as it's disallowed in XML.
239 * Using it at the front also gives us a little extra robustness
240 * since it shouldn't match when butted up against identifier-like
243 * Must not consist of all title characters, or else it will change
244 * the behaviour of <nowiki> in a link.
246 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
247 # Changed to \x7f to allow XML double-parsing -- TS
248 $this->mUniqPrefix
= "\x7fUNIQ" . self
::getRandomString();
249 $this->mStripState
= new StripState( $this->mUniqPrefix
);
252 # Clear these on every parse, bug 4549
253 $this->mTplExpandCache
= $this->mTplRedirCache
= $this->mTplDomCache
= array();
255 $this->mShowToc
= true;
256 $this->mForceTocPosition
= false;
257 $this->mIncludeSizes
= array(
261 $this->mPPNodeCount
= 0;
262 $this->mDefaultSort
= false;
263 $this->mHeadings
= array();
264 $this->mDoubleUnderscores
= array();
265 $this->mExpensiveFunctionCount
= 0;
268 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
269 $this->mPreprocessor
= null;
272 wfRunHooks( 'ParserClearState', array( &$this ) );
273 wfProfileOut( __METHOD__
);
277 * Convert wikitext to HTML
278 * Do not call this function recursively.
280 * @param $text String: text we want to parse
281 * @param $title Title object
282 * @param $options ParserOptions
283 * @param $linestart boolean
284 * @param $clearState boolean
285 * @param $revid Int: number to pass in {{REVISIONID}}
286 * @return ParserOutput a ParserOutput
288 public function parse( $text, Title
$title, ParserOptions
$options, $linestart = true, $clearState = true, $revid = null ) {
290 * First pass--just handle <nowiki> sections, pass the rest off
291 * to internalParse() which does all the real work.
294 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
295 $fname = __METHOD__
.'-' . wfGetCaller();
296 wfProfileIn( __METHOD__
);
297 wfProfileIn( $fname );
299 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
301 $oldRevisionId = $this->mRevisionId
;
302 $oldRevisionObject = $this->mRevisionObject
;
303 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
304 $oldRevisionUser = $this->mRevisionUser
;
305 if ( $revid !== null ) {
306 $this->mRevisionId
= $revid;
307 $this->mRevisionObject
= null;
308 $this->mRevisionTimestamp
= null;
309 $this->mRevisionUser
= null;
312 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
314 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
315 $text = $this->internalParse( $text );
317 $text = $this->mStripState
->unstripGeneral( $text );
319 # Clean up special characters, only run once, next-to-last before doBlockLevels
321 # french spaces, last one Guillemet-left
322 # only if there is something before the space
323 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1 \\2',
324 # french spaces, Guillemet-right
325 '/(\\302\\253) /' => '\\1 ',
326 '/ (!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
328 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
330 $text = $this->doBlockLevels( $text, $linestart );
332 $this->replaceLinkHolders( $text );
335 * The page doesn't get language converted if
337 * b) Content isn't converted
338 * c) It's a conversion table
340 if ( !( $wgDisableLangConversion
341 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
342 ||
$this->mTitle
->isConversionTable() ) ) {
344 # The position of the convert() call should not be changed. it
345 # assumes that the links are all replaced and the only thing left
346 # is the <nowiki> mark.
348 $text = $wgContLang->convert( $text );
352 * A converted title will be provided in the output object if title and
353 * content conversion are enabled, the article text does not contain
354 * a conversion-suppressing double-underscore tag, and no
355 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
356 * automatic link conversion.
358 if ( !( $wgDisableLangConversion
359 ||
$wgDisableTitleConversion
360 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
361 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
362 ||
$this->mOutput
->getDisplayTitle() !== false ) )
364 $convruletitle = $wgContLang->getConvRuleTitle();
365 if ( $convruletitle ) {
366 $this->mOutput
->setTitleText( $convruletitle );
368 $titleText = $wgContLang->convertTitle( $title );
369 $this->mOutput
->setTitleText( $titleText );
373 $text = $this->mStripState
->unstripNoWiki( $text );
375 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
377 $text = $this->replaceTransparentTags( $text );
378 $text = $this->mStripState
->unstripGeneral( $text );
380 $text = Sanitizer
::normalizeCharReferences( $text );
382 if ( ( $wgUseTidy && $this->mOptions
->getTidy() ) ||
$wgAlwaysUseTidy ) {
383 $text = MWTidy
::tidy( $text );
385 # attempt to sanitize at least some nesting problems
386 # (bug #2702 and quite a few others)
388 # ''Something [http://www.cool.com cool''] -->
389 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
390 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
391 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
392 # fix up an anchor inside another anchor, only
393 # at least for a single single nested link (bug 3695)
394 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
395 '\\1\\2</a>\\3</a>\\1\\4</a>',
396 # fix div inside inline elements- doBlockLevels won't wrap a line which
397 # contains a div, so fix it up here; replace
398 # div with escaped text
399 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
400 '\\1\\3<div\\5>\\6</div>\\8\\9',
401 # remove empty italic or bold tag pairs, some
402 # introduced by rules above
403 '/<([bi])><\/\\1>/' => '',
406 $text = preg_replace(
407 array_keys( $tidyregs ),
408 array_values( $tidyregs ),
411 global $wgExpensiveParserFunctionLimit;
412 if ( $this->mExpensiveFunctionCount
> $wgExpensiveParserFunctionLimit ) {
413 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount
, $wgExpensiveParserFunctionLimit );
416 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
418 # Information on include size limits, for the benefit of users who try to skirt them
419 if ( $this->mOptions
->getEnableLimitReport() ) {
420 $max = $this->mOptions
->getMaxIncludeSize();
421 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
423 "NewPP limit report\n" .
424 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
425 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
426 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
428 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
429 $text .= "\n<!-- \n$limitReport-->\n";
431 $this->mOutput
->setText( $text );
433 $this->mRevisionId
= $oldRevisionId;
434 $this->mRevisionObject
= $oldRevisionObject;
435 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
436 $this->mRevisionUser
= $oldRevisionUser;
437 wfProfileOut( $fname );
438 wfProfileOut( __METHOD__
);
440 return $this->mOutput
;
444 * Recursive parser entry point that can be called from an extension tag
447 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
449 * @param $text String: text extension wants to have parsed
450 * @param $frame PPFrame: The frame to use for expanding any template variables
452 function recursiveTagParse( $text, $frame=false ) {
453 wfProfileIn( __METHOD__
);
454 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
455 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
456 $text = $this->internalParse( $text, false, $frame );
457 wfProfileOut( __METHOD__
);
462 * Expand templates and variables in the text, producing valid, static wikitext.
463 * Also removes comments.
465 function preprocess( $text, Title
$title, ParserOptions
$options, $revid = null ) {
466 wfProfileIn( __METHOD__
);
467 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
468 if ( $revid !== null ) {
469 $this->mRevisionId
= $revid;
471 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
472 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
473 $text = $this->replaceVariables( $text );
474 $text = $this->mStripState
->unstripBoth( $text );
475 wfProfileOut( __METHOD__
);
480 * Process the wikitext for the ?preload= feature. (bug 5210)
482 * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
483 * comments, templates, arguments, tags hooks and parser functions are untouched.
485 public function getPreloadText( $text, Title
$title, ParserOptions
$options ) {
486 # Parser (re)initialisation
487 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
489 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
490 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
491 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
492 $text = $this->mStripState
->unstripBoth( $text );
497 * Get a random string
501 static public function getRandomString() {
502 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
506 * Set the current user.
507 * Should only be used when doing pre-save transform.
509 * @param $user Mixed: User object or null (to reset)
511 function setUser( $user ) {
512 $this->mUser
= $user;
516 * Accessor for mUniqPrefix.
520 public function uniqPrefix() {
521 if ( !isset( $this->mUniqPrefix
) ) {
522 # @todo Fixme: this is probably *horribly wrong*
523 # LanguageConverter seems to want $wgParser's uniqPrefix, however
524 # if this is called for a parser cache hit, the parser may not
525 # have ever been initialized in the first place.
526 # Not really sure what the heck is supposed to be going on here.
528 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
530 return $this->mUniqPrefix
;
534 * Set the context title
536 function setTitle( $t ) {
537 if ( !$t ||
$t instanceof FakeTitle
) {
538 $t = Title
::newFromText( 'NO TITLE' );
541 if ( strval( $t->getFragment() ) !== '' ) {
542 # Strip the fragment to avoid various odd effects
543 $this->mTitle
= clone $t;
544 $this->mTitle
->setFragment( '' );
551 * Accessor for the Title object
553 * @return Title object
555 function getTitle() {
556 return $this->mTitle
;
560 * Accessor/mutator for the Title object
562 * @param $x New Title object or null to just get the current one
563 * @return Title object
565 function Title( $x = null ) {
566 return wfSetVar( $this->mTitle
, $x );
570 * Set the output type
572 * @param $ot Integer: new value
574 function setOutputType( $ot ) {
575 $this->mOutputType
= $ot;
578 'html' => $ot == self
::OT_HTML
,
579 'wiki' => $ot == self
::OT_WIKI
,
580 'pre' => $ot == self
::OT_PREPROCESS
,
581 'plain' => $ot == self
::OT_PLAIN
,
586 * Accessor/mutator for the output type
588 * @param $x New value or null to just get the current one
591 function OutputType( $x = null ) {
592 return wfSetVar( $this->mOutputType
, $x );
596 * Get the ParserOutput object
598 * @return ParserOutput object
600 function getOutput() {
601 return $this->mOutput
;
605 * Get the ParserOptions object
607 * @return ParserOptions object
609 function getOptions() {
610 return $this->mOptions
;
614 * Accessor/mutator for the ParserOptions object
616 * @param $x New value or null to just get the current one
617 * @return Current ParserOptions object
619 function Options( $x = null ) {
620 return wfSetVar( $this->mOptions
, $x );
623 function nextLinkID() {
624 return $this->mLinkID++
;
627 function setLinkID( $id ) {
628 $this->mLinkID
= $id;
634 function getFunctionLang() {
635 global $wgLang, $wgContLang;
637 $target = $this->mOptions
->getTargetLanguage();
638 if ( $target !== null ) {
641 return $this->mOptions
->getInterfaceMessage() ?
$wgLang : $wgContLang;
646 * Get a User object either from $this->mUser, if set, or from the
647 * ParserOptions object otherwise
649 * @return User object
652 if ( !is_null( $this->mUser
) ) {
655 return $this->mOptions
->getUser();
659 * Get a preprocessor object
661 * @return Preprocessor instance
663 function getPreprocessor() {
664 if ( !isset( $this->mPreprocessor
) ) {
665 $class = $this->mPreprocessorClass
;
666 $this->mPreprocessor
= new $class( $this );
668 return $this->mPreprocessor
;
672 * Replaces all occurrences of HTML-style comments and the given tags
673 * in the text with a random marker and returns the next text. The output
674 * parameter $matches will be an associative array filled with data in
676 * 'UNIQ-xxxxx' => array(
679 * array( 'param' => 'x' ),
680 * '<element param="x">tag content</element>' ) )
682 * @param $elements list of element names. Comments are always extracted.
683 * @param $text Source text string.
684 * @param $matches Out parameter, Array: extracted tags
685 * @param $uniq_prefix
686 * @return String: stripped text
690 public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
695 $taglist = implode( '|', $elements );
696 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
698 while ( $text != '' ) {
699 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
701 if ( count( $p ) < 5 ) {
704 if ( count( $p ) > 5 ) {
718 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
719 $stripped .= $marker;
721 if ( $close === '/>' ) {
722 # Empty element tag, <tag />
727 if ( $element === '!--' ) {
730 $end = "/(<\\/$element\\s*>)/i";
732 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
734 if ( count( $q ) < 3 ) {
735 # No end tag -- let it run out to the end of the text.
744 $matches[$marker] = array( $element,
746 Sanitizer
::decodeTagAttributes( $attributes ),
747 "<$element$attributes$close$content$tail" );
753 * Get a list of strippable XML-like elements
755 function getStripList() {
756 return $this->mStripList
;
760 * @deprecated use replaceVariables
762 function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
767 * Restores pre, math, and other extensions removed by strip()
769 * always call unstripNoWiki() after this one
771 * @deprecated use $this->mStripState->unstrip()
773 function unstrip( $text, $state ) {
774 return $state->unstripGeneral( $text );
778 * Always call this after unstrip() to preserve the order
781 * @deprecated use $this->mStripState->unstrip()
783 function unstripNoWiki( $text, $state ) {
784 return $state->unstripNoWiki( $text );
788 * @deprecated use $this->mStripState->unstripBoth()
790 function unstripForHTML( $text ) {
791 return $this->mStripState
->unstripBoth( $text );
795 * Add an item to the strip state
796 * Returns the unique tag which must be inserted into the stripped text
797 * The tag will be replaced with the original text in unstrip()
801 function insertStripItem( $text ) {
802 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
803 $this->mMarkerIndex++
;
804 $this->mStripState
->addGeneral( $rnd, $text );
809 * Interface with html tidy
810 * @deprecated Use MWTidy::tidy()
812 public static function tidy( $text ) {
813 wfDeprecated( __METHOD__
);
814 return MWTidy
::tidy( $text );
818 * parse the wiki syntax used to render tables
822 function doTableStuff( $text ) {
823 wfProfileIn( __METHOD__
);
825 $lines = StringUtils
::explode( "\n", $text );
829 foreach ( $lines as $outLine ) {
830 $line = trim( $outLine );
832 # empty line, go to next line,
833 # but only append \n if outside of table
835 $output .= $outLine . "\n";
838 $firstChars = $line[0];
839 if ( strlen( $line ) > 1 ) {
840 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ?
$line[1] : '';
844 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
846 $table =& $this->last( $tables );
847 $table[0] = array(); // first row
848 $currentRow =& $table[0];
849 $table['indent'] = strlen( $matches[1] );
851 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
852 $attributes = Sanitizer
::fixTagAttributes( $attributes , 'table' );
854 if ( $attributes !== '' ) {
855 $table['attributes'] = $attributes;
857 } else if ( !isset( $tables[0] ) ) {
858 // we're outside the table
860 $out .= $outLine . "\n";
861 } else if ( $firstChars === '|}' ) {
862 // trim the |} code from the line
863 $line = substr ( $line , 2 );
865 // Shorthand for last row
866 $lastRow =& $this->last( $table );
868 // a thead at the end becomes a tfoot, unless there is only one row
869 // Do this before deleting empty last lines to allow headers at the bottom of tables
870 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
871 $lastRow['type'] = 'tfoot';
872 for ( $i = 0; isset( $lastRow[$i] ); $i++
) {
873 $lastRow[$i]['type'] = 'th';
877 // Delete empty last lines
878 if ( empty( $lastRow ) ) {
882 $curtable = array_pop( $tables );
884 #Add a line-ending before the table, but only if there isn't one already
885 if ( substr( $out, -1 ) !== "\n" ) {
888 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
890 if ( count( $tables ) > 0 ) {
891 $table =& $this->last( $tables );
892 $currentRow =& $this->last( $table );
893 $currentElement =& $this->last( $currentRow );
895 $output =& $currentElement['content'];
902 } else if ( $firstChars === '|-' ) {
903 // start a new row element
904 // but only when we haven't started one already
905 if ( count( $currentRow ) != 0 ) {
907 $currentRow =& $this->last( $table );
909 // Get the attributes, there's nothing else useful in $line now
910 $line = substr ( $line , 2 );
911 $attributes = $this->mStripState
->unstripBoth( $line );
912 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
913 if ( $attributes !== '' ) {
914 $currentRow['attributes'] = $attributes;
917 } else if ( $firstChars === '|+' ) {
919 $line = substr ( $line , 2 );
921 $c = $this->getCellAttr( $line , 'caption' );
922 $table['caption'] = array();
923 $table['caption']['content'] = $c[0];
924 if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
927 $output =& $table['caption'];
928 } else if ( $firstChars === '|' ||
$firstChars === '!' ||
$firstChars === '!+' ) {
929 // Which kind of cells are we dealing with
931 $line = substr ( $line , 1 );
933 if ( $firstChars === '!' ||
$firstChars === '!+' ) {
934 $line = str_replace ( '!!' , '||' , $line );
938 // Split up multiple cells on the same line.
939 $cells = StringUtils
::explodeMarkup( '||' , $line );
940 $line = ''; // save memory
942 // decide whether thead to tbody
943 if ( !array_key_exists( 'type', $currentRow ) ) {
944 $currentRow['type'] = ( $firstChars === '!' ) ?
'thead' : 'tbody' ;
945 } else if ( $firstChars === '|' ) {
946 $currentRow['type'] = 'tbody';
949 // Loop through each table cell
950 foreach ( $cells as $cell ) {
952 $currentRow[] = array();
953 $currentElement =& $this->last( $currentRow );
955 $currentElement['type'] = $currentTag;
957 $c = $this->getCellAttr( $cell , $currentTag );
958 $currentElement['content'] = $c[0];
959 if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
962 $output =& $currentElement['content'];
965 $output .= "\n$outLine";
969 # Remove trailing line-ending (b/c)
970 if ( substr( $out, -1 ) === "\n" ) {
971 $out = substr( $out, 0, -1 );
974 # Close any unclosed tables
975 if ( isset( $tables ) && count( $tables ) > 0 ) {
976 for ( $i = 0; $i < count( $tables ); $i++
) {
977 $curtable = array_pop( $tables );
978 $curtable = $this->generateTableHTML( $curtable );
979 #Add a line-ending before the table, but only if there isn't one already
980 if ( substr( $out, -1 ) !== "\n" && $curtable !== "" ) {
987 wfProfileOut( __METHOD__
);
994 * Helper function for doTableStuff() separating the contents of cells from
995 * attributes. Particularly useful as there's a possible bug and this action
1000 function getCellAttr ( $cell, $tagName ) {
1004 $cell = trim ( $cell );
1006 // A cell could contain both parameters and data
1007 $cellData = explode ( '|' , $cell , 2 );
1009 // Bug 553: Note that a '|' inside an invalid link should not
1010 // be mistaken as delimiting cell parameters
1011 if ( strpos( $cellData[0], '[[' ) !== false ) {
1012 $content = trim ( $cell );
1014 else if ( count ( $cellData ) == 1 ) {
1015 $content = trim ( $cellData[0] );
1018 $attributes = $this->mStripState
->unstripBoth( $cellData[0] );
1019 $attributes = Sanitizer
::fixTagAttributes( $attributes , $tagName );
1021 $content = trim ( $cellData[1] );
1023 return array( $content, $attributes );
1028 * Helper function for doTableStuff(). This converts the structured array into html.
1032 function generateTableHTML ( &$table ) {
1034 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1035 $return .= '<table';
1036 $return .= isset( $table['attributes'] ) ?
$table['attributes'] : '';
1038 unset( $table['attributes'] );
1040 if ( isset( $table['caption'] ) ) {
1041 $return .= "\n<caption";
1042 $return .= isset( $table['caption']['attributes'] ) ?
$table['caption']['attributes'] : '';
1044 $return .= $table['caption']['content'];
1045 $return .= "\n</caption>";
1051 // If we only have tbodies, mark table as simple
1052 for ( $i = 0; isset( $table[$i] ); $i++
) {
1053 if ( !count( $table[$i] ) ) continue;
1054 if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1055 if ( !$lastSection ) {
1056 $lastSection = $table[$i]['type'];
1057 } else if ( $lastSection != $table[$i]['type'] ) {
1063 for ( $i = 0; isset( $table[$i] ); $i++
) {
1064 // Check for empty tables
1065 if ( count( $table[$i] ) ) {
1070 if ( $table[$i]['type'] != $lastSection && !$simple ) {
1071 $return .= "\n<" . $table[$i]['type'] . '>';
1075 $return .= isset( $table[$i]['attributes'] ) ?
$table[$i]['attributes'] : '';
1077 for ( $j = 0; isset( $table[$i][$j] ); $j++
) {
1078 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1079 $return .= "\n<" . $table[$i][$j]['type'];
1080 $return .= isset( $table[$i][$j]['attributes'] ) ?
$table[$i][$j]['attributes'] : '';
1083 $return .= $table[$i][$j]['content'];
1084 if ( $table[$i][$j]['content'] != '' )
1087 $return .= '</' . $table[$i][$j]['type'] . '>';
1088 unset( $table[$i][$j] );
1090 $return .= "\n</tr>";
1092 if ( ( !isset( $table[$i +
1] ) && !$simple ) ||
( isset( $table[$i +
1] ) && ( $table[$i]['type'] != $table[$i +
1]['type'] ) ) ) {
1093 $return .= '</' . $table[$i]['type'] . '>';
1095 $lastSection = $table[$i]['type'];
1096 unset( $table[$i] );
1099 if ( isset( $table['caption'] ) ) {
1100 $return .= "\n<tr><td></td></tr>";
1105 $return .= "\n</table>";
1106 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1112 * like end() but only works on the numeric array index and php's internal pointers
1113 * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1114 * ignores associative elements and will create a 0 key will a NULL value if there were
1115 * no numric elements and an array itself if not previously defined.
1119 function &last ( &$arr ) {
1120 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) { }
1125 * Helper function for parse() that transforms wiki markup into
1126 * HTML. Only called for $mOutputType == self::OT_HTML.
1130 function internalParse( $text, $isMain = true, $frame=false ) {
1131 wfProfileIn( __METHOD__
);
1135 # Hook to suspend the parser in this state
1136 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState
) ) ) {
1137 wfProfileOut( __METHOD__
);
1141 # if $frame is provided, then use $frame for replacing any variables
1143 # use frame depth to infer how include/noinclude tags should be handled
1144 # depth=0 means this is the top-level document; otherwise it's an included document
1145 if ( !$frame->depth
) {
1148 $flag = Parser
::PTD_FOR_INCLUSION
;
1150 $dom = $this->preprocessToDom( $text, $flag );
1151 $text = $frame->expand( $dom );
1153 # if $frame is not provided, then use old-style replaceVariables
1154 $text = $this->replaceVariables( $text );
1157 $text = Sanitizer
::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks
) );
1158 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState
) );
1160 # Tables need to come after variable replacement for things to work
1161 # properly; putting them before other transformations should keep
1162 # exciting things like link expansions from showing up in surprising
1164 $text = $this->doTableStuff( $text );
1166 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1168 $text = $this->doDoubleUnderscore( $text );
1170 $text = $this->doHeadings( $text );
1171 if ( $this->mOptions
->getUseDynamicDates() ) {
1172 $df = DateFormatter
::getInstance();
1173 $text = $df->reformat( $this->mOptions
->getDateFormat(), $text );
1175 $text = $this->replaceInternalLinks( $text );
1176 $text = $this->doAllQuotes( $text );
1177 $text = $this->replaceExternalLinks( $text );
1179 # replaceInternalLinks may sometimes leave behind
1180 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1181 $text = str_replace( $this->mUniqPrefix
.'NOPARSE', '', $text );
1183 $text = $this->doMagicLinks( $text );
1184 $text = $this->formatHeadings( $text, $origText, $isMain );
1186 wfProfileOut( __METHOD__
);
1191 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1192 * magic external links.
1197 function doMagicLinks( $text ) {
1198 wfProfileIn( __METHOD__
);
1199 $prots = $this->mUrlProtocols
;
1200 $urlChar = self
::EXT_LINK_URL_CLASS
;
1201 $text = preg_replace_callback(
1203 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1204 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1205 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1206 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1207 ISBN\s+(\b # m[5]: ISBN, capture number
1208 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1209 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1210 [0-9Xx] # check digit
1212 )!x', array( &$this, 'magicLinkCallback' ), $text );
1213 wfProfileOut( __METHOD__
);
1217 function magicLinkCallback( $m ) {
1218 if ( isset( $m[1] ) && $m[1] !== '' ) {
1221 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1224 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1225 # Free external link
1226 return $this->makeFreeExternalLink( $m[0] );
1227 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1229 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1232 $CssClass = 'mw-magiclink-rfc';
1234 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1236 $urlmsg = 'pubmedurl';
1237 $CssClass = 'mw-magiclink-pmid';
1240 throw new MWException( __METHOD__
.': unrecognised match type "' .
1241 substr( $m[0], 0, 20 ) . '"' );
1243 $url = wfMsgForContent( $urlmsg, $id );
1244 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1245 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1248 $num = strtr( $isbn, array(
1253 $titleObj = SpecialPage
::getTitleFor( 'Booksources', $num );
1255 $titleObj->escapeLocalUrl() .
1256 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1263 * Make a free external link, given a user-supplied URL
1267 function makeFreeExternalLink( $url ) {
1269 wfProfileIn( __METHOD__
);
1273 # The characters '<' and '>' (which were escaped by
1274 # removeHTMLtags()) should not be included in
1275 # URLs, per RFC 2396.
1277 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1278 $trail = substr( $url, $m2[0][1] ) . $trail;
1279 $url = substr( $url, 0, $m2[0][1] );
1282 # Move trailing punctuation to $trail
1284 # If there is no left bracket, then consider right brackets fair game too
1285 if ( strpos( $url, '(' ) === false ) {
1289 $numSepChars = strspn( strrev( $url ), $sep );
1290 if ( $numSepChars ) {
1291 $trail = substr( $url, -$numSepChars ) . $trail;
1292 $url = substr( $url, 0, -$numSepChars );
1295 $url = Sanitizer
::cleanUrl( $url );
1297 # Is this an external image?
1298 $text = $this->maybeMakeExternalImage( $url );
1299 if ( $text === false ) {
1300 # Not an image, make a link
1301 $text = Linker
::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1302 $this->getExternalLinkAttribs( $url ) );
1303 # Register it in the output object...
1304 # Replace unnecessary URL escape codes with their equivalent characters
1305 $pasteurized = self
::replaceUnusualEscapes( $url );
1306 $this->mOutput
->addExternalLink( $pasteurized );
1308 wfProfileOut( __METHOD__
);
1309 return $text . $trail;
1314 * Parse headers and return html
1318 function doHeadings( $text ) {
1319 wfProfileIn( __METHOD__
);
1320 for ( $i = 6; $i >= 1; --$i ) {
1321 $h = str_repeat( '=', $i );
1322 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1323 "<h$i>\\1</h$i>", $text );
1325 wfProfileOut( __METHOD__
);
1330 * Replace single quotes with HTML markup
1332 * @return string the altered text
1334 function doAllQuotes( $text ) {
1335 wfProfileIn( __METHOD__
);
1337 $lines = StringUtils
::explode( "\n", $text );
1338 foreach ( $lines as $line ) {
1339 $outtext .= $this->doQuotes( $line ) . "\n";
1341 $outtext = substr( $outtext, 0,-1 );
1342 wfProfileOut( __METHOD__
);
1347 * Helper function for doAllQuotes()
1349 public function doQuotes( $text ) {
1350 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1351 if ( count( $arr ) == 1 ) {
1354 # First, do some preliminary work. This may shift some apostrophes from
1355 # being mark-up to being text. It also counts the number of occurrences
1356 # of bold and italics mark-ups.
1359 for ( $i = 0; $i < count( $arr ); $i++
) {
1360 if ( ( $i %
2 ) == 1 ) {
1361 # If there are ever four apostrophes, assume the first is supposed to
1362 # be text, and the remaining three constitute mark-up for bold text.
1363 if ( strlen( $arr[$i] ) == 4 ) {
1366 } elseif ( strlen( $arr[$i] ) > 5 ) {
1367 # If there are more than 5 apostrophes in a row, assume they're all
1368 # text except for the last 5.
1369 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1372 # Count the number of occurrences of bold and italics mark-ups.
1373 # We are not counting sequences of five apostrophes.
1374 if ( strlen( $arr[$i] ) == 2 ) {
1376 } elseif ( strlen( $arr[$i] ) == 3 ) {
1378 } elseif ( strlen( $arr[$i] ) == 5 ) {
1385 # If there is an odd number of both bold and italics, it is likely
1386 # that one of the bold ones was meant to be an apostrophe followed
1387 # by italics. Which one we cannot know for certain, but it is more
1388 # likely to be one that has a single-letter word before it.
1389 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1391 $firstsingleletterword = -1;
1392 $firstmultiletterword = -1;
1394 foreach ( $arr as $r ) {
1395 if ( ( $i %
2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1396 $x1 = substr( $arr[$i-1], -1 );
1397 $x2 = substr( $arr[$i-1], -2, 1 );
1398 if ( $x1 === ' ' ) {
1399 if ( $firstspace == -1 ) {
1402 } elseif ( $x2 === ' ') {
1403 if ( $firstsingleletterword == -1 ) {
1404 $firstsingleletterword = $i;
1407 if ( $firstmultiletterword == -1 ) {
1408 $firstmultiletterword = $i;
1415 # If there is a single-letter word, use it!
1416 if ( $firstsingleletterword > -1 ) {
1417 $arr[$firstsingleletterword] = "''";
1418 $arr[$firstsingleletterword-1] .= "'";
1419 } elseif ( $firstmultiletterword > -1 ) {
1420 # If not, but there's a multi-letter word, use that one.
1421 $arr[$firstmultiletterword] = "''";
1422 $arr[$firstmultiletterword-1] .= "'";
1423 } elseif ( $firstspace > -1 ) {
1424 # ... otherwise use the first one that has neither.
1425 # (notice that it is possible for all three to be -1 if, for example,
1426 # there is only one pentuple-apostrophe in the line)
1427 $arr[$firstspace] = "''";
1428 $arr[$firstspace-1] .= "'";
1432 # Now let's actually convert our apostrophic mush to HTML!
1437 foreach ( $arr as $r ) {
1438 if ( ( $i %
2 ) == 0 ) {
1439 if ( $state === 'both' ) {
1445 if ( strlen( $r ) == 2 ) {
1446 if ( $state === 'i' ) {
1447 $output .= '</i>'; $state = '';
1448 } elseif ( $state === 'bi' ) {
1449 $output .= '</i>'; $state = 'b';
1450 } elseif ( $state === 'ib' ) {
1451 $output .= '</b></i><b>'; $state = 'b';
1452 } elseif ( $state === 'both' ) {
1453 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1454 } else { # $state can be 'b' or ''
1455 $output .= '<i>'; $state .= 'i';
1457 } elseif ( strlen( $r ) == 3 ) {
1458 if ( $state === 'b' ) {
1459 $output .= '</b>'; $state = '';
1460 } elseif ( $state === 'bi' ) {
1461 $output .= '</i></b><i>'; $state = 'i';
1462 } elseif ( $state === 'ib' ) {
1463 $output .= '</b>'; $state = 'i';
1464 } elseif ( $state === 'both' ) {
1465 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1466 } else { # $state can be 'i' or ''
1467 $output .= '<b>'; $state .= 'b';
1469 } elseif ( strlen( $r ) == 5 ) {
1470 if ( $state === 'b' ) {
1471 $output .= '</b><i>'; $state = 'i';
1472 } elseif ( $state === 'i' ) {
1473 $output .= '</i><b>'; $state = 'b';
1474 } elseif ( $state === 'bi' ) {
1475 $output .= '</i></b>'; $state = '';
1476 } elseif ( $state === 'ib' ) {
1477 $output .= '</b></i>'; $state = '';
1478 } elseif ( $state === 'both' ) {
1479 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1480 } else { # ($state == '')
1481 $buffer = ''; $state = 'both';
1487 # Now close all remaining tags. Notice that the order is important.
1488 if ( $state === 'b' ||
$state === 'ib' ) {
1491 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1494 if ( $state === 'bi' ) {
1497 # There might be lonely ''''', so make sure we have a buffer
1498 if ( $state === 'both' && $buffer ) {
1499 $output .= '<b><i>'.$buffer.'</i></b>';
1506 * Replace external links (REL)
1508 * Note: this is all very hackish and the order of execution matters a lot.
1509 * Make sure to run maintenance/parserTests.php if you change this code.
1513 function replaceExternalLinks( $text ) {
1515 wfProfileIn( __METHOD__
);
1517 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1518 $s = array_shift( $bits );
1521 while ( $i<count( $bits ) ) {
1523 $protocol = $bits[$i++
];
1524 $text = $bits[$i++
];
1525 $trail = $bits[$i++
];
1527 # The characters '<' and '>' (which were escaped by
1528 # removeHTMLtags()) should not be included in
1529 # URLs, per RFC 2396.
1531 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1532 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1533 $url = substr( $url, 0, $m2[0][1] );
1536 # If the link text is an image URL, replace it with an <img> tag
1537 # This happened by accident in the original parser, but some people used it extensively
1538 $img = $this->maybeMakeExternalImage( $text );
1539 if ( $img !== false ) {
1545 # Set linktype for CSS - if URL==text, link is essentially free
1546 $linktype = ( $text === $url ) ?
'free' : 'text';
1548 # No link text, e.g. [http://domain.tld/some.link]
1549 if ( $text == '' ) {
1550 # Autonumber if allowed. See bug #5918
1551 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1552 $langObj = $this->getFunctionLang();
1553 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1554 $linktype = 'autonumber';
1556 # Otherwise just use the URL
1557 $text = htmlspecialchars( $url );
1561 # Have link text, e.g. [http://domain.tld/some.link text]s
1563 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1566 $text = $wgContLang->markNoConversion( $text );
1568 $url = Sanitizer
::cleanUrl( $url );
1570 # Use the encoded URL
1571 # This means that users can paste URLs directly into the text
1572 # Funny characters like ö aren't valid in URLs anyway
1573 # This was changed in August 2004
1574 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
1575 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1577 # Register link in the output object.
1578 # Replace unnecessary URL escape codes with the referenced character
1579 # This prevents spammers from hiding links from the filters
1580 $pasteurized = self
::replaceUnusualEscapes( $url );
1581 $this->mOutput
->addExternalLink( $pasteurized );
1584 wfProfileOut( __METHOD__
);
1589 * Get an associative array of additional HTML attributes appropriate for a
1590 * particular external link. This currently may include rel => nofollow
1591 * (depending on configuration, namespace, and the URL's domain) and/or a
1592 * target attribute (depending on configuration).
1594 * @param $url String: optional URL, to extract the domain from for rel =>
1595 * nofollow if appropriate
1596 * @return Array: associative array of HTML attributes
1598 function getExternalLinkAttribs( $url = false ) {
1600 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1601 $ns = $this->mTitle
->getNamespace();
1602 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1603 $attribs['rel'] = 'nofollow';
1605 global $wgNoFollowDomainExceptions;
1606 if ( $wgNoFollowDomainExceptions ) {
1607 $bits = wfParseUrl( $url );
1608 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1609 foreach ( $wgNoFollowDomainExceptions as $domain ) {
1610 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1611 unset( $attribs['rel'] );
1618 if ( $this->mOptions
->getExternalLinkTarget() ) {
1619 $attribs['target'] = $this->mOptions
->getExternalLinkTarget();
1626 * Replace unusual URL escape codes with their equivalent characters
1628 * @param $url String
1631 * @todo This can merge genuinely required bits in the path or query string,
1632 * breaking legit URLs. A proper fix would treat the various parts of
1633 * the URL differently; as a workaround, just use the output for
1634 * statistical records, not for actual linking/output.
1636 static function replaceUnusualEscapes( $url ) {
1637 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1638 array( __CLASS__
, 'replaceUnusualEscapesCallback' ), $url );
1642 * Callback function used in replaceUnusualEscapes().
1643 * Replaces unusual URL escape codes with their equivalent character
1645 private static function replaceUnusualEscapesCallback( $matches ) {
1646 $char = urldecode( $matches[0] );
1647 $ord = ord( $char );
1648 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1649 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1650 # No, shouldn't be escaped
1653 # Yes, leave it escaped
1659 * make an image if it's allowed, either through the global
1660 * option, through the exception, or through the on-wiki whitelist
1663 function maybeMakeExternalImage( $url ) {
1664 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
1665 $imagesexception = !empty( $imagesfrom );
1667 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1668 if ( $imagesexception && is_array( $imagesfrom ) ) {
1669 $imagematch = false;
1670 foreach ( $imagesfrom as $match ) {
1671 if ( strpos( $url, $match ) === 0 ) {
1676 } elseif ( $imagesexception ) {
1677 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1679 $imagematch = false;
1681 if ( $this->mOptions
->getAllowExternalImages()
1682 ||
( $imagesexception && $imagematch ) ) {
1683 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1685 $text = Linker
::makeExternalImage( $url );
1688 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
1689 && preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1690 $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1691 foreach ( $whitelist as $entry ) {
1692 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1693 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
1696 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1697 # Image matches a whitelist entry
1698 $text = Linker
::makeExternalImage( $url );
1707 * Process [[ ]] wikilinks
1708 * @return String: processed text
1712 function replaceInternalLinks( $s ) {
1713 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
1718 * Process [[ ]] wikilinks (RIL)
1719 * @return LinkHolderArray
1723 function replaceInternalLinks2( &$s ) {
1726 wfProfileIn( __METHOD__
);
1728 wfProfileIn( __METHOD__
.'-setup' );
1729 static $tc = FALSE, $e1, $e1_img;
1730 # the % is needed to support urlencoded titles as well
1732 $tc = Title
::legalChars() . '#%';
1733 # Match a link having the form [[namespace:link|alternate]]trail
1734 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1735 # Match cases where there is no "]]", which might still be images
1736 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1739 $holders = new LinkHolderArray( $this );
1741 # split the entire text string on occurences of [[
1742 $a = StringUtils
::explode( '[[', ' ' . $s );
1743 # get the first element (all text up to first [[), and remove the space we added
1746 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1747 $s = substr( $s, 1 );
1749 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1751 if ( $useLinkPrefixExtension ) {
1752 # Match the end of a line for a word that's not followed by whitespace,
1753 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1754 $e2 = wfMsgForContent( 'linkprefix' );
1757 if ( is_null( $this->mTitle
) ) {
1758 wfProfileOut( __METHOD__
.'-setup' );
1759 wfProfileOut( __METHOD__
);
1760 throw new MWException( __METHOD__
.": \$this->mTitle is null\n" );
1762 $nottalk = !$this->mTitle
->isTalkPage();
1764 if ( $useLinkPrefixExtension ) {
1766 if ( preg_match( $e2, $s, $m ) ) {
1767 $first_prefix = $m[2];
1769 $first_prefix = false;
1775 if ( $wgContLang->hasVariants() ) {
1776 $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle
->getPrefixedText() );
1778 $selflink = array( $this->mTitle
->getPrefixedText() );
1780 $useSubpages = $this->areSubpagesAllowed();
1781 wfProfileOut( __METHOD__
.'-setup' );
1783 # Loop for each link
1784 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1785 # Check for excessive memory usage
1786 if ( $holders->isBig() ) {
1788 # Do the existence check, replace the link holders and clear the array
1789 $holders->replace( $s );
1793 if ( $useLinkPrefixExtension ) {
1794 wfProfileIn( __METHOD__
.'-prefixhandling' );
1795 if ( preg_match( $e2, $s, $m ) ) {
1802 if ( $first_prefix ) {
1803 $prefix = $first_prefix;
1804 $first_prefix = false;
1806 wfProfileOut( __METHOD__
.'-prefixhandling' );
1809 $might_be_img = false;
1811 wfProfileIn( __METHOD__
."-e1" );
1812 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1814 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1815 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1816 # the real problem is with the $e1 regex
1819 # Still some problems for cases where the ] is meant to be outside punctuation,
1820 # and no image is in sight. See bug 2095.
1822 if ( $text !== '' &&
1823 substr( $m[3], 0, 1 ) === ']' &&
1824 strpos( $text, '[' ) !== false
1827 $text .= ']'; # so that replaceExternalLinks($text) works later
1828 $m[3] = substr( $m[3], 1 );
1830 # fix up urlencoded title texts
1831 if ( strpos( $m[1], '%' ) !== false ) {
1832 # Should anchors '#' also be rejected?
1833 $m[1] = str_replace( array('<', '>'), array('<', '>'), rawurldecode( $m[1] ) );
1836 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1837 $might_be_img = true;
1839 if ( strpos( $m[1], '%' ) !== false ) {
1840 $m[1] = rawurldecode( $m[1] );
1843 } else { # Invalid form; output directly
1844 $s .= $prefix . '[[' . $line ;
1845 wfProfileOut( __METHOD__
."-e1" );
1848 wfProfileOut( __METHOD__
."-e1" );
1849 wfProfileIn( __METHOD__
."-misc" );
1851 # Don't allow internal links to pages containing
1852 # PROTO: where PROTO is a valid URL protocol; these
1853 # should be external links.
1854 if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1855 $s .= $prefix . '[[' . $line ;
1856 wfProfileOut( __METHOD__
."-misc" );
1860 # Make subpage if necessary
1861 if ( $useSubpages ) {
1862 $link = $this->maybeDoSubpageLink( $m[1], $text );
1867 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1869 # Strip off leading ':'
1870 $link = substr( $link, 1 );
1873 wfProfileOut( __METHOD__
."-misc" );
1874 wfProfileIn( __METHOD__
."-title" );
1875 $nt = Title
::newFromText( $this->mStripState
->unstripNoWiki( $link ) );
1876 if ( $nt === null ) {
1877 $s .= $prefix . '[[' . $line;
1878 wfProfileOut( __METHOD__
."-title" );
1882 $ns = $nt->getNamespace();
1883 $iw = $nt->getInterWiki();
1884 wfProfileOut( __METHOD__
."-title" );
1886 if ( $might_be_img ) { # if this is actually an invalid link
1887 wfProfileIn( __METHOD__
."-might_be_img" );
1888 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
1891 # look at the next 'line' to see if we can close it there
1893 $next_line = $a->current();
1894 if ( $next_line === false ||
$next_line === null ) {
1897 $m = explode( ']]', $next_line, 3 );
1898 if ( count( $m ) == 3 ) {
1899 # the first ]] closes the inner link, the second the image
1901 $text .= "[[{$m[0]}]]{$m[1]}";
1904 } elseif ( count( $m ) == 2 ) {
1905 # if there's exactly one ]] that's fine, we'll keep looking
1906 $text .= "[[{$m[0]}]]{$m[1]}";
1908 # if $next_line is invalid too, we need look no further
1909 $text .= '[[' . $next_line;
1914 # we couldn't find the end of this imageLink, so output it raw
1915 # but don't ignore what might be perfectly normal links in the text we've examined
1916 $holders->merge( $this->replaceInternalLinks2( $text ) );
1917 $s .= "{$prefix}[[$link|$text";
1918 # note: no $trail, because without an end, there *is* no trail
1919 wfProfileOut( __METHOD__
."-might_be_img" );
1922 } else { # it's not an image, so output it raw
1923 $s .= "{$prefix}[[$link|$text";
1924 # note: no $trail, because without an end, there *is* no trail
1925 wfProfileOut( __METHOD__
."-might_be_img" );
1928 wfProfileOut( __METHOD__
."-might_be_img" );
1931 $wasblank = ( $text == '' );
1935 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1936 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1937 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1938 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1939 $text = $this->doQuotes( $text );
1942 # Link not escaped by : , create the various objects
1946 wfProfileIn( __METHOD__
."-interwiki" );
1947 if ( $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1948 $this->mOutput
->addLanguageLink( $nt->getFullText() );
1949 $s = rtrim( $s . $prefix );
1950 $s .= trim( $trail, "\n" ) == '' ?
'': $prefix . $trail;
1951 wfProfileOut( __METHOD__
."-interwiki" );
1954 wfProfileOut( __METHOD__
."-interwiki" );
1956 if ( $ns == NS_FILE
) {
1957 wfProfileIn( __METHOD__
."-image" );
1958 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
1960 # if no parameters were passed, $text
1961 # becomes something like "File:Foo.png",
1962 # which we don't want to pass on to the
1966 # recursively parse links inside the image caption
1967 # actually, this will parse them in any other parameters, too,
1968 # but it might be hard to fix that, and it doesn't matter ATM
1969 $text = $this->replaceExternalLinks( $text );
1970 $holders->merge( $this->replaceInternalLinks2( $text ) );
1972 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1973 $s .= $prefix . $this->armorLinks(
1974 $this->makeImage( $nt, $text, $holders ) ) . $trail;
1976 $s .= $prefix . $trail;
1978 wfProfileOut( __METHOD__
."-image" );
1982 if ( $ns == NS_CATEGORY
) {
1983 wfProfileIn( __METHOD__
."-category" );
1984 $s = rtrim( $s . "\n" ); # bug 87
1987 $sortkey = $this->getDefaultSort();
1991 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
1992 $sortkey = str_replace( "\n", '', $sortkey );
1993 $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1994 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
1997 * Strip the whitespace Category links produce, see bug 87
1998 * @todo We might want to use trim($tmp, "\n") here.
2000 $s .= trim( $prefix . $trail, "\n" ) == '' ?
'' : $prefix . $trail;
2002 wfProfileOut( __METHOD__
."-category" );
2007 # Self-link checking
2008 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL
) {
2009 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2010 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2015 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2016 # FIXME: Should do batch file existence checks, see comment below
2017 if ( $ns == NS_MEDIA
) {
2018 wfProfileIn( __METHOD__
."-media" );
2019 # Give extensions a chance to select the file revision for us
2020 $time = $sha1 = $descQuery = false;
2021 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2022 array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2023 # Fetch and register the file (file title may be different via hooks)
2024 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2025 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2026 $s .= $prefix . $this->armorLinks(
2027 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2028 wfProfileOut( __METHOD__
."-media" );
2032 wfProfileIn( __METHOD__
."-always_known" );
2033 # Some titles, such as valid special pages or files in foreign repos, should
2034 # be shown as bluelinks even though they're not included in the page table
2036 # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2037 # batch file existence checks for NS_FILE and NS_MEDIA
2038 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2039 $this->mOutput
->addLink( $nt );
2040 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2042 # Links will be added to the output link list after checking
2043 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2045 wfProfileOut( __METHOD__
."-always_known" );
2047 wfProfileOut( __METHOD__
);
2052 * Make a link placeholder. The text returned can be later resolved to a real link with
2053 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
2054 * parsing of interwiki links, and secondly to allow all existence checks and
2055 * article length checks (for stub links) to be bundled into a single query.
2059 function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2060 return $this->mLinkHolders
->makeHolder( $nt, $text, $query, $trail, $prefix );
2064 * Render a forced-blue link inline; protect against double expansion of
2065 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2066 * Since this little disaster has to split off the trail text to avoid
2067 * breaking URLs in the following text without breaking trails on the
2068 * wiki links, it's been made into a horrible function.
2071 * @param $text String
2072 * @param $query Array or String
2073 * @param $trail String
2074 * @param $prefix String
2075 * @return String: HTML-wikitext mix oh yuck
2077 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2078 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2080 if ( is_string( $query ) ) {
2081 $query = wfCgiToArray( $query );
2083 if ( $text == '' ) {
2084 $text = htmlspecialchars( $nt->getPrefixedText() );
2087 $link = Linker
::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2089 return $this->armorLinks( $link ) . $trail;
2093 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2094 * going to go through further parsing steps before inline URL expansion.
2096 * Not needed quite as much as it used to be since free links are a bit
2097 * more sensible these days. But bracketed links are still an issue.
2099 * @param $text String: more-or-less HTML
2100 * @return String: less-or-more HTML with NOPARSE bits
2102 function armorLinks( $text ) {
2103 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2104 "{$this->mUniqPrefix}NOPARSE$1", $text );
2108 * Return true if subpage links should be expanded on this page.
2111 function areSubpagesAllowed() {
2112 # Some namespaces don't allow subpages
2113 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2117 * Handle link to subpage if necessary
2119 * @param $target String: the source of the link
2120 * @param &$text String: the link text, modified as necessary
2121 * @return string the full name of the link
2124 function maybeDoSubpageLink( $target, &$text ) {
2125 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2129 * Used by doBlockLevels()
2132 function closeParagraph() {
2134 if ( $this->mLastSection
!= '' ) {
2135 $result = '</' . $this->mLastSection
. ">\n";
2137 $this->mInPre
= false;
2138 $this->mLastSection
= '';
2143 * getCommon() returns the length of the longest common substring
2144 * of both arguments, starting at the beginning of both.
2147 function getCommon( $st1, $st2 ) {
2148 $fl = strlen( $st1 );
2149 $shorter = strlen( $st2 );
2150 if ( $fl < $shorter ) {
2154 for ( $i = 0; $i < $shorter; ++
$i ) {
2155 if ( $st1[$i] != $st2[$i] ) {
2163 * These next three functions open, continue, and close the list
2164 * element appropriate to the prefix character passed into them.
2167 function openList( $char ) {
2168 $result = $this->closeParagraph();
2170 if ( '*' === $char ) {
2171 $result .= '<ul><li>';
2172 } elseif ( '#' === $char ) {
2173 $result .= '<ol><li>';
2174 } elseif ( ':' === $char ) {
2175 $result .= '<dl><dd>';
2176 } elseif ( ';' === $char ) {
2177 $result .= '<dl><dt>';
2178 $this->mDTopen
= true;
2180 $result = '<!-- ERR 1 -->';
2188 * @param $char String
2191 function nextItem( $char ) {
2192 if ( '*' === $char ||
'#' === $char ) {
2194 } elseif ( ':' === $char ||
';' === $char ) {
2196 if ( $this->mDTopen
) {
2199 if ( ';' === $char ) {
2200 $this->mDTopen
= true;
2201 return $close . '<dt>';
2203 $this->mDTopen
= false;
2204 return $close . '<dd>';
2207 return '<!-- ERR 2 -->';
2212 * @param $char String
2215 function closeList( $char ) {
2216 if ( '*' === $char ) {
2217 $text = '</li></ul>';
2218 } elseif ( '#' === $char ) {
2219 $text = '</li></ol>';
2220 } elseif ( ':' === $char ) {
2221 if ( $this->mDTopen
) {
2222 $this->mDTopen
= false;
2223 $text = '</dt></dl>';
2225 $text = '</dd></dl>';
2228 return '<!-- ERR 3 -->';
2235 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2237 * @param $text String
2238 * @param $linestart Boolean: whether or not this is at the start of a line.
2240 * @return string the lists rendered as HTML
2242 function doBlockLevels( $text, $linestart ) {
2243 wfProfileIn( __METHOD__
);
2245 # Parsing through the text line by line. The main thing
2246 # happening here is handling of block-level elements p, pre,
2247 # and making lists from lines starting with * # : etc.
2249 $textLines = StringUtils
::explode( "\n", $text );
2251 $lastPrefix = $output = '';
2252 $this->mDTopen
= $inBlockElem = false;
2254 $paragraphStack = false;
2256 foreach ( $textLines as $oLine ) {
2258 if ( !$linestart ) {
2268 $lastPrefixLength = strlen( $lastPrefix );
2269 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2270 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2271 # If not in a <pre> element, scan for and figure out what prefixes are there.
2272 if ( !$this->mInPre
) {
2273 # Multiple prefixes may abut each other for nested lists.
2274 $prefixLength = strspn( $oLine, '*#:;' );
2275 $prefix = substr( $oLine, 0, $prefixLength );
2278 # ; and : are both from definition-lists, so they're equivalent
2279 # for the purposes of determining whether or not we need to open/close
2281 $prefix2 = str_replace( ';', ':', $prefix );
2282 $t = substr( $oLine, $prefixLength );
2283 $this->mInPre
= (bool)$preOpenMatch;
2285 # Don't interpret any other prefixes in preformatted text
2287 $prefix = $prefix2 = '';
2292 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2293 # Same as the last item, so no need to deal with nesting or opening stuff
2294 $output .= $this->nextItem( substr( $prefix, -1 ) );
2295 $paragraphStack = false;
2297 if ( substr( $prefix, -1 ) === ';') {
2298 # The one nasty exception: definition lists work like this:
2299 # ; title : definition text
2300 # So we check for : in the remainder text to split up the
2301 # title and definition, without b0rking links.
2303 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2305 $output .= $term . $this->nextItem( ':' );
2308 } elseif ( $prefixLength ||
$lastPrefixLength ) {
2309 # We need to open or close prefixes, or both.
2311 # Either open or close a level...
2312 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2313 $paragraphStack = false;
2315 # Close all the prefixes which aren't shared.
2316 while ( $commonPrefixLength < $lastPrefixLength ) {
2317 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2318 --$lastPrefixLength;
2321 # Continue the current prefix if appropriate.
2322 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2323 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2326 # Open prefixes where appropriate.
2327 while ( $prefixLength > $commonPrefixLength ) {
2328 $char = substr( $prefix, $commonPrefixLength, 1 );
2329 $output .= $this->openList( $char );
2331 if ( ';' === $char ) {
2332 # FIXME: This is dupe of code above
2333 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2335 $output .= $term . $this->nextItem( ':' );
2338 ++
$commonPrefixLength;
2340 $lastPrefix = $prefix2;
2343 # If we have no prefixes, go to paragraph mode.
2344 if ( 0 == $prefixLength ) {
2345 wfProfileIn( __METHOD__
."-paragraph" );
2346 # No prefix (not in list)--go to paragraph mode
2347 # XXX: use a stack for nestable elements like span, table and div
2348 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2349 $closematch = preg_match(
2350 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2351 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix
.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2352 if ( $openmatch or $closematch ) {
2353 $paragraphStack = false;
2354 # TODO bug 5718: paragraph closed
2355 $output .= $this->closeParagraph();
2356 if ( $preOpenMatch and !$preCloseMatch ) {
2357 $this->mInPre
= true;
2359 $inBlockElem = !$closematch;
2360 } elseif ( !$inBlockElem && !$this->mInPre
) {
2361 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection
=== 'pre' ||
trim( $t ) != '' ) ) {
2363 if ( $this->mLastSection
!== 'pre' ) {
2364 $paragraphStack = false;
2365 $output .= $this->closeParagraph().'<pre>';
2366 $this->mLastSection
= 'pre';
2368 $t = substr( $t, 1 );
2371 if ( trim( $t ) === '' ) {
2372 if ( $paragraphStack ) {
2373 $output .= $paragraphStack.'<br />';
2374 $paragraphStack = false;
2375 $this->mLastSection
= 'p';
2377 if ( $this->mLastSection
!== 'p' ) {
2378 $output .= $this->closeParagraph();
2379 $this->mLastSection
= '';
2380 $paragraphStack = '<p>';
2382 $paragraphStack = '</p><p>';
2386 if ( $paragraphStack ) {
2387 $output .= $paragraphStack;
2388 $paragraphStack = false;
2389 $this->mLastSection
= 'p';
2390 } elseif ( $this->mLastSection
!== 'p' ) {
2391 $output .= $this->closeParagraph().'<p>';
2392 $this->mLastSection
= 'p';
2397 wfProfileOut( __METHOD__
."-paragraph" );
2399 # somewhere above we forget to get out of pre block (bug 785)
2400 if ( $preCloseMatch && $this->mInPre
) {
2401 $this->mInPre
= false;
2403 if ( $paragraphStack === false ) {
2407 while ( $prefixLength ) {
2408 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2411 if ( $this->mLastSection
!= '' ) {
2412 $output .= '</' . $this->mLastSection
. '>';
2413 $this->mLastSection
= '';
2416 wfProfileOut( __METHOD__
);
2421 * Split up a string on ':', ignoring any occurences inside tags
2422 * to prevent illegal overlapping.
2424 * @param $str String: the string to split
2425 * @param &$before String: set to everything before the ':'
2426 * @param &$after String: set to everything after the ':'
2427 * return String: the position of the ':', or false if none found
2429 function findColonNoLinks( $str, &$before, &$after ) {
2430 wfProfileIn( __METHOD__
);
2432 $pos = strpos( $str, ':' );
2433 if ( $pos === false ) {
2435 wfProfileOut( __METHOD__
);
2439 $lt = strpos( $str, '<' );
2440 if ( $lt === false ||
$lt > $pos ) {
2441 # Easy; no tag nesting to worry about
2442 $before = substr( $str, 0, $pos );
2443 $after = substr( $str, $pos+
1 );
2444 wfProfileOut( __METHOD__
);
2448 # Ugly state machine to walk through avoiding tags.
2449 $state = self
::COLON_STATE_TEXT
;
2451 $len = strlen( $str );
2452 for( $i = 0; $i < $len; $i++
) {
2456 # (Using the number is a performance hack for common cases)
2457 case 0: # self::COLON_STATE_TEXT:
2460 # Could be either a <start> tag or an </end> tag
2461 $state = self
::COLON_STATE_TAGSTART
;
2464 if ( $stack == 0 ) {
2466 $before = substr( $str, 0, $i );
2467 $after = substr( $str, $i +
1 );
2468 wfProfileOut( __METHOD__
);
2471 # Embedded in a tag; don't break it.
2474 # Skip ahead looking for something interesting
2475 $colon = strpos( $str, ':', $i );
2476 if ( $colon === false ) {
2477 # Nothing else interesting
2478 wfProfileOut( __METHOD__
);
2481 $lt = strpos( $str, '<', $i );
2482 if ( $stack === 0 ) {
2483 if ( $lt === false ||
$colon < $lt ) {
2485 $before = substr( $str, 0, $colon );
2486 $after = substr( $str, $colon +
1 );
2487 wfProfileOut( __METHOD__
);
2491 if ( $lt === false ) {
2492 # Nothing else interesting to find; abort!
2493 # We're nested, but there's no close tags left. Abort!
2496 # Skip ahead to next tag start
2498 $state = self
::COLON_STATE_TAGSTART
;
2501 case 1: # self::COLON_STATE_TAG:
2506 $state = self
::COLON_STATE_TEXT
;
2509 # Slash may be followed by >?
2510 $state = self
::COLON_STATE_TAGSLASH
;
2516 case 2: # self::COLON_STATE_TAGSTART:
2519 $state = self
::COLON_STATE_CLOSETAG
;
2522 $state = self
::COLON_STATE_COMMENT
;
2525 # Illegal early close? This shouldn't happen D:
2526 $state = self
::COLON_STATE_TEXT
;
2529 $state = self
::COLON_STATE_TAG
;
2532 case 3: # self::COLON_STATE_CLOSETAG:
2537 wfDebug( __METHOD__
.": Invalid input; too many close tags\n" );
2538 wfProfileOut( __METHOD__
);
2541 $state = self
::COLON_STATE_TEXT
;
2544 case self
::COLON_STATE_TAGSLASH
:
2546 # Yes, a self-closed tag <blah/>
2547 $state = self
::COLON_STATE_TEXT
;
2549 # Probably we're jumping the gun, and this is an attribute
2550 $state = self
::COLON_STATE_TAG
;
2553 case 5: # self::COLON_STATE_COMMENT:
2555 $state = self
::COLON_STATE_COMMENTDASH
;
2558 case self
::COLON_STATE_COMMENTDASH
:
2560 $state = self
::COLON_STATE_COMMENTDASHDASH
;
2562 $state = self
::COLON_STATE_COMMENT
;
2565 case self
::COLON_STATE_COMMENTDASHDASH
:
2567 $state = self
::COLON_STATE_TEXT
;
2569 $state = self
::COLON_STATE_COMMENT
;
2573 throw new MWException( "State machine error in " . __METHOD__
);
2577 wfDebug( __METHOD__
.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2578 wfProfileOut( __METHOD__
);
2581 wfProfileOut( __METHOD__
);
2586 * Return value of a magic variable (like PAGENAME)
2590 function getVariableValue( $index, $frame=false ) {
2591 global $wgContLang, $wgSitename, $wgServer;
2592 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2595 * Some of these require message or data lookups and can be
2596 * expensive to check many times.
2598 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache
) ) ) {
2599 if ( isset( $this->mVarCache
[$index] ) ) {
2600 return $this->mVarCache
[$index];
2604 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2605 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2608 global $wgLocaltimezone;
2609 if ( isset( $wgLocaltimezone ) ) {
2610 $oldtz = date_default_timezone_get();
2611 date_default_timezone_set( $wgLocaltimezone );
2614 $localTimestamp = date( 'YmdHis', $ts );
2615 $localMonth = date( 'm', $ts );
2616 $localMonth1 = date( 'n', $ts );
2617 $localMonthName = date( 'n', $ts );
2618 $localDay = date( 'j', $ts );
2619 $localDay2 = date( 'd', $ts );
2620 $localDayOfWeek = date( 'w', $ts );
2621 $localWeek = date( 'W', $ts );
2622 $localYear = date( 'Y', $ts );
2623 $localHour = date( 'H', $ts );
2624 if ( isset( $wgLocaltimezone ) ) {
2625 date_default_timezone_set( $oldtz );
2629 case 'currentmonth':
2630 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2632 case 'currentmonth1':
2633 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2635 case 'currentmonthname':
2636 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2638 case 'currentmonthnamegen':
2639 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2641 case 'currentmonthabbrev':
2642 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2645 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2648 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2651 $value = $wgContLang->formatNum( $localMonth );
2654 $value = $wgContLang->formatNum( $localMonth1 );
2656 case 'localmonthname':
2657 $value = $wgContLang->getMonthName( $localMonthName );
2659 case 'localmonthnamegen':
2660 $value = $wgContLang->getMonthNameGen( $localMonthName );
2662 case 'localmonthabbrev':
2663 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2666 $value = $wgContLang->formatNum( $localDay );
2669 $value = $wgContLang->formatNum( $localDay2 );
2672 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2675 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2677 case 'fullpagename':
2678 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2680 case 'fullpagenamee':
2681 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2684 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2686 case 'subpagenamee':
2687 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2689 case 'basepagename':
2690 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2692 case 'basepagenamee':
2693 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle
->getBaseText() ) ) );
2695 case 'talkpagename':
2696 if ( $this->mTitle
->canTalk() ) {
2697 $talkPage = $this->mTitle
->getTalkPage();
2698 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2703 case 'talkpagenamee':
2704 if ( $this->mTitle
->canTalk() ) {
2705 $talkPage = $this->mTitle
->getTalkPage();
2706 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2711 case 'subjectpagename':
2712 $subjPage = $this->mTitle
->getSubjectPage();
2713 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2715 case 'subjectpagenamee':
2716 $subjPage = $this->mTitle
->getSubjectPage();
2717 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2720 # Let the edit saving system know we should parse the page
2721 # *after* a revision ID has been assigned.
2722 $this->mOutput
->setFlag( 'vary-revision' );
2723 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision...\n" );
2724 $value = $this->mRevisionId
;
2727 # Let the edit saving system know we should parse the page
2728 # *after* a revision ID has been assigned. This is for null edits.
2729 $this->mOutput
->setFlag( 'vary-revision' );
2730 wfDebug( __METHOD__
. ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2731 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2733 case 'revisionday2':
2734 # Let the edit saving system know we should parse the page
2735 # *after* a revision ID has been assigned. This is for null edits.
2736 $this->mOutput
->setFlag( 'vary-revision' );
2737 wfDebug( __METHOD__
. ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2738 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2740 case 'revisionmonth':
2741 # Let the edit saving system know we should parse the page
2742 # *after* a revision ID has been assigned. This is for null edits.
2743 $this->mOutput
->setFlag( 'vary-revision' );
2744 wfDebug( __METHOD__
. ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2745 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2747 case 'revisionmonth1':
2748 # Let the edit saving system know we should parse the page
2749 # *after* a revision ID has been assigned. This is for null edits.
2750 $this->mOutput
->setFlag( 'vary-revision' );
2751 wfDebug( __METHOD__
. ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2752 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2754 case 'revisionyear':
2755 # Let the edit saving system know we should parse the page
2756 # *after* a revision ID has been assigned. This is for null edits.
2757 $this->mOutput
->setFlag( 'vary-revision' );
2758 wfDebug( __METHOD__
. ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2759 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2761 case 'revisiontimestamp':
2762 # Let the edit saving system know we should parse the page
2763 # *after* a revision ID has been assigned. This is for null edits.
2764 $this->mOutput
->setFlag( 'vary-revision' );
2765 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2766 $value = $this->getRevisionTimestamp();
2768 case 'revisionuser':
2769 # Let the edit saving system know we should parse the page
2770 # *after* a revision ID has been assigned. This is for null edits.
2771 $this->mOutput
->setFlag( 'vary-revision' );
2772 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2773 $value = $this->getRevisionUser();
2776 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2779 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2782 $value = $this->mTitle
->canTalk() ?
str_replace( '_',' ',$this->mTitle
->getTalkNsText() ) : '';
2785 $value = $this->mTitle
->canTalk() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2787 case 'subjectspace':
2788 $value = $this->mTitle
->getSubjectNsText();
2790 case 'subjectspacee':
2791 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2793 case 'currentdayname':
2794 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) +
1 );
2797 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2800 $value = $wgContLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2803 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2806 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2807 # int to remove the padding
2808 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2811 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2813 case 'localdayname':
2814 $value = $wgContLang->getWeekdayName( $localDayOfWeek +
1 );
2817 $value = $wgContLang->formatNum( $localYear, true );
2820 $value = $wgContLang->time( $localTimestamp, false, false );
2823 $value = $wgContLang->formatNum( $localHour, true );
2826 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2827 # int to remove the padding
2828 $value = $wgContLang->formatNum( (int)$localWeek );
2831 $value = $wgContLang->formatNum( $localDayOfWeek );
2833 case 'numberofarticles':
2834 $value = $wgContLang->formatNum( SiteStats
::articles() );
2836 case 'numberoffiles':
2837 $value = $wgContLang->formatNum( SiteStats
::images() );
2839 case 'numberofusers':
2840 $value = $wgContLang->formatNum( SiteStats
::users() );
2842 case 'numberofactiveusers':
2843 $value = $wgContLang->formatNum( SiteStats
::activeUsers() );
2845 case 'numberofpages':
2846 $value = $wgContLang->formatNum( SiteStats
::pages() );
2848 case 'numberofadmins':
2849 $value = $wgContLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2851 case 'numberofedits':
2852 $value = $wgContLang->formatNum( SiteStats
::edits() );
2854 case 'numberofviews':
2855 $value = $wgContLang->formatNum( SiteStats
::views() );
2857 case 'currenttimestamp':
2858 $value = wfTimestamp( TS_MW
, $ts );
2860 case 'localtimestamp':
2861 $value = $localTimestamp;
2863 case 'currentversion':
2864 $value = SpecialVersion
::getVersion();
2867 return $wgArticlePath;
2873 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2874 $serverName = parse_url( $wgServer, PHP_URL_HOST
);
2875 wfRestoreWarnings();
2876 return $serverName ?
$serverName : $wgServer;
2878 return $wgScriptPath;
2880 return $wgStylePath;
2881 case 'directionmark':
2882 return $wgContLang->getDirMark();
2883 case 'contentlanguage':
2884 global $wgLanguageCode;
2885 return $wgLanguageCode;
2888 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache
, &$index, &$ret, &$frame ) ) ) {
2896 $this->mVarCache
[$index] = $value;
2902 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2906 function initialiseVariables() {
2907 wfProfileIn( __METHOD__
);
2908 $variableIDs = MagicWord
::getVariableIDs();
2909 $substIDs = MagicWord
::getSubstIDs();
2911 $this->mVariables
= new MagicWordArray( $variableIDs );
2912 $this->mSubstWords
= new MagicWordArray( $substIDs );
2913 wfProfileOut( __METHOD__
);
2917 * Preprocess some wikitext and return the document tree.
2918 * This is the ghost of replace_variables().
2920 * @param $text String: The text to parse
2921 * @param $flags Integer: bitwise combination of:
2922 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2923 * included. Default is to assume a direct page view.
2925 * The generated DOM tree must depend only on the input text and the flags.
2926 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2928 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2929 * change in the DOM tree for a given text, must be passed through the section identifier
2930 * in the section edit link and thus back to extractSections().
2932 * The output of this function is currently only cached in process memory, but a persistent
2933 * cache may be implemented at a later date which takes further advantage of these strict
2934 * dependency requirements.
2938 function preprocessToDom( $text, $flags = 0 ) {
2939 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2944 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2946 public static function splitWhitespace( $s ) {
2947 $ltrimmed = ltrim( $s );
2948 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2949 $trimmed = rtrim( $ltrimmed );
2950 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2952 $w2 = substr( $ltrimmed, -$diff );
2956 return array( $w1, $trimmed, $w2 );
2960 * Replace magic variables, templates, and template arguments
2961 * with the appropriate text. Templates are substituted recursively,
2962 * taking care to avoid infinite loops.
2964 * Note that the substitution depends on value of $mOutputType:
2965 * self::OT_WIKI: only {{subst:}} templates
2966 * self::OT_PREPROCESS: templates but not extension tags
2967 * self::OT_HTML: all templates and extension tags
2969 * @param $text String: the text to transform
2970 * @param $frame PPFrame Object describing the arguments passed to the template.
2971 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2972 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2973 * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2976 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2977 # Is there any text? Also, Prevent too big inclusions!
2978 if ( strlen( $text ) < 1 ||
strlen( $text ) > $this->mOptions
->getMaxIncludeSize() ) {
2981 wfProfileIn( __METHOD__
);
2983 if ( $frame === false ) {
2984 $frame = $this->getPreprocessor()->newFrame();
2985 } elseif ( !( $frame instanceof PPFrame
) ) {
2986 wfDebug( __METHOD__
." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2987 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2990 $dom = $this->preprocessToDom( $text );
2991 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
2992 $text = $frame->expand( $dom, $flags );
2994 wfProfileOut( __METHOD__
);
2998 # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2999 static function createAssocArgs( $args ) {
3000 $assocArgs = array();
3002 foreach ( $args as $arg ) {
3003 $eqpos = strpos( $arg, '=' );
3004 if ( $eqpos === false ) {
3005 $assocArgs[$index++
] = $arg;
3007 $name = trim( substr( $arg, 0, $eqpos ) );
3008 $value = trim( substr( $arg, $eqpos+
1 ) );
3009 if ( $value === false ) {
3012 if ( $name !== false ) {
3013 $assocArgs[$name] = $value;
3022 * Warn the user when a parser limitation is reached
3023 * Will warn at most once the user per limitation type
3025 * @param $limitationType String: should be one of:
3026 * 'expensive-parserfunction' (corresponding messages:
3027 * 'expensive-parserfunction-warning',
3028 * 'expensive-parserfunction-category')
3029 * 'post-expand-template-argument' (corresponding messages:
3030 * 'post-expand-template-argument-warning',
3031 * 'post-expand-template-argument-category')
3032 * 'post-expand-template-inclusion' (corresponding messages:
3033 * 'post-expand-template-inclusion-warning',
3034 * 'post-expand-template-inclusion-category')
3035 * @param $current Current value
3036 * @param $max Maximum allowed, when an explicit limit has been
3037 * exceeded, provide the values (optional)
3039 function limitationWarn( $limitationType, $current=null, $max=null) {
3040 # does no harm if $current and $max are present but are unnecessary for the message
3041 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3042 $this->mOutput
->addWarning( $warning );
3043 $this->addTrackingCategory( "$limitationType-category" );
3047 * Return the text of a template, after recursively
3048 * replacing any variables or templates within the template.
3050 * @param $piece Array: the parts of the template
3051 * $piece['title']: the title, i.e. the part before the |
3052 * $piece['parts']: the parameter array
3053 * $piece['lineStart']: whether the brace was at the start of a line
3054 * @param $frame PPFrame The current frame, contains template arguments
3055 * @return String: the text of the template
3058 function braceSubstitution( $piece, $frame ) {
3059 global $wgContLang, $wgNonincludableNamespaces;
3060 wfProfileIn( __METHOD__
);
3061 wfProfileIn( __METHOD__
.'-setup' );
3064 $found = false; # $text has been filled
3065 $nowiki = false; # wiki markup in $text should be escaped
3066 $isHTML = false; # $text is HTML, armour it against wikitext transformation
3067 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3068 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
3069 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
3071 # Title object, where $text came from
3074 # $part1 is the bit before the first |, and must contain only title characters.
3075 # Various prefixes will be stripped from it later.
3076 $titleWithSpaces = $frame->expand( $piece['title'] );
3077 $part1 = trim( $titleWithSpaces );
3080 # Original title text preserved for various purposes
3081 $originalTitle = $part1;
3083 # $args is a list of argument nodes, starting from index 0, not including $part1
3084 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3085 $args = ( null == $piece['parts'] ) ?
array() : $piece['parts'];
3086 wfProfileOut( __METHOD__
.'-setup' );
3089 wfProfileIn( __METHOD__
.'-modifiers' );
3092 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3094 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3095 # Decide whether to expand template or keep wikitext as-is.
3096 if ( $this->ot
['wiki'] ) {
3097 if ( $substMatch === false ) {
3098 $literal = true; # literal when in PST with no prefix
3100 $literal = false; # expand when in PST with subst: or safesubst:
3103 if ( $substMatch == 'subst' ) {
3104 $literal = true; # literal when not in PST with plain subst:
3106 $literal = false; # expand when not in PST with safesubst: or no prefix
3110 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3117 if ( !$found && $args->getLength() == 0 ) {
3118 $id = $this->mVariables
->matchStartToEnd( $part1 );
3119 if ( $id !== false ) {
3120 $text = $this->getVariableValue( $id, $frame );
3121 if ( MagicWord
::getCacheTTL( $id ) > -1 ) {
3122 $this->mOutput
->updateCacheExpiry( MagicWord
::getCacheTTL( $id ) );
3128 # MSG, MSGNW and RAW
3131 $mwMsgnw = MagicWord
::get( 'msgnw' );
3132 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3135 # Remove obsolete MSG:
3136 $mwMsg = MagicWord
::get( 'msg' );
3137 $mwMsg->matchStartAndRemove( $part1 );
3141 $mwRaw = MagicWord
::get( 'raw' );
3142 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3143 $forceRawInterwiki = true;
3146 wfProfileOut( __METHOD__
.'-modifiers' );
3150 wfProfileIn( __METHOD__
. '-pfunc' );
3152 $colonPos = strpos( $part1, ':' );
3153 if ( $colonPos !== false ) {
3154 # Case sensitive functions
3155 $function = substr( $part1, 0, $colonPos );
3156 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3157 $function = $this->mFunctionSynonyms
[1][$function];
3159 # Case insensitive functions
3160 $function = $wgContLang->lc( $function );
3161 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3162 $function = $this->mFunctionSynonyms
[0][$function];
3168 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3169 $initialArgs = array( &$this );
3170 $funcArgs = array( trim( substr( $part1, $colonPos +
1 ) ) );
3171 if ( $flags & SFH_OBJECT_ARGS
) {
3172 # Add a frame parameter, and pass the arguments as an array
3173 $allArgs = $initialArgs;
3174 $allArgs[] = $frame;
3175 for ( $i = 0; $i < $args->getLength(); $i++
) {
3176 $funcArgs[] = $args->item( $i );
3178 $allArgs[] = $funcArgs;
3180 # Convert arguments to plain text
3181 for ( $i = 0; $i < $args->getLength(); $i++
) {
3182 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3184 $allArgs = array_merge( $initialArgs, $funcArgs );
3187 # Workaround for PHP bug 35229 and similar
3188 if ( !is_callable( $callback ) ) {
3189 wfProfileOut( __METHOD__
. '-pfunc' );
3190 wfProfileOut( __METHOD__
);
3191 throw new MWException( "Tag hook for $function is not callable\n" );
3193 $result = call_user_func_array( $callback, $allArgs );
3196 $preprocessFlags = 0;
3198 if ( is_array( $result ) ) {
3199 if ( isset( $result[0] ) ) {
3201 unset( $result[0] );
3204 # Extract flags into the local scope
3205 # This allows callers to set flags such as nowiki, found, etc.
3211 $text = $this->preprocessToDom( $text, $preprocessFlags );
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 ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3234 $wgContLang->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 && $title ) {
3249 wfProfileIn( __METHOD__
. '-loadtpl' );
3250 if ( !$title->isExternal() ) {
3251 if ( $title->getNamespace() == NS_SPECIAL
3252 && $this->mOptions
->getAllowSpecialInclusion()
3253 && $this->ot
['html'] )
3255 $text = SpecialPageFactory
::capturePath( $title );
3256 if ( is_string( $text ) ) {
3259 $this->disableCache();
3261 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3262 $found = false; # access denied
3263 wfDebug( __METHOD__
.": template inclusion denied for " . $title->getPrefixedDBkey() );
3265 list( $text, $title ) = $this->getTemplateDom( $title );
3266 if ( $text !== false ) {
3272 # If the title is valid but undisplayable, make a link to it
3273 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3274 $text = "[[:$titleText]]";
3277 } elseif ( $title->isTrans() ) {
3278 # Interwiki transclusion
3279 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3280 $text = $this->interwikiTransclude( $title, 'render' );
3283 $text = $this->interwikiTransclude( $title, 'raw' );
3284 # Preprocess it like a template
3285 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3291 # Do infinite loop check
3292 # This has to be done after redirect resolution to avoid infinite loops via redirects
3293 if ( !$frame->loopCheck( $title ) ) {
3295 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3296 wfDebug( __METHOD__
.": template loop broken at '$titleText'\n" );
3298 wfProfileOut( __METHOD__
. '-loadtpl' );
3301 # If we haven't found text to substitute by now, we're done
3302 # Recover the source wikitext and return it
3304 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3305 wfProfileOut( __METHOD__
);
3306 return array( 'object' => $text );
3309 # Expand DOM-style return values in a child frame
3310 if ( $isChildObj ) {
3311 # Clean up argument array
3312 $newFrame = $frame->newChild( $args, $title );
3315 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3316 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3317 # Expansion is eligible for the empty-frame cache
3318 if ( isset( $this->mTplExpandCache
[$titleText] ) ) {
3319 $text = $this->mTplExpandCache
[$titleText];
3321 $text = $newFrame->expand( $text );
3322 $this->mTplExpandCache
[$titleText] = $text;
3325 # Uncached expansion
3326 $text = $newFrame->expand( $text );
3329 if ( $isLocalObj && $nowiki ) {
3330 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3331 $isLocalObj = false;
3334 # Replace raw HTML by a placeholder
3335 # Add a blank line preceding, to prevent it from mucking up
3336 # immediately preceding headings
3338 $text = "\n\n" . $this->insertStripItem( $text );
3339 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3340 # Escape nowiki-style return values
3341 $text = wfEscapeWikiText( $text );
3342 } elseif ( is_string( $text )
3343 && !$piece['lineStart']
3344 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3346 # Bug 529: if the template begins with a table or block-level
3347 # element, it should be treated as beginning a new line.
3348 # This behaviour is somewhat controversial.
3349 $text = "\n" . $text;
3352 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3353 # Error, oversize inclusion
3354 if ( $titleText !== false ) {
3355 # Make a working, properly escaped link if possible (bug 23588)
3356 $text = "[[:$titleText]]";
3358 # This will probably not be a working link, but at least it may
3359 # provide some hint of where the problem is
3360 preg_replace( '/^:/', '', $originalTitle );
3361 $text = "[[:$originalTitle]]";
3363 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3364 $this->limitationWarn( 'post-expand-template-inclusion' );
3367 if ( $isLocalObj ) {
3368 $ret = array( 'object' => $text );
3370 $ret = array( 'text' => $text );
3373 wfProfileOut( __METHOD__
);
3378 * Get the semi-parsed DOM representation of a template with a given title,
3379 * and its redirect destination title. Cached.
3381 function getTemplateDom( $title ) {
3382 $cacheTitle = $title;
3383 $titleText = $title->getPrefixedDBkey();
3385 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3386 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3387 $title = Title
::makeTitle( $ns, $dbk );
3388 $titleText = $title->getPrefixedDBkey();
3390 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3391 return array( $this->mTplDomCache
[$titleText], $title );
3394 # Cache miss, go to the database
3395 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3397 if ( $text === false ) {
3398 $this->mTplDomCache
[$titleText] = false;
3399 return array( false, $title );
3402 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3403 $this->mTplDomCache
[ $titleText ] = $dom;
3405 if ( !$title->equals( $cacheTitle ) ) {
3406 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3407 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3410 return array( $dom, $title );
3414 * Fetch the unparsed text of a template and register a reference to it.
3415 * @param Title $title
3416 * @return Array ( string or false, Title )
3418 function fetchTemplateAndTitle( $title ) {
3419 $templateCb = $this->mOptions
->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3420 $stuff = call_user_func( $templateCb, $title, $this );
3421 $text = $stuff['text'];
3422 $finalTitle = isset( $stuff['finalTitle'] ) ?
$stuff['finalTitle'] : $title;
3423 if ( isset( $stuff['deps'] ) ) {
3424 foreach ( $stuff['deps'] as $dep ) {
3425 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3428 return array( $text, $finalTitle );
3432 * Fetch the unparsed text of a template and register a reference to it.
3433 * @param Title $title
3434 * @return mixed string or false
3436 function fetchTemplate( $title ) {
3437 $rv = $this->fetchTemplateAndTitle( $title );
3442 * Static function to get a template
3443 * Can be overridden via ParserOptions::setTemplateCallback().
3445 static function statelessFetchTemplate( $title, $parser=false ) {
3446 $text = $skip = false;
3447 $finalTitle = $title;
3450 # Loop to fetch the article, with up to 1 redirect
3451 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3452 # Give extensions a chance to select the revision instead
3453 $id = false; # Assume current
3454 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3455 array( $parser, $title, &$skip, &$id ) );
3461 'page_id' => $title->getArticleID(),
3468 ? Revision
::newFromId( $id )
3469 : Revision
::newFromTitle( $title );
3470 $rev_id = $rev ?
$rev->getId() : 0;
3471 # If there is no current revision, there is no page
3472 if ( $id === false && !$rev ) {
3473 $linkCache = LinkCache
::singleton();
3474 $linkCache->addBadLinkObj( $title );
3479 'page_id' => $title->getArticleID(),
3480 'rev_id' => $rev_id );
3481 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3482 # We fetched a rev from a different title; register it too...
3484 'title' => $rev->getTitle(),
3485 'page_id' => $rev->getPage(),
3486 'rev_id' => $rev_id );
3490 $text = $rev->getText();
3491 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3493 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3494 if ( !$message->exists() ) {
3498 $text = $message->plain();
3502 if ( $text === false ) {
3506 $finalTitle = $title;
3507 $title = Title
::newFromRedirect( $text );
3511 'finalTitle' => $finalTitle,
3516 * Fetch a file and its title and register a reference to it.
3517 * @param Title $title
3518 * @param string $time MW timestamp
3519 * @param string $sha1 base 36 SHA-1
3520 * @return mixed File or false
3522 function fetchFile( $title, $time = false, $sha1 = false ) {
3523 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3528 * Fetch a file and its title and register a reference to it.
3529 * @param Title $title
3530 * @param string $time MW timestamp
3531 * @param string $sha1 base 36 SHA-1
3532 * @return Array ( File or false, Title of file )
3534 function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3535 if ( $time === '0' ) {
3536 $file = false; // broken thumbnail forced by hook
3537 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3538 $file = RepoGroup
::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3539 } else { // get by (name,timestamp)
3540 $file = wfFindFile( $title, array( 'time' => $time ) );
3542 $time = $file ?
$file->getTimestamp() : false;
3543 $sha1 = $file ?
$file->getSha1() : false;
3544 # Register the file as a dependency...
3545 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3546 if ( $file && !$title->equals( $file->getTitle() ) ) {
3547 # We fetched a rev from a different title; register it too...
3548 $this->mOutput
->addImage( $file->getTitle()->getDBkey(), $time, $sha1 );
3549 # Update fetched file title
3550 $title = $file->getTitle();
3552 return array( $file, $title );
3556 * Transclude an interwiki link.
3558 function interwikiTransclude( $title, $action ) {
3559 global $wgEnableScaryTranscluding;
3561 if ( !$wgEnableScaryTranscluding ) {
3562 return wfMsgForContent('scarytranscludedisabled');
3565 $url = $title->getFullUrl( "action=$action" );
3567 if ( strlen( $url ) > 255 ) {
3568 return wfMsgForContent( 'scarytranscludetoolong' );
3570 return $this->fetchScaryTemplateMaybeFromCache( $url );
3573 function fetchScaryTemplateMaybeFromCache( $url ) {
3574 global $wgTranscludeCacheExpiry;
3575 $dbr = wfGetDB( DB_SLAVE
);
3576 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3577 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3578 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3580 return $obj->tc_contents
;
3583 $text = Http
::get( $url );
3585 return wfMsgForContent( 'scarytranscludefailed', $url );
3588 $dbw = wfGetDB( DB_MASTER
);
3589 $dbw->replace( 'transcache', array('tc_url'), array(
3591 'tc_time' => $dbw->timestamp( time() ),
3592 'tc_contents' => $text)
3599 * Triple brace replacement -- used for template arguments
3602 function argSubstitution( $piece, $frame ) {
3603 wfProfileIn( __METHOD__
);
3606 $parts = $piece['parts'];
3607 $nameWithSpaces = $frame->expand( $piece['title'] );
3608 $argName = trim( $nameWithSpaces );
3610 $text = $frame->getArgument( $argName );
3611 if ( $text === false && $parts->getLength() > 0
3615 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3618 # No match in frame, use the supplied default
3619 $object = $parts->item( 0 )->getChildren();
3621 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3622 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3623 $this->limitationWarn( 'post-expand-template-argument' );
3626 if ( $text === false && $object === false ) {
3628 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3630 if ( $error !== false ) {
3633 if ( $object !== false ) {
3634 $ret = array( 'object' => $object );
3636 $ret = array( 'text' => $text );
3639 wfProfileOut( __METHOD__
);
3644 * Return the text to be used for a given extension tag.
3645 * This is the ghost of strip().
3647 * @param $params Associative array of parameters:
3648 * name PPNode for the tag name
3649 * attr PPNode for unparsed text where tag attributes are thought to be
3650 * attributes Optional associative array of parsed attributes
3651 * inner Contents of extension element
3652 * noClose Original text did not have a close tag
3653 * @param $frame PPFrame
3655 function extensionSubstitution( $params, $frame ) {
3656 $name = $frame->expand( $params['name'] );
3657 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3658 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3659 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3661 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower($name)] ) &&
3662 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3663 if ( $isFunctionTag ) {
3664 $markerType = 'none';
3666 $markerType = 'general';
3668 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3669 $name = strtolower( $name );
3670 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3671 if ( isset( $params['attributes'] ) ) {
3672 $attributes = $attributes +
$params['attributes'];
3675 if ( isset( $this->mTagHooks
[$name] ) ) {
3676 # Workaround for PHP bug 35229 and similar
3677 if ( !is_callable( $this->mTagHooks
[$name] ) ) {
3678 throw new MWException( "Tag hook for $name is not callable\n" );
3680 $output = call_user_func_array( $this->mTagHooks
[$name],
3681 array( $content, $attributes, $this, $frame ) );
3682 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
3683 list( $callback, $flags ) = $this->mFunctionTagHooks
[$name];
3684 if ( !is_callable( $callback ) ) {
3685 throw new MWException( "Tag hook for $name is not callable\n" );
3688 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3690 $output = '<span class="error">Invalid tag extension name: ' .
3691 htmlspecialchars( $name ) . '</span>';
3694 if ( is_array( $output ) ) {
3695 # Extract flags to local scope (to override $markerType)
3697 $output = $flags[0];
3702 if ( is_null( $attrText ) ) {
3705 if ( isset( $params['attributes'] ) ) {
3706 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3707 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3708 htmlspecialchars( $attrValue ) . '"';
3711 if ( $content === null ) {
3712 $output = "<$name$attrText/>";
3714 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
3715 $output = "<$name$attrText>$content$close";
3719 if ( $markerType === 'none' ) {
3721 } elseif ( $markerType === 'nowiki' ) {
3722 $this->mStripState
->addNoWiki( $marker, $output );
3723 } elseif ( $markerType === 'general' ) {
3724 $this->mStripState
->addGeneral( $marker, $output );
3726 throw new MWException( __METHOD__
.': invalid marker type' );
3732 * Increment an include size counter
3734 * @param $type String: the type of expansion
3735 * @param $size Integer: the size of the text
3736 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3738 function incrementIncludeSize( $type, $size ) {
3739 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
3742 $this->mIncludeSizes
[$type] +
= $size;
3748 * Increment the expensive function count
3750 * @return Boolean: false if the limit has been exceeded
3752 function incrementExpensiveFunctionCount() {
3753 global $wgExpensiveParserFunctionLimit;
3754 $this->mExpensiveFunctionCount++
;
3755 if ( $this->mExpensiveFunctionCount
<= $wgExpensiveParserFunctionLimit ) {
3762 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3763 * Fills $this->mDoubleUnderscores, returns the modified text
3765 function doDoubleUnderscore( $text ) {
3766 wfProfileIn( __METHOD__
);
3768 # The position of __TOC__ needs to be recorded
3769 $mw = MagicWord
::get( 'toc' );
3770 if ( $mw->match( $text ) ) {
3771 $this->mShowToc
= true;
3772 $this->mForceTocPosition
= true;
3774 # Set a placeholder. At the end we'll fill it in with the TOC.
3775 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3777 # Only keep the first one.
3778 $text = $mw->replace( '', $text );
3781 # Now match and remove the rest of them
3782 $mwa = MagicWord
::getDoubleUnderscoreArray();
3783 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
3785 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
3786 $this->mOutput
->mNoGallery
= true;
3788 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
3789 $this->mShowToc
= false;
3791 if ( isset( $this->mDoubleUnderscores
['hiddencat'] ) && $this->mTitle
->getNamespace() == NS_CATEGORY
) {
3792 $this->addTrackingCategory( 'hidden-category-category' );
3794 # (bug 8068) Allow control over whether robots index a page.
3796 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3797 # is not desirable, the last one on the page should win.
3798 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
3799 $this->mOutput
->setIndexPolicy( 'noindex' );
3800 $this->addTrackingCategory( 'noindex-category' );
3802 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
3803 $this->mOutput
->setIndexPolicy( 'index' );
3804 $this->addTrackingCategory( 'index-category' );
3807 # Cache all double underscores in the database
3808 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
3809 $this->mOutput
->setProperty( $key, '' );
3812 wfProfileOut( __METHOD__
);
3817 * Add a tracking category, getting the title from a system message,
3818 * or print a debug message if the title is invalid.
3820 * @param $msg String: message key
3821 * @return Boolean: whether the addition was successful
3823 protected function addTrackingCategory( $msg ) {
3824 $cat = wfMsgForContent( $msg );
3826 # Allow tracking categories to be disabled by setting them to "-"
3827 if ( $cat === '-' ) {
3831 $containerCategory = Title
::makeTitleSafe( NS_CATEGORY
, $cat );
3832 if ( $containerCategory ) {
3833 $this->mOutput
->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3836 wfDebug( __METHOD__
.": [[MediaWiki:$msg]] is not a valid title!\n" );
3842 * This function accomplishes several tasks:
3843 * 1) Auto-number headings if that option is enabled
3844 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3845 * 3) Add a Table of contents on the top for users who have enabled the option
3846 * 4) Auto-anchor headings
3848 * It loops through all headlines, collects the necessary data, then splits up the
3849 * string and re-inserts the newly formatted headlines.
3851 * @param $text String
3852 * @param $origText String: original, untouched wikitext
3853 * @param $isMain Boolean
3856 function formatHeadings( $text, $origText, $isMain=true ) {
3857 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3859 # Inhibit editsection links if requested in the page
3860 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
3863 $showEditLink = $this->mOptions
->getEditSection();
3865 if ( $showEditLink ) {
3866 $this->mOutput
->setEditSectionTokens( true );
3869 # Get all headlines for numbering them and adding funky stuff like [edit]
3870 # links - this is for later, but we need the number of headlines right now
3872 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3874 # if there are fewer than 4 headlines in the article, do not show TOC
3875 # unless it's been explicitly enabled.
3876 $enoughToc = $this->mShowToc
&&
3877 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
3879 # Allow user to stipulate that a page should have a "new section"
3880 # link added via __NEWSECTIONLINK__
3881 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
3882 $this->mOutput
->setNewSection( true );
3885 # Allow user to remove the "new section"
3886 # link via __NONEWSECTIONLINK__
3887 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
3888 $this->mOutput
->hideNewSection( true );
3891 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3892 # override above conditions and always show TOC above first header
3893 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
3894 $this->mShowToc
= true;
3902 # Ugh .. the TOC should have neat indentation levels which can be
3903 # passed to the skin functions. These are determined here
3907 $sublevelCount = array();
3908 $levelCount = array();
3913 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self
::MARKER_SUFFIX
;
3914 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
3915 $oldType = $this->mOutputType
;
3916 $this->setOutputType( self
::OT_WIKI
);
3917 $frame = $this->getPreprocessor()->newFrame();
3918 $root = $this->preprocessToDom( $origText );
3919 $node = $root->getFirstChild();
3924 foreach ( $matches[3] as $headline ) {
3925 $isTemplate = false;
3927 $sectionIndex = false;
3929 $markerMatches = array();
3930 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3931 $serial = $markerMatches[1];
3932 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
3933 $isTemplate = ( $titleText != $baseTitleText );
3934 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3938 $prevlevel = $level;
3940 $level = $matches[1][$headlineCount];
3942 if ( $level > $prevlevel ) {
3943 # Increase TOC level
3945 $sublevelCount[$toclevel] = 0;
3946 if ( $toclevel<$wgMaxTocLevel ) {
3947 $prevtoclevel = $toclevel;
3948 $toc .= Linker
::tocIndent();
3951 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3952 # Decrease TOC level, find level to jump to
3954 for ( $i = $toclevel; $i > 0; $i-- ) {
3955 if ( $levelCount[$i] == $level ) {
3956 # Found last matching level
3959 } elseif ( $levelCount[$i] < $level ) {
3960 # Found first matching level below current level
3968 if ( $toclevel<$wgMaxTocLevel ) {
3969 if ( $prevtoclevel < $wgMaxTocLevel ) {
3970 # Unindent only if the previous toc level was shown :p
3971 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
3972 $prevtoclevel = $toclevel;
3974 $toc .= Linker
::tocLineEnd();
3978 # No change in level, end TOC line
3979 if ( $toclevel<$wgMaxTocLevel ) {
3980 $toc .= Linker
::tocLineEnd();
3984 $levelCount[$toclevel] = $level;
3986 # count number of headlines for each level
3987 @$sublevelCount[$toclevel]++
;
3989 for( $i = 1; $i <= $toclevel; $i++
) {
3990 if ( !empty( $sublevelCount[$i] ) ) {
3994 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3999 # The safe header is a version of the header text safe to use for links
4000 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4001 $safeHeadline = $this->mStripState
->unstripBoth( $headline );
4003 # Remove link placeholders by the link text.
4004 # <!--LINK number-->
4006 # link text with suffix
4007 $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4009 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4010 $tocline = preg_replace(
4011 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4012 array( '', '<$1>' ),
4015 $tocline = trim( $tocline );
4017 # For the anchor, strip out HTML-y stuff period
4018 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4019 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4021 # Save headline for section edit hint before it's escaped
4022 $headlineHint = $safeHeadline;
4024 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4025 # For reverse compatibility, provide an id that's
4026 # HTML4-compatible, like we used to.
4028 # It may be worth noting, academically, that it's possible for
4029 # the legacy anchor to conflict with a non-legacy headline
4030 # anchor on the page. In this case likely the "correct" thing
4031 # would be to either drop the legacy anchors or make sure
4032 # they're numbered first. However, this would require people
4033 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4034 # manually, so let's not bother worrying about it.
4035 $legacyHeadline = Sanitizer
::escapeId( $safeHeadline,
4036 array( 'noninitial', 'legacy' ) );
4037 $safeHeadline = Sanitizer
::escapeId( $safeHeadline );
4039 if ( $legacyHeadline == $safeHeadline ) {
4040 # No reason to have both (in fact, we can't)
4041 $legacyHeadline = false;
4044 $legacyHeadline = false;
4045 $safeHeadline = Sanitizer
::escapeId( $safeHeadline,
4049 # HTML names must be case-insensitively unique (bug 10721).
4050 # This does not apply to Unicode characters per
4051 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4052 # FIXME: We may be changing them depending on the current locale.
4053 $arrayKey = strtolower( $safeHeadline );
4054 if ( $legacyHeadline === false ) {
4055 $legacyArrayKey = false;
4057 $legacyArrayKey = strtolower( $legacyHeadline );
4060 # count how many in assoc. array so we can track dupes in anchors
4061 if ( isset( $refers[$arrayKey] ) ) {
4062 $refers[$arrayKey]++
;
4064 $refers[$arrayKey] = 1;
4066 if ( isset( $refers[$legacyArrayKey] ) ) {
4067 $refers[$legacyArrayKey]++
;
4069 $refers[$legacyArrayKey] = 1;
4072 # Don't number the heading if it is the only one (looks silly)
4073 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4074 # the two are different if the line contains a link
4075 $headline = $numbering . ' ' . $headline;
4078 # Create the anchor for linking from the TOC to the section
4079 $anchor = $safeHeadline;
4080 $legacyAnchor = $legacyHeadline;
4081 if ( $refers[$arrayKey] > 1 ) {
4082 $anchor .= '_' . $refers[$arrayKey];
4084 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4085 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4087 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) ||
$toclevel < $wgMaxTocLevel ) ) {
4088 $toc .= Linker
::tocLine( $anchor, $tocline,
4089 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4092 # Add the section to the section tree
4093 # Find the DOM node for this header
4094 while ( $node && !$isTemplate ) {
4095 if ( $node->getName() === 'h' ) {
4096 $bits = $node->splitHeading();
4097 if ( $bits['i'] == $sectionIndex ) {
4101 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4102 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4103 $node = $node->getNextSibling();
4106 'toclevel' => $toclevel,
4109 'number' => $numbering,
4110 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4111 'fromtitle' => $titleText,
4112 'byteoffset' => ( $isTemplate ?
null : $byteOffset ),
4113 'anchor' => $anchor,
4116 # give headline the correct <h#> tag
4117 if ( $showEditLink && $sectionIndex !== false ) {
4118 // Output edit section links as markers with styles that can be customized by skins
4119 if ( $isTemplate ) {
4120 # Put a T flag in the section identifier, to indicate to extractSections()
4121 # that sections inside <includeonly> should be counted.
4122 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4124 $editlinkArgs = array( $this->mTitle
->getPrefixedText(), $sectionIndex, $headlineHint );
4126 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4127 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4128 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4129 // so we don't have to worry about a user trying to input one of these markers directly.
4130 // We use a page and section attribute to stop the language converter from converting these important bits
4131 // of data, but put the headline hint inside a content block because the language converter is supposed to
4132 // be able to convert that piece of data.
4133 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4134 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4135 if ( isset($editlinkArgs[2]) ) {
4136 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4143 $head[$headlineCount] = Linker
::makeHeadline( $level,
4144 $matches['attrib'][$headlineCount], $anchor, $headline,
4145 $editlink, $legacyAnchor );
4150 $this->setOutputType( $oldType );
4152 # Never ever show TOC if no headers
4153 if ( $numVisible < 1 ) {
4158 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4159 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4161 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLang() );
4162 $this->mOutput
->setTOCHTML( $toc );
4166 $this->mOutput
->setSections( $tocraw );
4169 # split up and insert constructed headlines
4171 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4174 foreach ( $blocks as $block ) {
4175 if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4176 # This is the [edit] link that appears for the top block of text when
4177 # section editing is enabled
4179 # Disabled because it broke block formatting
4180 # For example, a bullet point in the top line
4181 # $full .= $sk->editSectionLink(0);
4184 if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition
) {
4185 # Top anchor now in skin
4189 if ( !empty( $head[$i] ) ) {
4194 if ( $this->mForceTocPosition
) {
4195 return str_replace( '<!--MWTOC-->', $toc, $full );
4202 * Transform wiki markup when saving a page by doing \r\n -> \n
4203 * conversion, substitting signatures, {{subst:}} templates, etc.
4205 * @param $text String: the text to transform
4206 * @param $title Title: the Title object for the current article
4207 * @param $user User: the User object describing the current user
4208 * @param $options ParserOptions: parsing options
4209 * @param $clearState Boolean: whether to clear the parser state first
4210 * @return String: the altered wiki markup
4212 public function preSaveTransform( $text, Title
$title, User
$user, ParserOptions
$options, $clearState = true ) {
4213 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4214 $this->setUser( $user );
4219 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4220 if( $options->getPreSaveTransform() ) {
4221 $text = $this->pstPass2( $text, $user );
4223 $text = $this->mStripState
->unstripBoth( $text );
4225 $this->setUser( null ); #Reset
4231 * Pre-save transform helper function
4234 function pstPass2( $text, $user ) {
4235 global $wgContLang, $wgLocaltimezone;
4237 # Note: This is the timestamp saved as hardcoded wikitext to
4238 # the database, we use $wgContLang here in order to give
4239 # everyone the same signature and use the default one rather
4240 # than the one selected in each user's preferences.
4241 # (see also bug 12815)
4242 $ts = $this->mOptions
->getTimestamp();
4243 if ( isset( $wgLocaltimezone ) ) {
4244 $tz = $wgLocaltimezone;
4246 $tz = date_default_timezone_get();
4249 $unixts = wfTimestamp( TS_UNIX
, $ts );
4250 $oldtz = date_default_timezone_get();
4251 date_default_timezone_set( $tz );
4252 $ts = date( 'YmdHis', $unixts );
4253 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4255 # Allow translation of timezones through wiki. date() can return
4256 # whatever crap the system uses, localised or not, so we cannot
4257 # ship premade translations.
4258 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4259 $msg = wfMessage( $key )->inContentLanguage();
4260 if ( $msg->exists() ) {
4261 $tzMsg = $msg->text();
4264 date_default_timezone_set( $oldtz );
4266 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4268 # Variable replacement
4269 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4270 $text = $this->replaceVariables( $text );
4272 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4273 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4276 $sigText = $this->getUserSig( $user );
4277 $text = strtr( $text, array(
4279 '~~~~' => "$sigText $d",
4283 # Context links: [[|name]] and [[name (context)|]]
4284 global $wgLegalTitleChars;
4285 $tc = "[$wgLegalTitleChars]";
4286 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4288 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4289 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(($tc+))\\|]]/"; # [[ns:page(context)|]]
4290 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4291 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4293 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4294 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4295 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4296 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4298 $t = $this->mTitle
->getText();
4300 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4301 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4302 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4303 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4305 # if there's no context, don't bother duplicating the title
4306 $text = preg_replace( $p2, '[[\\1]]', $text );
4309 # Trim trailing whitespace
4310 $text = rtrim( $text );
4316 * Fetch the user's signature text, if any, and normalize to
4317 * validated, ready-to-insert wikitext.
4318 * If you have pre-fetched the nickname or the fancySig option, you can
4319 * specify them here to save a database query.
4320 * Do not reuse this parser instance after calling getUserSig(),
4321 * as it may have changed if it's the $wgParser.
4324 * @param $nickname String: nickname to use or false to use user's default nickname
4325 * @param $fancySig Boolean: whether the nicknname is the complete signature
4326 * or null to use default value
4329 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4330 global $wgMaxSigChars;
4332 $username = $user->getName();
4334 # If not given, retrieve from the user object.
4335 if ( $nickname === false )
4336 $nickname = $user->getOption( 'nickname' );
4338 if ( is_null( $fancySig ) ) {
4339 $fancySig = $user->getBoolOption( 'fancysig' );
4342 $nickname = $nickname == null ?
$username : $nickname;
4344 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4345 $nickname = $username;
4346 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4347 } elseif ( $fancySig !== false ) {
4348 # Sig. might contain markup; validate this
4349 if ( $this->validateSig( $nickname ) !== false ) {
4350 # Validated; clean up (if needed) and return it
4351 return $this->cleanSig( $nickname, true );
4353 # Failed to validate; fall back to the default
4354 $nickname = $username;
4355 wfDebug( __METHOD__
.": $username has bad XML tags in signature.\n" );
4359 # Make sure nickname doesnt get a sig in a sig
4360 $nickname = $this->cleanSigInSig( $nickname );
4362 # If we're still here, make it a link to the user page
4363 $userText = wfEscapeWikiText( $username );
4364 $nickText = wfEscapeWikiText( $nickname );
4365 if ( $user->isAnon() ) {
4366 return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4368 return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4373 * Check that the user's signature contains no bad XML
4375 * @param $text String
4376 * @return mixed An expanded string, or false if invalid.
4378 function validateSig( $text ) {
4379 return( Xml
::isWellFormedXmlFragment( $text ) ?
$text : false );
4383 * Clean up signature text
4385 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4386 * 2) Substitute all transclusions
4388 * @param $text String
4389 * @param $parsing Whether we're cleaning (preferences save) or parsing
4390 * @return String: signature text
4392 function cleanSig( $text, $parsing = false ) {
4395 $this->mOptions
= new ParserOptions
;
4396 $this->clearState();
4397 $this->setTitle( $wgTitle );
4398 $this->setOutputType
= self
::OT_PREPROCESS
;
4401 # Option to disable this feature
4402 if ( !$this->mOptions
->getCleanSignatures() ) {
4406 # FIXME: regex doesn't respect extension tags or nowiki
4407 # => Move this logic to braceSubstitution()
4408 $substWord = MagicWord
::get( 'subst' );
4409 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4410 $substText = '{{' . $substWord->getSynonym( 0 );
4412 $text = preg_replace( $substRegex, $substText, $text );
4413 $text = $this->cleanSigInSig( $text );
4414 $dom = $this->preprocessToDom( $text );
4415 $frame = $this->getPreprocessor()->newFrame();
4416 $text = $frame->expand( $dom );
4419 $text = $this->mStripState
->unstripBoth( $text );
4426 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4428 * @param $text String
4429 * @return String: signature text with /~{3,5}/ removed
4431 function cleanSigInSig( $text ) {
4432 $text = preg_replace( '/~{3,5}/', '', $text );
4437 * Set up some variables which are usually set up in parse()
4438 * so that an external function can call some class members with confidence
4440 public function startExternalParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4441 $this->startParse( $title, $options, $outputType, $clearState );
4444 private function startParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4445 $this->setTitle( $title );
4446 $this->mOptions
= $options;
4447 $this->setOutputType( $outputType );
4448 if ( $clearState ) {
4449 $this->clearState();
4454 * Wrapper for preprocess()
4456 * @param $text String: the text to preprocess
4457 * @param $options ParserOptions: options
4458 * @param $title Title object or null to use $wgTitle
4461 public function transformMsg( $text, $options, $title = null ) {
4462 static $executing = false;
4464 # Guard against infinite recursion
4470 wfProfileIn( __METHOD__
);
4476 # It's not uncommon having a null $wgTitle in scripts. See r80898
4477 # Create a ghost title in such case
4478 $title = Title
::newFromText( 'Dwimmerlaik' );
4480 $text = $this->preprocess( $text, $title, $options );
4483 wfProfileOut( __METHOD__
);
4488 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4489 * The callback should have the following form:
4490 * function myParserHook( $text, $params, $parser ) { ... }
4492 * Transform and return $text. Use $parser for any required context, e.g. use
4493 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4495 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4496 * @param $callback Mixed: the callback function (and object) to use for the tag
4497 * @return The old value of the mTagHooks array associated with the hook
4499 public function setHook( $tag, $callback ) {
4500 $tag = strtolower( $tag );
4501 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4502 $oldVal = isset( $this->mTagHooks
[$tag] ) ?
$this->mTagHooks
[$tag] : null;
4503 $this->mTagHooks
[$tag] = $callback;
4504 if ( !in_array( $tag, $this->mStripList
) ) {
4505 $this->mStripList
[] = $tag;
4511 function setTransparentTagHook( $tag, $callback ) {
4512 $tag = strtolower( $tag );
4513 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4514 $oldVal = isset( $this->mTransparentTagHooks
[$tag] ) ?
$this->mTransparentTagHooks
[$tag] : null;
4515 $this->mTransparentTagHooks
[$tag] = $callback;
4521 * Remove all tag hooks
4523 function clearTagHooks() {
4524 $this->mTagHooks
= array();
4525 $this->mStripList
= $this->mDefaultStripList
;
4529 * Create a function, e.g. {{sum:1|2|3}}
4530 * The callback function should have the form:
4531 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4533 * Or with SFH_OBJECT_ARGS:
4534 * function myParserFunction( $parser, $frame, $args ) { ... }
4536 * The callback may either return the text result of the function, or an array with the text
4537 * in element 0, and a number of flags in the other elements. The names of the flags are
4538 * specified in the keys. Valid flags are:
4539 * found The text returned is valid, stop processing the template. This
4541 * nowiki Wiki markup in the return value should be escaped
4542 * isHTML The returned text is HTML, armour it against wikitext transformation
4544 * @param $id String: The magic word ID
4545 * @param $callback Mixed: the callback function (and object) to use
4546 * @param $flags Integer: a combination of the following flags:
4547 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4549 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4550 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4551 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4552 * the arguments, and to control the way they are expanded.
4554 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4555 * arguments, for instance:
4556 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4558 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4559 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4560 * working if/when this is changed.
4562 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4565 * Please read the documentation in includes/parser/Preprocessor.php for more information
4566 * about the methods available in PPFrame and PPNode.
4568 * @return The old callback function for this name, if any
4570 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4573 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4574 $this->mFunctionHooks
[$id] = array( $callback, $flags );
4576 # Add to function cache
4577 $mw = MagicWord
::get( $id );
4579 throw new MWException( __METHOD__
.'() expecting a magic word identifier.' );
4581 $synonyms = $mw->getSynonyms();
4582 $sensitive = intval( $mw->isCaseSensitive() );
4584 foreach ( $synonyms as $syn ) {
4586 if ( !$sensitive ) {
4587 $syn = $wgContLang->lc( $syn );
4590 if ( !( $flags & SFH_NO_HASH
) ) {
4593 # Remove trailing colon
4594 if ( substr( $syn, -1, 1 ) === ':' ) {
4595 $syn = substr( $syn, 0, -1 );
4597 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4603 * Get all registered function hook identifiers
4607 function getFunctionHooks() {
4608 return array_keys( $this->mFunctionHooks
);
4612 * Create a tag function, e.g. <test>some stuff</test>.
4613 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4614 * Unlike parser functions, their content is not preprocessed.
4616 function setFunctionTagHook( $tag, $callback, $flags ) {
4617 $tag = strtolower( $tag );
4618 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4619 $old = isset( $this->mFunctionTagHooks
[$tag] ) ?
4620 $this->mFunctionTagHooks
[$tag] : null;
4621 $this->mFunctionTagHooks
[$tag] = array( $callback, $flags );
4623 if ( !in_array( $tag, $this->mStripList
) ) {
4624 $this->mStripList
[] = $tag;
4631 * FIXME: update documentation. makeLinkObj() is deprecated.
4632 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4633 * Placeholders created in Skin::makeLinkObj()
4634 * Returns an array of link CSS classes, indexed by PDBK.
4636 function replaceLinkHolders( &$text, $options = 0 ) {
4637 return $this->mLinkHolders
->replace( $text );
4641 * Replace <!--LINK--> link placeholders with plain text of links
4642 * (not HTML-formatted).
4644 * @param $text String
4647 function replaceLinkHoldersText( $text ) {
4648 return $this->mLinkHolders
->replaceText( $text );
4652 * Renders an image gallery from a text with one line per image.
4653 * text labels may be given by using |-style alternative text. E.g.
4654 * Image:one.jpg|The number "1"
4655 * Image:tree.jpg|A tree
4656 * given as text will return the HTML of a gallery with two images,
4657 * labeled 'The number "1"' and
4660 function renderImageGallery( $text, $params ) {
4661 $ig = new ImageGallery();
4662 $ig->setContextTitle( $this->mTitle
);
4663 $ig->setShowBytes( false );
4664 $ig->setShowFilename( false );
4665 $ig->setParser( $this );
4666 $ig->setHideBadImages();
4667 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'table' ) );
4669 if ( isset( $params['showfilename'] ) ) {
4670 $ig->setShowFilename( true );
4672 $ig->setShowFilename( false );
4674 if ( isset( $params['caption'] ) ) {
4675 $caption = $params['caption'];
4676 $caption = htmlspecialchars( $caption );
4677 $caption = $this->replaceInternalLinks( $caption );
4678 $ig->setCaptionHtml( $caption );
4680 if ( isset( $params['perrow'] ) ) {
4681 $ig->setPerRow( $params['perrow'] );
4683 if ( isset( $params['widths'] ) ) {
4684 $ig->setWidths( $params['widths'] );
4686 if ( isset( $params['heights'] ) ) {
4687 $ig->setHeights( $params['heights'] );
4690 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4692 $lines = StringUtils
::explode( "\n", $text );
4693 foreach ( $lines as $line ) {
4694 # match lines like these:
4695 # Image:someimage.jpg|This is some image
4697 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4699 if ( count( $matches ) == 0 ) {
4703 if ( strpos( $matches[0], '%' ) !== false ) {
4704 $matches[1] = rawurldecode( $matches[1] );
4706 $tp = Title
::newFromText( $matches[1], NS_FILE
);
4708 if ( is_null( $nt ) ) {
4709 # Bogus title. Ignore these so we don't bomb out later.
4712 if ( isset( $matches[3] ) ) {
4713 $label = $matches[3];
4718 $html = $this->recursiveTagParse( trim( $label ) );
4720 $ig->add( $nt, $html );
4722 return $ig->toHTML();
4725 function getImageParams( $handler ) {
4727 $handlerClass = get_class( $handler );
4731 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
4732 # Initialise static lists
4733 static $internalParamNames = array(
4734 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4735 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4736 'bottom', 'text-bottom' ),
4737 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4738 'upright', 'border', 'link', 'alt' ),
4740 static $internalParamMap;
4741 if ( !$internalParamMap ) {
4742 $internalParamMap = array();
4743 foreach ( $internalParamNames as $type => $names ) {
4744 foreach ( $names as $name ) {
4745 $magicName = str_replace( '-', '_', "img_$name" );
4746 $internalParamMap[$magicName] = array( $type, $name );
4751 # Add handler params
4752 $paramMap = $internalParamMap;
4754 $handlerParamMap = $handler->getParamMap();
4755 foreach ( $handlerParamMap as $magic => $paramName ) {
4756 $paramMap[$magic] = array( 'handler', $paramName );
4759 $this->mImageParams
[$handlerClass] = $paramMap;
4760 $this->mImageParamsMagicArray
[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4762 return array( $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] );
4766 * Parse image options text and use it to make an image
4768 * @param $title Title
4769 * @param $options String
4770 * @param $holders LinkHolderArray
4771 * @return string HTML
4773 function makeImage( $title, $options, $holders = false ) {
4774 # Check if the options text is of the form "options|alt text"
4776 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4777 # * left no resizing, just left align. label is used for alt= only
4778 # * right same, but right aligned
4779 # * none same, but not aligned
4780 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4781 # * center center the image
4782 # * frame Keep original image size, no magnify-button.
4783 # * framed Same as "frame"
4784 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4785 # * upright reduce width for upright images, rounded to full __0 px
4786 # * border draw a 1px border around the image
4787 # * alt Text for HTML alt attribute (defaults to empty)
4788 # * link Set the target of the image link. Can be external, interwiki, or local
4789 # vertical-align values (no % or length right now):
4799 $parts = StringUtils
::explode( "|", $options );
4801 # Give extensions a chance to select the file revision for us
4802 $time = $sha1 = $descQuery = false;
4803 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4804 array( $this, $title, &$time, &$sha1, &$descQuery ) );
4805 # Fetch and register the file (file title may be different via hooks)
4806 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4809 $handler = $file ?
$file->getHandler() : false;
4811 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4814 $this->addTrackingCategory( 'broken-file-category' );
4817 # Process the input parameters
4819 $params = array( 'frame' => array(), 'handler' => array(),
4820 'horizAlign' => array(), 'vertAlign' => array() );
4821 foreach ( $parts as $part ) {
4822 $part = trim( $part );
4823 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4825 if ( isset( $paramMap[$magicName] ) ) {
4826 list( $type, $paramName ) = $paramMap[$magicName];
4828 # Special case; width and height come in one variable together
4829 if ( $type === 'handler' && $paramName === 'width' ) {
4831 # (bug 13500) In both cases (width/height and width only),
4832 # permit trailing "px" for backward compatibility.
4833 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4834 $width = intval( $m[1] );
4835 $height = intval( $m[2] );
4836 if ( $handler->validateParam( 'width', $width ) ) {
4837 $params[$type]['width'] = $width;
4840 if ( $handler->validateParam( 'height', $height ) ) {
4841 $params[$type]['height'] = $height;
4844 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4845 $width = intval( $value );
4846 if ( $handler->validateParam( 'width', $width ) ) {
4847 $params[$type]['width'] = $width;
4850 } # else no validation -- bug 13436
4852 if ( $type === 'handler' ) {
4853 # Validate handler parameter
4854 $validated = $handler->validateParam( $paramName, $value );
4856 # Validate internal parameters
4857 switch( $paramName ) {
4860 # @todo Fixme: possibly check validity here for
4861 # manualthumb? downstream behavior seems odd with
4862 # missing manual thumbs.
4864 $value = $this->stripAltText( $value, $holders );
4867 $chars = self
::EXT_LINK_URL_CLASS
;
4868 $prots = $this->mUrlProtocols
;
4869 if ( $value === '' ) {
4870 $paramName = 'no-link';
4873 } elseif ( preg_match( "/^$prots/", $value ) ) {
4874 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4875 $paramName = 'link-url';
4876 $this->mOutput
->addExternalLink( $value );
4877 if ( $this->mOptions
->getExternalLinkTarget() ) {
4878 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
4883 $linkTitle = Title
::newFromText( $value );
4885 $paramName = 'link-title';
4886 $value = $linkTitle;
4887 $this->mOutput
->addLink( $linkTitle );
4893 # Most other things appear to be empty or numeric...
4894 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
4899 $params[$type][$paramName] = $value;
4903 if ( !$validated ) {
4908 # Process alignment parameters
4909 if ( $params['horizAlign'] ) {
4910 $params['frame']['align'] = key( $params['horizAlign'] );
4912 if ( $params['vertAlign'] ) {
4913 $params['frame']['valign'] = key( $params['vertAlign'] );
4916 $params['frame']['caption'] = $caption;
4918 # Will the image be presented in a frame, with the caption below?
4919 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4920 isset( $params['frame']['framed'] ) ||
4921 isset( $params['frame']['thumbnail'] ) ||
4922 isset( $params['frame']['manualthumb'] );
4924 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
4925 # came to also set the caption, ordinary text after the image -- which
4926 # makes no sense, because that just repeats the text multiple times in
4927 # screen readers. It *also* came to set the title attribute.
4929 # Now that we have an alt attribute, we should not set the alt text to
4930 # equal the caption: that's worse than useless, it just repeats the
4931 # text. This is the framed/thumbnail case. If there's no caption, we
4932 # use the unnamed parameter for alt text as well, just for the time be-
4933 # ing, if the unnamed param is set and the alt param is not.
4935 # For the future, we need to figure out if we want to tweak this more,
4936 # e.g., introducing a title= parameter for the title; ignoring the un-
4937 # named parameter entirely for images without a caption; adding an ex-
4938 # plicit caption= parameter and preserving the old magic unnamed para-
4940 if ( $imageIsFramed ) { # Framed image
4941 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4942 # No caption or alt text, add the filename as the alt text so
4943 # that screen readers at least get some description of the image
4944 $params['frame']['alt'] = $title->getText();
4946 # Do not set $params['frame']['title'] because tooltips don't make sense
4948 } else { # Inline image
4949 if ( !isset( $params['frame']['alt'] ) ) {
4950 # No alt text, use the "caption" for the alt text
4951 if ( $caption !== '') {
4952 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4954 # No caption, fall back to using the filename for the
4956 $params['frame']['alt'] = $title->getText();
4959 # Use the "caption" for the tooltip text
4960 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4963 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4965 # Linker does the rest
4966 $ret = Linker
::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
4967 $time, $descQuery, $this->mOptions
->getThumbSize() );
4969 # Give the handler a chance to modify the parser object
4971 $handler->parserTransformHook( $this, $file );
4977 protected function stripAltText( $caption, $holders ) {
4978 # Strip bad stuff out of the title (tooltip). We can't just use
4979 # replaceLinkHoldersText() here, because if this function is called
4980 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4982 $tooltip = $holders->replaceText( $caption );
4984 $tooltip = $this->replaceLinkHoldersText( $caption );
4987 # make sure there are no placeholders in thumbnail attributes
4988 # that are later expanded to html- so expand them now and
4990 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
4991 $tooltip = Sanitizer
::stripAllTags( $tooltip );
4997 * Set a flag in the output object indicating that the content is dynamic and
4998 * shouldn't be cached.
5000 function disableCache() {
5001 wfDebug( "Parser output marked as uncacheable.\n" );
5002 $this->mOutput
->setCacheTime( -1 ); // old style, for compatibility
5003 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5007 * Callback from the Sanitizer for expanding items found in HTML attribute
5008 * values, so they can be safely tested and escaped.
5010 * @param $text String
5011 * @param $frame PPFrame
5015 function attributeStripCallback( &$text, $frame = false ) {
5016 $text = $this->replaceVariables( $text, $frame );
5017 $text = $this->mStripState
->unstripBoth( $text );
5024 function getTags() {
5025 return array_merge( array_keys( $this->mTransparentTagHooks
), array_keys( $this->mTagHooks
) );
5029 * Replace transparent tags in $text with the values given by the callbacks.
5031 * Transparent tag hooks are like regular XML-style tag hooks, except they
5032 * operate late in the transformation sequence, on HTML instead of wikitext.
5034 function replaceTransparentTags( $text ) {
5036 $elements = array_keys( $this->mTransparentTagHooks
);
5037 $text = $this->extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix
);
5039 foreach ( $matches as $marker => $data ) {
5040 list( $element, $content, $params, $tag ) = $data;
5041 $tagName = strtolower( $element );
5042 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5043 $output = call_user_func_array( $this->mTransparentTagHooks
[$tagName], array( $content, $params, $this ) );
5047 $this->mStripState
->addGeneral( $marker, $output );
5053 * Break wikitext input into sections, and either pull or replace
5054 * some particular section's text.
5056 * External callers should use the getSection and replaceSection methods.
5058 * @param $text String: Page wikitext
5059 * @param $section String: a section identifier string of the form:
5060 * <flag1> - <flag2> - ... - <section number>
5062 * Currently the only recognised flag is "T", which means the target section number
5063 * was derived during a template inclusion parse, in other words this is a template
5064 * section edit link. If no flags are given, it was an ordinary section edit link.
5065 * This flag is required to avoid a section numbering mismatch when a section is
5066 * enclosed by <includeonly> (bug 6563).
5068 * The section number 0 pulls the text before the first heading; other numbers will
5069 * pull the given section along with its lower-level subsections. If the section is
5070 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5072 * @param $mode String: one of "get" or "replace"
5073 * @param $newText String: replacement text for section data.
5074 * @return String: for "get", the extracted section text.
5075 * for "replace", the whole page with the section replaced.
5077 private function extractSections( $text, $section, $mode, $newText='' ) {
5078 global $wgTitle; # not generally used but removes an ugly failure mode
5079 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5081 $frame = $this->getPreprocessor()->newFrame();
5083 # Process section extraction flags
5085 $sectionParts = explode( '-', $section );
5086 $sectionIndex = array_pop( $sectionParts );
5087 foreach ( $sectionParts as $part ) {
5088 if ( $part === 'T' ) {
5089 $flags |
= self
::PTD_FOR_INCLUSION
;
5092 # Preprocess the text
5093 $root = $this->preprocessToDom( $text, $flags );
5095 # <h> nodes indicate section breaks
5096 # They can only occur at the top level, so we can find them by iterating the root's children
5097 $node = $root->getFirstChild();
5099 # Find the target section
5100 if ( $sectionIndex == 0 ) {
5101 # Section zero doesn't nest, level=big
5102 $targetLevel = 1000;
5104 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5109 if ( $node->getName() === 'h' ) {
5110 $bits = $node->splitHeading();
5111 if ( $bits['i'] == $sectionIndex ) {
5112 $targetLevel = $bits['level'];
5116 if ( $mode === 'replace' ) {
5117 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5119 $node = $node->getNextSibling();
5125 if ( $mode === 'get' ) {
5132 # Find the end of the section, including nested sections
5134 if ( $node->getName() === 'h' ) {
5135 $bits = $node->splitHeading();
5136 $curLevel = $bits['level'];
5137 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5141 if ( $mode === 'get' ) {
5142 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5144 $node = $node->getNextSibling();
5147 # Write out the remainder (in replace mode only)
5148 if ( $mode === 'replace' ) {
5149 # Output the replacement text
5150 # Add two newlines on -- trailing whitespace in $newText is conventionally
5151 # stripped by the editor, so we need both newlines to restore the paragraph gap
5152 # Only add trailing whitespace if there is newText
5153 if ( $newText != "" ) {
5154 $outText .= $newText . "\n\n";
5158 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5159 $node = $node->getNextSibling();
5163 if ( is_string( $outText ) ) {
5164 # Re-insert stripped tags
5165 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5172 * This function returns the text of a section, specified by a number ($section).
5173 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5174 * the first section before any such heading (section 0).
5176 * If a section contains subsections, these are also returned.
5178 * @param $text String: text to look in
5179 * @param $section String: section identifier
5180 * @param $deftext String: default to return if section is not found
5181 * @return string text of the requested section
5183 public function getSection( $text, $section, $deftext='' ) {
5184 return $this->extractSections( $text, $section, "get", $deftext );
5188 * This function returns $oldtext after the content of the section
5189 * specified by $section has been replaced with $text.
5191 * @param $text String: former text of the article
5192 * @param $section Numeric: section identifier
5193 * @param $text String: replacing text
5194 * #return String: modified text
5196 public function replaceSection( $oldtext, $section, $text ) {
5197 return $this->extractSections( $oldtext, $section, "replace", $text );
5201 * Get the ID of the revision we are parsing
5203 * @return Mixed: integer or null
5205 function getRevisionId() {
5206 return $this->mRevisionId
;
5210 * Get the revision object for $this->mRevisionId
5212 * @return either a Revision object or null
5214 protected function getRevisionObject() {
5215 if ( !is_null( $this->mRevisionObject
) ) {
5216 return $this->mRevisionObject
;
5218 if ( is_null( $this->mRevisionId
) ) {
5222 $this->mRevisionObject
= Revision
::newFromId( $this->mRevisionId
);
5223 return $this->mRevisionObject
;
5227 * Get the timestamp associated with the current revision, adjusted for
5228 * the default server-local timestamp
5230 function getRevisionTimestamp() {
5231 if ( is_null( $this->mRevisionTimestamp
) ) {
5232 wfProfileIn( __METHOD__
);
5234 $revObject = $this->getRevisionObject();
5235 $timestamp = $revObject ?
$revObject->getTimestamp() : false;
5237 if( $timestamp !== false ) {
5240 # The cryptic '' timezone parameter tells to use the site-default
5241 # timezone offset instead of the user settings.
5243 # Since this value will be saved into the parser cache, served
5244 # to other users, and potentially even used inside links and such,
5245 # it needs to be consistent for all visitors.
5246 $this->mRevisionTimestamp
= $wgContLang->userAdjust( $timestamp, '' );
5249 wfProfileOut( __METHOD__
);
5251 return $this->mRevisionTimestamp
;
5255 * Get the name of the user that edited the last revision
5257 * @return String: user name
5259 function getRevisionUser() {
5260 if( is_null( $this->mRevisionUser
) ) {
5261 $revObject = $this->getRevisionObject();
5263 # if this template is subst: the revision id will be blank,
5264 # so just use the current user's name
5266 $this->mRevisionUser
= $revObject->getUserText();
5267 } elseif( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5268 $this->mRevisionUser
= $this->getUser()->getName();
5271 return $this->mRevisionUser
;
5275 * Mutator for $mDefaultSort
5277 * @param $sort New value
5279 public function setDefaultSort( $sort ) {
5280 $this->mDefaultSort
= $sort;
5281 $this->mOutput
->setProperty( 'defaultsort', $sort );
5285 * Accessor for $mDefaultSort
5286 * Will use the empty string if none is set.
5288 * This value is treated as a prefix, so the
5289 * empty string is equivalent to sorting by
5294 public function getDefaultSort() {
5295 if ( $this->mDefaultSort
!== false ) {
5296 return $this->mDefaultSort
;
5303 * Accessor for $mDefaultSort
5304 * Unlike getDefaultSort(), will return false if none is set
5306 * @return string or false
5308 public function getCustomDefaultSort() {
5309 return $this->mDefaultSort
;
5313 * Try to guess the section anchor name based on a wikitext fragment
5314 * presumably extracted from a heading, for example "Header" from
5317 public function guessSectionNameFromWikiText( $text ) {
5318 # Strip out wikitext links(they break the anchor)
5319 $text = $this->stripSectionName( $text );
5320 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5321 return '#' . Sanitizer
::escapeId( $text, 'noninitial' );
5325 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5326 * instead. For use in redirects, since IE6 interprets Redirect: headers
5327 * as something other than UTF-8 (apparently?), resulting in breakage.
5329 * @param $text String: The section name
5330 * @return string An anchor
5332 public function guessLegacySectionNameFromWikiText( $text ) {
5333 # Strip out wikitext links(they break the anchor)
5334 $text = $this->stripSectionName( $text );
5335 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5336 return '#' . Sanitizer
::escapeId( $text, array( 'noninitial', 'legacy' ) );
5340 * Strips a text string of wikitext for use in a section anchor
5342 * Accepts a text string and then removes all wikitext from the
5343 * string and leaves only the resultant text (i.e. the result of
5344 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5345 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5346 * to create valid section anchors by mimicing the output of the
5347 * parser when headings are parsed.
5349 * @param $text String: text string to be stripped of wikitext
5350 * for use in a Section anchor
5351 * @return Filtered text string
5353 public function stripSectionName( $text ) {
5354 # Strip internal link markup
5355 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5356 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5358 # Strip external link markup (FIXME: Not Tolerant to blank link text
5359 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5360 # on how many empty links there are on the page - need to figure that out.
5361 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5363 # Parse wikitext quotes (italics & bold)
5364 $text = $this->doQuotes( $text );
5367 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
5372 * strip/replaceVariables/unstrip for preprocessor regression testing
5374 function testSrvus( $text, $title, ParserOptions
$options, $outputType = self
::OT_HTML
) {
5375 if ( !$title instanceof Title
) {
5376 $title = Title
::newFromText( $title );
5378 $this->startParse( $title, $options, $outputType, true );
5380 $text = $this->replaceVariables( $text );
5381 $text = $this->mStripState
->unstripBoth( $text );
5382 $text = Sanitizer
::removeHTMLtags( $text );
5386 function testPst( $text, $title, $options ) {
5388 if ( !$title instanceof Title
) {
5389 $title = Title
::newFromText( $title );
5391 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5394 function testPreprocess( $text, $title, $options ) {
5395 if ( !$title instanceof Title
) {
5396 $title = Title
::newFromText( $title );
5398 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
5402 * Call a callback function on all regions of the given text that are not
5403 * inside strip markers, and replace those regions with the return value
5404 * of the callback. For example, with input:
5408 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5409 * two strings will be replaced with the value returned by the callback in
5412 function markerSkipCallback( $s, $callback ) {
5415 while ( $i < strlen( $s ) ) {
5416 $markerStart = strpos( $s, $this->mUniqPrefix
, $i );
5417 if ( $markerStart === false ) {
5418 $out .= call_user_func( $callback, substr( $s, $i ) );
5421 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5422 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
5423 if ( $markerEnd === false ) {
5424 $out .= substr( $s, $markerStart );
5427 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
5428 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5437 * Save the parser state required to convert the given half-parsed text to
5438 * HTML. "Half-parsed" in this context means the output of
5439 * recursiveTagParse() or internalParse(). This output has strip markers
5440 * from replaceVariables (extensionSubstitution() etc.), and link
5441 * placeholders from replaceLinkHolders().
5443 * Returns an array which can be serialized and stored persistently. This
5444 * array can later be loaded into another parser instance with
5445 * unserializeHalfParsedText(). The text can then be safely incorporated into
5446 * the return value of a parser hook.
5448 function serializeHalfParsedText( $text ) {
5449 wfProfileIn( __METHOD__
);
5452 'version' => self
::HALF_PARSED_VERSION
,
5453 'stripState' => $this->mStripState
->getSubState( $text ),
5454 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
5456 wfProfileOut( __METHOD__
);
5461 * Load the parser state given in the $data array, which is assumed to
5462 * have been generated by serializeHalfParsedText(). The text contents is
5463 * extracted from the array, and its markers are transformed into markers
5464 * appropriate for the current Parser instance. This transformed text is
5465 * returned, and can be safely included in the return value of a parser
5468 * If the $data array has been stored persistently, the caller should first
5469 * check whether it is still valid, by calling isValidHalfParsedText().
5471 * @param $data Serialized data
5474 function unserializeHalfParsedText( $data ) {
5475 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
5476 throw new MWException( __METHOD__
.': invalid version' );
5479 # First, extract the strip state.
5480 $texts = array( $data['text'] );
5481 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
5483 # Now renumber links
5484 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
5486 # Should be good to go.
5491 * Returns true if the given array, presumed to be generated by
5492 * serializeHalfParsedText(), is compatible with the current version of the
5495 * @param $data Array.
5497 function isValidHalfParsedText( $data ) {
5498 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;