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 * $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
41 * $wgNamespacesWithSubpages, $wgAllowExternalImages*,
42 * $wgLocaltimezone, $wgAllowSpecialInclusion*,
45 * * only within ParserOptions
52 * Update this version number when the ParserOutput format
53 * changes in an incompatible way, so the parser cache
54 * can automatically discard old data.
56 const VERSION
= '1.6.4';
59 * Update this version number when the output of serialiseHalfParsedText()
60 * changes in an incompatible way
62 const HALF_PARSED_VERSION
= 2;
64 # Flags for Parser::setFunctionHook
65 # Also available as global constants from Defines.php
66 const SFH_NO_HASH
= 1;
67 const SFH_OBJECT_ARGS
= 2;
69 # Constants needed for external link processing
70 # Everything except bracket, space, or control characters
71 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():
116 var $mOutput, $mAutonumber, $mDTopen;
123 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
124 var $mLinkHolders, $mLinkID;
125 var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
126 var $mTplExpandCache; # empty-frame expansion cache
127 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
128 var $mExpensiveFunctionCount; # number of expensive parser function calls
133 var $mUser; # User object; only used when doing pre-save transform
136 # These are variables reset at least once per parse regardless of $clearState
146 var $mTitle; # Title context, used for self-link rendering and similar things
147 var $mOutputType; # Output type, one of the OT_xxx constants
148 var $ot; # Shortcut alias, see setOutputType()
149 var $mRevisionObject; # The revision object of the specified revision ID
150 var $mRevisionId; # ID to display in {{REVISIONID}} tags
151 var $mRevisionTimestamp; # The timestamp of the specified revision ID
152 var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
153 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
160 function __construct( $conf = array() ) {
161 $this->mConf
= $conf;
162 $this->mUrlProtocols
= wfUrlProtocols();
163 $this->mExtLinkBracketedRegex
= '/\[(\b(' . wfUrlProtocols() . ')'.
164 '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
165 if ( isset( $conf['preprocessorClass'] ) ) {
166 $this->mPreprocessorClass
= $conf['preprocessorClass'];
167 } elseif ( extension_loaded( 'domxml' ) ) {
168 # PECL extension that conflicts with the core DOM extension (bug 13770)
169 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
170 $this->mPreprocessorClass
= 'Preprocessor_Hash';
171 } elseif ( extension_loaded( 'dom' ) ) {
172 $this->mPreprocessorClass
= 'Preprocessor_DOM';
174 $this->mPreprocessorClass
= 'Preprocessor_Hash';
179 * Reduce memory usage to reduce the impact of circular references
181 function __destruct() {
182 if ( isset( $this->mLinkHolders
) ) {
183 $this->mLinkHolders
->__destruct();
185 foreach ( $this as $name => $value ) {
186 unset( $this->$name );
191 * Do various kinds of initialisation on the first call of the parser
193 function firstCallInit() {
194 if ( !$this->mFirstCall
) {
197 $this->mFirstCall
= false;
199 wfProfileIn( __METHOD__
);
201 CoreParserFunctions
::register( $this );
202 CoreTagHooks
::register( $this );
203 $this->initialiseVariables();
205 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
206 wfProfileOut( __METHOD__
);
214 function clearState() {
215 wfProfileIn( __METHOD__
);
216 if ( $this->mFirstCall
) {
217 $this->firstCallInit();
219 $this->mOutput
= new ParserOutput
;
220 $this->mOptions
->registerWatcher( array( $this->mOutput
, 'recordOption' ) );
221 $this->mAutonumber
= 0;
222 $this->mLastSection
= '';
223 $this->mDTopen
= false;
224 $this->mIncludeCount
= array();
225 $this->mArgStack
= false;
226 $this->mInPre
= false;
227 $this->mLinkHolders
= new LinkHolderArray( $this );
229 $this->mRevisionObject
= $this->mRevisionTimestamp
=
230 $this->mRevisionId
= $this->mRevisionUser
= null;
231 $this->mVarCache
= array();
235 * Prefix for temporary replacement strings for the multipass parser.
236 * \x07 should never appear in input as it's disallowed in XML.
237 * Using it at the front also gives us a little extra robustness
238 * since it shouldn't match when butted up against identifier-like
241 * Must not consist of all title characters, or else it will change
242 * the behaviour of <nowiki> in a link.
244 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
245 # Changed to \x7f to allow XML double-parsing -- TS
246 $this->mUniqPrefix
= "\x7fUNIQ" . self
::getRandomString();
247 $this->mStripState
= new StripState( $this->mUniqPrefix
);
250 # Clear these on every parse, bug 4549
251 $this->mTplExpandCache
= $this->mTplRedirCache
= $this->mTplDomCache
= array();
253 $this->mShowToc
= true;
254 $this->mForceTocPosition
= false;
255 $this->mIncludeSizes
= array(
259 $this->mPPNodeCount
= 0;
260 $this->mDefaultSort
= false;
261 $this->mHeadings
= array();
262 $this->mDoubleUnderscores
= array();
263 $this->mExpensiveFunctionCount
= 0;
266 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
267 $this->mPreprocessor
= null;
270 wfRunHooks( 'ParserClearState', array( &$this ) );
271 wfProfileOut( __METHOD__
);
275 * Convert wikitext to HTML
276 * Do not call this function recursively.
278 * @param $text String: text we want to parse
279 * @param $title A title object
280 * @param $options ParserOptions
281 * @param $linestart boolean
282 * @param $clearState boolean
283 * @param $revid Int: number to pass in {{REVISIONID}}
284 * @return ParserOutput a ParserOutput
286 public function parse( $text, Title
$title, ParserOptions
$options, $linestart = true, $clearState = true, $revid = null ) {
288 * First pass--just handle <nowiki> sections, pass the rest off
289 * to internalParse() which does all the real work.
292 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
293 $fname = __METHOD__
.'-' . wfGetCaller();
294 wfProfileIn( __METHOD__
);
295 wfProfileIn( $fname );
297 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
299 $oldRevisionId = $this->mRevisionId
;
300 $oldRevisionObject = $this->mRevisionObject
;
301 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
302 $oldRevisionUser = $this->mRevisionUser
;
303 if ( $revid !== null ) {
304 $this->mRevisionId
= $revid;
305 $this->mRevisionObject
= null;
306 $this->mRevisionTimestamp
= null;
307 $this->mRevisionUser
= null;
310 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
312 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
313 $text = $this->internalParse( $text );
315 $text = $this->mStripState
->unstripGeneral( $text );
317 # Clean up special characters, only run once, next-to-last before doBlockLevels
319 # french spaces, last one Guillemet-left
320 # only if there is something before the space
321 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1 \\2',
322 # french spaces, Guillemet-right
323 '/(\\302\\253) /' => '\\1 ',
324 '/ (!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
326 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
328 $text = $this->doBlockLevels( $text, $linestart );
330 $this->replaceLinkHolders( $text );
333 * The page doesn't get language converted if
335 * b) Content isn't converted
336 * c) It's a conversion table
338 if ( !( $wgDisableLangConversion
339 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
340 ||
$this->mTitle
->isConversionTable() ) ) {
342 # The position of the convert() call should not be changed. it
343 # assumes that the links are all replaced and the only thing left
344 # is the <nowiki> mark.
346 $text = $wgContLang->convert( $text );
350 * A converted title will be provided in the output object if title and
351 * content conversion are enabled, the article text does not contain
352 * a conversion-suppressing double-underscore tag, and no
353 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
354 * automatic link conversion.
356 if ( !( $wgDisableLangConversion
357 ||
$wgDisableTitleConversion
358 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
359 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
360 ||
$this->mOutput
->getDisplayTitle() !== false ) )
362 $convruletitle = $wgContLang->getConvRuleTitle();
363 if ( $convruletitle ) {
364 $this->mOutput
->setTitleText( $convruletitle );
366 $titleText = $wgContLang->convertTitle( $title );
367 $this->mOutput
->setTitleText( $titleText );
371 $text = $this->mStripState
->unstripNoWiki( $text );
373 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
375 $text = $this->replaceTransparentTags( $text );
376 $text = $this->mStripState
->unstripGeneral( $text );
378 $text = Sanitizer
::normalizeCharReferences( $text );
380 if ( ( $wgUseTidy && $this->mOptions
->getTidy() ) ||
$wgAlwaysUseTidy ) {
381 $text = MWTidy
::tidy( $text );
383 # attempt to sanitize at least some nesting problems
384 # (bug #2702 and quite a few others)
386 # ''Something [http://www.cool.com cool''] -->
387 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
388 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
389 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
390 # fix up an anchor inside another anchor, only
391 # at least for a single single nested link (bug 3695)
392 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
393 '\\1\\2</a>\\3</a>\\1\\4</a>',
394 # fix div inside inline elements- doBlockLevels won't wrap a line which
395 # contains a div, so fix it up here; replace
396 # div with escaped text
397 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
398 '\\1\\3<div\\5>\\6</div>\\8\\9',
399 # remove empty italic or bold tag pairs, some
400 # introduced by rules above
401 '/<([bi])><\/\\1>/' => '',
404 $text = preg_replace(
405 array_keys( $tidyregs ),
406 array_values( $tidyregs ),
409 global $wgExpensiveParserFunctionLimit;
410 if ( $this->mExpensiveFunctionCount
> $wgExpensiveParserFunctionLimit ) {
411 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount
, $wgExpensiveParserFunctionLimit );
414 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
416 # Information on include size limits, for the benefit of users who try to skirt them
417 if ( $this->mOptions
->getEnableLimitReport() ) {
418 $max = $this->mOptions
->getMaxIncludeSize();
419 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
421 "NewPP limit report\n" .
422 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
423 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
424 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
426 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
427 $text .= "\n<!-- \n$limitReport-->\n";
429 $this->mOutput
->setText( $text );
431 $this->mRevisionId
= $oldRevisionId;
432 $this->mRevisionObject
= $oldRevisionObject;
433 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
434 $this->mRevisionUser
= $oldRevisionUser;
435 wfProfileOut( $fname );
436 wfProfileOut( __METHOD__
);
438 return $this->mOutput
;
442 * Recursive parser entry point that can be called from an extension tag
445 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
447 * @param $text String: text extension wants to have parsed
448 * @param $frame PPFrame: The frame to use for expanding any template variables
450 function recursiveTagParse( $text, $frame=false ) {
451 wfProfileIn( __METHOD__
);
452 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
453 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
454 $text = $this->internalParse( $text, false, $frame );
455 wfProfileOut( __METHOD__
);
460 * Expand templates and variables in the text, producing valid, static wikitext.
461 * Also removes comments.
463 function preprocess( $text, Title
$title, ParserOptions
$options, $revid = null ) {
464 wfProfileIn( __METHOD__
);
465 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
466 if ( $revid !== null ) {
467 $this->mRevisionId
= $revid;
469 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
470 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
471 $text = $this->replaceVariables( $text );
472 $text = $this->mStripState
->unstripBoth( $text );
473 wfProfileOut( __METHOD__
);
478 * Process the wikitext for the ?preload= feature. (bug 5210)
480 * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
481 * comments, templates, arguments, tags hooks and parser functions are untouched.
483 public function getPreloadText( $text, Title
$title, ParserOptions
$options ) {
484 # Parser (re)initialisation
485 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
487 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
488 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
489 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
490 $text = $this->mStripState
->unstripBoth( $text );
495 * Get a random string
499 static public function getRandomString() {
500 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
504 * Set the current user.
505 * Should only be used when doing pre-save transform.
507 * @param $user Mixed: User object or null (to reset)
509 function setUser( $user ) {
510 $this->mUser
= $user;
514 * Accessor for mUniqPrefix.
518 public function uniqPrefix() {
519 if ( !isset( $this->mUniqPrefix
) ) {
520 # @todo Fixme: this is probably *horribly wrong*
521 # LanguageConverter seems to want $wgParser's uniqPrefix, however
522 # if this is called for a parser cache hit, the parser may not
523 # have ever been initialized in the first place.
524 # Not really sure what the heck is supposed to be going on here.
526 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
528 return $this->mUniqPrefix
;
532 * Set the context title
534 function setTitle( $t ) {
535 if ( !$t ||
$t instanceof FakeTitle
) {
536 $t = Title
::newFromText( 'NO TITLE' );
539 if ( strval( $t->getFragment() ) !== '' ) {
540 # Strip the fragment to avoid various odd effects
541 $this->mTitle
= clone $t;
542 $this->mTitle
->setFragment( '' );
549 * Accessor for the Title object
551 * @return Title object
553 function getTitle() {
554 return $this->mTitle
;
558 * Accessor/mutator for the Title object
560 * @param $x New Title object or null to just get the current one
561 * @return Title object
563 function Title( $x = null ) {
564 return wfSetVar( $this->mTitle
, $x );
568 * Set the output type
570 * @param $ot Integer: new value
572 function setOutputType( $ot ) {
573 $this->mOutputType
= $ot;
576 'html' => $ot == self
::OT_HTML
,
577 'wiki' => $ot == self
::OT_WIKI
,
578 'pre' => $ot == self
::OT_PREPROCESS
,
579 'plain' => $ot == self
::OT_PLAIN
,
584 * Accessor/mutator for the output type
586 * @param $x New value or null to just get the current one
589 function OutputType( $x = null ) {
590 return wfSetVar( $this->mOutputType
, $x );
594 * Get the ParserOutput object
596 * @return ParserOutput object
598 function getOutput() {
599 return $this->mOutput
;
603 * Get the ParserOptions object
605 * @return ParserOptions object
607 function getOptions() {
608 return $this->mOptions
;
612 * Accessor/mutator for the ParserOptions object
614 * @param $x New value or null to just get the current one
615 * @return Current ParserOptions object
617 function Options( $x = null ) {
618 return wfSetVar( $this->mOptions
, $x );
621 function nextLinkID() {
622 return $this->mLinkID++
;
625 function setLinkID( $id ) {
626 $this->mLinkID
= $id;
632 function getFunctionLang() {
633 global $wgLang, $wgContLang;
635 $target = $this->mOptions
->getTargetLanguage();
636 if ( $target !== null ) {
639 return $this->mOptions
->getInterfaceMessage() ?
$wgLang : $wgContLang;
644 * Get a User object either from $this->mUser, if set, or from the
645 * ParserOptions object otherwise
647 * @return User object
650 if ( !is_null( $this->mUser
) ) {
653 return $this->mOptions
->getUser();
657 * Get a preprocessor object
659 * @return Preprocessor instance
661 function getPreprocessor() {
662 if ( !isset( $this->mPreprocessor
) ) {
663 $class = $this->mPreprocessorClass
;
664 $this->mPreprocessor
= new $class( $this );
666 return $this->mPreprocessor
;
670 * Replaces all occurrences of HTML-style comments and the given tags
671 * in the text with a random marker and returns the next text. The output
672 * parameter $matches will be an associative array filled with data in
674 * 'UNIQ-xxxxx' => array(
677 * array( 'param' => 'x' ),
678 * '<element param="x">tag content</element>' ) )
680 * @param $elements list of element names. Comments are always extracted.
681 * @param $text Source text string.
682 * @param $matches Out parameter, Array: extracted tags
683 * @param $uniq_prefix
684 * @return String: stripped text
688 public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
693 $taglist = implode( '|', $elements );
694 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
696 while ( $text != '' ) {
697 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
699 if ( count( $p ) < 5 ) {
702 if ( count( $p ) > 5 ) {
716 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
717 $stripped .= $marker;
719 if ( $close === '/>' ) {
720 # Empty element tag, <tag />
725 if ( $element === '!--' ) {
728 $end = "/(<\\/$element\\s*>)/i";
730 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
732 if ( count( $q ) < 3 ) {
733 # No end tag -- let it run out to the end of the text.
742 $matches[$marker] = array( $element,
744 Sanitizer
::decodeTagAttributes( $attributes ),
745 "<$element$attributes$close$content$tail" );
751 * Get a list of strippable XML-like elements
753 function getStripList() {
754 return $this->mStripList
;
758 * @deprecated use replaceVariables
760 function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
765 * Restores pre, math, and other extensions removed by strip()
767 * always call unstripNoWiki() after this one
769 * @deprecated use $this->mStripState->unstrip()
771 function unstrip( $text, $state ) {
772 return $state->unstripGeneral( $text );
776 * Always call this after unstrip() to preserve the order
779 * @deprecated use $this->mStripState->unstrip()
781 function unstripNoWiki( $text, $state ) {
782 return $state->unstripNoWiki( $text );
786 * @deprecated use $this->mStripState->unstripBoth()
788 function unstripForHTML( $text ) {
789 return $this->mStripState
->unstripBoth( $text );
793 * Add an item to the strip state
794 * Returns the unique tag which must be inserted into the stripped text
795 * The tag will be replaced with the original text in unstrip()
799 function insertStripItem( $text ) {
800 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
801 $this->mMarkerIndex++
;
802 $this->mStripState
->addGeneral( $rnd, $text );
807 * Interface with html tidy
808 * @deprecated Use MWTidy::tidy()
810 public static function tidy( $text ) {
811 wfDeprecated( __METHOD__
);
812 return MWTidy
::tidy( $text );
816 * parse the wiki syntax used to render tables
820 function doTableStuff( $text ) {
821 wfProfileIn( __METHOD__
);
823 $lines = StringUtils
::explode( "\n", $text );
825 $td_history = array(); # Is currently a td tag open?
826 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
827 $tr_history = array(); # Is currently a tr tag open?
828 $tr_attributes = array(); # history of tr attributes
829 $has_opened_tr = array(); # Did this table open a <tr> element?
830 $indent_level = 0; # indent level of the table
832 foreach ( $lines as $outLine ) {
833 $line = trim( $outLine );
835 if ( $line === '' ) { # empty line, go to next line
836 $out .= $outLine."\n";
840 $first_character = $line[0];
843 if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
844 # First check if we are starting a new table
845 $indent_level = strlen( $matches[1] );
847 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
848 $attributes = Sanitizer
::fixTagAttributes( $attributes , 'table' );
850 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
851 array_push( $td_history , false );
852 array_push( $last_tag_history , '' );
853 array_push( $tr_history , false );
854 array_push( $tr_attributes , '' );
855 array_push( $has_opened_tr , false );
856 } elseif ( count( $td_history ) == 0 ) {
857 # Don't do any of the following
858 $out .= $outLine."\n";
860 } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
861 # We are ending a table
862 $line = '</table>' . substr( $line , 2 );
863 $last_tag = array_pop( $last_tag_history );
865 if ( !array_pop( $has_opened_tr ) ) {
866 $line = "<tr><td></td></tr>{$line}";
869 if ( array_pop( $tr_history ) ) {
870 $line = "</tr>{$line}";
873 if ( array_pop( $td_history ) ) {
874 $line = "</{$last_tag}>{$line}";
876 array_pop( $tr_attributes );
877 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
878 } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
879 # Now we have a table row
880 $line = preg_replace( '#^\|-+#', '', $line );
882 # Whats after the tag is now only attributes
883 $attributes = $this->mStripState
->unstripBoth( $line );
884 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
885 array_pop( $tr_attributes );
886 array_push( $tr_attributes, $attributes );
889 $last_tag = array_pop( $last_tag_history );
890 array_pop( $has_opened_tr );
891 array_push( $has_opened_tr , true );
893 if ( array_pop( $tr_history ) ) {
897 if ( array_pop( $td_history ) ) {
898 $line = "</{$last_tag}>{$line}";
902 array_push( $tr_history , false );
903 array_push( $td_history , false );
904 array_push( $last_tag_history , '' );
905 } elseif ( $first_character === '|' ||
$first_character === '!' ||
substr( $line , 0 , 2 ) === '|+' ) {
906 # This might be cell elements, td, th or captions
907 if ( substr( $line , 0 , 2 ) === '|+' ) {
908 $first_character = '+';
909 $line = substr( $line , 1 );
912 $line = substr( $line , 1 );
914 if ( $first_character === '!' ) {
915 $line = str_replace( '!!' , '||' , $line );
918 # Split up multiple cells on the same line.
919 # FIXME : This can result in improper nesting of tags processed
920 # by earlier parser steps, but should avoid splitting up eg
921 # attribute values containing literal "||".
922 $cells = StringUtils
::explodeMarkup( '||' , $line );
926 # Loop through each table cell
927 foreach ( $cells as $cell ) {
929 if ( $first_character !== '+' ) {
930 $tr_after = array_pop( $tr_attributes );
931 if ( !array_pop( $tr_history ) ) {
932 $previous = "<tr{$tr_after}>\n";
934 array_push( $tr_history , true );
935 array_push( $tr_attributes , '' );
936 array_pop( $has_opened_tr );
937 array_push( $has_opened_tr , true );
940 $last_tag = array_pop( $last_tag_history );
942 if ( array_pop( $td_history ) ) {
943 $previous = "</{$last_tag}>\n{$previous}";
946 if ( $first_character === '|' ) {
948 } elseif ( $first_character === '!' ) {
950 } elseif ( $first_character === '+' ) {
951 $last_tag = 'caption';
956 array_push( $last_tag_history , $last_tag );
958 # A cell could contain both parameters and data
959 $cell_data = explode( '|' , $cell , 2 );
961 # Bug 553: Note that a '|' inside an invalid link should not
962 # be mistaken as delimiting cell parameters
963 if ( strpos( $cell_data[0], '[[' ) !== false ) {
964 $cell = "{$previous}<{$last_tag}>{$cell}";
965 } elseif ( count( $cell_data ) == 1 ) {
966 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
968 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
969 $attributes = Sanitizer
::fixTagAttributes( $attributes , $last_tag );
970 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
974 array_push( $td_history , true );
977 $out .= $outLine . "\n";
980 # Closing open td, tr && table
981 while ( count( $td_history ) > 0 ) {
982 if ( array_pop( $td_history ) ) {
985 if ( array_pop( $tr_history ) ) {
988 if ( !array_pop( $has_opened_tr ) ) {
989 $out .= "<tr><td></td></tr>\n" ;
992 $out .= "</table>\n";
995 # Remove trailing line-ending (b/c)
996 if ( substr( $out, -1 ) === "\n" ) {
997 $out = substr( $out, 0, -1 );
1000 # special case: don't return empty table
1001 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1005 wfProfileOut( __METHOD__
);
1011 * Helper function for parse() that transforms wiki markup into
1012 * HTML. Only called for $mOutputType == self::OT_HTML.
1016 function internalParse( $text, $isMain = true, $frame=false ) {
1017 wfProfileIn( __METHOD__
);
1021 # Hook to suspend the parser in this state
1022 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState
) ) ) {
1023 wfProfileOut( __METHOD__
);
1027 # if $frame is provided, then use $frame for replacing any variables
1029 # use frame depth to infer how include/noinclude tags should be handled
1030 # depth=0 means this is the top-level document; otherwise it's an included document
1031 if ( !$frame->depth
) {
1034 $flag = Parser
::PTD_FOR_INCLUSION
;
1036 $dom = $this->preprocessToDom( $text, $flag );
1037 $text = $frame->expand( $dom );
1039 # if $frame is not provided, then use old-style replaceVariables
1040 $text = $this->replaceVariables( $text );
1043 $text = Sanitizer
::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks
) );
1044 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState
) );
1046 # Tables need to come after variable replacement for things to work
1047 # properly; putting them before other transformations should keep
1048 # exciting things like link expansions from showing up in surprising
1050 $text = $this->doTableStuff( $text );
1052 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1054 $text = $this->doDoubleUnderscore( $text );
1056 $text = $this->doHeadings( $text );
1057 if ( $this->mOptions
->getUseDynamicDates() ) {
1058 $df = DateFormatter
::getInstance();
1059 $text = $df->reformat( $this->mOptions
->getDateFormat(), $text );
1061 $text = $this->replaceInternalLinks( $text );
1062 $text = $this->doAllQuotes( $text );
1063 $text = $this->replaceExternalLinks( $text );
1065 # replaceInternalLinks may sometimes leave behind
1066 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1067 $text = str_replace( $this->mUniqPrefix
.'NOPARSE', '', $text );
1069 $text = $this->doMagicLinks( $text );
1070 $text = $this->formatHeadings( $text, $origText, $isMain );
1072 wfProfileOut( __METHOD__
);
1077 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1078 * magic external links.
1083 function doMagicLinks( $text ) {
1084 wfProfileIn( __METHOD__
);
1085 $prots = $this->mUrlProtocols
;
1086 $urlChar = self
::EXT_LINK_URL_CLASS
;
1087 $text = preg_replace_callback(
1089 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1090 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1091 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1092 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1093 ISBN\s+(\b # m[5]: ISBN, capture number
1094 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1095 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1096 [0-9Xx] # check digit
1098 )!x', array( &$this, 'magicLinkCallback' ), $text );
1099 wfProfileOut( __METHOD__
);
1103 function magicLinkCallback( $m ) {
1104 if ( isset( $m[1] ) && $m[1] !== '' ) {
1107 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1110 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1111 # Free external link
1112 return $this->makeFreeExternalLink( $m[0] );
1113 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1115 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1118 $CssClass = 'mw-magiclink-rfc';
1120 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1122 $urlmsg = 'pubmedurl';
1123 $CssClass = 'mw-magiclink-pmid';
1126 throw new MWException( __METHOD__
.': unrecognised match type "' .
1127 substr( $m[0], 0, 20 ) . '"' );
1129 $url = wfMsgForContent( $urlmsg, $id );
1130 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1131 return $sk->makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1132 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1135 $num = strtr( $isbn, array(
1140 $titleObj = SpecialPage
::getTitleFor( 'Booksources', $num );
1142 $titleObj->escapeLocalUrl() .
1143 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1150 * Make a free external link, given a user-supplied URL
1154 function makeFreeExternalLink( $url ) {
1156 wfProfileIn( __METHOD__
);
1158 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1161 # The characters '<' and '>' (which were escaped by
1162 # removeHTMLtags()) should not be included in
1163 # URLs, per RFC 2396.
1165 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1166 $trail = substr( $url, $m2[0][1] ) . $trail;
1167 $url = substr( $url, 0, $m2[0][1] );
1170 # Move trailing punctuation to $trail
1172 # If there is no left bracket, then consider right brackets fair game too
1173 if ( strpos( $url, '(' ) === false ) {
1177 $numSepChars = strspn( strrev( $url ), $sep );
1178 if ( $numSepChars ) {
1179 $trail = substr( $url, -$numSepChars ) . $trail;
1180 $url = substr( $url, 0, -$numSepChars );
1183 $url = Sanitizer
::cleanUrl( $url );
1185 # Is this an external image?
1186 $text = $this->maybeMakeExternalImage( $url );
1187 if ( $text === false ) {
1188 # Not an image, make a link
1189 $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1190 $this->getExternalLinkAttribs( $url ) );
1191 # Register it in the output object...
1192 # Replace unnecessary URL escape codes with their equivalent characters
1193 $pasteurized = self
::replaceUnusualEscapes( $url );
1194 $this->mOutput
->addExternalLink( $pasteurized );
1196 wfProfileOut( __METHOD__
);
1197 return $text . $trail;
1202 * Parse headers and return html
1206 function doHeadings( $text ) {
1207 wfProfileIn( __METHOD__
);
1208 for ( $i = 6; $i >= 1; --$i ) {
1209 $h = str_repeat( '=', $i );
1210 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1211 "<h$i>\\1</h$i>", $text );
1213 wfProfileOut( __METHOD__
);
1218 * Replace single quotes with HTML markup
1220 * @return string the altered text
1222 function doAllQuotes( $text ) {
1223 wfProfileIn( __METHOD__
);
1225 $lines = StringUtils
::explode( "\n", $text );
1226 foreach ( $lines as $line ) {
1227 $outtext .= $this->doQuotes( $line ) . "\n";
1229 $outtext = substr( $outtext, 0,-1 );
1230 wfProfileOut( __METHOD__
);
1235 * Helper function for doAllQuotes()
1237 public function doQuotes( $text ) {
1238 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1239 if ( count( $arr ) == 1 ) {
1242 # First, do some preliminary work. This may shift some apostrophes from
1243 # being mark-up to being text. It also counts the number of occurrences
1244 # of bold and italics mark-ups.
1247 for ( $i = 0; $i < count( $arr ); $i++
) {
1248 if ( ( $i %
2 ) == 1 ) {
1249 # If there are ever four apostrophes, assume the first is supposed to
1250 # be text, and the remaining three constitute mark-up for bold text.
1251 if ( strlen( $arr[$i] ) == 4 ) {
1254 } elseif ( strlen( $arr[$i] ) > 5 ) {
1255 # If there are more than 5 apostrophes in a row, assume they're all
1256 # text except for the last 5.
1257 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1260 # Count the number of occurrences of bold and italics mark-ups.
1261 # We are not counting sequences of five apostrophes.
1262 if ( strlen( $arr[$i] ) == 2 ) {
1264 } elseif ( strlen( $arr[$i] ) == 3 ) {
1266 } elseif ( strlen( $arr[$i] ) == 5 ) {
1273 # If there is an odd number of both bold and italics, it is likely
1274 # that one of the bold ones was meant to be an apostrophe followed
1275 # by italics. Which one we cannot know for certain, but it is more
1276 # likely to be one that has a single-letter word before it.
1277 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1279 $firstsingleletterword = -1;
1280 $firstmultiletterword = -1;
1282 foreach ( $arr as $r ) {
1283 if ( ( $i %
2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1284 $x1 = substr( $arr[$i-1], -1 );
1285 $x2 = substr( $arr[$i-1], -2, 1 );
1286 if ( $x1 === ' ' ) {
1287 if ( $firstspace == -1 ) {
1290 } elseif ( $x2 === ' ') {
1291 if ( $firstsingleletterword == -1 ) {
1292 $firstsingleletterword = $i;
1295 if ( $firstmultiletterword == -1 ) {
1296 $firstmultiletterword = $i;
1303 # If there is a single-letter word, use it!
1304 if ( $firstsingleletterword > -1 ) {
1305 $arr[$firstsingleletterword] = "''";
1306 $arr[$firstsingleletterword-1] .= "'";
1307 } elseif ( $firstmultiletterword > -1 ) {
1308 # If not, but there's a multi-letter word, use that one.
1309 $arr[$firstmultiletterword] = "''";
1310 $arr[$firstmultiletterword-1] .= "'";
1311 } elseif ( $firstspace > -1 ) {
1312 # ... otherwise use the first one that has neither.
1313 # (notice that it is possible for all three to be -1 if, for example,
1314 # there is only one pentuple-apostrophe in the line)
1315 $arr[$firstspace] = "''";
1316 $arr[$firstspace-1] .= "'";
1320 # Now let's actually convert our apostrophic mush to HTML!
1325 foreach ( $arr as $r ) {
1326 if ( ( $i %
2 ) == 0 ) {
1327 if ( $state === 'both' ) {
1333 if ( strlen( $r ) == 2 ) {
1334 if ( $state === 'i' ) {
1335 $output .= '</i>'; $state = '';
1336 } elseif ( $state === 'bi' ) {
1337 $output .= '</i>'; $state = 'b';
1338 } elseif ( $state === 'ib' ) {
1339 $output .= '</b></i><b>'; $state = 'b';
1340 } elseif ( $state === 'both' ) {
1341 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1342 } else { # $state can be 'b' or ''
1343 $output .= '<i>'; $state .= 'i';
1345 } elseif ( strlen( $r ) == 3 ) {
1346 if ( $state === 'b' ) {
1347 $output .= '</b>'; $state = '';
1348 } elseif ( $state === 'bi' ) {
1349 $output .= '</i></b><i>'; $state = 'i';
1350 } elseif ( $state === 'ib' ) {
1351 $output .= '</b>'; $state = 'i';
1352 } elseif ( $state === 'both' ) {
1353 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1354 } else { # $state can be 'i' or ''
1355 $output .= '<b>'; $state .= 'b';
1357 } elseif ( strlen( $r ) == 5 ) {
1358 if ( $state === 'b' ) {
1359 $output .= '</b><i>'; $state = 'i';
1360 } elseif ( $state === 'i' ) {
1361 $output .= '</i><b>'; $state = 'b';
1362 } elseif ( $state === 'bi' ) {
1363 $output .= '</i></b>'; $state = '';
1364 } elseif ( $state === 'ib' ) {
1365 $output .= '</b></i>'; $state = '';
1366 } elseif ( $state === 'both' ) {
1367 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1368 } else { # ($state == '')
1369 $buffer = ''; $state = 'both';
1375 # Now close all remaining tags. Notice that the order is important.
1376 if ( $state === 'b' ||
$state === 'ib' ) {
1379 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1382 if ( $state === 'bi' ) {
1385 # There might be lonely ''''', so make sure we have a buffer
1386 if ( $state === 'both' && $buffer ) {
1387 $output .= '<b><i>'.$buffer.'</i></b>';
1394 * Replace external links (REL)
1396 * Note: this is all very hackish and the order of execution matters a lot.
1397 * Make sure to run maintenance/parserTests.php if you change this code.
1401 function replaceExternalLinks( $text ) {
1403 wfProfileIn( __METHOD__
);
1405 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1407 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1408 $s = array_shift( $bits );
1411 while ( $i<count( $bits ) ) {
1413 $protocol = $bits[$i++
];
1414 $text = $bits[$i++
];
1415 $trail = $bits[$i++
];
1417 # The characters '<' and '>' (which were escaped by
1418 # removeHTMLtags()) should not be included in
1419 # URLs, per RFC 2396.
1421 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1422 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1423 $url = substr( $url, 0, $m2[0][1] );
1426 # If the link text is an image URL, replace it with an <img> tag
1427 # This happened by accident in the original parser, but some people used it extensively
1428 $img = $this->maybeMakeExternalImage( $text );
1429 if ( $img !== false ) {
1435 # Set linktype for CSS - if URL==text, link is essentially free
1436 $linktype = ( $text === $url ) ?
'free' : 'text';
1438 # No link text, e.g. [http://domain.tld/some.link]
1439 if ( $text == '' ) {
1440 # Autonumber if allowed. See bug #5918
1441 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1442 $langObj = $this->getFunctionLang();
1443 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1444 $linktype = 'autonumber';
1446 # Otherwise just use the URL
1447 $text = htmlspecialchars( $url );
1451 # Have link text, e.g. [http://domain.tld/some.link text]s
1453 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1456 $text = $wgContLang->markNoConversion( $text );
1458 $url = Sanitizer
::cleanUrl( $url );
1460 # Use the encoded URL
1461 # This means that users can paste URLs directly into the text
1462 # Funny characters like ö aren't valid in URLs anyway
1463 # This was changed in August 2004
1464 $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1465 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1467 # Register link in the output object.
1468 # Replace unnecessary URL escape codes with the referenced character
1469 # This prevents spammers from hiding links from the filters
1470 $pasteurized = self
::replaceUnusualEscapes( $url );
1471 $this->mOutput
->addExternalLink( $pasteurized );
1474 wfProfileOut( __METHOD__
);
1479 * Get an associative array of additional HTML attributes appropriate for a
1480 * particular external link. This currently may include rel => nofollow
1481 * (depending on configuration, namespace, and the URL's domain) and/or a
1482 * target attribute (depending on configuration).
1484 * @param $url String: optional URL, to extract the domain from for rel =>
1485 * nofollow if appropriate
1486 * @return Array: associative array of HTML attributes
1488 function getExternalLinkAttribs( $url = false ) {
1490 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1491 $ns = $this->mTitle
->getNamespace();
1492 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1493 $attribs['rel'] = 'nofollow';
1495 global $wgNoFollowDomainExceptions;
1496 if ( $wgNoFollowDomainExceptions ) {
1497 $bits = wfParseUrl( $url );
1498 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1499 foreach ( $wgNoFollowDomainExceptions as $domain ) {
1500 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1501 unset( $attribs['rel'] );
1508 if ( $this->mOptions
->getExternalLinkTarget() ) {
1509 $attribs['target'] = $this->mOptions
->getExternalLinkTarget();
1516 * Replace unusual URL escape codes with their equivalent characters
1518 * @param $url String
1521 * @todo This can merge genuinely required bits in the path or query string,
1522 * breaking legit URLs. A proper fix would treat the various parts of
1523 * the URL differently; as a workaround, just use the output for
1524 * statistical records, not for actual linking/output.
1526 static function replaceUnusualEscapes( $url ) {
1527 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1528 array( __CLASS__
, 'replaceUnusualEscapesCallback' ), $url );
1532 * Callback function used in replaceUnusualEscapes().
1533 * Replaces unusual URL escape codes with their equivalent character
1535 private static function replaceUnusualEscapesCallback( $matches ) {
1536 $char = urldecode( $matches[0] );
1537 $ord = ord( $char );
1538 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1539 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1540 # No, shouldn't be escaped
1543 # Yes, leave it escaped
1549 * make an image if it's allowed, either through the global
1550 * option, through the exception, or through the on-wiki whitelist
1553 function maybeMakeExternalImage( $url ) {
1554 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1555 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
1556 $imagesexception = !empty( $imagesfrom );
1558 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1559 if ( $imagesexception && is_array( $imagesfrom ) ) {
1560 $imagematch = false;
1561 foreach ( $imagesfrom as $match ) {
1562 if ( strpos( $url, $match ) === 0 ) {
1567 } elseif ( $imagesexception ) {
1568 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1570 $imagematch = false;
1572 if ( $this->mOptions
->getAllowExternalImages()
1573 ||
( $imagesexception && $imagematch ) ) {
1574 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1576 $text = $sk->makeExternalImage( $url );
1579 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
1580 && preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1581 $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1582 foreach ( $whitelist as $entry ) {
1583 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1584 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
1587 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1588 # Image matches a whitelist entry
1589 $text = $sk->makeExternalImage( $url );
1598 * Process [[ ]] wikilinks
1599 * @return String: processed text
1603 function replaceInternalLinks( $s ) {
1604 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
1609 * Process [[ ]] wikilinks (RIL)
1610 * @return LinkHolderArray
1614 function replaceInternalLinks2( &$s ) {
1617 wfProfileIn( __METHOD__
);
1619 wfProfileIn( __METHOD__
.'-setup' );
1620 static $tc = FALSE, $e1, $e1_img;
1621 # the % is needed to support urlencoded titles as well
1623 $tc = Title
::legalChars() . '#%';
1624 # Match a link having the form [[namespace:link|alternate]]trail
1625 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1626 # Match cases where there is no "]]", which might still be images
1627 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1630 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1631 $holders = new LinkHolderArray( $this );
1633 # split the entire text string on occurences of [[
1634 $a = StringUtils
::explode( '[[', ' ' . $s );
1635 # get the first element (all text up to first [[), and remove the space we added
1638 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1639 $s = substr( $s, 1 );
1641 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1643 if ( $useLinkPrefixExtension ) {
1644 # Match the end of a line for a word that's not followed by whitespace,
1645 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1646 $e2 = wfMsgForContent( 'linkprefix' );
1649 if ( is_null( $this->mTitle
) ) {
1650 wfProfileOut( __METHOD__
.'-setup' );
1651 wfProfileOut( __METHOD__
);
1652 throw new MWException( __METHOD__
.": \$this->mTitle is null\n" );
1654 $nottalk = !$this->mTitle
->isTalkPage();
1656 if ( $useLinkPrefixExtension ) {
1658 if ( preg_match( $e2, $s, $m ) ) {
1659 $first_prefix = $m[2];
1661 $first_prefix = false;
1667 if ( $wgContLang->hasVariants() ) {
1668 $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle
->getPrefixedText() );
1670 $selflink = array( $this->mTitle
->getPrefixedText() );
1672 $useSubpages = $this->areSubpagesAllowed();
1673 wfProfileOut( __METHOD__
.'-setup' );
1675 # Loop for each link
1676 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1677 # Check for excessive memory usage
1678 if ( $holders->isBig() ) {
1680 # Do the existence check, replace the link holders and clear the array
1681 $holders->replace( $s );
1685 if ( $useLinkPrefixExtension ) {
1686 wfProfileIn( __METHOD__
.'-prefixhandling' );
1687 if ( preg_match( $e2, $s, $m ) ) {
1694 if ( $first_prefix ) {
1695 $prefix = $first_prefix;
1696 $first_prefix = false;
1698 wfProfileOut( __METHOD__
.'-prefixhandling' );
1701 $might_be_img = false;
1703 wfProfileIn( __METHOD__
."-e1" );
1704 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1706 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1707 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1708 # the real problem is with the $e1 regex
1711 # Still some problems for cases where the ] is meant to be outside punctuation,
1712 # and no image is in sight. See bug 2095.
1714 if ( $text !== '' &&
1715 substr( $m[3], 0, 1 ) === ']' &&
1716 strpos( $text, '[' ) !== false
1719 $text .= ']'; # so that replaceExternalLinks($text) works later
1720 $m[3] = substr( $m[3], 1 );
1722 # fix up urlencoded title texts
1723 if ( strpos( $m[1], '%' ) !== false ) {
1724 # Should anchors '#' also be rejected?
1725 $m[1] = str_replace( array('<', '>'), array('<', '>'), rawurldecode( $m[1] ) );
1728 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1729 $might_be_img = true;
1731 if ( strpos( $m[1], '%' ) !== false ) {
1732 $m[1] = rawurldecode( $m[1] );
1735 } else { # Invalid form; output directly
1736 $s .= $prefix . '[[' . $line ;
1737 wfProfileOut( __METHOD__
."-e1" );
1740 wfProfileOut( __METHOD__
."-e1" );
1741 wfProfileIn( __METHOD__
."-misc" );
1743 # Don't allow internal links to pages containing
1744 # PROTO: where PROTO is a valid URL protocol; these
1745 # should be external links.
1746 if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1747 $s .= $prefix . '[[' . $line ;
1748 wfProfileOut( __METHOD__
."-misc" );
1752 # Make subpage if necessary
1753 if ( $useSubpages ) {
1754 $link = $this->maybeDoSubpageLink( $m[1], $text );
1759 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1761 # Strip off leading ':'
1762 $link = substr( $link, 1 );
1765 wfProfileOut( __METHOD__
."-misc" );
1766 wfProfileIn( __METHOD__
."-title" );
1767 $nt = Title
::newFromText( $this->mStripState
->unstripNoWiki( $link ) );
1768 if ( $nt === null ) {
1769 $s .= $prefix . '[[' . $line;
1770 wfProfileOut( __METHOD__
."-title" );
1774 $ns = $nt->getNamespace();
1775 $iw = $nt->getInterWiki();
1776 wfProfileOut( __METHOD__
."-title" );
1778 if ( $might_be_img ) { # if this is actually an invalid link
1779 wfProfileIn( __METHOD__
."-might_be_img" );
1780 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
1783 # look at the next 'line' to see if we can close it there
1785 $next_line = $a->current();
1786 if ( $next_line === false ||
$next_line === null ) {
1789 $m = explode( ']]', $next_line, 3 );
1790 if ( count( $m ) == 3 ) {
1791 # the first ]] closes the inner link, the second the image
1793 $text .= "[[{$m[0]}]]{$m[1]}";
1796 } elseif ( count( $m ) == 2 ) {
1797 # if there's exactly one ]] that's fine, we'll keep looking
1798 $text .= "[[{$m[0]}]]{$m[1]}";
1800 # if $next_line is invalid too, we need look no further
1801 $text .= '[[' . $next_line;
1806 # we couldn't find the end of this imageLink, so output it raw
1807 # but don't ignore what might be perfectly normal links in the text we've examined
1808 $holders->merge( $this->replaceInternalLinks2( $text ) );
1809 $s .= "{$prefix}[[$link|$text";
1810 # note: no $trail, because without an end, there *is* no trail
1811 wfProfileOut( __METHOD__
."-might_be_img" );
1814 } else { # it's not an image, so output it raw
1815 $s .= "{$prefix}[[$link|$text";
1816 # note: no $trail, because without an end, there *is* no trail
1817 wfProfileOut( __METHOD__
."-might_be_img" );
1820 wfProfileOut( __METHOD__
."-might_be_img" );
1823 $wasblank = ( $text == '' );
1827 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1828 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1829 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1830 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1831 $text = $this->doQuotes( $text );
1834 # Link not escaped by : , create the various objects
1838 wfProfileIn( __METHOD__
."-interwiki" );
1839 if ( $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1840 $this->mOutput
->addLanguageLink( $nt->getFullText() );
1841 $s = rtrim( $s . $prefix );
1842 $s .= trim( $trail, "\n" ) == '' ?
'': $prefix . $trail;
1843 wfProfileOut( __METHOD__
."-interwiki" );
1846 wfProfileOut( __METHOD__
."-interwiki" );
1848 if ( $ns == NS_FILE
) {
1849 wfProfileIn( __METHOD__
."-image" );
1850 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
1852 # if no parameters were passed, $text
1853 # becomes something like "File:Foo.png",
1854 # which we don't want to pass on to the
1858 # recursively parse links inside the image caption
1859 # actually, this will parse them in any other parameters, too,
1860 # but it might be hard to fix that, and it doesn't matter ATM
1861 $text = $this->replaceExternalLinks( $text );
1862 $holders->merge( $this->replaceInternalLinks2( $text ) );
1864 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1865 $s .= $prefix . $this->armorLinks(
1866 $this->makeImage( $nt, $text, $holders ) ) . $trail;
1868 $s .= $prefix . $trail;
1870 wfProfileOut( __METHOD__
."-image" );
1875 if ( $ns == NS_CATEGORY
) {
1876 wfProfileIn( __METHOD__
."-category" );
1877 $s = rtrim( $s . "\n" ); # bug 87
1880 $sortkey = $this->getDefaultSort();
1884 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
1885 $sortkey = str_replace( "\n", '', $sortkey );
1886 $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1887 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
1890 * Strip the whitespace Category links produce, see bug 87
1891 * @todo We might want to use trim($tmp, "\n") here.
1893 $s .= trim( $prefix . $trail, "\n" ) == '' ?
'' : $prefix . $trail;
1895 wfProfileOut( __METHOD__
."-category" );
1900 # Self-link checking
1901 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL
) {
1902 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1903 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1908 # NS_MEDIA is a pseudo-namespace for linking directly to a file
1909 # FIXME: Should do batch file existence checks, see comment below
1910 if ( $ns == NS_MEDIA
) {
1911 wfProfileIn( __METHOD__
."-media" );
1912 # Give extensions a chance to select the file revision for us
1913 $skip = $time = $sha1 = $descQuery = false;
1914 wfRunHooks( 'BeforeParserMakeImageLinkObj',
1915 array( &$this, &$nt, &$skip, &$time, &$descQuery, &$sha1 ) );
1917 $this->mOutput
->addImage( $nt->getDBkey() ); // register
1918 $link = $sk->link( $nt );
1920 # Fetch and register the file
1921 $file = $this->fetchFile( $nt, $time, $sha1 );
1923 $nt = $file->getTitle(); // file title may be different (via hooks)
1925 $link = $sk->makeMediaLinkFile( $nt, $file, $text );
1927 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1928 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1929 wfProfileOut( __METHOD__
."-media" );
1933 wfProfileIn( __METHOD__
."-always_known" );
1934 # Some titles, such as valid special pages or files in foreign repos, should
1935 # be shown as bluelinks even though they're not included in the page table
1937 # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1938 # batch file existence checks for NS_FILE and NS_MEDIA
1939 if ( $iw == '' && $nt->isAlwaysKnown() ) {
1940 $this->mOutput
->addLink( $nt );
1941 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1943 # Links will be added to the output link list after checking
1944 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1946 wfProfileOut( __METHOD__
."-always_known" );
1948 wfProfileOut( __METHOD__
);
1953 * Make a link placeholder. The text returned can be later resolved to a real link with
1954 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1955 * parsing of interwiki links, and secondly to allow all existence checks and
1956 * article length checks (for stub links) to be bundled into a single query.
1960 function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
1961 return $this->mLinkHolders
->makeHolder( $nt, $text, $query, $trail, $prefix );
1965 * Render a forced-blue link inline; protect against double expansion of
1966 * URLs if we're in a mode that prepends full URL prefixes to internal links.
1967 * Since this little disaster has to split off the trail text to avoid
1968 * breaking URLs in the following text without breaking trails on the
1969 * wiki links, it's been made into a horrible function.
1972 * @param $text String
1973 * @param $query Array or String
1974 * @param $trail String
1975 * @param $prefix String
1976 * @return String: HTML-wikitext mix oh yuck
1978 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
1979 list( $inside, $trail ) = Linker
::splitTrail( $trail );
1981 if ( is_string( $query ) ) {
1982 $query = wfCgiToArray( $query );
1984 if ( $text == '' ) {
1985 $text = htmlspecialchars( $nt->getPrefixedText() );
1988 $sk = $this->mOptions
->getSkin( $this->mTitle
);
1989 $link = $sk->linkKnown( $nt, "$prefix$text$inside", array(), $query );
1991 return $this->armorLinks( $link ) . $trail;
1995 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1996 * going to go through further parsing steps before inline URL expansion.
1998 * Not needed quite as much as it used to be since free links are a bit
1999 * more sensible these days. But bracketed links are still an issue.
2001 * @param $text String: more-or-less HTML
2002 * @return String: less-or-more HTML with NOPARSE bits
2004 function armorLinks( $text ) {
2005 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2006 "{$this->mUniqPrefix}NOPARSE$1", $text );
2010 * Return true if subpage links should be expanded on this page.
2013 function areSubpagesAllowed() {
2014 # Some namespaces don't allow subpages
2015 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2019 * Handle link to subpage if necessary
2021 * @param $target String: the source of the link
2022 * @param &$text String: the link text, modified as necessary
2023 * @return string the full name of the link
2026 function maybeDoSubpageLink( $target, &$text ) {
2027 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2031 * Used by doBlockLevels()
2034 function closeParagraph() {
2036 if ( $this->mLastSection
!= '' ) {
2037 $result = '</' . $this->mLastSection
. ">\n";
2039 $this->mInPre
= false;
2040 $this->mLastSection
= '';
2045 * getCommon() returns the length of the longest common substring
2046 * of both arguments, starting at the beginning of both.
2049 function getCommon( $st1, $st2 ) {
2050 $fl = strlen( $st1 );
2051 $shorter = strlen( $st2 );
2052 if ( $fl < $shorter ) {
2056 for ( $i = 0; $i < $shorter; ++
$i ) {
2057 if ( $st1{$i} != $st2{$i} ) {
2065 * These next three functions open, continue, and close the list
2066 * element appropriate to the prefix character passed into them.
2069 function openList( $char ) {
2070 $result = $this->closeParagraph();
2072 if ( '*' === $char ) {
2073 $result .= '<ul><li>';
2074 } elseif ( '#' === $char ) {
2075 $result .= '<ol><li>';
2076 } elseif ( ':' === $char ) {
2077 $result .= '<dl><dd>';
2078 } elseif ( ';' === $char ) {
2079 $result .= '<dl><dt>';
2080 $this->mDTopen
= true;
2082 $result = '<!-- ERR 1 -->';
2090 * @param $char String
2093 function nextItem( $char ) {
2094 if ( '*' === $char ||
'#' === $char ) {
2096 } elseif ( ':' === $char ||
';' === $char ) {
2098 if ( $this->mDTopen
) {
2101 if ( ';' === $char ) {
2102 $this->mDTopen
= true;
2103 return $close . '<dt>';
2105 $this->mDTopen
= false;
2106 return $close . '<dd>';
2109 return '<!-- ERR 2 -->';
2114 * @param $char String
2117 function closeList( $char ) {
2118 if ( '*' === $char ) {
2119 $text = '</li></ul>';
2120 } elseif ( '#' === $char ) {
2121 $text = '</li></ol>';
2122 } elseif ( ':' === $char ) {
2123 if ( $this->mDTopen
) {
2124 $this->mDTopen
= false;
2125 $text = '</dt></dl>';
2127 $text = '</dd></dl>';
2130 return '<!-- ERR 3 -->';
2137 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2139 * @param $text String
2140 * @param $linestart Boolean: whether or not this is at the start of a line.
2142 * @return string the lists rendered as HTML
2144 function doBlockLevels( $text, $linestart ) {
2145 wfProfileIn( __METHOD__
);
2147 # Parsing through the text line by line. The main thing
2148 # happening here is handling of block-level elements p, pre,
2149 # and making lists from lines starting with * # : etc.
2151 $textLines = StringUtils
::explode( "\n", $text );
2153 $lastPrefix = $output = '';
2154 $this->mDTopen
= $inBlockElem = false;
2156 $paragraphStack = false;
2158 foreach ( $textLines as $oLine ) {
2160 if ( !$linestart ) {
2170 $lastPrefixLength = strlen( $lastPrefix );
2171 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2172 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2173 # If not in a <pre> element, scan for and figure out what prefixes are there.
2174 if ( !$this->mInPre
) {
2175 # Multiple prefixes may abut each other for nested lists.
2176 $prefixLength = strspn( $oLine, '*#:;' );
2177 $prefix = substr( $oLine, 0, $prefixLength );
2180 # ; and : are both from definition-lists, so they're equivalent
2181 # for the purposes of determining whether or not we need to open/close
2183 $prefix2 = str_replace( ';', ':', $prefix );
2184 $t = substr( $oLine, $prefixLength );
2185 $this->mInPre
= (bool)$preOpenMatch;
2187 # Don't interpret any other prefixes in preformatted text
2189 $prefix = $prefix2 = '';
2194 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2195 # Same as the last item, so no need to deal with nesting or opening stuff
2196 $output .= $this->nextItem( substr( $prefix, -1 ) );
2197 $paragraphStack = false;
2199 if ( substr( $prefix, -1 ) === ';') {
2200 # The one nasty exception: definition lists work like this:
2201 # ; title : definition text
2202 # So we check for : in the remainder text to split up the
2203 # title and definition, without b0rking links.
2205 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2207 $output .= $term . $this->nextItem( ':' );
2210 } elseif ( $prefixLength ||
$lastPrefixLength ) {
2211 # We need to open or close prefixes, or both.
2213 # Either open or close a level...
2214 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2215 $paragraphStack = false;
2217 # Close all the prefixes which aren't shared.
2218 while ( $commonPrefixLength < $lastPrefixLength ) {
2219 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2220 --$lastPrefixLength;
2223 # Continue the current prefix if appropriate.
2224 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2225 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2228 # Open prefixes where appropriate.
2229 while ( $prefixLength > $commonPrefixLength ) {
2230 $char = substr( $prefix, $commonPrefixLength, 1 );
2231 $output .= $this->openList( $char );
2233 if ( ';' === $char ) {
2234 # FIXME: This is dupe of code above
2235 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2237 $output .= $term . $this->nextItem( ':' );
2240 ++
$commonPrefixLength;
2242 $lastPrefix = $prefix2;
2245 # If we have no prefixes, go to paragraph mode.
2246 if ( 0 == $prefixLength ) {
2247 wfProfileIn( __METHOD__
."-paragraph" );
2248 # No prefix (not in list)--go to paragraph mode
2249 # XXX: use a stack for nestable elements like span, table and div
2250 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2251 $closematch = preg_match(
2252 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2253 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix
.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2254 if ( $openmatch or $closematch ) {
2255 $paragraphStack = false;
2256 #Â TODO bug 5718: paragraph closed
2257 $output .= $this->closeParagraph();
2258 if ( $preOpenMatch and !$preCloseMatch ) {
2259 $this->mInPre
= true;
2261 $inBlockElem = !$closematch;
2262 } elseif ( !$inBlockElem && !$this->mInPre
) {
2263 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection
=== 'pre' ||
trim( $t ) != '' ) ) {
2265 if ( $this->mLastSection
!== 'pre' ) {
2266 $paragraphStack = false;
2267 $output .= $this->closeParagraph().'<pre>';
2268 $this->mLastSection
= 'pre';
2270 $t = substr( $t, 1 );
2273 if ( trim( $t ) === '' ) {
2274 if ( $paragraphStack ) {
2275 $output .= $paragraphStack.'<br />';
2276 $paragraphStack = false;
2277 $this->mLastSection
= 'p';
2279 if ( $this->mLastSection
!== 'p' ) {
2280 $output .= $this->closeParagraph();
2281 $this->mLastSection
= '';
2282 $paragraphStack = '<p>';
2284 $paragraphStack = '</p><p>';
2288 if ( $paragraphStack ) {
2289 $output .= $paragraphStack;
2290 $paragraphStack = false;
2291 $this->mLastSection
= 'p';
2292 } elseif ( $this->mLastSection
!== 'p' ) {
2293 $output .= $this->closeParagraph().'<p>';
2294 $this->mLastSection
= 'p';
2299 wfProfileOut( __METHOD__
."-paragraph" );
2301 # somewhere above we forget to get out of pre block (bug 785)
2302 if ( $preCloseMatch && $this->mInPre
) {
2303 $this->mInPre
= false;
2305 if ( $paragraphStack === false ) {
2309 while ( $prefixLength ) {
2310 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2313 if ( $this->mLastSection
!= '' ) {
2314 $output .= '</' . $this->mLastSection
. '>';
2315 $this->mLastSection
= '';
2318 wfProfileOut( __METHOD__
);
2323 * Split up a string on ':', ignoring any occurences inside tags
2324 * to prevent illegal overlapping.
2326 * @param $str String: the string to split
2327 * @param &$before String: set to everything before the ':'
2328 * @param &$after String: set to everything after the ':'
2329 * return String: the position of the ':', or false if none found
2331 function findColonNoLinks( $str, &$before, &$after ) {
2332 wfProfileIn( __METHOD__
);
2334 $pos = strpos( $str, ':' );
2335 if ( $pos === false ) {
2337 wfProfileOut( __METHOD__
);
2341 $lt = strpos( $str, '<' );
2342 if ( $lt === false ||
$lt > $pos ) {
2343 # Easy; no tag nesting to worry about
2344 $before = substr( $str, 0, $pos );
2345 $after = substr( $str, $pos+
1 );
2346 wfProfileOut( __METHOD__
);
2350 # Ugly state machine to walk through avoiding tags.
2351 $state = self
::COLON_STATE_TEXT
;
2353 $len = strlen( $str );
2354 for( $i = 0; $i < $len; $i++
) {
2358 # (Using the number is a performance hack for common cases)
2359 case 0: # self::COLON_STATE_TEXT:
2362 # Could be either a <start> tag or an </end> tag
2363 $state = self
::COLON_STATE_TAGSTART
;
2366 if ( $stack == 0 ) {
2368 $before = substr( $str, 0, $i );
2369 $after = substr( $str, $i +
1 );
2370 wfProfileOut( __METHOD__
);
2373 # Embedded in a tag; don't break it.
2376 # Skip ahead looking for something interesting
2377 $colon = strpos( $str, ':', $i );
2378 if ( $colon === false ) {
2379 # Nothing else interesting
2380 wfProfileOut( __METHOD__
);
2383 $lt = strpos( $str, '<', $i );
2384 if ( $stack === 0 ) {
2385 if ( $lt === false ||
$colon < $lt ) {
2387 $before = substr( $str, 0, $colon );
2388 $after = substr( $str, $colon +
1 );
2389 wfProfileOut( __METHOD__
);
2393 if ( $lt === false ) {
2394 # Nothing else interesting to find; abort!
2395 # We're nested, but there's no close tags left. Abort!
2398 # Skip ahead to next tag start
2400 $state = self
::COLON_STATE_TAGSTART
;
2403 case 1: # self::COLON_STATE_TAG:
2408 $state = self
::COLON_STATE_TEXT
;
2411 # Slash may be followed by >?
2412 $state = self
::COLON_STATE_TAGSLASH
;
2418 case 2: # self::COLON_STATE_TAGSTART:
2421 $state = self
::COLON_STATE_CLOSETAG
;
2424 $state = self
::COLON_STATE_COMMENT
;
2427 # Illegal early close? This shouldn't happen D:
2428 $state = self
::COLON_STATE_TEXT
;
2431 $state = self
::COLON_STATE_TAG
;
2434 case 3: # self::COLON_STATE_CLOSETAG:
2439 wfDebug( __METHOD__
.": Invalid input; too many close tags\n" );
2440 wfProfileOut( __METHOD__
);
2443 $state = self
::COLON_STATE_TEXT
;
2446 case self
::COLON_STATE_TAGSLASH
:
2448 # Yes, a self-closed tag <blah/>
2449 $state = self
::COLON_STATE_TEXT
;
2451 # Probably we're jumping the gun, and this is an attribute
2452 $state = self
::COLON_STATE_TAG
;
2455 case 5: # self::COLON_STATE_COMMENT:
2457 $state = self
::COLON_STATE_COMMENTDASH
;
2460 case self
::COLON_STATE_COMMENTDASH
:
2462 $state = self
::COLON_STATE_COMMENTDASHDASH
;
2464 $state = self
::COLON_STATE_COMMENT
;
2467 case self
::COLON_STATE_COMMENTDASHDASH
:
2469 $state = self
::COLON_STATE_TEXT
;
2471 $state = self
::COLON_STATE_COMMENT
;
2475 throw new MWException( "State machine error in " . __METHOD__
);
2479 wfDebug( __METHOD__
.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2480 wfProfileOut( __METHOD__
);
2483 wfProfileOut( __METHOD__
);
2488 * Return value of a magic variable (like PAGENAME)
2492 function getVariableValue( $index, $frame=false ) {
2493 global $wgContLang, $wgSitename, $wgServer;
2494 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2497 * Some of these require message or data lookups and can be
2498 * expensive to check many times.
2500 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache
) ) ) {
2501 if ( isset( $this->mVarCache
[$index] ) ) {
2502 return $this->mVarCache
[$index];
2506 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2507 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2510 global $wgLocaltimezone;
2511 if ( isset( $wgLocaltimezone ) ) {
2512 $oldtz = date_default_timezone_get();
2513 date_default_timezone_set( $wgLocaltimezone );
2516 $localTimestamp = date( 'YmdHis', $ts );
2517 $localMonth = date( 'm', $ts );
2518 $localMonth1 = date( 'n', $ts );
2519 $localMonthName = date( 'n', $ts );
2520 $localDay = date( 'j', $ts );
2521 $localDay2 = date( 'd', $ts );
2522 $localDayOfWeek = date( 'w', $ts );
2523 $localWeek = date( 'W', $ts );
2524 $localYear = date( 'Y', $ts );
2525 $localHour = date( 'H', $ts );
2526 if ( isset( $wgLocaltimezone ) ) {
2527 date_default_timezone_set( $oldtz );
2531 case 'currentmonth':
2532 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2534 case 'currentmonth1':
2535 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2537 case 'currentmonthname':
2538 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2540 case 'currentmonthnamegen':
2541 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2543 case 'currentmonthabbrev':
2544 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2547 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2550 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2553 $value = $wgContLang->formatNum( $localMonth );
2556 $value = $wgContLang->formatNum( $localMonth1 );
2558 case 'localmonthname':
2559 $value = $wgContLang->getMonthName( $localMonthName );
2561 case 'localmonthnamegen':
2562 $value = $wgContLang->getMonthNameGen( $localMonthName );
2564 case 'localmonthabbrev':
2565 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2568 $value = $wgContLang->formatNum( $localDay );
2571 $value = $wgContLang->formatNum( $localDay2 );
2574 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2577 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2579 case 'fullpagename':
2580 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2582 case 'fullpagenamee':
2583 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2586 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2588 case 'subpagenamee':
2589 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2591 case 'basepagename':
2592 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2594 case 'basepagenamee':
2595 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle
->getBaseText() ) ) );
2597 case 'talkpagename':
2598 if ( $this->mTitle
->canTalk() ) {
2599 $talkPage = $this->mTitle
->getTalkPage();
2600 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2605 case 'talkpagenamee':
2606 if ( $this->mTitle
->canTalk() ) {
2607 $talkPage = $this->mTitle
->getTalkPage();
2608 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2613 case 'subjectpagename':
2614 $subjPage = $this->mTitle
->getSubjectPage();
2615 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2617 case 'subjectpagenamee':
2618 $subjPage = $this->mTitle
->getSubjectPage();
2619 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2622 # Let the edit saving system know we should parse the page
2623 # *after* a revision ID has been assigned.
2624 $this->mOutput
->setFlag( 'vary-revision' );
2625 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision...\n" );
2626 $value = $this->mRevisionId
;
2629 # Let the edit saving system know we should parse the page
2630 # *after* a revision ID has been assigned. This is for null edits.
2631 $this->mOutput
->setFlag( 'vary-revision' );
2632 wfDebug( __METHOD__
. ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2633 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2635 case 'revisionday2':
2636 # Let the edit saving system know we should parse the page
2637 # *after* a revision ID has been assigned. This is for null edits.
2638 $this->mOutput
->setFlag( 'vary-revision' );
2639 wfDebug( __METHOD__
. ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2640 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2642 case 'revisionmonth':
2643 # Let the edit saving system know we should parse the page
2644 # *after* a revision ID has been assigned. This is for null edits.
2645 $this->mOutput
->setFlag( 'vary-revision' );
2646 wfDebug( __METHOD__
. ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2647 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2649 case 'revisionmonth1':
2650 # Let the edit saving system know we should parse the page
2651 # *after* a revision ID has been assigned. This is for null edits.
2652 $this->mOutput
->setFlag( 'vary-revision' );
2653 wfDebug( __METHOD__
. ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2654 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2656 case 'revisionyear':
2657 # Let the edit saving system know we should parse the page
2658 # *after* a revision ID has been assigned. This is for null edits.
2659 $this->mOutput
->setFlag( 'vary-revision' );
2660 wfDebug( __METHOD__
. ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2661 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2663 case 'revisiontimestamp':
2664 # Let the edit saving system know we should parse the page
2665 # *after* a revision ID has been assigned. This is for null edits.
2666 $this->mOutput
->setFlag( 'vary-revision' );
2667 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2668 $value = $this->getRevisionTimestamp();
2670 case 'revisionuser':
2671 # Let the edit saving system know we should parse the page
2672 # *after* a revision ID has been assigned. This is for null edits.
2673 $this->mOutput
->setFlag( 'vary-revision' );
2674 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2675 $value = $this->getRevisionUser();
2678 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2681 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2684 $value = $this->mTitle
->canTalk() ?
str_replace( '_',' ',$this->mTitle
->getTalkNsText() ) : '';
2687 $value = $this->mTitle
->canTalk() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2689 case 'subjectspace':
2690 $value = $this->mTitle
->getSubjectNsText();
2692 case 'subjectspacee':
2693 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2695 case 'currentdayname':
2696 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) +
1 );
2699 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2702 $value = $wgContLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2705 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2708 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2709 # int to remove the padding
2710 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2713 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2715 case 'localdayname':
2716 $value = $wgContLang->getWeekdayName( $localDayOfWeek +
1 );
2719 $value = $wgContLang->formatNum( $localYear, true );
2722 $value = $wgContLang->time( $localTimestamp, false, false );
2725 $value = $wgContLang->formatNum( $localHour, true );
2728 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2729 # int to remove the padding
2730 $value = $wgContLang->formatNum( (int)$localWeek );
2733 $value = $wgContLang->formatNum( $localDayOfWeek );
2735 case 'numberofarticles':
2736 $value = $wgContLang->formatNum( SiteStats
::articles() );
2738 case 'numberoffiles':
2739 $value = $wgContLang->formatNum( SiteStats
::images() );
2741 case 'numberofusers':
2742 $value = $wgContLang->formatNum( SiteStats
::users() );
2744 case 'numberofactiveusers':
2745 $value = $wgContLang->formatNum( SiteStats
::activeUsers() );
2747 case 'numberofpages':
2748 $value = $wgContLang->formatNum( SiteStats
::pages() );
2750 case 'numberofadmins':
2751 $value = $wgContLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2753 case 'numberofedits':
2754 $value = $wgContLang->formatNum( SiteStats
::edits() );
2756 case 'numberofviews':
2757 $value = $wgContLang->formatNum( SiteStats
::views() );
2759 case 'currenttimestamp':
2760 $value = wfTimestamp( TS_MW
, $ts );
2762 case 'localtimestamp':
2763 $value = $localTimestamp;
2765 case 'currentversion':
2766 $value = SpecialVersion
::getVersion();
2769 return $wgArticlePath;
2775 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2776 $serverName = parse_url( $wgServer, PHP_URL_HOST
);
2777 wfRestoreWarnings();
2778 return $serverName ?
$serverName : $wgServer;
2780 return $wgScriptPath;
2782 return $wgStylePath;
2783 case 'directionmark':
2784 return $wgContLang->getDirMark();
2785 case 'contentlanguage':
2786 global $wgLanguageCode;
2787 return $wgLanguageCode;
2790 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache
, &$index, &$ret, &$frame ) ) ) {
2798 $this->mVarCache
[$index] = $value;
2804 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2808 function initialiseVariables() {
2809 wfProfileIn( __METHOD__
);
2810 $variableIDs = MagicWord
::getVariableIDs();
2811 $substIDs = MagicWord
::getSubstIDs();
2813 $this->mVariables
= new MagicWordArray( $variableIDs );
2814 $this->mSubstWords
= new MagicWordArray( $substIDs );
2815 wfProfileOut( __METHOD__
);
2819 * Preprocess some wikitext and return the document tree.
2820 * This is the ghost of replace_variables().
2822 * @param $text String: The text to parse
2823 * @param $flags Integer: bitwise combination of:
2824 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2825 * included. Default is to assume a direct page view.
2827 * The generated DOM tree must depend only on the input text and the flags.
2828 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2830 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2831 * change in the DOM tree for a given text, must be passed through the section identifier
2832 * in the section edit link and thus back to extractSections().
2834 * The output of this function is currently only cached in process memory, but a persistent
2835 * cache may be implemented at a later date which takes further advantage of these strict
2836 * dependency requirements.
2840 function preprocessToDom( $text, $flags = 0 ) {
2841 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2846 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2848 public static function splitWhitespace( $s ) {
2849 $ltrimmed = ltrim( $s );
2850 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2851 $trimmed = rtrim( $ltrimmed );
2852 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2854 $w2 = substr( $ltrimmed, -$diff );
2858 return array( $w1, $trimmed, $w2 );
2862 * Replace magic variables, templates, and template arguments
2863 * with the appropriate text. Templates are substituted recursively,
2864 * taking care to avoid infinite loops.
2866 * Note that the substitution depends on value of $mOutputType:
2867 * self::OT_WIKI: only {{subst:}} templates
2868 * self::OT_PREPROCESS: templates but not extension tags
2869 * self::OT_HTML: all templates and extension tags
2871 * @param $text String: the text to transform
2872 * @param $frame PPFrame Object describing the arguments passed to the template.
2873 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2874 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2875 * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2878 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2879 # Is there any text? Also, Prevent too big inclusions!
2880 if ( strlen( $text ) < 1 ||
strlen( $text ) > $this->mOptions
->getMaxIncludeSize() ) {
2883 wfProfileIn( __METHOD__
);
2885 if ( $frame === false ) {
2886 $frame = $this->getPreprocessor()->newFrame();
2887 } elseif ( !( $frame instanceof PPFrame
) ) {
2888 wfDebug( __METHOD__
." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2889 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2892 $dom = $this->preprocessToDom( $text );
2893 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
2894 $text = $frame->expand( $dom, $flags );
2896 wfProfileOut( __METHOD__
);
2900 # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2901 static function createAssocArgs( $args ) {
2902 $assocArgs = array();
2904 foreach ( $args as $arg ) {
2905 $eqpos = strpos( $arg, '=' );
2906 if ( $eqpos === false ) {
2907 $assocArgs[$index++
] = $arg;
2909 $name = trim( substr( $arg, 0, $eqpos ) );
2910 $value = trim( substr( $arg, $eqpos+
1 ) );
2911 if ( $value === false ) {
2914 if ( $name !== false ) {
2915 $assocArgs[$name] = $value;
2924 * Warn the user when a parser limitation is reached
2925 * Will warn at most once the user per limitation type
2927 * @param $limitationType String: should be one of:
2928 * 'expensive-parserfunction' (corresponding messages:
2929 * 'expensive-parserfunction-warning',
2930 * 'expensive-parserfunction-category')
2931 * 'post-expand-template-argument' (corresponding messages:
2932 * 'post-expand-template-argument-warning',
2933 * 'post-expand-template-argument-category')
2934 * 'post-expand-template-inclusion' (corresponding messages:
2935 * 'post-expand-template-inclusion-warning',
2936 * 'post-expand-template-inclusion-category')
2937 * @param $current Current value
2938 * @param $max Maximum allowed, when an explicit limit has been
2939 * exceeded, provide the values (optional)
2941 function limitationWarn( $limitationType, $current=null, $max=null) {
2942 # does no harm if $current and $max are present but are unnecessary for the message
2943 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2944 $this->mOutput
->addWarning( $warning );
2945 $this->addTrackingCategory( "$limitationType-category" );
2949 * Return the text of a template, after recursively
2950 * replacing any variables or templates within the template.
2952 * @param $piece Array: the parts of the template
2953 * $piece['title']: the title, i.e. the part before the |
2954 * $piece['parts']: the parameter array
2955 * $piece['lineStart']: whether the brace was at the start of a line
2956 * @param $frame PPFrame The current frame, contains template arguments
2957 * @return String: the text of the template
2960 function braceSubstitution( $piece, $frame ) {
2961 global $wgContLang, $wgNonincludableNamespaces;
2962 wfProfileIn( __METHOD__
);
2963 wfProfileIn( __METHOD__
.'-setup' );
2966 $found = false; # $text has been filled
2967 $nowiki = false; # wiki markup in $text should be escaped
2968 $isHTML = false; # $text is HTML, armour it against wikitext transformation
2969 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2970 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
2971 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
2973 # Title object, where $text came from
2976 # $part1 is the bit before the first |, and must contain only title characters.
2977 # Various prefixes will be stripped from it later.
2978 $titleWithSpaces = $frame->expand( $piece['title'] );
2979 $part1 = trim( $titleWithSpaces );
2982 # Original title text preserved for various purposes
2983 $originalTitle = $part1;
2985 # $args is a list of argument nodes, starting from index 0, not including $part1
2986 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2987 $args = ( null == $piece['parts'] ) ?
array() : $piece['parts'];
2988 wfProfileOut( __METHOD__
.'-setup' );
2991 wfProfileIn( __METHOD__
.'-modifiers' );
2994 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
2996 # Possibilities for substMatch: "subst", "safesubst" or FALSE
2997 # Decide whether to expand template or keep wikitext as-is.
2998 if ( $this->ot
['wiki'] ) {
2999 if ( $substMatch === false ) {
3000 $literal = true; # literal when in PST with no prefix
3002 $literal = false; # expand when in PST with subst: or safesubst:
3005 if ( $substMatch == 'subst' ) {
3006 $literal = true; # literal when not in PST with plain subst:
3008 $literal = false; # expand when not in PST with safesubst: or no prefix
3012 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3019 if ( !$found && $args->getLength() == 0 ) {
3020 $id = $this->mVariables
->matchStartToEnd( $part1 );
3021 if ( $id !== false ) {
3022 $text = $this->getVariableValue( $id, $frame );
3023 if ( MagicWord
::getCacheTTL( $id ) > -1 ) {
3024 $this->mOutput
->updateCacheExpiry( MagicWord
::getCacheTTL( $id ) );
3030 # MSG, MSGNW and RAW
3033 $mwMsgnw = MagicWord
::get( 'msgnw' );
3034 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3037 # Remove obsolete MSG:
3038 $mwMsg = MagicWord
::get( 'msg' );
3039 $mwMsg->matchStartAndRemove( $part1 );
3043 $mwRaw = MagicWord
::get( 'raw' );
3044 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3045 $forceRawInterwiki = true;
3048 wfProfileOut( __METHOD__
.'-modifiers' );
3052 wfProfileIn( __METHOD__
. '-pfunc' );
3054 $colonPos = strpos( $part1, ':' );
3055 if ( $colonPos !== false ) {
3056 # Case sensitive functions
3057 $function = substr( $part1, 0, $colonPos );
3058 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3059 $function = $this->mFunctionSynonyms
[1][$function];
3061 # Case insensitive functions
3062 $function = $wgContLang->lc( $function );
3063 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3064 $function = $this->mFunctionSynonyms
[0][$function];
3070 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3071 $initialArgs = array( &$this );
3072 $funcArgs = array( trim( substr( $part1, $colonPos +
1 ) ) );
3073 if ( $flags & SFH_OBJECT_ARGS
) {
3074 # Add a frame parameter, and pass the arguments as an array
3075 $allArgs = $initialArgs;
3076 $allArgs[] = $frame;
3077 for ( $i = 0; $i < $args->getLength(); $i++
) {
3078 $funcArgs[] = $args->item( $i );
3080 $allArgs[] = $funcArgs;
3082 # Convert arguments to plain text
3083 for ( $i = 0; $i < $args->getLength(); $i++
) {
3084 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3086 $allArgs = array_merge( $initialArgs, $funcArgs );
3089 # Workaround for PHP bug 35229 and similar
3090 if ( !is_callable( $callback ) ) {
3091 wfProfileOut( __METHOD__
. '-pfunc' );
3092 wfProfileOut( __METHOD__
);
3093 throw new MWException( "Tag hook for $function is not callable\n" );
3095 $result = call_user_func_array( $callback, $allArgs );
3098 $preprocessFlags = 0;
3100 if ( is_array( $result ) ) {
3101 if ( isset( $result[0] ) ) {
3103 unset( $result[0] );
3106 # Extract flags into the local scope
3107 # This allows callers to set flags such as nowiki, found, etc.
3113 $text = $this->preprocessToDom( $text, $preprocessFlags );
3118 wfProfileOut( __METHOD__
. '-pfunc' );
3121 # Finish mangling title and then check for loops.
3122 # Set $title to a Title object and $titleText to the PDBK
3125 # Split the title into page and subpage
3127 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3128 if ( $subpage !== '' ) {
3129 $ns = $this->mTitle
->getNamespace();
3131 $title = Title
::newFromText( $part1, $ns );
3133 $titleText = $title->getPrefixedText();
3134 # Check for language variants if the template is not found
3135 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3136 $wgContLang->findVariantLink( $part1, $title, true );
3138 # Do recursion depth check
3139 $limit = $this->mOptions
->getMaxTemplateDepth();
3140 if ( $frame->depth
>= $limit ) {
3142 $text = '<span class="error">'
3143 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3149 # Load from database
3150 if ( !$found && $title ) {
3151 wfProfileIn( __METHOD__
. '-loadtpl' );
3152 if ( !$title->isExternal() ) {
3153 if ( $title->getNamespace() == NS_SPECIAL
3154 && $this->mOptions
->getAllowSpecialInclusion()
3155 && $this->ot
['html'] )
3157 $text = SpecialPage
::capturePath( $title );
3158 if ( is_string( $text ) ) {
3161 $this->disableCache();
3163 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3164 $found = false; # access denied
3165 wfDebug( __METHOD__
.": template inclusion denied for " . $title->getPrefixedDBkey() );
3167 list( $text, $title ) = $this->getTemplateDom( $title );
3168 if ( $text !== false ) {
3174 # If the title is valid but undisplayable, make a link to it
3175 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3176 $text = "[[:$titleText]]";
3179 } elseif ( $title->isTrans() ) {
3180 # Interwiki transclusion
3181 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3182 $text = $this->interwikiTransclude( $title, 'render' );
3185 $text = $this->interwikiTransclude( $title, 'raw' );
3186 # Preprocess it like a template
3187 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3193 # Do infinite loop check
3194 # This has to be done after redirect resolution to avoid infinite loops via redirects
3195 if ( !$frame->loopCheck( $title ) ) {
3197 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3198 wfDebug( __METHOD__
.": template loop broken at '$titleText'\n" );
3200 wfProfileOut( __METHOD__
. '-loadtpl' );
3203 # If we haven't found text to substitute by now, we're done
3204 # Recover the source wikitext and return it
3206 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3207 wfProfileOut( __METHOD__
);
3208 return array( 'object' => $text );
3211 # Expand DOM-style return values in a child frame
3212 if ( $isChildObj ) {
3213 # Clean up argument array
3214 $newFrame = $frame->newChild( $args, $title );
3217 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3218 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3219 # Expansion is eligible for the empty-frame cache
3220 if ( isset( $this->mTplExpandCache
[$titleText] ) ) {
3221 $text = $this->mTplExpandCache
[$titleText];
3223 $text = $newFrame->expand( $text );
3224 $this->mTplExpandCache
[$titleText] = $text;
3227 # Uncached expansion
3228 $text = $newFrame->expand( $text );
3231 if ( $isLocalObj && $nowiki ) {
3232 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3233 $isLocalObj = false;
3236 # Replace raw HTML by a placeholder
3237 # Add a blank line preceding, to prevent it from mucking up
3238 # immediately preceding headings
3240 $text = "\n\n" . $this->insertStripItem( $text );
3241 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3242 # Escape nowiki-style return values
3243 $text = wfEscapeWikiText( $text );
3244 } elseif ( is_string( $text )
3245 && !$piece['lineStart']
3246 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3248 # Bug 529: if the template begins with a table or block-level
3249 # element, it should be treated as beginning a new line.
3250 # This behaviour is somewhat controversial.
3251 $text = "\n" . $text;
3254 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3255 # Error, oversize inclusion
3256 if ( $titleText !== false ) {
3257 # Make a working, properly escaped link if possible (bug 23588)
3258 $text = "[[:$titleText]]";
3260 # This will probably not be a working link, but at least it may
3261 # provide some hint of where the problem is
3262 preg_replace( '/^:/', '', $originalTitle );
3263 $text = "[[:$originalTitle]]";
3265 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3266 $this->limitationWarn( 'post-expand-template-inclusion' );
3269 if ( $isLocalObj ) {
3270 $ret = array( 'object' => $text );
3272 $ret = array( 'text' => $text );
3275 wfProfileOut( __METHOD__
);
3280 * Get the semi-parsed DOM representation of a template with a given title,
3281 * and its redirect destination title. Cached.
3283 function getTemplateDom( $title ) {
3284 $cacheTitle = $title;
3285 $titleText = $title->getPrefixedDBkey();
3287 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3288 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3289 $title = Title
::makeTitle( $ns, $dbk );
3290 $titleText = $title->getPrefixedDBkey();
3292 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3293 return array( $this->mTplDomCache
[$titleText], $title );
3296 # Cache miss, go to the database
3297 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3299 if ( $text === false ) {
3300 $this->mTplDomCache
[$titleText] = false;
3301 return array( false, $title );
3304 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3305 $this->mTplDomCache
[ $titleText ] = $dom;
3307 if ( !$title->equals( $cacheTitle ) ) {
3308 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3309 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3312 return array( $dom, $title );
3316 * Fetch the unparsed text of a template and register a reference to it.
3318 function fetchTemplateAndTitle( $title ) {
3319 $templateCb = $this->mOptions
->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3320 $stuff = call_user_func( $templateCb, $title, $this );
3321 $text = $stuff['text'];
3322 $finalTitle = isset( $stuff['finalTitle'] ) ?
$stuff['finalTitle'] : $title;
3323 if ( isset( $stuff['deps'] ) ) {
3324 foreach ( $stuff['deps'] as $dep ) {
3325 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3328 return array( $text, $finalTitle );
3331 function fetchTemplate( $title ) {
3332 $rv = $this->fetchTemplateAndTitle( $title );
3337 * Static function to get a template
3338 * Can be overridden via ParserOptions::setTemplateCallback().
3340 static function statelessFetchTemplate( $title, $parser=false ) {
3341 $text = $skip = false;
3342 $finalTitle = $title;
3345 # Loop to fetch the article, with up to 1 redirect
3346 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3347 # Give extensions a chance to select the revision instead
3348 $id = false; # Assume current
3349 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3350 array( $parser, &$title, &$skip, &$id ) );
3356 'page_id' => $title->getArticleID(),
3361 $rev = $id ? Revision
::newFromId( $id ) : Revision
::newFromTitle( $title );
3362 $rev_id = $rev ?
$rev->getId() : 0;
3363 # If there is no current revision, there is no page
3364 if ( $id === false && !$rev ) {
3365 $linkCache = LinkCache
::singleton();
3366 $linkCache->addBadLinkObj( $title );
3371 'page_id' => $title->getArticleID(),
3372 'rev_id' => $rev_id );
3375 $text = $rev->getText();
3376 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3378 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3379 if ( !$message->exists() ) {
3383 $text = $message->plain();
3387 if ( $text === false ) {
3391 $finalTitle = $title;
3392 $title = Title
::newFromRedirect( $text );
3396 'finalTitle' => $finalTitle,
3401 * Fetch a file and register a reference to it.
3402 * @TODO: register and track file version info too
3404 function fetchFile( $title, $time = false, $sha1 = false ) {
3405 if ( $sha1 ) { // get by (sha1,timestamp)
3406 $file = RepoGroup
::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3408 $title = $file->getTitle(); // file title may not match $title
3410 } else { // get by (name,timestamp)
3411 $file = wfFindFile( $title, array( 'time' => $time ) );
3413 $this->mOutput
->addImage( $title->getDBkey() );
3418 * Transclude an interwiki link.
3420 function interwikiTransclude( $title, $action ) {
3421 global $wgEnableScaryTranscluding;
3423 if ( !$wgEnableScaryTranscluding ) {
3424 return wfMsgForContent('scarytranscludedisabled');
3427 $url = $title->getFullUrl( "action=$action" );
3429 if ( strlen( $url ) > 255 ) {
3430 return wfMsgForContent( 'scarytranscludetoolong' );
3432 return $this->fetchScaryTemplateMaybeFromCache( $url );
3435 function fetchScaryTemplateMaybeFromCache( $url ) {
3436 global $wgTranscludeCacheExpiry;
3437 $dbr = wfGetDB( DB_SLAVE
);
3438 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3439 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3440 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3442 return $obj->tc_contents
;
3445 $text = Http
::get( $url );
3447 return wfMsgForContent( 'scarytranscludefailed', $url );
3450 $dbw = wfGetDB( DB_MASTER
);
3451 $dbw->replace( 'transcache', array('tc_url'), array(
3453 'tc_time' => $dbw->timestamp( time() ),
3454 'tc_contents' => $text)
3461 * Triple brace replacement -- used for template arguments
3464 function argSubstitution( $piece, $frame ) {
3465 wfProfileIn( __METHOD__
);
3468 $parts = $piece['parts'];
3469 $nameWithSpaces = $frame->expand( $piece['title'] );
3470 $argName = trim( $nameWithSpaces );
3472 $text = $frame->getArgument( $argName );
3473 if ( $text === false && $parts->getLength() > 0
3477 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3480 # No match in frame, use the supplied default
3481 $object = $parts->item( 0 )->getChildren();
3483 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3484 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3485 $this->limitationWarn( 'post-expand-template-argument' );
3488 if ( $text === false && $object === false ) {
3490 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3492 if ( $error !== false ) {
3495 if ( $object !== false ) {
3496 $ret = array( 'object' => $object );
3498 $ret = array( 'text' => $text );
3501 wfProfileOut( __METHOD__
);
3506 * Return the text to be used for a given extension tag.
3507 * This is the ghost of strip().
3509 * @param $params Associative array of parameters:
3510 * name PPNode for the tag name
3511 * attr PPNode for unparsed text where tag attributes are thought to be
3512 * attributes Optional associative array of parsed attributes
3513 * inner Contents of extension element
3514 * noClose Original text did not have a close tag
3515 * @param $frame PPFrame
3517 function extensionSubstitution( $params, $frame ) {
3518 $name = $frame->expand( $params['name'] );
3519 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3520 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3521 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3523 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower($name)] ) &&
3524 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3525 if ( $isFunctionTag ) {
3526 $markerType = 'none';
3528 $markerType = 'general';
3530 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3531 $name = strtolower( $name );
3532 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3533 if ( isset( $params['attributes'] ) ) {
3534 $attributes = $attributes +
$params['attributes'];
3537 if ( isset( $this->mTagHooks
[$name] ) ) {
3538 # Workaround for PHP bug 35229 and similar
3539 if ( !is_callable( $this->mTagHooks
[$name] ) ) {
3540 throw new MWException( "Tag hook for $name is not callable\n" );
3542 $output = call_user_func_array( $this->mTagHooks
[$name],
3543 array( $content, $attributes, $this, $frame ) );
3544 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
3545 list( $callback, $flags ) = $this->mFunctionTagHooks
[$name];
3546 if ( !is_callable( $callback ) ) {
3547 throw new MWException( "Tag hook for $name is not callable\n" );
3550 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3552 $output = '<span class="error">Invalid tag extension name: ' .
3553 htmlspecialchars( $name ) . '</span>';
3556 if ( is_array( $output ) ) {
3557 # Extract flags to local scope (to override $markerType)
3559 $output = $flags[0];
3564 if ( is_null( $attrText ) ) {
3567 if ( isset( $params['attributes'] ) ) {
3568 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3569 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3570 htmlspecialchars( $attrValue ) . '"';
3573 if ( $content === null ) {
3574 $output = "<$name$attrText/>";
3576 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
3577 $output = "<$name$attrText>$content$close";
3581 if ( $markerType === 'none' ) {
3583 } elseif ( $markerType === 'nowiki' ) {
3584 $this->mStripState
->addNoWiki( $marker, $output );
3585 } elseif ( $markerType === 'general' ) {
3586 $this->mStripState
->addGeneral( $marker, $output );
3588 throw new MWException( __METHOD__
.': invalid marker type' );
3594 * Increment an include size counter
3596 * @param $type String: the type of expansion
3597 * @param $size Integer: the size of the text
3598 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3600 function incrementIncludeSize( $type, $size ) {
3601 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
3604 $this->mIncludeSizes
[$type] +
= $size;
3610 * Increment the expensive function count
3612 * @return Boolean: false if the limit has been exceeded
3614 function incrementExpensiveFunctionCount() {
3615 global $wgExpensiveParserFunctionLimit;
3616 $this->mExpensiveFunctionCount++
;
3617 if ( $this->mExpensiveFunctionCount
<= $wgExpensiveParserFunctionLimit ) {
3624 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3625 * Fills $this->mDoubleUnderscores, returns the modified text
3627 function doDoubleUnderscore( $text ) {
3628 wfProfileIn( __METHOD__
);
3630 # The position of __TOC__ needs to be recorded
3631 $mw = MagicWord
::get( 'toc' );
3632 if ( $mw->match( $text ) ) {
3633 $this->mShowToc
= true;
3634 $this->mForceTocPosition
= true;
3636 # Set a placeholder. At the end we'll fill it in with the TOC.
3637 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3639 # Only keep the first one.
3640 $text = $mw->replace( '', $text );
3643 # Now match and remove the rest of them
3644 $mwa = MagicWord
::getDoubleUnderscoreArray();
3645 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
3647 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
3648 $this->mOutput
->mNoGallery
= true;
3650 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
3651 $this->mShowToc
= false;
3653 if ( isset( $this->mDoubleUnderscores
['hiddencat'] ) && $this->mTitle
->getNamespace() == NS_CATEGORY
) {
3654 $this->addTrackingCategory( 'hidden-category-category' );
3656 # (bug 8068) Allow control over whether robots index a page.
3658 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3659 # is not desirable, the last one on the page should win.
3660 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
3661 $this->mOutput
->setIndexPolicy( 'noindex' );
3662 $this->addTrackingCategory( 'noindex-category' );
3664 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
3665 $this->mOutput
->setIndexPolicy( 'index' );
3666 $this->addTrackingCategory( 'index-category' );
3669 # Cache all double underscores in the database
3670 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
3671 $this->mOutput
->setProperty( $key, '' );
3674 wfProfileOut( __METHOD__
);
3679 * Add a tracking category, getting the title from a system message,
3680 * or print a debug message if the title is invalid.
3682 * @param $msg String: message key
3683 * @return Boolean: whether the addition was successful
3685 protected function addTrackingCategory( $msg ) {
3686 $cat = wfMsgForContent( $msg );
3688 # Allow tracking categories to be disabled by setting them to "-"
3689 if ( $cat === '-' ) {
3693 $containerCategory = Title
::makeTitleSafe( NS_CATEGORY
, $cat );
3694 if ( $containerCategory ) {
3695 $this->mOutput
->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3698 wfDebug( __METHOD__
.": [[MediaWiki:$msg]] is not a valid title!\n" );
3704 * This function accomplishes several tasks:
3705 * 1) Auto-number headings if that option is enabled
3706 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3707 * 3) Add a Table of contents on the top for users who have enabled the option
3708 * 4) Auto-anchor headings
3710 * It loops through all headlines, collects the necessary data, then splits up the
3711 * string and re-inserts the newly formatted headlines.
3713 * @param $text String
3714 * @param $origText String: original, untouched wikitext
3715 * @param $isMain Boolean
3718 function formatHeadings( $text, $origText, $isMain=true ) {
3719 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3721 # Inhibit editsection links if requested in the page
3722 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
3725 $showEditLink = $this->mOptions
->getEditSection();
3727 if ( $showEditLink ) {
3728 $this->mOutput
->setEditSectionTokens( true );
3731 # Get all headlines for numbering them and adding funky stuff like [edit]
3732 # links - this is for later, but we need the number of headlines right now
3734 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3736 # if there are fewer than 4 headlines in the article, do not show TOC
3737 # unless it's been explicitly enabled.
3738 $enoughToc = $this->mShowToc
&&
3739 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
3741 # Allow user to stipulate that a page should have a "new section"
3742 # link added via __NEWSECTIONLINK__
3743 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
3744 $this->mOutput
->setNewSection( true );
3747 # Allow user to remove the "new section"
3748 # link via __NONEWSECTIONLINK__
3749 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
3750 $this->mOutput
->hideNewSection( true );
3753 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3754 # override above conditions and always show TOC above first header
3755 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
3756 $this->mShowToc
= true;
3760 # We need this to perform operations on the HTML
3761 $sk = $this->mOptions
->getSkin( $this->mTitle
);
3767 # Ugh .. the TOC should have neat indentation levels which can be
3768 # passed to the skin functions. These are determined here
3772 $sublevelCount = array();
3773 $levelCount = array();
3778 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self
::MARKER_SUFFIX
;
3779 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
3780 $oldType = $this->mOutputType
;
3781 $this->setOutputType( self
::OT_WIKI
);
3782 $frame = $this->getPreprocessor()->newFrame();
3783 $root = $this->preprocessToDom( $origText );
3784 $node = $root->getFirstChild();
3789 foreach ( $matches[3] as $headline ) {
3790 $isTemplate = false;
3792 $sectionIndex = false;
3794 $markerMatches = array();
3795 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3796 $serial = $markerMatches[1];
3797 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
3798 $isTemplate = ( $titleText != $baseTitleText );
3799 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3803 $prevlevel = $level;
3805 $level = $matches[1][$headlineCount];
3807 if ( $level > $prevlevel ) {
3808 # Increase TOC level
3810 $sublevelCount[$toclevel] = 0;
3811 if ( $toclevel<$wgMaxTocLevel ) {
3812 $prevtoclevel = $toclevel;
3813 $toc .= $sk->tocIndent();
3816 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3817 # Decrease TOC level, find level to jump to
3819 for ( $i = $toclevel; $i > 0; $i-- ) {
3820 if ( $levelCount[$i] == $level ) {
3821 # Found last matching level
3824 } elseif ( $levelCount[$i] < $level ) {
3825 # Found first matching level below current level
3833 if ( $toclevel<$wgMaxTocLevel ) {
3834 if ( $prevtoclevel < $wgMaxTocLevel ) {
3835 # Unindent only if the previous toc level was shown :p
3836 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3837 $prevtoclevel = $toclevel;
3839 $toc .= $sk->tocLineEnd();
3843 # No change in level, end TOC line
3844 if ( $toclevel<$wgMaxTocLevel ) {
3845 $toc .= $sk->tocLineEnd();
3849 $levelCount[$toclevel] = $level;
3851 # count number of headlines for each level
3852 @$sublevelCount[$toclevel]++
;
3854 for( $i = 1; $i <= $toclevel; $i++
) {
3855 if ( !empty( $sublevelCount[$i] ) ) {
3859 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3864 # The safe header is a version of the header text safe to use for links
3865 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3866 $safeHeadline = $this->mStripState
->unstripBoth( $headline );
3868 # Remove link placeholders by the link text.
3869 # <!--LINK number-->
3871 # link text with suffix
3872 $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3874 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3875 $tocline = preg_replace(
3876 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3877 array( '', '<$1>' ),
3880 $tocline = trim( $tocline );
3882 # For the anchor, strip out HTML-y stuff period
3883 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3884 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
3886 # Save headline for section edit hint before it's escaped
3887 $headlineHint = $safeHeadline;
3889 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3890 # For reverse compatibility, provide an id that's
3891 # HTML4-compatible, like we used to.
3893 # It may be worth noting, academically, that it's possible for
3894 # the legacy anchor to conflict with a non-legacy headline
3895 # anchor on the page. In this case likely the "correct" thing
3896 # would be to either drop the legacy anchors or make sure
3897 # they're numbered first. However, this would require people
3898 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3899 # manually, so let's not bother worrying about it.
3900 $legacyHeadline = Sanitizer
::escapeId( $safeHeadline,
3901 array( 'noninitial', 'legacy' ) );
3902 $safeHeadline = Sanitizer
::escapeId( $safeHeadline );
3904 if ( $legacyHeadline == $safeHeadline ) {
3905 # No reason to have both (in fact, we can't)
3906 $legacyHeadline = false;
3909 $legacyHeadline = false;
3910 $safeHeadline = Sanitizer
::escapeId( $safeHeadline,
3914 # HTML names must be case-insensitively unique (bug 10721).
3915 # This does not apply to Unicode characters per
3916 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3917 # FIXME: We may be changing them depending on the current locale.
3918 $arrayKey = strtolower( $safeHeadline );
3919 if ( $legacyHeadline === false ) {
3920 $legacyArrayKey = false;
3922 $legacyArrayKey = strtolower( $legacyHeadline );
3925 # count how many in assoc. array so we can track dupes in anchors
3926 if ( isset( $refers[$arrayKey] ) ) {
3927 $refers[$arrayKey]++
;
3929 $refers[$arrayKey] = 1;
3931 if ( isset( $refers[$legacyArrayKey] ) ) {
3932 $refers[$legacyArrayKey]++
;
3934 $refers[$legacyArrayKey] = 1;
3937 # Don't number the heading if it is the only one (looks silly)
3938 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
3939 # the two are different if the line contains a link
3940 $headline = $numbering . ' ' . $headline;
3943 # Create the anchor for linking from the TOC to the section
3944 $anchor = $safeHeadline;
3945 $legacyAnchor = $legacyHeadline;
3946 if ( $refers[$arrayKey] > 1 ) {
3947 $anchor .= '_' . $refers[$arrayKey];
3949 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3950 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3952 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) ||
$toclevel < $wgMaxTocLevel ) ) {
3953 $toc .= $sk->tocLine( $anchor, $tocline,
3954 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
3957 # Add the section to the section tree
3958 # Find the DOM node for this header
3959 while ( $node && !$isTemplate ) {
3960 if ( $node->getName() === 'h' ) {
3961 $bits = $node->splitHeading();
3962 if ( $bits['i'] == $sectionIndex ) {
3966 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
3967 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
3968 $node = $node->getNextSibling();
3971 'toclevel' => $toclevel,
3974 'number' => $numbering,
3975 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
3976 'fromtitle' => $titleText,
3977 'byteoffset' => ( $isTemplate ?
null : $byteOffset ),
3978 'anchor' => $anchor,
3981 # give headline the correct <h#> tag
3982 if ( $showEditLink && $sectionIndex !== false ) {
3983 // Output edit section links as markers with styles that can be customized by skins
3984 if ( $isTemplate ) {
3985 # Put a T flag in the section identifier, to indicate to extractSections()
3986 # that sections inside <includeonly> should be counted.
3987 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3989 $editlinkArgs = array( $this->mTitle
->getPrefixedText(), $sectionIndex, $headlineHint );
3991 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3992 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3993 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3994 // so we don't have to worry about a user trying to input one of these markers directly.
3995 // We use a page and section attribute to stop the language converter from converting these important bits
3996 // of data, but put the headline hint inside a content block because the language converter is supposed to
3997 // be able to convert that piece of data.
3998 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
3999 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4000 if ( isset($editlinkArgs[2]) ) {
4001 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4008 $head[$headlineCount] = $sk->makeHeadline( $level,
4009 $matches['attrib'][$headlineCount], $anchor, $headline,
4010 $editlink, $legacyAnchor );
4015 $this->setOutputType( $oldType );
4017 # Never ever show TOC if no headers
4018 if ( $numVisible < 1 ) {
4023 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4024 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4026 $toc = $sk->tocList( $toc, $this->mOptions
->getUserLang() );
4027 $this->mOutput
->setTOCHTML( $toc );
4031 $this->mOutput
->setSections( $tocraw );
4034 # split up and insert constructed headlines
4036 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4039 foreach ( $blocks as $block ) {
4040 if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4041 # This is the [edit] link that appears for the top block of text when
4042 # section editing is enabled
4044 # Disabled because it broke block formatting
4045 # For example, a bullet point in the top line
4046 # $full .= $sk->editSectionLink(0);
4049 if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition
) {
4050 # Top anchor now in skin
4054 if ( !empty( $head[$i] ) ) {
4059 if ( $this->mForceTocPosition
) {
4060 return str_replace( '<!--MWTOC-->', $toc, $full );
4067 * Transform wiki markup when saving a page by doing \r\n -> \n
4068 * conversion, substitting signatures, {{subst:}} templates, etc.
4070 * @param $text String: the text to transform
4071 * @param $title Title: the Title object for the current article
4072 * @param $user User: the User object describing the current user
4073 * @param $options ParserOptions: parsing options
4074 * @param $clearState Boolean: whether to clear the parser state first
4075 * @return String: the altered wiki markup
4077 public function preSaveTransform( $text, Title
$title, User
$user, ParserOptions
$options, $clearState = true ) {
4078 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4079 $this->setUser( $user );
4084 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4085 if( $options->getPreSaveTransform() ) {
4086 $text = $this->pstPass2( $text, $user );
4088 $text = $this->mStripState
->unstripBoth( $text );
4090 $this->setUser( null ); #Reset
4096 * Pre-save transform helper function
4099 function pstPass2( $text, $user ) {
4100 global $wgContLang, $wgLocaltimezone;
4102 # Note: This is the timestamp saved as hardcoded wikitext to
4103 # the database, we use $wgContLang here in order to give
4104 # everyone the same signature and use the default one rather
4105 # than the one selected in each user's preferences.
4106 # (see also bug 12815)
4107 $ts = $this->mOptions
->getTimestamp();
4108 if ( isset( $wgLocaltimezone ) ) {
4109 $tz = $wgLocaltimezone;
4111 $tz = date_default_timezone_get();
4114 $unixts = wfTimestamp( TS_UNIX
, $ts );
4115 $oldtz = date_default_timezone_get();
4116 date_default_timezone_set( $tz );
4117 $ts = date( 'YmdHis', $unixts );
4118 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4120 # Allow translation of timezones through wiki. date() can return
4121 # whatever crap the system uses, localised or not, so we cannot
4122 # ship premade translations.
4123 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4124 $msg = wfMessage( $key )->inContentLanguage();
4125 if ( $msg->exists() ) {
4126 $tzMsg = $msg->text();
4129 date_default_timezone_set( $oldtz );
4131 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4133 # Variable replacement
4134 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4135 $text = $this->replaceVariables( $text );
4137 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4138 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4141 $sigText = $this->getUserSig( $user );
4142 $text = strtr( $text, array(
4144 '~~~~' => "$sigText $d",
4148 # Context links: [[|name]] and [[name (context)|]]
4149 global $wgLegalTitleChars;
4150 $tc = "[$wgLegalTitleChars]";
4151 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4153 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4154 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(($tc+))\\|]]/"; # [[ns:page(context)|]]
4155 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4156 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4158 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4159 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4160 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4161 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4163 $t = $this->mTitle
->getText();
4165 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4166 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4167 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4168 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4170 # if there's no context, don't bother duplicating the title
4171 $text = preg_replace( $p2, '[[\\1]]', $text );
4174 # Trim trailing whitespace
4175 $text = rtrim( $text );
4181 * Fetch the user's signature text, if any, and normalize to
4182 * validated, ready-to-insert wikitext.
4183 * If you have pre-fetched the nickname or the fancySig option, you can
4184 * specify them here to save a database query.
4185 * Do not reuse this parser instance after calling getUserSig(),
4186 * as it may have changed if it's the $wgParser.
4189 * @param $nickname String: nickname to use or false to use user's default nickname
4190 * @param $fancySig Boolean: whether the nicknname is the complete signature
4191 * or null to use default value
4194 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4195 global $wgMaxSigChars;
4197 $username = $user->getName();
4199 # If not given, retrieve from the user object.
4200 if ( $nickname === false )
4201 $nickname = $user->getOption( 'nickname' );
4203 if ( is_null( $fancySig ) ) {
4204 $fancySig = $user->getBoolOption( 'fancysig' );
4207 $nickname = $nickname == null ?
$username : $nickname;
4209 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4210 $nickname = $username;
4211 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4212 } elseif ( $fancySig !== false ) {
4213 # Sig. might contain markup; validate this
4214 if ( $this->validateSig( $nickname ) !== false ) {
4215 # Validated; clean up (if needed) and return it
4216 return $this->cleanSig( $nickname, true );
4218 # Failed to validate; fall back to the default
4219 $nickname = $username;
4220 wfDebug( __METHOD__
.": $username has bad XML tags in signature.\n" );
4224 # Make sure nickname doesnt get a sig in a sig
4225 $nickname = $this->cleanSigInSig( $nickname );
4227 # If we're still here, make it a link to the user page
4228 $userText = wfEscapeWikiText( $username );
4229 $nickText = wfEscapeWikiText( $nickname );
4230 if ( $user->isAnon() ) {
4231 return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4233 return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4238 * Check that the user's signature contains no bad XML
4240 * @param $text String
4241 * @return mixed An expanded string, or false if invalid.
4243 function validateSig( $text ) {
4244 return( Xml
::isWellFormedXmlFragment( $text ) ?
$text : false );
4248 * Clean up signature text
4250 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4251 * 2) Substitute all transclusions
4253 * @param $text String
4254 * @param $parsing Whether we're cleaning (preferences save) or parsing
4255 * @return String: signature text
4257 function cleanSig( $text, $parsing = false ) {
4260 $this->mOptions
= new ParserOptions
;
4261 $this->clearState();
4262 $this->setTitle( $wgTitle );
4263 $this->setOutputType
= self
::OT_PREPROCESS
;
4266 # Option to disable this feature
4267 if ( !$this->mOptions
->getCleanSignatures() ) {
4271 # FIXME: regex doesn't respect extension tags or nowiki
4272 # => Move this logic to braceSubstitution()
4273 $substWord = MagicWord
::get( 'subst' );
4274 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4275 $substText = '{{' . $substWord->getSynonym( 0 );
4277 $text = preg_replace( $substRegex, $substText, $text );
4278 $text = $this->cleanSigInSig( $text );
4279 $dom = $this->preprocessToDom( $text );
4280 $frame = $this->getPreprocessor()->newFrame();
4281 $text = $frame->expand( $dom );
4284 $text = $this->mStripState
->unstripBoth( $text );
4291 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4293 * @param $text String
4294 * @return String: signature text with /~{3,5}/ removed
4296 function cleanSigInSig( $text ) {
4297 $text = preg_replace( '/~{3,5}/', '', $text );
4302 * Set up some variables which are usually set up in parse()
4303 * so that an external function can call some class members with confidence
4305 public function startExternalParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4306 $this->startParse( $title, $options, $outputType, $clearState );
4309 private function startParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4310 $this->setTitle( $title );
4311 $this->mOptions
= $options;
4312 $this->setOutputType( $outputType );
4313 if ( $clearState ) {
4314 $this->clearState();
4319 * Wrapper for preprocess()
4321 * @param $text String: the text to preprocess
4322 * @param $options ParserOptions: options
4323 * @param $title Title object or null to use $wgTitle
4326 public function transformMsg( $text, $options, $title = null ) {
4327 static $executing = false;
4329 # Guard against infinite recursion
4335 wfProfileIn( __METHOD__
);
4341 # It's not uncommon having a null $wgTitle in scripts. See r80898
4342 # Create a ghost title in such case
4343 $title = Title
::newFromText( 'Dwimmerlaik' );
4345 $text = $this->preprocess( $text, $title, $options );
4348 wfProfileOut( __METHOD__
);
4353 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4354 * The callback should have the following form:
4355 * function myParserHook( $text, $params, $parser ) { ... }
4357 * Transform and return $text. Use $parser for any required context, e.g. use
4358 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4360 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4361 * @param $callback Mixed: the callback function (and object) to use for the tag
4362 * @return The old value of the mTagHooks array associated with the hook
4364 public function setHook( $tag, $callback ) {
4365 $tag = strtolower( $tag );
4366 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4367 $oldVal = isset( $this->mTagHooks
[$tag] ) ?
$this->mTagHooks
[$tag] : null;
4368 $this->mTagHooks
[$tag] = $callback;
4369 if ( !in_array( $tag, $this->mStripList
) ) {
4370 $this->mStripList
[] = $tag;
4376 function setTransparentTagHook( $tag, $callback ) {
4377 $tag = strtolower( $tag );
4378 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4379 $oldVal = isset( $this->mTransparentTagHooks
[$tag] ) ?
$this->mTransparentTagHooks
[$tag] : null;
4380 $this->mTransparentTagHooks
[$tag] = $callback;
4386 * Remove all tag hooks
4388 function clearTagHooks() {
4389 $this->mTagHooks
= array();
4390 $this->mStripList
= $this->mDefaultStripList
;
4394 * Create a function, e.g. {{sum:1|2|3}}
4395 * The callback function should have the form:
4396 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4398 * Or with SFH_OBJECT_ARGS:
4399 * function myParserFunction( $parser, $frame, $args ) { ... }
4401 * The callback may either return the text result of the function, or an array with the text
4402 * in element 0, and a number of flags in the other elements. The names of the flags are
4403 * specified in the keys. Valid flags are:
4404 * found The text returned is valid, stop processing the template. This
4406 * nowiki Wiki markup in the return value should be escaped
4407 * isHTML The returned text is HTML, armour it against wikitext transformation
4409 * @param $id String: The magic word ID
4410 * @param $callback Mixed: the callback function (and object) to use
4411 * @param $flags Integer: a combination of the following flags:
4412 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4414 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4415 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4416 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4417 * the arguments, and to control the way they are expanded.
4419 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4420 * arguments, for instance:
4421 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4423 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4424 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4425 * working if/when this is changed.
4427 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4430 * Please read the documentation in includes/parser/Preprocessor.php for more information
4431 * about the methods available in PPFrame and PPNode.
4433 * @return The old callback function for this name, if any
4435 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4438 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4439 $this->mFunctionHooks
[$id] = array( $callback, $flags );
4441 # Add to function cache
4442 $mw = MagicWord
::get( $id );
4444 throw new MWException( __METHOD__
.'() expecting a magic word identifier.' );
4446 $synonyms = $mw->getSynonyms();
4447 $sensitive = intval( $mw->isCaseSensitive() );
4449 foreach ( $synonyms as $syn ) {
4451 if ( !$sensitive ) {
4452 $syn = $wgContLang->lc( $syn );
4455 if ( !( $flags & SFH_NO_HASH
) ) {
4458 # Remove trailing colon
4459 if ( substr( $syn, -1, 1 ) === ':' ) {
4460 $syn = substr( $syn, 0, -1 );
4462 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4468 * Get all registered function hook identifiers
4472 function getFunctionHooks() {
4473 return array_keys( $this->mFunctionHooks
);
4477 * Create a tag function, e.g. <test>some stuff</test>.
4478 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4479 * Unlike parser functions, their content is not preprocessed.
4481 function setFunctionTagHook( $tag, $callback, $flags ) {
4482 $tag = strtolower( $tag );
4483 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4484 $old = isset( $this->mFunctionTagHooks
[$tag] ) ?
4485 $this->mFunctionTagHooks
[$tag] : null;
4486 $this->mFunctionTagHooks
[$tag] = array( $callback, $flags );
4488 if ( !in_array( $tag, $this->mStripList
) ) {
4489 $this->mStripList
[] = $tag;
4496 * FIXME: update documentation. makeLinkObj() is deprecated.
4497 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4498 * Placeholders created in Skin::makeLinkObj()
4499 * Returns an array of link CSS classes, indexed by PDBK.
4501 function replaceLinkHolders( &$text, $options = 0 ) {
4502 return $this->mLinkHolders
->replace( $text );
4506 * Replace <!--LINK--> link placeholders with plain text of links
4507 * (not HTML-formatted).
4509 * @param $text String
4512 function replaceLinkHoldersText( $text ) {
4513 return $this->mLinkHolders
->replaceText( $text );
4517 * Renders an image gallery from a text with one line per image.
4518 * text labels may be given by using |-style alternative text. E.g.
4519 * Image:one.jpg|The number "1"
4520 * Image:tree.jpg|A tree
4521 * given as text will return the HTML of a gallery with two images,
4522 * labeled 'The number "1"' and
4525 function renderImageGallery( $text, $params ) {
4526 $ig = new ImageGallery();
4527 $ig->setContextTitle( $this->mTitle
);
4528 $ig->setShowBytes( false );
4529 $ig->setShowFilename( false );
4530 $ig->setParser( $this );
4531 $ig->setHideBadImages();
4532 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'table' ) );
4533 $ig->useSkin( $this->mOptions
->getSkin( $this->mTitle
) );
4535 if ( isset( $params['showfilename'] ) ) {
4536 $ig->setShowFilename( true );
4538 $ig->setShowFilename( false );
4540 if ( isset( $params['caption'] ) ) {
4541 $caption = $params['caption'];
4542 $caption = htmlspecialchars( $caption );
4543 $caption = $this->replaceInternalLinks( $caption );
4544 $ig->setCaptionHtml( $caption );
4546 if ( isset( $params['perrow'] ) ) {
4547 $ig->setPerRow( $params['perrow'] );
4549 if ( isset( $params['widths'] ) ) {
4550 $ig->setWidths( $params['widths'] );
4552 if ( isset( $params['heights'] ) ) {
4553 $ig->setHeights( $params['heights'] );
4556 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4558 $lines = StringUtils
::explode( "\n", $text );
4559 foreach ( $lines as $line ) {
4560 # match lines like these:
4561 # Image:someimage.jpg|This is some image
4563 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4565 if ( count( $matches ) == 0 ) {
4569 if ( strpos( $matches[0], '%' ) !== false ) {
4570 $matches[1] = rawurldecode( $matches[1] );
4572 $tp = Title
::newFromText( $matches[1], NS_FILE
);
4574 if ( is_null( $nt ) ) {
4575 # Bogus title. Ignore these so we don't bomb out later.
4578 if ( isset( $matches[3] ) ) {
4579 $label = $matches[3];
4584 $html = $this->recursiveTagParse( trim( $label ) );
4586 $ig->add( $nt, $html );
4588 return $ig->toHTML();
4591 function getImageParams( $handler ) {
4593 $handlerClass = get_class( $handler );
4597 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
4598 # Initialise static lists
4599 static $internalParamNames = array(
4600 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4601 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4602 'bottom', 'text-bottom' ),
4603 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4604 'upright', 'border', 'link', 'alt' ),
4606 static $internalParamMap;
4607 if ( !$internalParamMap ) {
4608 $internalParamMap = array();
4609 foreach ( $internalParamNames as $type => $names ) {
4610 foreach ( $names as $name ) {
4611 $magicName = str_replace( '-', '_', "img_$name" );
4612 $internalParamMap[$magicName] = array( $type, $name );
4617 # Add handler params
4618 $paramMap = $internalParamMap;
4620 $handlerParamMap = $handler->getParamMap();
4621 foreach ( $handlerParamMap as $magic => $paramName ) {
4622 $paramMap[$magic] = array( 'handler', $paramName );
4625 $this->mImageParams
[$handlerClass] = $paramMap;
4626 $this->mImageParamsMagicArray
[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4628 return array( $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] );
4632 * Parse image options text and use it to make an image
4634 * @param $title Title
4635 * @param $options String
4636 * @param $holders LinkHolderArray
4637 * @return string HTML
4639 function makeImage( $title, $options, $holders = false ) {
4640 # Check if the options text is of the form "options|alt text"
4642 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4643 # * left no resizing, just left align. label is used for alt= only
4644 # * right same, but right aligned
4645 # * none same, but not aligned
4646 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4647 # * center center the image
4648 # * frame Keep original image size, no magnify-button.
4649 # * framed Same as "frame"
4650 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4651 # * upright reduce width for upright images, rounded to full __0 px
4652 # * border draw a 1px border around the image
4653 # * alt Text for HTML alt attribute (defaults to empty)
4654 # * link Set the target of the image link. Can be external, interwiki, or local
4655 # vertical-align values (no % or length right now):
4665 $parts = StringUtils
::explode( "|", $options );
4666 $sk = $this->mOptions
->getSkin( $this->mTitle
);
4668 # Give extensions a chance to select the file revision for us
4669 $skip = $time = $sha1 = $descQuery = false;
4670 wfRunHooks( 'BeforeParserMakeImageLinkObj',
4671 array( &$this, &$title, &$skip, &$time, &$descQuery, &$sha1 ) );
4673 $this->mOutput
->addImage( $title->getDBkey() ); // register
4674 return $sk->link( $title );
4676 # Fetch and register the file
4677 $file = $this->fetchFile( $title, $time, $sha1 );
4679 $title = $file->getTitle(); // file title may be different (via hooks)
4682 $handler = $file ?
$file->getHandler() : false;
4684 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4686 # Process the input parameters
4688 $params = array( 'frame' => array(), 'handler' => array(),
4689 'horizAlign' => array(), 'vertAlign' => array() );
4690 foreach ( $parts as $part ) {
4691 $part = trim( $part );
4692 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4694 if ( isset( $paramMap[$magicName] ) ) {
4695 list( $type, $paramName ) = $paramMap[$magicName];
4697 # Special case; width and height come in one variable together
4698 if ( $type === 'handler' && $paramName === 'width' ) {
4700 # (bug 13500) In both cases (width/height and width only),
4701 # permit trailing "px" for backward compatibility.
4702 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4703 $width = intval( $m[1] );
4704 $height = intval( $m[2] );
4705 if ( $handler->validateParam( 'width', $width ) ) {
4706 $params[$type]['width'] = $width;
4709 if ( $handler->validateParam( 'height', $height ) ) {
4710 $params[$type]['height'] = $height;
4713 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4714 $width = intval( $value );
4715 if ( $handler->validateParam( 'width', $width ) ) {
4716 $params[$type]['width'] = $width;
4719 } # else no validation -- bug 13436
4721 if ( $type === 'handler' ) {
4722 # Validate handler parameter
4723 $validated = $handler->validateParam( $paramName, $value );
4725 # Validate internal parameters
4726 switch( $paramName ) {
4729 # @todo Fixme: possibly check validity here for
4730 # manualthumb? downstream behavior seems odd with
4731 # missing manual thumbs.
4733 $value = $this->stripAltText( $value, $holders );
4736 $chars = self
::EXT_LINK_URL_CLASS
;
4737 $prots = $this->mUrlProtocols
;
4738 if ( $value === '' ) {
4739 $paramName = 'no-link';
4742 } elseif ( preg_match( "/^$prots/", $value ) ) {
4743 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4744 $paramName = 'link-url';
4745 $this->mOutput
->addExternalLink( $value );
4746 if ( $this->mOptions
->getExternalLinkTarget() ) {
4747 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
4752 $linkTitle = Title
::newFromText( $value );
4754 $paramName = 'link-title';
4755 $value = $linkTitle;
4756 $this->mOutput
->addLink( $linkTitle );
4762 # Most other things appear to be empty or numeric...
4763 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
4768 $params[$type][$paramName] = $value;
4772 if ( !$validated ) {
4777 # Process alignment parameters
4778 if ( $params['horizAlign'] ) {
4779 $params['frame']['align'] = key( $params['horizAlign'] );
4781 if ( $params['vertAlign'] ) {
4782 $params['frame']['valign'] = key( $params['vertAlign'] );
4785 $params['frame']['caption'] = $caption;
4787 # Will the image be presented in a frame, with the caption below?
4788 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4789 isset( $params['frame']['framed'] ) ||
4790 isset( $params['frame']['thumbnail'] ) ||
4791 isset( $params['frame']['manualthumb'] );
4793 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
4794 # came to also set the caption, ordinary text after the image -- which
4795 # makes no sense, because that just repeats the text multiple times in
4796 # screen readers. It *also* came to set the title attribute.
4798 # Now that we have an alt attribute, we should not set the alt text to
4799 # equal the caption: that's worse than useless, it just repeats the
4800 # text. This is the framed/thumbnail case. If there's no caption, we
4801 # use the unnamed parameter for alt text as well, just for the time be-
4802 # ing, if the unnamed param is set and the alt param is not.
4804 # For the future, we need to figure out if we want to tweak this more,
4805 # e.g., introducing a title= parameter for the title; ignoring the un-
4806 # named parameter entirely for images without a caption; adding an ex-
4807 # plicit caption= parameter and preserving the old magic unnamed para-
4809 if ( $imageIsFramed ) { # Framed image
4810 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4811 # No caption or alt text, add the filename as the alt text so
4812 # that screen readers at least get some description of the image
4813 $params['frame']['alt'] = $title->getText();
4815 # Do not set $params['frame']['title'] because tooltips don't make sense
4817 } else { # Inline image
4818 if ( !isset( $params['frame']['alt'] ) ) {
4819 # No alt text, use the "caption" for the alt text
4820 if ( $caption !== '') {
4821 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4823 # No caption, fall back to using the filename for the
4825 $params['frame']['alt'] = $title->getText();
4828 # Use the "caption" for the tooltip text
4829 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4832 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4834 # Linker does the rest
4835 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'],
4836 $time, $descQuery, $this->mOptions
->getThumbSize() );
4838 # Give the handler a chance to modify the parser object
4840 $handler->parserTransformHook( $this, $file );
4846 protected function stripAltText( $caption, $holders ) {
4847 # Strip bad stuff out of the title (tooltip). We can't just use
4848 # replaceLinkHoldersText() here, because if this function is called
4849 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4851 $tooltip = $holders->replaceText( $caption );
4853 $tooltip = $this->replaceLinkHoldersText( $caption );
4856 # make sure there are no placeholders in thumbnail attributes
4857 # that are later expanded to html- so expand them now and
4859 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
4860 $tooltip = Sanitizer
::stripAllTags( $tooltip );
4866 * Set a flag in the output object indicating that the content is dynamic and
4867 * shouldn't be cached.
4869 function disableCache() {
4870 wfDebug( "Parser output marked as uncacheable.\n" );
4871 $this->mOutput
->setCacheTime( -1 ); // old style, for compatibility
4872 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
4876 * Callback from the Sanitizer for expanding items found in HTML attribute
4877 * values, so they can be safely tested and escaped.
4879 * @param $text String
4880 * @param $frame PPFrame
4884 function attributeStripCallback( &$text, $frame = false ) {
4885 $text = $this->replaceVariables( $text, $frame );
4886 $text = $this->mStripState
->unstripBoth( $text );
4893 function getTags() {
4894 return array_merge( array_keys( $this->mTransparentTagHooks
), array_keys( $this->mTagHooks
) );
4898 * Replace transparent tags in $text with the values given by the callbacks.
4900 * Transparent tag hooks are like regular XML-style tag hooks, except they
4901 * operate late in the transformation sequence, on HTML instead of wikitext.
4903 function replaceTransparentTags( $text ) {
4905 $elements = array_keys( $this->mTransparentTagHooks
);
4906 $text = $this->extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix
);
4908 foreach ( $matches as $marker => $data ) {
4909 list( $element, $content, $params, $tag ) = $data;
4910 $tagName = strtolower( $element );
4911 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
4912 $output = call_user_func_array( $this->mTransparentTagHooks
[$tagName], array( $content, $params, $this ) );
4916 $this->mStripState
->addGeneral( $marker, $output );
4922 * Break wikitext input into sections, and either pull or replace
4923 * some particular section's text.
4925 * External callers should use the getSection and replaceSection methods.
4927 * @param $text String: Page wikitext
4928 * @param $section String: a section identifier string of the form:
4929 * <flag1> - <flag2> - ... - <section number>
4931 * Currently the only recognised flag is "T", which means the target section number
4932 * was derived during a template inclusion parse, in other words this is a template
4933 * section edit link. If no flags are given, it was an ordinary section edit link.
4934 * This flag is required to avoid a section numbering mismatch when a section is
4935 * enclosed by <includeonly> (bug 6563).
4937 * The section number 0 pulls the text before the first heading; other numbers will
4938 * pull the given section along with its lower-level subsections. If the section is
4939 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4941 * @param $mode String: one of "get" or "replace"
4942 * @param $newText String: replacement text for section data.
4943 * @return String: for "get", the extracted section text.
4944 * for "replace", the whole page with the section replaced.
4946 private function extractSections( $text, $section, $mode, $newText='' ) {
4947 global $wgTitle; # not generally used but removes an ugly failure mode
4948 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
4950 $frame = $this->getPreprocessor()->newFrame();
4952 # Process section extraction flags
4954 $sectionParts = explode( '-', $section );
4955 $sectionIndex = array_pop( $sectionParts );
4956 foreach ( $sectionParts as $part ) {
4957 if ( $part === 'T' ) {
4958 $flags |
= self
::PTD_FOR_INCLUSION
;
4961 # Preprocess the text
4962 $root = $this->preprocessToDom( $text, $flags );
4964 # <h> nodes indicate section breaks
4965 # They can only occur at the top level, so we can find them by iterating the root's children
4966 $node = $root->getFirstChild();
4968 # Find the target section
4969 if ( $sectionIndex == 0 ) {
4970 # Section zero doesn't nest, level=big
4971 $targetLevel = 1000;
4973 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
4978 if ( $node->getName() === 'h' ) {
4979 $bits = $node->splitHeading();
4980 if ( $bits['i'] == $sectionIndex ) {
4981 $targetLevel = $bits['level'];
4985 if ( $mode === 'replace' ) {
4986 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
4988 $node = $node->getNextSibling();
4994 if ( $mode === 'get' ) {
5001 # Find the end of the section, including nested sections
5003 if ( $node->getName() === 'h' ) {
5004 $bits = $node->splitHeading();
5005 $curLevel = $bits['level'];
5006 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5010 if ( $mode === 'get' ) {
5011 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5013 $node = $node->getNextSibling();
5016 # Write out the remainder (in replace mode only)
5017 if ( $mode === 'replace' ) {
5018 # Output the replacement text
5019 # Add two newlines on -- trailing whitespace in $newText is conventionally
5020 # stripped by the editor, so we need both newlines to restore the paragraph gap
5021 # Only add trailing whitespace if there is newText
5022 if ( $newText != "" ) {
5023 $outText .= $newText . "\n\n";
5027 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5028 $node = $node->getNextSibling();
5032 if ( is_string( $outText ) ) {
5033 # Re-insert stripped tags
5034 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5041 * This function returns the text of a section, specified by a number ($section).
5042 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5043 * the first section before any such heading (section 0).
5045 * If a section contains subsections, these are also returned.
5047 * @param $text String: text to look in
5048 * @param $section String: section identifier
5049 * @param $deftext String: default to return if section is not found
5050 * @return string text of the requested section
5052 public function getSection( $text, $section, $deftext='' ) {
5053 return $this->extractSections( $text, $section, "get", $deftext );
5057 * This function returns $oldtext after the content of the section
5058 * specified by $section has been replaced with $text.
5060 * @param $text String: former text of the article
5061 * @param $section Numeric: section identifier
5062 * @param $text String: replacing text
5063 * #return String: modified text
5065 public function replaceSection( $oldtext, $section, $text ) {
5066 return $this->extractSections( $oldtext, $section, "replace", $text );
5070 * Get the ID of the revision we are parsing
5072 * @return Mixed: integer or null
5074 function getRevisionId() {
5075 return $this->mRevisionId
;
5079 * Get the revision object for $this->mRevisionId
5081 * @return either a Revision object or null
5083 protected function getRevisionObject() {
5084 if ( !is_null( $this->mRevisionObject
) ) {
5085 return $this->mRevisionObject
;
5087 if ( is_null( $this->mRevisionId
) ) {
5091 $this->mRevisionObject
= Revision
::newFromId( $this->mRevisionId
);
5092 return $this->mRevisionObject
;
5096 * Get the timestamp associated with the current revision, adjusted for
5097 * the default server-local timestamp
5099 function getRevisionTimestamp() {
5100 if ( is_null( $this->mRevisionTimestamp
) ) {
5101 wfProfileIn( __METHOD__
);
5103 $revObject = $this->getRevisionObject();
5104 $timestamp = $revObject ?
$revObject->getTimestamp() : false;
5106 if( $timestamp !== false ) {
5109 # The cryptic '' timezone parameter tells to use the site-default
5110 # timezone offset instead of the user settings.
5112 # Since this value will be saved into the parser cache, served
5113 # to other users, and potentially even used inside links and such,
5114 # it needs to be consistent for all visitors.
5115 $this->mRevisionTimestamp
= $wgContLang->userAdjust( $timestamp, '' );
5118 wfProfileOut( __METHOD__
);
5120 return $this->mRevisionTimestamp
;
5124 * Get the name of the user that edited the last revision
5126 * @return String: user name
5128 function getRevisionUser() {
5129 if( is_null( $this->mRevisionUser
) ) {
5130 $revObject = $this->getRevisionObject();
5132 # if this template is subst: the revision id will be blank,
5133 # so just use the current user's name
5135 $this->mRevisionUser
= $revObject->getUserText();
5136 } elseif( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5137 $this->mRevisionUser
= $this->getUser()->getName();
5140 return $this->mRevisionUser
;
5144 * Mutator for $mDefaultSort
5146 * @param $sort New value
5148 public function setDefaultSort( $sort ) {
5149 $this->mDefaultSort
= $sort;
5150 $this->mOutput
->setProperty( 'defaultsort', $sort );
5154 * Accessor for $mDefaultSort
5155 * Will use the empty string if none is set.
5157 * This value is treated as a prefix, so the
5158 * empty string is equivalent to sorting by
5163 public function getDefaultSort() {
5164 if ( $this->mDefaultSort
!== false ) {
5165 return $this->mDefaultSort
;
5172 * Accessor for $mDefaultSort
5173 * Unlike getDefaultSort(), will return false if none is set
5175 * @return string or false
5177 public function getCustomDefaultSort() {
5178 return $this->mDefaultSort
;
5182 * Try to guess the section anchor name based on a wikitext fragment
5183 * presumably extracted from a heading, for example "Header" from
5186 public function guessSectionNameFromWikiText( $text ) {
5187 # Strip out wikitext links(they break the anchor)
5188 $text = $this->stripSectionName( $text );
5189 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5190 return '#' . Sanitizer
::escapeId( $text, 'noninitial' );
5194 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5195 * instead. For use in redirects, since IE6 interprets Redirect: headers
5196 * as something other than UTF-8 (apparently?), resulting in breakage.
5198 * @param $text String: The section name
5199 * @return string An anchor
5201 public function guessLegacySectionNameFromWikiText( $text ) {
5202 # Strip out wikitext links(they break the anchor)
5203 $text = $this->stripSectionName( $text );
5204 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5205 return '#' . Sanitizer
::escapeId( $text, array( 'noninitial', 'legacy' ) );
5209 * Strips a text string of wikitext for use in a section anchor
5211 * Accepts a text string and then removes all wikitext from the
5212 * string and leaves only the resultant text (i.e. the result of
5213 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5214 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5215 * to create valid section anchors by mimicing the output of the
5216 * parser when headings are parsed.
5218 * @param $text String: text string to be stripped of wikitext
5219 * for use in a Section anchor
5220 * @return Filtered text string
5222 public function stripSectionName( $text ) {
5223 # Strip internal link markup
5224 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5225 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5227 # Strip external link markup (FIXME: Not Tolerant to blank link text
5228 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5229 # on how many empty links there are on the page - need to figure that out.
5230 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5232 # Parse wikitext quotes (italics & bold)
5233 $text = $this->doQuotes( $text );
5236 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
5241 * strip/replaceVariables/unstrip for preprocessor regression testing
5243 function testSrvus( $text, $title, ParserOptions
$options, $outputType = self
::OT_HTML
) {
5244 if ( !$title instanceof Title
) {
5245 $title = Title
::newFromText( $title );
5247 $this->startParse( $title, $options, $outputType, true );
5249 $text = $this->replaceVariables( $text );
5250 $text = $this->mStripState
->unstripBoth( $text );
5251 $text = Sanitizer
::removeHTMLtags( $text );
5255 function testPst( $text, $title, $options ) {
5257 if ( !$title instanceof Title
) {
5258 $title = Title
::newFromText( $title );
5260 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5263 function testPreprocess( $text, $title, $options ) {
5264 if ( !$title instanceof Title
) {
5265 $title = Title
::newFromText( $title );
5267 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
5271 * Call a callback function on all regions of the given text that are not
5272 * inside strip markers, and replace those regions with the return value
5273 * of the callback. For example, with input:
5277 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5278 * two strings will be replaced with the value returned by the callback in
5281 function markerSkipCallback( $s, $callback ) {
5284 while ( $i < strlen( $s ) ) {
5285 $markerStart = strpos( $s, $this->mUniqPrefix
, $i );
5286 if ( $markerStart === false ) {
5287 $out .= call_user_func( $callback, substr( $s, $i ) );
5290 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5291 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
5292 if ( $markerEnd === false ) {
5293 $out .= substr( $s, $markerStart );
5296 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
5297 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5306 * Save the parser state required to convert the given half-parsed text to
5307 * HTML. "Half-parsed" in this context means the output of
5308 * recursiveTagParse() or internalParse(). This output has strip markers
5309 * from replaceVariables (extensionSubstitution() etc.), and link
5310 * placeholders from replaceLinkHolders().
5312 * Returns an array which can be serialized and stored persistently. This
5313 * array can later be loaded into another parser instance with
5314 * unserializeHalfParsedText(). The text can then be safely incorporated into
5315 * the return value of a parser hook.
5317 function serializeHalfParsedText( $text ) {
5318 wfProfileIn( __METHOD__
);
5321 'version' => self
::HALF_PARSED_VERSION
,
5322 'stripState' => $this->mStripState
->getSubState( $text ),
5323 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
5325 wfProfileOut( __METHOD__
);
5330 * Load the parser state given in the $data array, which is assumed to
5331 * have been generated by serializeHalfParsedText(). The text contents is
5332 * extracted from the array, and its markers are transformed into markers
5333 * appropriate for the current Parser instance. This transformed text is
5334 * returned, and can be safely included in the return value of a parser
5337 * If the $data array has been stored persistently, the caller should first
5338 * check whether it is still valid, by calling isValidHalfParsedText().
5340 * @param $data Serialized data
5343 function unserializeHalfParsedText( $data ) {
5344 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
5345 throw new MWException( __METHOD__
.': invalid version' );
5348 # First, extract the strip state.
5349 $texts = array( $data['text'] );
5350 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
5352 # Now renumber links
5353 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
5355 # Should be good to go.
5360 * Returns true if the given array, presumed to be generated by
5361 * serializeHalfParsedText(), is compatible with the current version of the
5364 * @param $data Array.
5366 function isValidHalfParsedText( $data ) {
5367 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;