3 * @defgroup Parser Parser
7 * File for Parser and related classes
12 * PHP Parser - Processes wiki markup (which uses a more user-friendly
13 * syntax, such as "[[link]]" for making links), and provides a one-way
14 * transformation of that wiki markup it into XHTML output / markup
15 * (which in turn the browser understands, and can display).
18 * There are five main entry points into the Parser class:
20 * produces HTML output
22 * produces altered wiki markup.
24 * removes HTML comments and expands templates
25 * cleanSig() / cleanSigInSig()
26 * Cleans a signature before saving it to preferences
28 * Return the content of a section from an article for section editing
30 * Replaces a section by number inside an article
32 * Removes <noinclude> sections, and <includeonly> tags.
35 * objects: $wgLang, $wgContLang
37 * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
40 * $wgUseDynamicDates*, $wgInterwikiMagic*,
41 * $wgNamespacesWithSubpages, $wgAllowExternalImages*,
42 * $wgLocaltimezone, $wgAllowSpecialInclusion*,
45 * * only within ParserOptions
52 * Update this version number when the ParserOutput format
53 * changes in an incompatible way, so the parser cache
54 * can automatically discard old data.
56 const VERSION
= '1.6.5';
59 * Update this version number when the output of serialiseHalfParsedText()
60 * changes in an incompatible way
62 const HALF_PARSED_VERSION
= 2;
64 # Flags for Parser::setFunctionHook
65 # Also available as global constants from Defines.php
66 const SFH_NO_HASH
= 1;
67 const SFH_OBJECT_ARGS
= 2;
69 # Constants needed for external link processing
70 # Everything except bracket, space, or control characters
71 const EXT_LINK_URL_CLASS
= '(?:[^\]\[<>"\\x00-\\x20\\x7F]|(?:\[\]))';
72 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
73 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
75 # State constants for the definition list colon extraction
76 const COLON_STATE_TEXT
= 0;
77 const COLON_STATE_TAG
= 1;
78 const COLON_STATE_TAGSTART
= 2;
79 const COLON_STATE_CLOSETAG
= 3;
80 const COLON_STATE_TAGSLASH
= 4;
81 const COLON_STATE_COMMENT
= 5;
82 const COLON_STATE_COMMENTDASH
= 6;
83 const COLON_STATE_COMMENTDASHDASH
= 7;
85 # Flags for preprocessToDom
86 const PTD_FOR_INCLUSION
= 1;
88 # Allowed values for $this->mOutputType
89 # Parameter to startExternalParse().
90 const OT_HTML
= 1; # like parse()
91 const OT_WIKI
= 2; # like preSaveTransform()
92 const OT_PREPROCESS
= 3; # like preprocess()
94 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
96 # Marker Suffix needs to be accessible staticly.
97 const MARKER_SUFFIX
= "-QINU\x7f";
100 var $mTagHooks = array();
101 var $mTransparentTagHooks = array();
102 var $mFunctionHooks = array();
103 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
104 var $mFunctionTagHooks = array();
105 var $mStripList = array();
106 var $mDefaultStripList = array();
107 var $mVarCache = array();
108 var $mImageParams = array();
109 var $mImageParamsMagicArray = array();
110 var $mMarkerIndex = 0;
111 var $mFirstCall = true;
112 var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
113 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
115 # Cleared with clearState():
120 var $mAutonumber, $mDTopen;
127 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
128 var $mLinkHolders, $mLinkID;
129 var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
130 var $mTplExpandCache; # empty-frame expansion cache
131 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
132 var $mExpensiveFunctionCount; # number of expensive parser function calls
137 var $mUser; # User object; only used when doing pre-save transform
140 # These are variables reset at least once per parse regardless of $clearState
150 var $mTitle; # Title context, used for self-link rendering and similar things
151 var $mOutputType; # Output type, one of the OT_xxx constants
152 var $ot; # Shortcut alias, see setOutputType()
153 var $mRevisionObject; # The revision object of the specified revision ID
154 var $mRevisionId; # ID to display in {{REVISIONID}} tags
155 var $mRevisionTimestamp; # The timestamp of the specified revision ID
156 var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
157 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
162 public function __construct( $conf = array() ) {
163 $this->mConf
= $conf;
164 $this->mUrlProtocols
= wfUrlProtocols();
165 $this->mExtLinkBracketedRegex
= '/\[(\b(' . wfUrlProtocols() . ')'.
166 '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
167 if ( isset( $conf['preprocessorClass'] ) ) {
168 $this->mPreprocessorClass
= $conf['preprocessorClass'];
169 } elseif ( extension_loaded( 'domxml' ) ) {
170 # PECL extension that conflicts with the core DOM extension (bug 13770)
171 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
172 $this->mPreprocessorClass
= 'Preprocessor_Hash';
173 } elseif ( extension_loaded( 'dom' ) ) {
174 $this->mPreprocessorClass
= 'Preprocessor_DOM';
176 $this->mPreprocessorClass
= 'Preprocessor_Hash';
181 * Reduce memory usage to reduce the impact of circular references
183 function __destruct() {
184 if ( isset( $this->mLinkHolders
) ) {
185 unset( $this->mLinkHolders
);
187 foreach ( $this as $name => $value ) {
188 unset( $this->$name );
193 * Do various kinds of initialisation on the first call of the parser
195 function firstCallInit() {
196 if ( !$this->mFirstCall
) {
199 $this->mFirstCall
= false;
201 wfProfileIn( __METHOD__
);
203 CoreParserFunctions
::register( $this );
204 CoreTagHooks
::register( $this );
205 $this->initialiseVariables();
207 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
208 wfProfileOut( __METHOD__
);
216 function clearState() {
217 wfProfileIn( __METHOD__
);
218 if ( $this->mFirstCall
) {
219 $this->firstCallInit();
221 $this->mOutput
= new ParserOutput
;
222 $this->mOptions
->registerWatcher( array( $this->mOutput
, 'recordOption' ) );
223 $this->mAutonumber
= 0;
224 $this->mLastSection
= '';
225 $this->mDTopen
= false;
226 $this->mIncludeCount
= array();
227 $this->mArgStack
= false;
228 $this->mInPre
= false;
229 $this->mLinkHolders
= new LinkHolderArray( $this );
231 $this->mRevisionObject
= $this->mRevisionTimestamp
=
232 $this->mRevisionId
= $this->mRevisionUser
= null;
233 $this->mVarCache
= array();
237 * Prefix for temporary replacement strings for the multipass parser.
238 * \x07 should never appear in input as it's disallowed in XML.
239 * Using it at the front also gives us a little extra robustness
240 * since it shouldn't match when butted up against identifier-like
243 * Must not consist of all title characters, or else it will change
244 * the behaviour of <nowiki> in a link.
246 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
247 # Changed to \x7f to allow XML double-parsing -- TS
248 $this->mUniqPrefix
= "\x7fUNIQ" . self
::getRandomString();
249 $this->mStripState
= new StripState( $this->mUniqPrefix
);
252 # Clear these on every parse, bug 4549
253 $this->mTplExpandCache
= $this->mTplRedirCache
= $this->mTplDomCache
= array();
255 $this->mShowToc
= true;
256 $this->mForceTocPosition
= false;
257 $this->mIncludeSizes
= array(
261 $this->mPPNodeCount
= 0;
262 $this->mDefaultSort
= false;
263 $this->mHeadings
= array();
264 $this->mDoubleUnderscores
= array();
265 $this->mExpensiveFunctionCount
= 0;
268 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
269 $this->mPreprocessor
= null;
272 wfRunHooks( 'ParserClearState', array( &$this ) );
273 wfProfileOut( __METHOD__
);
277 * Convert wikitext to HTML
278 * Do not call this function recursively.
280 * @param $text String: text we want to parse
281 * @param $title Title object
282 * @param $options ParserOptions
283 * @param $linestart boolean
284 * @param $clearState boolean
285 * @param $revid Int: number to pass in {{REVISIONID}}
286 * @return ParserOutput a ParserOutput
288 public function parse( $text, Title
$title, ParserOptions
$options, $linestart = true, $clearState = true, $revid = null ) {
290 * First pass--just handle <nowiki> sections, pass the rest off
291 * to internalParse() which does all the real work.
294 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
295 $fname = __METHOD__
.'-' . wfGetCaller();
296 wfProfileIn( __METHOD__
);
297 wfProfileIn( $fname );
299 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
301 $oldRevisionId = $this->mRevisionId
;
302 $oldRevisionObject = $this->mRevisionObject
;
303 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
304 $oldRevisionUser = $this->mRevisionUser
;
305 if ( $revid !== null ) {
306 $this->mRevisionId
= $revid;
307 $this->mRevisionObject
= null;
308 $this->mRevisionTimestamp
= null;
309 $this->mRevisionUser
= null;
312 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
314 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
315 $text = $this->internalParse( $text );
317 $text = $this->mStripState
->unstripGeneral( $text );
319 # Clean up special characters, only run once, next-to-last before doBlockLevels
321 # french spaces, last one Guillemet-left
322 # only if there is something before the space
323 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1 \\2',
324 # french spaces, Guillemet-right
325 '/(\\302\\253) /' => '\\1 ',
326 '/ (!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
328 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
330 $text = $this->doBlockLevels( $text, $linestart );
332 $this->replaceLinkHolders( $text );
335 * The page doesn't get language converted if
337 * b) Content isn't converted
338 * c) It's a conversion table
340 if ( !( $wgDisableLangConversion
341 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
342 ||
$this->mTitle
->isConversionTable() ) ) {
344 # The position of the convert() call should not be changed. it
345 # assumes that the links are all replaced and the only thing left
346 # is the <nowiki> mark.
348 $text = $wgContLang->convert( $text );
352 * A converted title will be provided in the output object if title and
353 * content conversion are enabled, the article text does not contain
354 * a conversion-suppressing double-underscore tag, and no
355 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
356 * automatic link conversion.
358 if ( !( $wgDisableLangConversion
359 ||
$wgDisableTitleConversion
360 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
361 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
362 ||
$this->mOutput
->getDisplayTitle() !== false ) )
364 $convruletitle = $wgContLang->getConvRuleTitle();
365 if ( $convruletitle ) {
366 $this->mOutput
->setTitleText( $convruletitle );
368 $titleText = $wgContLang->convertTitle( $title );
369 $this->mOutput
->setTitleText( $titleText );
373 $text = $this->mStripState
->unstripNoWiki( $text );
375 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
377 $text = $this->replaceTransparentTags( $text );
378 $text = $this->mStripState
->unstripGeneral( $text );
380 $text = Sanitizer
::normalizeCharReferences( $text );
382 if ( ( $wgUseTidy && $this->mOptions
->getTidy() ) ||
$wgAlwaysUseTidy ) {
383 $text = MWTidy
::tidy( $text );
385 # attempt to sanitize at least some nesting problems
386 # (bug #2702 and quite a few others)
388 # ''Something [http://www.cool.com cool''] -->
389 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
390 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
391 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
392 # fix up an anchor inside another anchor, only
393 # at least for a single single nested link (bug 3695)
394 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
395 '\\1\\2</a>\\3</a>\\1\\4</a>',
396 # fix div inside inline elements- doBlockLevels won't wrap a line which
397 # contains a div, so fix it up here; replace
398 # div with escaped text
399 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
400 '\\1\\3<div\\5>\\6</div>\\8\\9',
401 # remove empty italic or bold tag pairs, some
402 # introduced by rules above
403 '/<([bi])><\/\\1>/' => '',
406 $text = preg_replace(
407 array_keys( $tidyregs ),
408 array_values( $tidyregs ),
411 global $wgExpensiveParserFunctionLimit;
412 if ( $this->mExpensiveFunctionCount
> $wgExpensiveParserFunctionLimit ) {
413 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount
, $wgExpensiveParserFunctionLimit );
416 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
418 # Information on include size limits, for the benefit of users who try to skirt them
419 if ( $this->mOptions
->getEnableLimitReport() ) {
420 $max = $this->mOptions
->getMaxIncludeSize();
421 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
423 "NewPP limit report\n" .
424 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
425 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
426 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
428 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
429 $text .= "\n<!-- \n$limitReport-->\n";
431 $this->mOutput
->setText( $text );
433 $this->mRevisionId
= $oldRevisionId;
434 $this->mRevisionObject
= $oldRevisionObject;
435 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
436 $this->mRevisionUser
= $oldRevisionUser;
437 wfProfileOut( $fname );
438 wfProfileOut( __METHOD__
);
440 return $this->mOutput
;
444 * Recursive parser entry point that can be called from an extension tag
447 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
449 * @param $text String: text extension wants to have parsed
450 * @param $frame PPFrame: The frame to use for expanding any template variables
452 function recursiveTagParse( $text, $frame=false ) {
453 wfProfileIn( __METHOD__
);
454 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
455 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
456 $text = $this->internalParse( $text, false, $frame );
457 wfProfileOut( __METHOD__
);
462 * Expand templates and variables in the text, producing valid, static wikitext.
463 * Also removes comments.
465 function preprocess( $text, Title
$title, ParserOptions
$options, $revid = null ) {
466 wfProfileIn( __METHOD__
);
467 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
468 if ( $revid !== null ) {
469 $this->mRevisionId
= $revid;
471 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState
) );
472 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState
) );
473 $text = $this->replaceVariables( $text );
474 $text = $this->mStripState
->unstripBoth( $text );
475 wfProfileOut( __METHOD__
);
480 * Process the wikitext for the ?preload= feature. (bug 5210)
482 * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
483 * comments, templates, arguments, tags hooks and parser functions are untouched.
485 public function getPreloadText( $text, Title
$title, ParserOptions
$options ) {
486 # Parser (re)initialisation
487 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
489 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
490 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
491 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
492 $text = $this->mStripState
->unstripBoth( $text );
497 * Get a random string
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
686 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
691 $taglist = implode( '|', $elements );
692 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
694 while ( $text != '' ) {
695 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
697 if ( count( $p ) < 5 ) {
700 if ( count( $p ) > 5 ) {
714 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
715 $stripped .= $marker;
717 if ( $close === '/>' ) {
718 # Empty element tag, <tag />
723 if ( $element === '!--' ) {
726 $end = "/(<\\/$element\\s*>)/i";
728 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
730 if ( count( $q ) < 3 ) {
731 # No end tag -- let it run out to the end of the text.
740 $matches[$marker] = array( $element,
742 Sanitizer
::decodeTagAttributes( $attributes ),
743 "<$element$attributes$close$content$tail" );
749 * Get a list of strippable XML-like elements
751 function getStripList() {
752 return $this->mStripList
;
756 * Add an item to the strip state
757 * Returns the unique tag which must be inserted into the stripped text
758 * The tag will be replaced with the original text in unstrip()
762 function insertStripItem( $text ) {
763 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
764 $this->mMarkerIndex++
;
765 $this->mStripState
->addGeneral( $rnd, $text );
770 * parse the wiki syntax used to render tables
774 function doTableStuff( $text ) {
775 wfProfileIn( __METHOD__
);
777 $lines = StringUtils
::explode( "\n", $text );
781 foreach ( $lines as $outLine ) {
782 $line = trim( $outLine );
784 # empty line, go to next line,
785 # but only append \n if outside of table
787 $output .= $outLine . "\n";
790 $firstChars = $line[0];
791 if ( strlen( $line ) > 1 ) {
792 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ?
$line[1] : '';
796 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
798 $table =& $this->last( $tables );
799 $table[0] = array(); // first row
800 $currentRow =& $table[0];
801 $table['indent'] = strlen( $matches[1] );
803 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
804 $attributes = Sanitizer
::fixTagAttributes( $attributes , 'table' );
806 if ( $attributes !== '' ) {
807 $table['attributes'] = $attributes;
809 } else if ( !isset( $tables[0] ) ) {
810 // we're outside the table
812 $out .= $outLine . "\n";
813 } else if ( $firstChars === '|}' ) {
814 // trim the |} code from the line
815 $line = substr ( $line , 2 );
817 // Shorthand for last row
818 $lastRow =& $this->last( $table );
820 // a thead at the end becomes a tfoot, unless there is only one row
821 // Do this before deleting empty last lines to allow headers at the bottom of tables
822 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
823 $lastRow['type'] = 'tfoot';
824 for ( $i = 0; isset( $lastRow[$i] ); $i++
) {
825 $lastRow[$i]['type'] = 'th';
829 // Delete empty last lines
830 if ( empty( $lastRow ) ) {
834 $curtable = array_pop( $tables );
836 #Add a line-ending before the table, but only if there isn't one already
837 if ( substr( $out, -1 ) !== "\n" ) {
840 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
842 if ( count( $tables ) > 0 ) {
843 $table =& $this->last( $tables );
844 $currentRow =& $this->last( $table );
845 $currentElement =& $this->last( $currentRow );
847 $output =& $currentElement['content'];
854 } else if ( $firstChars === '|-' ) {
855 // start a new row element
856 // but only when we haven't started one already
857 if ( count( $currentRow ) != 0 ) {
859 $currentRow =& $this->last( $table );
861 // Get the attributes, there's nothing else useful in $line now
862 $line = substr ( $line , 2 );
863 $attributes = $this->mStripState
->unstripBoth( $line );
864 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
865 if ( $attributes !== '' ) {
866 $currentRow['attributes'] = $attributes;
869 } else if ( $firstChars === '|+' ) {
870 // a table caption, but only proceed if there isn't one already
871 if ( !isset ( $table['caption'] ) ) {
872 $line = substr ( $line , 2 );
874 $c = $this->getCellAttr( $line , 'caption' );
875 $table['caption'] = array();
876 $table['caption']['content'] = $c[0];
877 if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
879 $output =& $table['caption']['content'];
881 } else if ( $firstChars === '|' ||
$firstChars === '!' ||
$firstChars === '!+' ) {
882 // Which kind of cells are we dealing with
884 $line = substr ( $line , 1 );
886 if ( $firstChars === '!' ||
$firstChars === '!+' ) {
887 $line = str_replace ( '!!' , '||' , $line );
891 // Split up multiple cells on the same line.
892 $cells = StringUtils
::explodeMarkup( '||' , $line );
893 $line = ''; // save memory
895 // decide whether thead to tbody
896 if ( !array_key_exists( 'type', $currentRow ) ) {
897 $currentRow['type'] = ( $firstChars === '!' ) ?
'thead' : 'tbody' ;
898 } else if ( $firstChars === '|' ) {
899 $currentRow['type'] = 'tbody';
902 // Loop through each table cell
903 foreach ( $cells as $cell ) {
905 $currentRow[] = array();
906 $currentElement =& $this->last( $currentRow );
908 $currentElement['type'] = $currentTag;
910 $c = $this->getCellAttr( $cell , $currentTag );
911 $currentElement['content'] = $c[0];
912 if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
915 $output =& $currentElement['content'];
918 $output .= "\n$outLine";
922 # Remove trailing line-ending (b/c)
923 if ( substr( $out, -1 ) === "\n" ) {
924 $out = substr( $out, 0, -1 );
927 # Close any unclosed tables
928 if ( isset( $tables ) && count( $tables ) > 0 ) {
929 for ( $i = 0; $i < count( $tables ); $i++
) {
930 $curtable = array_pop( $tables );
931 $curtable = $this->generateTableHTML( $curtable );
932 #Add a line-ending before the table, but only if there isn't one already
933 if ( substr( $out, -1 ) !== "\n" && $curtable !== "" ) {
940 wfProfileOut( __METHOD__
);
946 * Helper function for doTableStuff() separating the contents of cells from
947 * attributes. Particularly useful as there's a possible bug and this action
952 function getCellAttr ( $cell, $tagName ) {
956 $cell = trim ( $cell );
958 // A cell could contain both parameters and data
959 $cellData = 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( $cellData[0], '[[' ) !== false ) {
964 $content = trim ( $cell );
966 else if ( count ( $cellData ) == 1 ) {
967 $content = trim ( $cellData[0] );
970 $attributes = $this->mStripState
->unstripBoth( $cellData[0] );
971 $attributes = Sanitizer
::fixTagAttributes( $attributes , $tagName );
973 $content = trim ( $cellData[1] );
975 return array( $content, $attributes );
980 * Helper function for doTableStuff(). This converts the structured array into html.
984 function generateTableHTML ( &$table ) {
986 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
988 $return .= isset( $table['attributes'] ) ?
$table['attributes'] : '';
990 unset( $table['attributes'] );
992 if ( isset( $table['caption'] ) ) {
993 $return .= "\n<caption";
994 $return .= isset( $table['caption']['attributes'] ) ?
$table['caption']['attributes'] : '';
996 $return .= $table['caption']['content'];
997 $return .= "\n</caption>";
1003 // If we only have tbodies, mark table as simple
1004 for ( $i = 0; isset( $table[$i] ); $i++
) {
1005 if ( !count( $table[$i] ) ) continue;
1006 if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1007 if ( !$lastSection ) {
1008 $lastSection = $table[$i]['type'];
1009 } else if ( $lastSection != $table[$i]['type'] ) {
1015 for ( $i = 0; isset( $table[$i] ); $i++
) {
1016 // Check for empty tables
1017 if ( count( $table[$i] ) ) {
1022 if ( $table[$i]['type'] != $lastSection && !$simple ) {
1023 $return .= "\n<" . $table[$i]['type'] . '>';
1027 $return .= isset( $table[$i]['attributes'] ) ?
$table[$i]['attributes'] : '';
1029 for ( $j = 0; isset( $table[$i][$j] ); $j++
) {
1030 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1031 $return .= "\n<" . $table[$i][$j]['type'];
1032 $return .= isset( $table[$i][$j]['attributes'] ) ?
$table[$i][$j]['attributes'] : '';
1035 $return .= $table[$i][$j]['content'];
1036 if ( $table[$i][$j]['content'] != '' )
1039 $return .= '</' . $table[$i][$j]['type'] . '>';
1040 unset( $table[$i][$j] );
1042 $return .= "\n</tr>";
1044 if ( ( !isset( $table[$i +
1] ) && !$simple ) ||
( isset( $table[$i +
1] ) && isset( $table[$i +
1]['type'] ) && $table[$i]['type'] != $table[$i +
1]['type'] ) ) {
1045 $return .= '</' . $table[$i]['type'] . '>';
1047 $lastSection = $table[$i]['type'];
1048 unset( $table[$i] );
1051 if ( isset( $table['caption'] ) ) {
1052 $return .= "\n<tr><td></td></tr>";
1057 $return .= "\n</table>";
1058 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1064 * like end() but only works on the numeric array index and php's internal pointers
1065 * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1066 * ignores associative elements and will create a 0 key will a NULL value if there were
1067 * no numric elements and an array itself if not previously defined.
1071 function &last ( &$arr ) {
1072 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) { }
1077 * Helper function for parse() that transforms wiki markup into
1078 * HTML. Only called for $mOutputType == self::OT_HTML.
1082 function internalParse( $text, $isMain = true, $frame=false ) {
1083 wfProfileIn( __METHOD__
);
1087 # Hook to suspend the parser in this state
1088 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState
) ) ) {
1089 wfProfileOut( __METHOD__
);
1093 # if $frame is provided, then use $frame for replacing any variables
1095 # use frame depth to infer how include/noinclude tags should be handled
1096 # depth=0 means this is the top-level document; otherwise it's an included document
1097 if ( !$frame->depth
) {
1100 $flag = Parser
::PTD_FOR_INCLUSION
;
1102 $dom = $this->preprocessToDom( $text, $flag );
1103 $text = $frame->expand( $dom );
1105 # if $frame is not provided, then use old-style replaceVariables
1106 $text = $this->replaceVariables( $text );
1109 $text = Sanitizer
::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks
) );
1110 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState
) );
1112 # Tables need to come after variable replacement for things to work
1113 # properly; putting them before other transformations should keep
1114 # exciting things like link expansions from showing up in surprising
1116 $text = $this->doTableStuff( $text );
1118 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1120 $text = $this->doDoubleUnderscore( $text );
1122 $text = $this->doHeadings( $text );
1123 if ( $this->mOptions
->getUseDynamicDates() ) {
1124 $df = DateFormatter
::getInstance();
1125 $text = $df->reformat( $this->mOptions
->getDateFormat(), $text );
1127 $text = $this->replaceInternalLinks( $text );
1128 $text = $this->doAllQuotes( $text );
1129 $text = $this->replaceExternalLinks( $text );
1131 # replaceInternalLinks may sometimes leave behind
1132 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1133 $text = str_replace( $this->mUniqPrefix
.'NOPARSE', '', $text );
1135 $text = $this->doMagicLinks( $text );
1136 $text = $this->formatHeadings( $text, $origText, $isMain );
1138 wfProfileOut( __METHOD__
);
1143 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1144 * magic external links.
1149 function doMagicLinks( $text ) {
1150 wfProfileIn( __METHOD__
);
1151 $prots = $this->mUrlProtocols
;
1152 $urlChar = self
::EXT_LINK_URL_CLASS
;
1153 $text = preg_replace_callback(
1155 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1156 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1157 (\\b(?:$prots)$urlChar+) | # m[3]: Free external links" . '
1158 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1159 ISBN\s+(\b # m[5]: ISBN, capture number
1160 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1161 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1162 [0-9Xx] # check digit
1164 )!x', array( &$this, 'magicLinkCallback' ), $text );
1165 wfProfileOut( __METHOD__
);
1169 function magicLinkCallback( $m ) {
1170 if ( isset( $m[1] ) && $m[1] !== '' ) {
1173 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1176 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1177 # Free external link
1178 return $this->makeFreeExternalLink( $m[0] );
1179 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1181 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1184 $CssClass = 'mw-magiclink-rfc';
1186 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1188 $urlmsg = 'pubmedurl';
1189 $CssClass = 'mw-magiclink-pmid';
1192 throw new MWException( __METHOD__
.': unrecognised match type "' .
1193 substr( $m[0], 0, 20 ) . '"' );
1195 $url = wfMsgForContent( $urlmsg, $id );
1196 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1197 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1200 $num = strtr( $isbn, array(
1205 $titleObj = SpecialPage
::getTitleFor( 'Booksources', $num );
1207 $titleObj->escapeLocalUrl() .
1208 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1215 * Make a free external link, given a user-supplied URL
1219 function makeFreeExternalLink( $url ) {
1221 wfProfileIn( __METHOD__
);
1225 # The characters '<' and '>' (which were escaped by
1226 # removeHTMLtags()) should not be included in
1227 # URLs, per RFC 2396.
1229 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1230 $trail = substr( $url, $m2[0][1] ) . $trail;
1231 $url = substr( $url, 0, $m2[0][1] );
1234 # Move trailing punctuation to $trail
1236 # If there is no left bracket, then consider right brackets fair game too
1237 if ( strpos( $url, '(' ) === false ) {
1241 $numSepChars = strspn( strrev( $url ), $sep );
1242 if ( $numSepChars ) {
1243 $trail = substr( $url, -$numSepChars ) . $trail;
1244 $url = substr( $url, 0, -$numSepChars );
1247 $url = Sanitizer
::cleanUrl( $url );
1249 # Is this an external image?
1250 $text = $this->maybeMakeExternalImage( $url );
1251 if ( $text === false ) {
1252 # Not an image, make a link
1253 $text = Linker
::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1254 $this->getExternalLinkAttribs( $url ) );
1255 # Register it in the output object...
1256 # Replace unnecessary URL escape codes with their equivalent characters
1257 $pasteurized = self
::replaceUnusualEscapes( $url );
1258 $this->mOutput
->addExternalLink( $pasteurized );
1260 wfProfileOut( __METHOD__
);
1261 return $text . $trail;
1266 * Parse headers and return html
1270 function doHeadings( $text ) {
1271 wfProfileIn( __METHOD__
);
1272 for ( $i = 6; $i >= 1; --$i ) {
1273 $h = str_repeat( '=', $i );
1274 $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1275 "<h$i>\\1</h$i>", $text );
1277 wfProfileOut( __METHOD__
);
1282 * Replace single quotes with HTML markup
1284 * @return string the altered text
1286 function doAllQuotes( $text ) {
1287 wfProfileIn( __METHOD__
);
1289 $lines = StringUtils
::explode( "\n", $text );
1290 foreach ( $lines as $line ) {
1291 $outtext .= $this->doQuotes( $line ) . "\n";
1293 $outtext = substr( $outtext, 0,-1 );
1294 wfProfileOut( __METHOD__
);
1299 * Helper function for doAllQuotes()
1301 public function doQuotes( $text ) {
1302 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1303 if ( count( $arr ) == 1 ) {
1306 # First, do some preliminary work. This may shift some apostrophes from
1307 # being mark-up to being text. It also counts the number of occurrences
1308 # of bold and italics mark-ups.
1311 for ( $i = 0; $i < count( $arr ); $i++
) {
1312 if ( ( $i %
2 ) == 1 ) {
1313 # If there are ever four apostrophes, assume the first is supposed to
1314 # be text, and the remaining three constitute mark-up for bold text.
1315 if ( strlen( $arr[$i] ) == 4 ) {
1318 } elseif ( strlen( $arr[$i] ) > 5 ) {
1319 # If there are more than 5 apostrophes in a row, assume they're all
1320 # text except for the last 5.
1321 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1324 # Count the number of occurrences of bold and italics mark-ups.
1325 # We are not counting sequences of five apostrophes.
1326 if ( strlen( $arr[$i] ) == 2 ) {
1328 } elseif ( strlen( $arr[$i] ) == 3 ) {
1330 } elseif ( strlen( $arr[$i] ) == 5 ) {
1337 # If there is an odd number of both bold and italics, it is likely
1338 # that one of the bold ones was meant to be an apostrophe followed
1339 # by italics. Which one we cannot know for certain, but it is more
1340 # likely to be one that has a single-letter word before it.
1341 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1343 $firstsingleletterword = -1;
1344 $firstmultiletterword = -1;
1346 foreach ( $arr as $r ) {
1347 if ( ( $i %
2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1348 $x1 = substr( $arr[$i-1], -1 );
1349 $x2 = substr( $arr[$i-1], -2, 1 );
1350 if ( $x1 === ' ' ) {
1351 if ( $firstspace == -1 ) {
1354 } elseif ( $x2 === ' ') {
1355 if ( $firstsingleletterword == -1 ) {
1356 $firstsingleletterword = $i;
1359 if ( $firstmultiletterword == -1 ) {
1360 $firstmultiletterword = $i;
1367 # If there is a single-letter word, use it!
1368 if ( $firstsingleletterword > -1 ) {
1369 $arr[$firstsingleletterword] = "''";
1370 $arr[$firstsingleletterword-1] .= "'";
1371 } elseif ( $firstmultiletterword > -1 ) {
1372 # If not, but there's a multi-letter word, use that one.
1373 $arr[$firstmultiletterword] = "''";
1374 $arr[$firstmultiletterword-1] .= "'";
1375 } elseif ( $firstspace > -1 ) {
1376 # ... otherwise use the first one that has neither.
1377 # (notice that it is possible for all three to be -1 if, for example,
1378 # there is only one pentuple-apostrophe in the line)
1379 $arr[$firstspace] = "''";
1380 $arr[$firstspace-1] .= "'";
1384 # Now let's actually convert our apostrophic mush to HTML!
1389 foreach ( $arr as $r ) {
1390 if ( ( $i %
2 ) == 0 ) {
1391 if ( $state === 'both' ) {
1397 if ( strlen( $r ) == 2 ) {
1398 if ( $state === 'i' ) {
1399 $output .= '</i>'; $state = '';
1400 } elseif ( $state === 'bi' ) {
1401 $output .= '</i>'; $state = 'b';
1402 } elseif ( $state === 'ib' ) {
1403 $output .= '</b></i><b>'; $state = 'b';
1404 } elseif ( $state === 'both' ) {
1405 $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1406 } else { # $state can be 'b' or ''
1407 $output .= '<i>'; $state .= 'i';
1409 } elseif ( strlen( $r ) == 3 ) {
1410 if ( $state === 'b' ) {
1411 $output .= '</b>'; $state = '';
1412 } elseif ( $state === 'bi' ) {
1413 $output .= '</i></b><i>'; $state = 'i';
1414 } elseif ( $state === 'ib' ) {
1415 $output .= '</b>'; $state = 'i';
1416 } elseif ( $state === 'both' ) {
1417 $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1418 } else { # $state can be 'i' or ''
1419 $output .= '<b>'; $state .= 'b';
1421 } elseif ( strlen( $r ) == 5 ) {
1422 if ( $state === 'b' ) {
1423 $output .= '</b><i>'; $state = 'i';
1424 } elseif ( $state === 'i' ) {
1425 $output .= '</i><b>'; $state = 'b';
1426 } elseif ( $state === 'bi' ) {
1427 $output .= '</i></b>'; $state = '';
1428 } elseif ( $state === 'ib' ) {
1429 $output .= '</b></i>'; $state = '';
1430 } elseif ( $state === 'both' ) {
1431 $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1432 } else { # ($state == '')
1433 $buffer = ''; $state = 'both';
1439 # Now close all remaining tags. Notice that the order is important.
1440 if ( $state === 'b' ||
$state === 'ib' ) {
1443 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1446 if ( $state === 'bi' ) {
1449 # There might be lonely ''''', so make sure we have a buffer
1450 if ( $state === 'both' && $buffer ) {
1451 $output .= '<b><i>'.$buffer.'</i></b>';
1458 * Replace external links (REL)
1460 * Note: this is all very hackish and the order of execution matters a lot.
1461 * Make sure to run maintenance/parserTests.php if you change this code.
1465 function replaceExternalLinks( $text ) {
1467 wfProfileIn( __METHOD__
);
1469 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1470 $s = array_shift( $bits );
1473 while ( $i<count( $bits ) ) {
1475 $protocol = $bits[$i++
];
1476 $text = $bits[$i++
];
1477 $trail = $bits[$i++
];
1479 # The characters '<' and '>' (which were escaped by
1480 # removeHTMLtags()) should not be included in
1481 # URLs, per RFC 2396.
1483 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1484 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1485 $url = substr( $url, 0, $m2[0][1] );
1488 # If the link text is an image URL, replace it with an <img> tag
1489 # This happened by accident in the original parser, but some people used it extensively
1490 $img = $this->maybeMakeExternalImage( $text );
1491 if ( $img !== false ) {
1497 # Set linktype for CSS - if URL==text, link is essentially free
1498 $linktype = ( $text === $url ) ?
'free' : 'text';
1500 # No link text, e.g. [http://domain.tld/some.link]
1501 if ( $text == '' ) {
1502 # Autonumber if allowed. See bug #5918
1503 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1504 $langObj = $this->getFunctionLang();
1505 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1506 $linktype = 'autonumber';
1508 # Otherwise just use the URL
1509 $text = htmlspecialchars( $url );
1513 # Have link text, e.g. [http://domain.tld/some.link text]s
1515 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1518 $text = $wgContLang->markNoConversion( $text );
1520 $url = Sanitizer
::cleanUrl( $url );
1522 # Use the encoded URL
1523 # This means that users can paste URLs directly into the text
1524 # Funny characters like ö aren't valid in URLs anyway
1525 # This was changed in August 2004
1526 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
1527 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1529 # Register link in the output object.
1530 # Replace unnecessary URL escape codes with the referenced character
1531 # This prevents spammers from hiding links from the filters
1532 $pasteurized = self
::replaceUnusualEscapes( $url );
1533 $this->mOutput
->addExternalLink( $pasteurized );
1536 wfProfileOut( __METHOD__
);
1541 * Get an associative array of additional HTML attributes appropriate for a
1542 * particular external link. This currently may include rel => nofollow
1543 * (depending on configuration, namespace, and the URL's domain) and/or a
1544 * target attribute (depending on configuration).
1546 * @param $url String: optional URL, to extract the domain from for rel =>
1547 * nofollow if appropriate
1548 * @return Array: associative array of HTML attributes
1550 function getExternalLinkAttribs( $url = false ) {
1552 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1553 $ns = $this->mTitle
->getNamespace();
1554 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1555 $attribs['rel'] = 'nofollow';
1557 global $wgNoFollowDomainExceptions;
1558 if ( $wgNoFollowDomainExceptions ) {
1559 $bits = wfParseUrl( $url );
1560 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1561 foreach ( $wgNoFollowDomainExceptions as $domain ) {
1562 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1563 unset( $attribs['rel'] );
1570 if ( $this->mOptions
->getExternalLinkTarget() ) {
1571 $attribs['target'] = $this->mOptions
->getExternalLinkTarget();
1577 * Replace unusual URL escape codes with their equivalent characters
1579 * @param $url String
1582 * @todo This can merge genuinely required bits in the path or query string,
1583 * breaking legit URLs. A proper fix would treat the various parts of
1584 * the URL differently; as a workaround, just use the output for
1585 * statistical records, not for actual linking/output.
1587 static function replaceUnusualEscapes( $url ) {
1588 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1589 array( __CLASS__
, 'replaceUnusualEscapesCallback' ), $url );
1593 * Callback function used in replaceUnusualEscapes().
1594 * Replaces unusual URL escape codes with their equivalent character
1596 private static function replaceUnusualEscapesCallback( $matches ) {
1597 $char = urldecode( $matches[0] );
1598 $ord = ord( $char );
1599 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1600 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1601 # No, shouldn't be escaped
1604 # Yes, leave it escaped
1610 * make an image if it's allowed, either through the global
1611 * option, through the exception, or through the on-wiki whitelist
1614 function maybeMakeExternalImage( $url ) {
1615 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
1616 $imagesexception = !empty( $imagesfrom );
1618 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1619 if ( $imagesexception && is_array( $imagesfrom ) ) {
1620 $imagematch = false;
1621 foreach ( $imagesfrom as $match ) {
1622 if ( strpos( $url, $match ) === 0 ) {
1627 } elseif ( $imagesexception ) {
1628 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1630 $imagematch = false;
1632 if ( $this->mOptions
->getAllowExternalImages()
1633 ||
( $imagesexception && $imagematch ) ) {
1634 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1636 $text = Linker
::makeExternalImage( $url );
1639 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
1640 && preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
1641 $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1642 foreach ( $whitelist as $entry ) {
1643 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1644 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
1647 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1648 # Image matches a whitelist entry
1649 $text = Linker
::makeExternalImage( $url );
1658 * Process [[ ]] wikilinks
1659 * @return String: processed text
1663 function replaceInternalLinks( $s ) {
1664 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
1669 * Process [[ ]] wikilinks (RIL)
1670 * @return LinkHolderArray
1674 function replaceInternalLinks2( &$s ) {
1677 wfProfileIn( __METHOD__
);
1679 wfProfileIn( __METHOD__
.'-setup' );
1680 static $tc = FALSE, $e1, $e1_img;
1681 # the % is needed to support urlencoded titles as well
1683 $tc = Title
::legalChars() . '#%';
1684 # Match a link having the form [[namespace:link|alternate]]trail
1685 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1686 # Match cases where there is no "]]", which might still be images
1687 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1690 $holders = new LinkHolderArray( $this );
1692 # split the entire text string on occurences of [[
1693 $a = StringUtils
::explode( '[[', ' ' . $s );
1694 # get the first element (all text up to first [[), and remove the space we added
1697 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1698 $s = substr( $s, 1 );
1700 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1702 if ( $useLinkPrefixExtension ) {
1703 # Match the end of a line for a word that's not followed by whitespace,
1704 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1705 $e2 = wfMsgForContent( 'linkprefix' );
1708 if ( is_null( $this->mTitle
) ) {
1709 wfProfileOut( __METHOD__
.'-setup' );
1710 wfProfileOut( __METHOD__
);
1711 throw new MWException( __METHOD__
.": \$this->mTitle is null\n" );
1713 $nottalk = !$this->mTitle
->isTalkPage();
1715 if ( $useLinkPrefixExtension ) {
1717 if ( preg_match( $e2, $s, $m ) ) {
1718 $first_prefix = $m[2];
1720 $first_prefix = false;
1726 if ( $wgContLang->hasVariants() ) {
1727 $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle
->getPrefixedText() );
1729 $selflink = array( $this->mTitle
->getPrefixedText() );
1731 $useSubpages = $this->areSubpagesAllowed();
1732 wfProfileOut( __METHOD__
.'-setup' );
1734 # Loop for each link
1735 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1736 # Check for excessive memory usage
1737 if ( $holders->isBig() ) {
1739 # Do the existence check, replace the link holders and clear the array
1740 $holders->replace( $s );
1744 if ( $useLinkPrefixExtension ) {
1745 wfProfileIn( __METHOD__
.'-prefixhandling' );
1746 if ( preg_match( $e2, $s, $m ) ) {
1753 if ( $first_prefix ) {
1754 $prefix = $first_prefix;
1755 $first_prefix = false;
1757 wfProfileOut( __METHOD__
.'-prefixhandling' );
1760 $might_be_img = false;
1762 wfProfileIn( __METHOD__
."-e1" );
1763 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1765 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1766 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1767 # the real problem is with the $e1 regex
1770 # Still some problems for cases where the ] is meant to be outside punctuation,
1771 # and no image is in sight. See bug 2095.
1773 if ( $text !== '' &&
1774 substr( $m[3], 0, 1 ) === ']' &&
1775 strpos( $text, '[' ) !== false
1778 $text .= ']'; # so that replaceExternalLinks($text) works later
1779 $m[3] = substr( $m[3], 1 );
1781 # fix up urlencoded title texts
1782 if ( strpos( $m[1], '%' ) !== false ) {
1783 # Should anchors '#' also be rejected?
1784 $m[1] = str_replace( array('<', '>'), array('<', '>'), rawurldecode( $m[1] ) );
1787 } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1788 $might_be_img = true;
1790 if ( strpos( $m[1], '%' ) !== false ) {
1791 $m[1] = rawurldecode( $m[1] );
1794 } else { # Invalid form; output directly
1795 $s .= $prefix . '[[' . $line ;
1796 wfProfileOut( __METHOD__
."-e1" );
1799 wfProfileOut( __METHOD__
."-e1" );
1800 wfProfileIn( __METHOD__
."-misc" );
1802 # Don't allow internal links to pages containing
1803 # PROTO: where PROTO is a valid URL protocol; these
1804 # should be external links.
1805 if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1806 $s .= $prefix . '[[' . $line ;
1807 wfProfileOut( __METHOD__
."-misc" );
1811 # Make subpage if necessary
1812 if ( $useSubpages ) {
1813 $link = $this->maybeDoSubpageLink( $m[1], $text );
1818 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1820 # Strip off leading ':'
1821 $link = substr( $link, 1 );
1824 wfProfileOut( __METHOD__
."-misc" );
1825 wfProfileIn( __METHOD__
."-title" );
1826 $nt = Title
::newFromText( $this->mStripState
->unstripNoWiki( $link ) );
1827 if ( $nt === null ) {
1828 $s .= $prefix . '[[' . $line;
1829 wfProfileOut( __METHOD__
."-title" );
1833 $ns = $nt->getNamespace();
1834 $iw = $nt->getInterWiki();
1835 wfProfileOut( __METHOD__
."-title" );
1837 if ( $might_be_img ) { # if this is actually an invalid link
1838 wfProfileIn( __METHOD__
."-might_be_img" );
1839 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
1842 # look at the next 'line' to see if we can close it there
1844 $next_line = $a->current();
1845 if ( $next_line === false ||
$next_line === null ) {
1848 $m = explode( ']]', $next_line, 3 );
1849 if ( count( $m ) == 3 ) {
1850 # the first ]] closes the inner link, the second the image
1852 $text .= "[[{$m[0]}]]{$m[1]}";
1855 } elseif ( count( $m ) == 2 ) {
1856 # if there's exactly one ]] that's fine, we'll keep looking
1857 $text .= "[[{$m[0]}]]{$m[1]}";
1859 # if $next_line is invalid too, we need look no further
1860 $text .= '[[' . $next_line;
1865 # we couldn't find the end of this imageLink, so output it raw
1866 # but don't ignore what might be perfectly normal links in the text we've examined
1867 $holders->merge( $this->replaceInternalLinks2( $text ) );
1868 $s .= "{$prefix}[[$link|$text";
1869 # note: no $trail, because without an end, there *is* no trail
1870 wfProfileOut( __METHOD__
."-might_be_img" );
1873 } else { # it's not an image, so output it raw
1874 $s .= "{$prefix}[[$link|$text";
1875 # note: no $trail, because without an end, there *is* no trail
1876 wfProfileOut( __METHOD__
."-might_be_img" );
1879 wfProfileOut( __METHOD__
."-might_be_img" );
1882 $wasblank = ( $text == '' );
1886 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1887 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1888 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1889 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1890 $text = $this->doQuotes( $text );
1893 # Link not escaped by : , create the various objects
1897 wfProfileIn( __METHOD__
."-interwiki" );
1898 if ( $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1899 $this->mOutput
->addLanguageLink( $nt->getFullText() );
1900 $s = rtrim( $s . $prefix );
1901 $s .= trim( $trail, "\n" ) == '' ?
'': $prefix . $trail;
1902 wfProfileOut( __METHOD__
."-interwiki" );
1905 wfProfileOut( __METHOD__
."-interwiki" );
1907 if ( $ns == NS_FILE
) {
1908 wfProfileIn( __METHOD__
."-image" );
1909 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
1911 # if no parameters were passed, $text
1912 # becomes something like "File:Foo.png",
1913 # which we don't want to pass on to the
1917 # recursively parse links inside the image caption
1918 # actually, this will parse them in any other parameters, too,
1919 # but it might be hard to fix that, and it doesn't matter ATM
1920 $text = $this->replaceExternalLinks( $text );
1921 $holders->merge( $this->replaceInternalLinks2( $text ) );
1923 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1924 $s .= $prefix . $this->armorLinks(
1925 $this->makeImage( $nt, $text, $holders ) ) . $trail;
1927 $s .= $prefix . $trail;
1929 wfProfileOut( __METHOD__
."-image" );
1933 if ( $ns == NS_CATEGORY
) {
1934 wfProfileIn( __METHOD__
."-category" );
1935 $s = rtrim( $s . "\n" ); # bug 87
1938 $sortkey = $this->getDefaultSort();
1942 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
1943 $sortkey = str_replace( "\n", '', $sortkey );
1944 $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1945 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
1948 * Strip the whitespace Category links produce, see bug 87
1949 * @todo We might want to use trim($tmp, "\n") here.
1951 $s .= trim( $prefix . $trail, "\n" ) == '' ?
'' : $prefix . $trail;
1953 wfProfileOut( __METHOD__
."-category" );
1958 # Self-link checking
1959 if ( $nt->getFragment() === '' && $ns != NS_SPECIAL
) {
1960 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1961 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
1966 # NS_MEDIA is a pseudo-namespace for linking directly to a file
1967 # FIXME: Should do batch file existence checks, see comment below
1968 if ( $ns == NS_MEDIA
) {
1969 wfProfileIn( __METHOD__
."-media" );
1970 # Give extensions a chance to select the file revision for us
1971 $time = $sha1 = $descQuery = false;
1972 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1973 array( $this, $nt, &$time, &$sha1, &$descQuery ) );
1974 # Fetch and register the file (file title may be different via hooks)
1975 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
1976 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1977 $s .= $prefix . $this->armorLinks(
1978 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1979 wfProfileOut( __METHOD__
."-media" );
1983 wfProfileIn( __METHOD__
."-always_known" );
1984 # Some titles, such as valid special pages or files in foreign repos, should
1985 # be shown as bluelinks even though they're not included in the page table
1987 # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1988 # batch file existence checks for NS_FILE and NS_MEDIA
1989 if ( $iw == '' && $nt->isAlwaysKnown() ) {
1990 $this->mOutput
->addLink( $nt );
1991 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1993 # Links will be added to the output link list after checking
1994 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1996 wfProfileOut( __METHOD__
."-always_known" );
1998 wfProfileOut( __METHOD__
);
2003 * Render a forced-blue link inline; protect against double expansion of
2004 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2005 * Since this little disaster has to split off the trail text to avoid
2006 * breaking URLs in the following text without breaking trails on the
2007 * wiki links, it's been made into a horrible function.
2010 * @param $text String
2011 * @param $query Array or String
2012 * @param $trail String
2013 * @param $prefix String
2014 * @return String: HTML-wikitext mix oh yuck
2016 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2017 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2019 if ( is_string( $query ) ) {
2020 $query = wfCgiToArray( $query );
2022 if ( $text == '' ) {
2023 $text = htmlspecialchars( $nt->getPrefixedText() );
2026 $link = Linker
::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2028 return $this->armorLinks( $link ) . $trail;
2032 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2033 * going to go through further parsing steps before inline URL expansion.
2035 * Not needed quite as much as it used to be since free links are a bit
2036 * more sensible these days. But bracketed links are still an issue.
2038 * @param $text String: more-or-less HTML
2039 * @return String: less-or-more HTML with NOPARSE bits
2041 function armorLinks( $text ) {
2042 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2043 "{$this->mUniqPrefix}NOPARSE$1", $text );
2047 * Return true if subpage links should be expanded on this page.
2050 function areSubpagesAllowed() {
2051 # Some namespaces don't allow subpages
2052 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2056 * Handle link to subpage if necessary
2058 * @param $target String: the source of the link
2059 * @param &$text String: the link text, modified as necessary
2060 * @return string the full name of the link
2063 function maybeDoSubpageLink( $target, &$text ) {
2064 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2068 * Used by doBlockLevels()
2073 function closeParagraph() {
2075 if ( $this->mLastSection
!= '' ) {
2076 $result = '</' . $this->mLastSection
. ">\n";
2078 $this->mInPre
= false;
2079 $this->mLastSection
= '';
2084 * getCommon() returns the length of the longest common substring
2085 * of both arguments, starting at the beginning of both.
2088 function getCommon( $st1, $st2 ) {
2089 $fl = strlen( $st1 );
2090 $shorter = strlen( $st2 );
2091 if ( $fl < $shorter ) {
2095 for ( $i = 0; $i < $shorter; ++
$i ) {
2096 if ( $st1[$i] != $st2[$i] ) {
2104 * These next three functions open, continue, and close the list
2105 * element appropriate to the prefix character passed into them.
2110 function openList( $char ) {
2111 $result = $this->closeParagraph();
2113 if ( '*' === $char ) {
2114 $result .= '<ul><li>';
2115 } elseif ( '#' === $char ) {
2116 $result .= '<ol><li>';
2117 } elseif ( ':' === $char ) {
2118 $result .= '<dl><dd>';
2119 } elseif ( ';' === $char ) {
2120 $result .= '<dl><dt>';
2121 $this->mDTopen
= true;
2123 $result = '<!-- ERR 1 -->';
2131 * @param $char String
2136 function nextItem( $char ) {
2137 if ( '*' === $char ||
'#' === $char ) {
2139 } elseif ( ':' === $char ||
';' === $char ) {
2141 if ( $this->mDTopen
) {
2144 if ( ';' === $char ) {
2145 $this->mDTopen
= true;
2146 return $close . '<dt>';
2148 $this->mDTopen
= false;
2149 return $close . '<dd>';
2152 return '<!-- ERR 2 -->';
2157 * @param $char String
2162 function closeList( $char ) {
2163 if ( '*' === $char ) {
2164 $text = '</li></ul>';
2165 } elseif ( '#' === $char ) {
2166 $text = '</li></ol>';
2167 } elseif ( ':' === $char ) {
2168 if ( $this->mDTopen
) {
2169 $this->mDTopen
= false;
2170 $text = '</dt></dl>';
2172 $text = '</dd></dl>';
2175 return '<!-- ERR 3 -->';
2182 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2184 * @param $text String
2185 * @param $linestart Boolean: whether or not this is at the start of a line.
2187 * @return string the lists rendered as HTML
2189 function doBlockLevels( $text, $linestart ) {
2190 wfProfileIn( __METHOD__
);
2192 # Parsing through the text line by line. The main thing
2193 # happening here is handling of block-level elements p, pre,
2194 # and making lists from lines starting with * # : etc.
2196 $textLines = StringUtils
::explode( "\n", $text );
2198 $lastPrefix = $output = '';
2199 $this->mDTopen
= $inBlockElem = false;
2201 $paragraphStack = false;
2203 foreach ( $textLines as $oLine ) {
2205 if ( !$linestart ) {
2215 $lastPrefixLength = strlen( $lastPrefix );
2216 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2217 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2218 # If not in a <pre> element, scan for and figure out what prefixes are there.
2219 if ( !$this->mInPre
) {
2220 # Multiple prefixes may abut each other for nested lists.
2221 $prefixLength = strspn( $oLine, '*#:;' );
2222 $prefix = substr( $oLine, 0, $prefixLength );
2225 # ; and : are both from definition-lists, so they're equivalent
2226 # for the purposes of determining whether or not we need to open/close
2228 $prefix2 = str_replace( ';', ':', $prefix );
2229 $t = substr( $oLine, $prefixLength );
2230 $this->mInPre
= (bool)$preOpenMatch;
2232 # Don't interpret any other prefixes in preformatted text
2234 $prefix = $prefix2 = '';
2239 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2240 # Same as the last item, so no need to deal with nesting or opening stuff
2241 $output .= $this->nextItem( substr( $prefix, -1 ) );
2242 $paragraphStack = false;
2244 if ( substr( $prefix, -1 ) === ';') {
2245 # The one nasty exception: definition lists work like this:
2246 # ; title : definition text
2247 # So we check for : in the remainder text to split up the
2248 # title and definition, without b0rking links.
2250 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2252 $output .= $term . $this->nextItem( ':' );
2255 } elseif ( $prefixLength ||
$lastPrefixLength ) {
2256 # We need to open or close prefixes, or both.
2258 # Either open or close a level...
2259 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2260 $paragraphStack = false;
2262 # Close all the prefixes which aren't shared.
2263 while ( $commonPrefixLength < $lastPrefixLength ) {
2264 $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2265 --$lastPrefixLength;
2268 # Continue the current prefix if appropriate.
2269 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2270 $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2273 # Open prefixes where appropriate.
2274 while ( $prefixLength > $commonPrefixLength ) {
2275 $char = substr( $prefix, $commonPrefixLength, 1 );
2276 $output .= $this->openList( $char );
2278 if ( ';' === $char ) {
2279 # FIXME: This is dupe of code above
2280 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2282 $output .= $term . $this->nextItem( ':' );
2285 ++
$commonPrefixLength;
2287 $lastPrefix = $prefix2;
2290 # If we have no prefixes, go to paragraph mode.
2291 if ( 0 == $prefixLength ) {
2292 wfProfileIn( __METHOD__
."-paragraph" );
2293 # No prefix (not in list)--go to paragraph mode
2294 # XXX: use a stack for nestable elements like span, table and div
2295 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2296 $closematch = preg_match(
2297 '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2298 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix
.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2299 if ( $openmatch or $closematch ) {
2300 $paragraphStack = false;
2301 # TODO bug 5718: paragraph closed
2302 $output .= $this->closeParagraph();
2303 if ( $preOpenMatch and !$preCloseMatch ) {
2304 $this->mInPre
= true;
2306 $inBlockElem = !$closematch;
2307 } elseif ( !$inBlockElem && !$this->mInPre
) {
2308 if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection
=== 'pre' ||
trim( $t ) != '' ) ) {
2310 if ( $this->mLastSection
!== 'pre' ) {
2311 $paragraphStack = false;
2312 $output .= $this->closeParagraph().'<pre>';
2313 $this->mLastSection
= 'pre';
2315 $t = substr( $t, 1 );
2318 if ( trim( $t ) === '' ) {
2319 if ( $paragraphStack ) {
2320 $output .= $paragraphStack.'<br />';
2321 $paragraphStack = false;
2322 $this->mLastSection
= 'p';
2324 if ( $this->mLastSection
!== 'p' ) {
2325 $output .= $this->closeParagraph();
2326 $this->mLastSection
= '';
2327 $paragraphStack = '<p>';
2329 $paragraphStack = '</p><p>';
2333 if ( $paragraphStack ) {
2334 $output .= $paragraphStack;
2335 $paragraphStack = false;
2336 $this->mLastSection
= 'p';
2337 } elseif ( $this->mLastSection
!== 'p' ) {
2338 $output .= $this->closeParagraph().'<p>';
2339 $this->mLastSection
= 'p';
2344 wfProfileOut( __METHOD__
."-paragraph" );
2346 # somewhere above we forget to get out of pre block (bug 785)
2347 if ( $preCloseMatch && $this->mInPre
) {
2348 $this->mInPre
= false;
2350 if ( $paragraphStack === false ) {
2354 while ( $prefixLength ) {
2355 $output .= $this->closeList( $prefix2[$prefixLength-1] );
2358 if ( $this->mLastSection
!= '' ) {
2359 $output .= '</' . $this->mLastSection
. '>';
2360 $this->mLastSection
= '';
2363 wfProfileOut( __METHOD__
);
2368 * Split up a string on ':', ignoring any occurences inside tags
2369 * to prevent illegal overlapping.
2371 * @param $str String: the string to split
2372 * @param &$before String: set to everything before the ':'
2373 * @param &$after String: set to everything after the ':'
2374 * return String: the position of the ':', or false if none found
2376 function findColonNoLinks( $str, &$before, &$after ) {
2377 wfProfileIn( __METHOD__
);
2379 $pos = strpos( $str, ':' );
2380 if ( $pos === false ) {
2382 wfProfileOut( __METHOD__
);
2386 $lt = strpos( $str, '<' );
2387 if ( $lt === false ||
$lt > $pos ) {
2388 # Easy; no tag nesting to worry about
2389 $before = substr( $str, 0, $pos );
2390 $after = substr( $str, $pos+
1 );
2391 wfProfileOut( __METHOD__
);
2395 # Ugly state machine to walk through avoiding tags.
2396 $state = self
::COLON_STATE_TEXT
;
2398 $len = strlen( $str );
2399 for( $i = 0; $i < $len; $i++
) {
2403 # (Using the number is a performance hack for common cases)
2404 case 0: # self::COLON_STATE_TEXT:
2407 # Could be either a <start> tag or an </end> tag
2408 $state = self
::COLON_STATE_TAGSTART
;
2411 if ( $stack == 0 ) {
2413 $before = substr( $str, 0, $i );
2414 $after = substr( $str, $i +
1 );
2415 wfProfileOut( __METHOD__
);
2418 # Embedded in a tag; don't break it.
2421 # Skip ahead looking for something interesting
2422 $colon = strpos( $str, ':', $i );
2423 if ( $colon === false ) {
2424 # Nothing else interesting
2425 wfProfileOut( __METHOD__
);
2428 $lt = strpos( $str, '<', $i );
2429 if ( $stack === 0 ) {
2430 if ( $lt === false ||
$colon < $lt ) {
2432 $before = substr( $str, 0, $colon );
2433 $after = substr( $str, $colon +
1 );
2434 wfProfileOut( __METHOD__
);
2438 if ( $lt === false ) {
2439 # Nothing else interesting to find; abort!
2440 # We're nested, but there's no close tags left. Abort!
2443 # Skip ahead to next tag start
2445 $state = self
::COLON_STATE_TAGSTART
;
2448 case 1: # self::COLON_STATE_TAG:
2453 $state = self
::COLON_STATE_TEXT
;
2456 # Slash may be followed by >?
2457 $state = self
::COLON_STATE_TAGSLASH
;
2463 case 2: # self::COLON_STATE_TAGSTART:
2466 $state = self
::COLON_STATE_CLOSETAG
;
2469 $state = self
::COLON_STATE_COMMENT
;
2472 # Illegal early close? This shouldn't happen D:
2473 $state = self
::COLON_STATE_TEXT
;
2476 $state = self
::COLON_STATE_TAG
;
2479 case 3: # self::COLON_STATE_CLOSETAG:
2484 wfDebug( __METHOD__
.": Invalid input; too many close tags\n" );
2485 wfProfileOut( __METHOD__
);
2488 $state = self
::COLON_STATE_TEXT
;
2491 case self
::COLON_STATE_TAGSLASH
:
2493 # Yes, a self-closed tag <blah/>
2494 $state = self
::COLON_STATE_TEXT
;
2496 # Probably we're jumping the gun, and this is an attribute
2497 $state = self
::COLON_STATE_TAG
;
2500 case 5: # self::COLON_STATE_COMMENT:
2502 $state = self
::COLON_STATE_COMMENTDASH
;
2505 case self
::COLON_STATE_COMMENTDASH
:
2507 $state = self
::COLON_STATE_COMMENTDASHDASH
;
2509 $state = self
::COLON_STATE_COMMENT
;
2512 case self
::COLON_STATE_COMMENTDASHDASH
:
2514 $state = self
::COLON_STATE_TEXT
;
2516 $state = self
::COLON_STATE_COMMENT
;
2520 throw new MWException( "State machine error in " . __METHOD__
);
2524 wfDebug( __METHOD__
.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2525 wfProfileOut( __METHOD__
);
2528 wfProfileOut( __METHOD__
);
2533 * Return value of a magic variable (like PAGENAME)
2537 * @param $index integer
2538 * @param $frame PPFrame
2540 function getVariableValue( $index, $frame=false ) {
2541 global $wgContLang, $wgSitename, $wgServer;
2542 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2545 * Some of these require message or data lookups and can be
2546 * expensive to check many times.
2548 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache
) ) ) {
2549 if ( isset( $this->mVarCache
[$index] ) ) {
2550 return $this->mVarCache
[$index];
2554 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2555 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2558 global $wgLocaltimezone;
2559 if ( isset( $wgLocaltimezone ) ) {
2560 $oldtz = date_default_timezone_get();
2561 date_default_timezone_set( $wgLocaltimezone );
2564 $localTimestamp = date( 'YmdHis', $ts );
2565 $localMonth = date( 'm', $ts );
2566 $localMonth1 = date( 'n', $ts );
2567 $localMonthName = date( 'n', $ts );
2568 $localDay = date( 'j', $ts );
2569 $localDay2 = date( 'd', $ts );
2570 $localDayOfWeek = date( 'w', $ts );
2571 $localWeek = date( 'W', $ts );
2572 $localYear = date( 'Y', $ts );
2573 $localHour = date( 'H', $ts );
2574 if ( isset( $wgLocaltimezone ) ) {
2575 date_default_timezone_set( $oldtz );
2579 case 'currentmonth':
2580 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2582 case 'currentmonth1':
2583 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2585 case 'currentmonthname':
2586 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2588 case 'currentmonthnamegen':
2589 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2591 case 'currentmonthabbrev':
2592 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2595 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2598 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2601 $value = $wgContLang->formatNum( $localMonth );
2604 $value = $wgContLang->formatNum( $localMonth1 );
2606 case 'localmonthname':
2607 $value = $wgContLang->getMonthName( $localMonthName );
2609 case 'localmonthnamegen':
2610 $value = $wgContLang->getMonthNameGen( $localMonthName );
2612 case 'localmonthabbrev':
2613 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2616 $value = $wgContLang->formatNum( $localDay );
2619 $value = $wgContLang->formatNum( $localDay2 );
2622 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2625 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2627 case 'fullpagename':
2628 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2630 case 'fullpagenamee':
2631 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2634 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2636 case 'subpagenamee':
2637 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2639 case 'basepagename':
2640 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2642 case 'basepagenamee':
2643 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle
->getBaseText() ) ) );
2645 case 'talkpagename':
2646 if ( $this->mTitle
->canTalk() ) {
2647 $talkPage = $this->mTitle
->getTalkPage();
2648 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2653 case 'talkpagenamee':
2654 if ( $this->mTitle
->canTalk() ) {
2655 $talkPage = $this->mTitle
->getTalkPage();
2656 $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2661 case 'subjectpagename':
2662 $subjPage = $this->mTitle
->getSubjectPage();
2663 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2665 case 'subjectpagenamee':
2666 $subjPage = $this->mTitle
->getSubjectPage();
2667 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2670 # Let the edit saving system know we should parse the page
2671 # *after* a revision ID has been assigned.
2672 $this->mOutput
->setFlag( 'vary-revision' );
2673 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision...\n" );
2674 $value = $this->mRevisionId
;
2677 # Let the edit saving system know we should parse the page
2678 # *after* a revision ID has been assigned. This is for null edits.
2679 $this->mOutput
->setFlag( 'vary-revision' );
2680 wfDebug( __METHOD__
. ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2681 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2683 case 'revisionday2':
2684 # Let the edit saving system know we should parse the page
2685 # *after* a revision ID has been assigned. This is for null edits.
2686 $this->mOutput
->setFlag( 'vary-revision' );
2687 wfDebug( __METHOD__
. ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2688 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2690 case 'revisionmonth':
2691 # Let the edit saving system know we should parse the page
2692 # *after* a revision ID has been assigned. This is for null edits.
2693 $this->mOutput
->setFlag( 'vary-revision' );
2694 wfDebug( __METHOD__
. ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2695 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2697 case 'revisionmonth1':
2698 # Let the edit saving system know we should parse the page
2699 # *after* a revision ID has been assigned. This is for null edits.
2700 $this->mOutput
->setFlag( 'vary-revision' );
2701 wfDebug( __METHOD__
. ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2702 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2704 case 'revisionyear':
2705 # Let the edit saving system know we should parse the page
2706 # *after* a revision ID has been assigned. This is for null edits.
2707 $this->mOutput
->setFlag( 'vary-revision' );
2708 wfDebug( __METHOD__
. ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2709 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2711 case 'revisiontimestamp':
2712 # Let the edit saving system know we should parse the page
2713 # *after* a revision ID has been assigned. This is for null edits.
2714 $this->mOutput
->setFlag( 'vary-revision' );
2715 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2716 $value = $this->getRevisionTimestamp();
2718 case 'revisionuser':
2719 # Let the edit saving system know we should parse the page
2720 # *after* a revision ID has been assigned. This is for null edits.
2721 $this->mOutput
->setFlag( 'vary-revision' );
2722 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2723 $value = $this->getRevisionUser();
2726 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2729 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle
->getNamespace() ) );
2732 $value = $this->mTitle
->canTalk() ?
str_replace( '_',' ',$this->mTitle
->getTalkNsText() ) : '';
2735 $value = $this->mTitle
->canTalk() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2737 case 'subjectspace':
2738 $value = $this->mTitle
->getSubjectNsText();
2740 case 'subjectspacee':
2741 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2743 case 'currentdayname':
2744 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) +
1 );
2747 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2750 $value = $wgContLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2753 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2756 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2757 # int to remove the padding
2758 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2761 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2763 case 'localdayname':
2764 $value = $wgContLang->getWeekdayName( $localDayOfWeek +
1 );
2767 $value = $wgContLang->formatNum( $localYear, true );
2770 $value = $wgContLang->time( $localTimestamp, false, false );
2773 $value = $wgContLang->formatNum( $localHour, true );
2776 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2777 # int to remove the padding
2778 $value = $wgContLang->formatNum( (int)$localWeek );
2781 $value = $wgContLang->formatNum( $localDayOfWeek );
2783 case 'numberofarticles':
2784 $value = $wgContLang->formatNum( SiteStats
::articles() );
2786 case 'numberoffiles':
2787 $value = $wgContLang->formatNum( SiteStats
::images() );
2789 case 'numberofusers':
2790 $value = $wgContLang->formatNum( SiteStats
::users() );
2792 case 'numberofactiveusers':
2793 $value = $wgContLang->formatNum( SiteStats
::activeUsers() );
2795 case 'numberofpages':
2796 $value = $wgContLang->formatNum( SiteStats
::pages() );
2798 case 'numberofadmins':
2799 $value = $wgContLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2801 case 'numberofedits':
2802 $value = $wgContLang->formatNum( SiteStats
::edits() );
2804 case 'numberofviews':
2805 $value = $wgContLang->formatNum( SiteStats
::views() );
2807 case 'currenttimestamp':
2808 $value = wfTimestamp( TS_MW
, $ts );
2810 case 'localtimestamp':
2811 $value = $localTimestamp;
2813 case 'currentversion':
2814 $value = SpecialVersion
::getVersion();
2817 return $wgArticlePath;
2823 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2824 $serverName = parse_url( $wgServer, PHP_URL_HOST
);
2825 wfRestoreWarnings();
2826 return $serverName ?
$serverName : $wgServer;
2828 return $wgScriptPath;
2830 return $wgStylePath;
2831 case 'directionmark':
2832 return $wgContLang->getDirMark();
2833 case 'contentlanguage':
2834 global $wgLanguageCode;
2835 return $wgLanguageCode;
2838 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache
, &$index, &$ret, &$frame ) ) ) {
2846 $this->mVarCache
[$index] = $value;
2852 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2856 function initialiseVariables() {
2857 wfProfileIn( __METHOD__
);
2858 $variableIDs = MagicWord
::getVariableIDs();
2859 $substIDs = MagicWord
::getSubstIDs();
2861 $this->mVariables
= new MagicWordArray( $variableIDs );
2862 $this->mSubstWords
= new MagicWordArray( $substIDs );
2863 wfProfileOut( __METHOD__
);
2867 * Preprocess some wikitext and return the document tree.
2868 * This is the ghost of replace_variables().
2870 * @param $text String: The text to parse
2871 * @param $flags Integer: bitwise combination of:
2872 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2873 * included. Default is to assume a direct page view.
2875 * The generated DOM tree must depend only on the input text and the flags.
2876 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2878 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2879 * change in the DOM tree for a given text, must be passed through the section identifier
2880 * in the section edit link and thus back to extractSections().
2882 * The output of this function is currently only cached in process memory, but a persistent
2883 * cache may be implemented at a later date which takes further advantage of these strict
2884 * dependency requirements.
2890 function preprocessToDom( $text, $flags = 0 ) {
2891 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2896 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2900 public static function splitWhitespace( $s ) {
2901 $ltrimmed = ltrim( $s );
2902 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2903 $trimmed = rtrim( $ltrimmed );
2904 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2906 $w2 = substr( $ltrimmed, -$diff );
2910 return array( $w1, $trimmed, $w2 );
2914 * Replace magic variables, templates, and template arguments
2915 * with the appropriate text. Templates are substituted recursively,
2916 * taking care to avoid infinite loops.
2918 * Note that the substitution depends on value of $mOutputType:
2919 * self::OT_WIKI: only {{subst:}} templates
2920 * self::OT_PREPROCESS: templates but not extension tags
2921 * self::OT_HTML: all templates and extension tags
2923 * @param $text String: the text to transform
2924 * @param $frame PPFrame Object describing the arguments passed to the template.
2925 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2926 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2927 * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2932 function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2933 # Is there any text? Also, Prevent too big inclusions!
2934 if ( strlen( $text ) < 1 ||
strlen( $text ) > $this->mOptions
->getMaxIncludeSize() ) {
2937 wfProfileIn( __METHOD__
);
2939 if ( $frame === false ) {
2940 $frame = $this->getPreprocessor()->newFrame();
2941 } elseif ( !( $frame instanceof PPFrame
) ) {
2942 wfDebug( __METHOD__
." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2943 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2946 $dom = $this->preprocessToDom( $text );
2947 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
2948 $text = $frame->expand( $dom, $flags );
2950 wfProfileOut( __METHOD__
);
2955 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2959 static function createAssocArgs( $args ) {
2960 $assocArgs = array();
2962 foreach ( $args as $arg ) {
2963 $eqpos = strpos( $arg, '=' );
2964 if ( $eqpos === false ) {
2965 $assocArgs[$index++
] = $arg;
2967 $name = trim( substr( $arg, 0, $eqpos ) );
2968 $value = trim( substr( $arg, $eqpos+
1 ) );
2969 if ( $value === false ) {
2972 if ( $name !== false ) {
2973 $assocArgs[$name] = $value;
2982 * Warn the user when a parser limitation is reached
2983 * Will warn at most once the user per limitation type
2985 * @param $limitationType String: should be one of:
2986 * 'expensive-parserfunction' (corresponding messages:
2987 * 'expensive-parserfunction-warning',
2988 * 'expensive-parserfunction-category')
2989 * 'post-expand-template-argument' (corresponding messages:
2990 * 'post-expand-template-argument-warning',
2991 * 'post-expand-template-argument-category')
2992 * 'post-expand-template-inclusion' (corresponding messages:
2993 * 'post-expand-template-inclusion-warning',
2994 * 'post-expand-template-inclusion-category')
2995 * @param $current Current value
2996 * @param $max Maximum allowed, when an explicit limit has been
2997 * exceeded, provide the values (optional)
2999 function limitationWarn( $limitationType, $current=null, $max=null) {
3000 # does no harm if $current and $max are present but are unnecessary for the message
3001 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3002 $this->mOutput
->addWarning( $warning );
3003 $this->addTrackingCategory( "$limitationType-category" );
3007 * Return the text of a template, after recursively
3008 * replacing any variables or templates within the template.
3010 * @param $piece Array: the parts of the template
3011 * $piece['title']: the title, i.e. the part before the |
3012 * $piece['parts']: the parameter array
3013 * $piece['lineStart']: whether the brace was at the start of a line
3014 * @param $frame PPFrame The current frame, contains template arguments
3015 * @return String: the text of the template
3018 function braceSubstitution( $piece, $frame ) {
3019 global $wgContLang, $wgNonincludableNamespaces;
3020 wfProfileIn( __METHOD__
);
3021 wfProfileIn( __METHOD__
.'-setup' );
3024 $found = false; # $text has been filled
3025 $nowiki = false; # wiki markup in $text should be escaped
3026 $isHTML = false; # $text is HTML, armour it against wikitext transformation
3027 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3028 $isChildObj = false; # $text is a DOM node needing expansion in a child frame
3029 $isLocalObj = false; # $text is a DOM node needing expansion in the current frame
3031 # Title object, where $text came from
3034 # $part1 is the bit before the first |, and must contain only title characters.
3035 # Various prefixes will be stripped from it later.
3036 $titleWithSpaces = $frame->expand( $piece['title'] );
3037 $part1 = trim( $titleWithSpaces );
3040 # Original title text preserved for various purposes
3041 $originalTitle = $part1;
3043 # $args is a list of argument nodes, starting from index 0, not including $part1
3044 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3045 $args = ( null == $piece['parts'] ) ?
array() : $piece['parts'];
3046 wfProfileOut( __METHOD__
.'-setup' );
3049 wfProfileIn( __METHOD__
.'-modifiers' );
3052 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3054 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3055 # Decide whether to expand template or keep wikitext as-is.
3056 if ( $this->ot
['wiki'] ) {
3057 if ( $substMatch === false ) {
3058 $literal = true; # literal when in PST with no prefix
3060 $literal = false; # expand when in PST with subst: or safesubst:
3063 if ( $substMatch == 'subst' ) {
3064 $literal = true; # literal when not in PST with plain subst:
3066 $literal = false; # expand when not in PST with safesubst: or no prefix
3070 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3077 if ( !$found && $args->getLength() == 0 ) {
3078 $id = $this->mVariables
->matchStartToEnd( $part1 );
3079 if ( $id !== false ) {
3080 $text = $this->getVariableValue( $id, $frame );
3081 if ( MagicWord
::getCacheTTL( $id ) > -1 ) {
3082 $this->mOutput
->updateCacheExpiry( MagicWord
::getCacheTTL( $id ) );
3088 # MSG, MSGNW and RAW
3091 $mwMsgnw = MagicWord
::get( 'msgnw' );
3092 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3095 # Remove obsolete MSG:
3096 $mwMsg = MagicWord
::get( 'msg' );
3097 $mwMsg->matchStartAndRemove( $part1 );
3101 $mwRaw = MagicWord
::get( 'raw' );
3102 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3103 $forceRawInterwiki = true;
3106 wfProfileOut( __METHOD__
.'-modifiers' );
3110 wfProfileIn( __METHOD__
. '-pfunc' );
3112 $colonPos = strpos( $part1, ':' );
3113 if ( $colonPos !== false ) {
3114 # Case sensitive functions
3115 $function = substr( $part1, 0, $colonPos );
3116 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3117 $function = $this->mFunctionSynonyms
[1][$function];
3119 # Case insensitive functions
3120 $function = $wgContLang->lc( $function );
3121 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3122 $function = $this->mFunctionSynonyms
[0][$function];
3128 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3129 $initialArgs = array( &$this );
3130 $funcArgs = array( trim( substr( $part1, $colonPos +
1 ) ) );
3131 if ( $flags & SFH_OBJECT_ARGS
) {
3132 # Add a frame parameter, and pass the arguments as an array
3133 $allArgs = $initialArgs;
3134 $allArgs[] = $frame;
3135 for ( $i = 0; $i < $args->getLength(); $i++
) {
3136 $funcArgs[] = $args->item( $i );
3138 $allArgs[] = $funcArgs;
3140 # Convert arguments to plain text
3141 for ( $i = 0; $i < $args->getLength(); $i++
) {
3142 $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3144 $allArgs = array_merge( $initialArgs, $funcArgs );
3147 # Workaround for PHP bug 35229 and similar
3148 if ( !is_callable( $callback ) ) {
3149 wfProfileOut( __METHOD__
. '-pfunc' );
3150 wfProfileOut( __METHOD__
);
3151 throw new MWException( "Tag hook for $function is not callable\n" );
3153 $result = call_user_func_array( $callback, $allArgs );
3156 $preprocessFlags = 0;
3158 if ( is_array( $result ) ) {
3159 if ( isset( $result[0] ) ) {
3161 unset( $result[0] );
3164 # Extract flags into the local scope
3165 # This allows callers to set flags such as nowiki, found, etc.
3171 $text = $this->preprocessToDom( $text, $preprocessFlags );
3176 wfProfileOut( __METHOD__
. '-pfunc' );
3179 # Finish mangling title and then check for loops.
3180 # Set $title to a Title object and $titleText to the PDBK
3183 # Split the title into page and subpage
3185 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3186 if ( $subpage !== '' ) {
3187 $ns = $this->mTitle
->getNamespace();
3189 $title = Title
::newFromText( $part1, $ns );
3191 $titleText = $title->getPrefixedText();
3192 # Check for language variants if the template is not found
3193 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3194 $wgContLang->findVariantLink( $part1, $title, true );
3196 # Do recursion depth check
3197 $limit = $this->mOptions
->getMaxTemplateDepth();
3198 if ( $frame->depth
>= $limit ) {
3200 $text = '<span class="error">'
3201 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3207 # Load from database
3208 if ( !$found && $title ) {
3209 wfProfileIn( __METHOD__
. '-loadtpl' );
3210 if ( !$title->isExternal() ) {
3211 if ( $title->getNamespace() == NS_SPECIAL
3212 && $this->mOptions
->getAllowSpecialInclusion()
3213 && $this->ot
['html'] )
3215 $text = SpecialPageFactory
::capturePath( $title );
3216 if ( is_string( $text ) ) {
3219 $this->disableCache();
3221 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3222 $found = false; # access denied
3223 wfDebug( __METHOD__
.": template inclusion denied for " . $title->getPrefixedDBkey() );
3225 list( $text, $title ) = $this->getTemplateDom( $title );
3226 if ( $text !== false ) {
3232 # If the title is valid but undisplayable, make a link to it
3233 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3234 $text = "[[:$titleText]]";
3237 } elseif ( $title->isTrans() ) {
3238 # Interwiki transclusion
3239 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3240 $text = $this->interwikiTransclude( $title, 'render' );
3243 $text = $this->interwikiTransclude( $title, 'raw' );
3244 # Preprocess it like a template
3245 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3251 # Do infinite loop check
3252 # This has to be done after redirect resolution to avoid infinite loops via redirects
3253 if ( !$frame->loopCheck( $title ) ) {
3255 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3256 wfDebug( __METHOD__
.": template loop broken at '$titleText'\n" );
3258 wfProfileOut( __METHOD__
. '-loadtpl' );
3261 # If we haven't found text to substitute by now, we're done
3262 # Recover the source wikitext and return it
3264 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3265 wfProfileOut( __METHOD__
);
3266 return array( 'object' => $text );
3269 # Expand DOM-style return values in a child frame
3270 if ( $isChildObj ) {
3271 # Clean up argument array
3272 $newFrame = $frame->newChild( $args, $title );
3275 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3276 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3277 # Expansion is eligible for the empty-frame cache
3278 if ( isset( $this->mTplExpandCache
[$titleText] ) ) {
3279 $text = $this->mTplExpandCache
[$titleText];
3281 $text = $newFrame->expand( $text );
3282 $this->mTplExpandCache
[$titleText] = $text;
3285 # Uncached expansion
3286 $text = $newFrame->expand( $text );
3289 if ( $isLocalObj && $nowiki ) {
3290 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3291 $isLocalObj = false;
3294 # Replace raw HTML by a placeholder
3295 # Add a blank line preceding, to prevent it from mucking up
3296 # immediately preceding headings
3298 $text = "\n\n" . $this->insertStripItem( $text );
3299 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3300 # Escape nowiki-style return values
3301 $text = wfEscapeWikiText( $text );
3302 } elseif ( is_string( $text )
3303 && !$piece['lineStart']
3304 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3306 # Bug 529: if the template begins with a table or block-level
3307 # element, it should be treated as beginning a new line.
3308 # This behaviour is somewhat controversial.
3309 $text = "\n" . $text;
3312 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3313 # Error, oversize inclusion
3314 if ( $titleText !== false ) {
3315 # Make a working, properly escaped link if possible (bug 23588)
3316 $text = "[[:$titleText]]";
3318 # This will probably not be a working link, but at least it may
3319 # provide some hint of where the problem is
3320 preg_replace( '/^:/', '', $originalTitle );
3321 $text = "[[:$originalTitle]]";
3323 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3324 $this->limitationWarn( 'post-expand-template-inclusion' );
3327 if ( $isLocalObj ) {
3328 $ret = array( 'object' => $text );
3330 $ret = array( 'text' => $text );
3333 wfProfileOut( __METHOD__
);
3338 * Get the semi-parsed DOM representation of a template with a given title,
3339 * and its redirect destination title. Cached.
3343 function getTemplateDom( $title ) {
3344 $cacheTitle = $title;
3345 $titleText = $title->getPrefixedDBkey();
3347 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3348 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3349 $title = Title
::makeTitle( $ns, $dbk );
3350 $titleText = $title->getPrefixedDBkey();
3352 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3353 return array( $this->mTplDomCache
[$titleText], $title );
3356 # Cache miss, go to the database
3357 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3359 if ( $text === false ) {
3360 $this->mTplDomCache
[$titleText] = false;
3361 return array( false, $title );
3364 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3365 $this->mTplDomCache
[ $titleText ] = $dom;
3367 if ( !$title->equals( $cacheTitle ) ) {
3368 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3369 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3372 return array( $dom, $title );
3376 * Fetch the unparsed text of a template and register a reference to it.
3377 * @param Title $title
3378 * @return Array ( string or false, Title )
3380 function fetchTemplateAndTitle( $title ) {
3381 $templateCb = $this->mOptions
->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3382 $stuff = call_user_func( $templateCb, $title, $this );
3383 $text = $stuff['text'];
3384 $finalTitle = isset( $stuff['finalTitle'] ) ?
$stuff['finalTitle'] : $title;
3385 if ( isset( $stuff['deps'] ) ) {
3386 foreach ( $stuff['deps'] as $dep ) {
3387 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3390 return array( $text, $finalTitle );
3394 * Fetch the unparsed text of a template and register a reference to it.
3395 * @param Title $title
3396 * @return mixed string or false
3398 function fetchTemplate( $title ) {
3399 $rv = $this->fetchTemplateAndTitle( $title );
3404 * Static function to get a template
3405 * Can be overridden via ParserOptions::setTemplateCallback().
3409 static function statelessFetchTemplate( $title, $parser = false ) {
3410 $text = $skip = false;
3411 $finalTitle = $title;
3414 # Loop to fetch the article, with up to 1 redirect
3415 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3416 # Give extensions a chance to select the revision instead
3417 $id = false; # Assume current
3418 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3419 array( $parser, $title, &$skip, &$id ) );
3425 'page_id' => $title->getArticleID(),
3432 ? Revision
::newFromId( $id )
3433 : Revision
::newFromTitle( $title );
3434 $rev_id = $rev ?
$rev->getId() : 0;
3435 # If there is no current revision, there is no page
3436 if ( $id === false && !$rev ) {
3437 $linkCache = LinkCache
::singleton();
3438 $linkCache->addBadLinkObj( $title );
3443 'page_id' => $title->getArticleID(),
3444 'rev_id' => $rev_id );
3445 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3446 # We fetched a rev from a different title; register it too...
3448 'title' => $rev->getTitle(),
3449 'page_id' => $rev->getPage(),
3450 'rev_id' => $rev_id );
3454 $text = $rev->getText();
3455 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3457 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3458 if ( !$message->exists() ) {
3462 $text = $message->plain();
3466 if ( $text === false ) {
3470 $finalTitle = $title;
3471 $title = Title
::newFromRedirect( $text );
3475 'finalTitle' => $finalTitle,
3480 * Fetch a file and its title and register a reference to it.
3481 * @param Title $title
3482 * @param string $time MW timestamp
3483 * @param string $sha1 base 36 SHA-1
3484 * @return mixed File or false
3486 function fetchFile( $title, $time = false, $sha1 = false ) {
3487 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3492 * Fetch a file and its title and register a reference to it.
3493 * @param Title $title
3494 * @param string $time MW timestamp
3495 * @param string $sha1 base 36 SHA-1
3496 * @return Array ( File or false, Title of file )
3498 function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3499 if ( $time === '0' ) {
3500 $file = false; // broken thumbnail forced by hook
3501 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3502 $file = RepoGroup
::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3503 } else { // get by (name,timestamp)
3504 $file = wfFindFile( $title, array( 'time' => $time ) );
3506 $time = $file ?
$file->getTimestamp() : false;
3507 $sha1 = $file ?
$file->getSha1() : false;
3508 # Register the file as a dependency...
3509 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3510 if ( $file && !$title->equals( $file->getTitle() ) ) {
3511 # We fetched a rev from a different title; register it too...
3512 $this->mOutput
->addImage( $file->getTitle()->getDBkey(), $time, $sha1 );
3513 # Update fetched file title
3514 $title = $file->getTitle();
3516 return array( $file, $title );
3520 * Transclude an interwiki link.
3524 function interwikiTransclude( $title, $action ) {
3525 global $wgEnableScaryTranscluding;
3527 if ( !$wgEnableScaryTranscluding ) {
3528 return wfMsgForContent('scarytranscludedisabled');
3531 $url = $title->getFullUrl( "action=$action" );
3533 if ( strlen( $url ) > 255 ) {
3534 return wfMsgForContent( 'scarytranscludetoolong' );
3536 return $this->fetchScaryTemplateMaybeFromCache( $url );
3540 * @param $url string
3541 * @return Mixed|String
3543 function fetchScaryTemplateMaybeFromCache( $url ) {
3544 global $wgTranscludeCacheExpiry;
3545 $dbr = wfGetDB( DB_SLAVE
);
3546 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3547 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3548 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3550 return $obj->tc_contents
;
3553 $text = Http
::get( $url );
3555 return wfMsgForContent( 'scarytranscludefailed', $url );
3558 $dbw = wfGetDB( DB_MASTER
);
3559 $dbw->replace( 'transcache', array('tc_url'), array(
3561 'tc_time' => $dbw->timestamp( time() ),
3562 'tc_contents' => $text)
3568 * Triple brace replacement -- used for template arguments
3571 * @param $peice array
3572 * @param $frame PPFrame
3576 function argSubstitution( $piece, $frame ) {
3577 wfProfileIn( __METHOD__
);
3580 $parts = $piece['parts'];
3581 $nameWithSpaces = $frame->expand( $piece['title'] );
3582 $argName = trim( $nameWithSpaces );
3584 $text = $frame->getArgument( $argName );
3585 if ( $text === false && $parts->getLength() > 0
3589 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3592 # No match in frame, use the supplied default
3593 $object = $parts->item( 0 )->getChildren();
3595 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3596 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3597 $this->limitationWarn( 'post-expand-template-argument' );
3600 if ( $text === false && $object === false ) {
3602 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3604 if ( $error !== false ) {
3607 if ( $object !== false ) {
3608 $ret = array( 'object' => $object );
3610 $ret = array( 'text' => $text );
3613 wfProfileOut( __METHOD__
);
3618 * Return the text to be used for a given extension tag.
3619 * This is the ghost of strip().
3621 * @param $params Associative array of parameters:
3622 * name PPNode for the tag name
3623 * attr PPNode for unparsed text where tag attributes are thought to be
3624 * attributes Optional associative array of parsed attributes
3625 * inner Contents of extension element
3626 * noClose Original text did not have a close tag
3627 * @param $frame PPFrame
3631 function extensionSubstitution( $params, $frame ) {
3632 $name = $frame->expand( $params['name'] );
3633 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3634 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3635 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3637 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower($name)] ) &&
3638 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3639 if ( $isFunctionTag ) {
3640 $markerType = 'none';
3642 $markerType = 'general';
3644 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3645 $name = strtolower( $name );
3646 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3647 if ( isset( $params['attributes'] ) ) {
3648 $attributes = $attributes +
$params['attributes'];
3651 if ( isset( $this->mTagHooks
[$name] ) ) {
3652 # Workaround for PHP bug 35229 and similar
3653 if ( !is_callable( $this->mTagHooks
[$name] ) ) {
3654 throw new MWException( "Tag hook for $name is not callable\n" );
3656 $output = call_user_func_array( $this->mTagHooks
[$name],
3657 array( $content, $attributes, $this, $frame ) );
3658 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
3659 list( $callback, $flags ) = $this->mFunctionTagHooks
[$name];
3660 if ( !is_callable( $callback ) ) {
3661 throw new MWException( "Tag hook for $name is not callable\n" );
3664 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3666 $output = '<span class="error">Invalid tag extension name: ' .
3667 htmlspecialchars( $name ) . '</span>';
3670 if ( is_array( $output ) ) {
3671 # Extract flags to local scope (to override $markerType)
3673 $output = $flags[0];
3678 if ( is_null( $attrText ) ) {
3681 if ( isset( $params['attributes'] ) ) {
3682 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3683 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3684 htmlspecialchars( $attrValue ) . '"';
3687 if ( $content === null ) {
3688 $output = "<$name$attrText/>";
3690 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
3691 $output = "<$name$attrText>$content$close";
3695 if ( $markerType === 'none' ) {
3697 } elseif ( $markerType === 'nowiki' ) {
3698 $this->mStripState
->addNoWiki( $marker, $output );
3699 } elseif ( $markerType === 'general' ) {
3700 $this->mStripState
->addGeneral( $marker, $output );
3702 throw new MWException( __METHOD__
.': invalid marker type' );
3708 * Increment an include size counter
3710 * @param $type String: the type of expansion
3711 * @param $size Integer: the size of the text
3712 * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3714 function incrementIncludeSize( $type, $size ) {
3715 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
3718 $this->mIncludeSizes
[$type] +
= $size;
3724 * Increment the expensive function count
3726 * @return Boolean: false if the limit has been exceeded
3728 function incrementExpensiveFunctionCount() {
3729 global $wgExpensiveParserFunctionLimit;
3730 $this->mExpensiveFunctionCount++
;
3731 if ( $this->mExpensiveFunctionCount
<= $wgExpensiveParserFunctionLimit ) {
3738 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3739 * Fills $this->mDoubleUnderscores, returns the modified text
3741 function doDoubleUnderscore( $text ) {
3742 wfProfileIn( __METHOD__
);
3744 # The position of __TOC__ needs to be recorded
3745 $mw = MagicWord
::get( 'toc' );
3746 if ( $mw->match( $text ) ) {
3747 $this->mShowToc
= true;
3748 $this->mForceTocPosition
= true;
3750 # Set a placeholder. At the end we'll fill it in with the TOC.
3751 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3753 # Only keep the first one.
3754 $text = $mw->replace( '', $text );
3757 # Now match and remove the rest of them
3758 $mwa = MagicWord
::getDoubleUnderscoreArray();
3759 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
3761 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
3762 $this->mOutput
->mNoGallery
= true;
3764 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
3765 $this->mShowToc
= false;
3767 if ( isset( $this->mDoubleUnderscores
['hiddencat'] ) && $this->mTitle
->getNamespace() == NS_CATEGORY
) {
3768 $this->addTrackingCategory( 'hidden-category-category' );
3770 # (bug 8068) Allow control over whether robots index a page.
3772 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3773 # is not desirable, the last one on the page should win.
3774 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
3775 $this->mOutput
->setIndexPolicy( 'noindex' );
3776 $this->addTrackingCategory( 'noindex-category' );
3778 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
3779 $this->mOutput
->setIndexPolicy( 'index' );
3780 $this->addTrackingCategory( 'index-category' );
3783 # Cache all double underscores in the database
3784 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
3785 $this->mOutput
->setProperty( $key, '' );
3788 wfProfileOut( __METHOD__
);
3793 * Add a tracking category, getting the title from a system message,
3794 * or print a debug message if the title is invalid.
3796 * @param $msg String: message key
3797 * @return Boolean: whether the addition was successful
3799 protected function addTrackingCategory( $msg ) {
3800 $cat = wfMsgForContent( $msg );
3802 # Allow tracking categories to be disabled by setting them to "-"
3803 if ( $cat === '-' ) {
3807 $containerCategory = Title
::makeTitleSafe( NS_CATEGORY
, $cat );
3808 if ( $containerCategory ) {
3809 $this->mOutput
->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3812 wfDebug( __METHOD__
.": [[MediaWiki:$msg]] is not a valid title!\n" );
3818 * This function accomplishes several tasks:
3819 * 1) Auto-number headings if that option is enabled
3820 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3821 * 3) Add a Table of contents on the top for users who have enabled the option
3822 * 4) Auto-anchor headings
3824 * It loops through all headlines, collects the necessary data, then splits up the
3825 * string and re-inserts the newly formatted headlines.
3827 * @param $text String
3828 * @param $origText String: original, untouched wikitext
3829 * @param $isMain Boolean
3832 function formatHeadings( $text, $origText, $isMain=true ) {
3833 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3835 # Inhibit editsection links if requested in the page
3836 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
3839 $showEditLink = $this->mOptions
->getEditSection();
3841 if ( $showEditLink ) {
3842 $this->mOutput
->setEditSectionTokens( true );
3845 # Get all headlines for numbering them and adding funky stuff like [edit]
3846 # links - this is for later, but we need the number of headlines right now
3848 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3850 # if there are fewer than 4 headlines in the article, do not show TOC
3851 # unless it's been explicitly enabled.
3852 $enoughToc = $this->mShowToc
&&
3853 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
3855 # Allow user to stipulate that a page should have a "new section"
3856 # link added via __NEWSECTIONLINK__
3857 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
3858 $this->mOutput
->setNewSection( true );
3861 # Allow user to remove the "new section"
3862 # link via __NONEWSECTIONLINK__
3863 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
3864 $this->mOutput
->hideNewSection( true );
3867 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3868 # override above conditions and always show TOC above first header
3869 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
3870 $this->mShowToc
= true;
3878 # Ugh .. the TOC should have neat indentation levels which can be
3879 # passed to the skin functions. These are determined here
3883 $sublevelCount = array();
3884 $levelCount = array();
3889 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self
::MARKER_SUFFIX
;
3890 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
3891 $oldType = $this->mOutputType
;
3892 $this->setOutputType( self
::OT_WIKI
);
3893 $frame = $this->getPreprocessor()->newFrame();
3894 $root = $this->preprocessToDom( $origText );
3895 $node = $root->getFirstChild();
3900 foreach ( $matches[3] as $headline ) {
3901 $isTemplate = false;
3903 $sectionIndex = false;
3905 $markerMatches = array();
3906 if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3907 $serial = $markerMatches[1];
3908 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
3909 $isTemplate = ( $titleText != $baseTitleText );
3910 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3914 $prevlevel = $level;
3916 $level = $matches[1][$headlineCount];
3918 if ( $level > $prevlevel ) {
3919 # Increase TOC level
3921 $sublevelCount[$toclevel] = 0;
3922 if ( $toclevel<$wgMaxTocLevel ) {
3923 $prevtoclevel = $toclevel;
3924 $toc .= Linker
::tocIndent();
3927 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3928 # Decrease TOC level, find level to jump to
3930 for ( $i = $toclevel; $i > 0; $i-- ) {
3931 if ( $levelCount[$i] == $level ) {
3932 # Found last matching level
3935 } elseif ( $levelCount[$i] < $level ) {
3936 # Found first matching level below current level
3944 if ( $toclevel<$wgMaxTocLevel ) {
3945 if ( $prevtoclevel < $wgMaxTocLevel ) {
3946 # Unindent only if the previous toc level was shown :p
3947 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
3948 $prevtoclevel = $toclevel;
3950 $toc .= Linker
::tocLineEnd();
3954 # No change in level, end TOC line
3955 if ( $toclevel<$wgMaxTocLevel ) {
3956 $toc .= Linker
::tocLineEnd();
3960 $levelCount[$toclevel] = $level;
3962 # count number of headlines for each level
3963 @$sublevelCount[$toclevel]++
;
3965 for( $i = 1; $i <= $toclevel; $i++
) {
3966 if ( !empty( $sublevelCount[$i] ) ) {
3970 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3975 # The safe header is a version of the header text safe to use for links
3976 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3977 $safeHeadline = $this->mStripState
->unstripBoth( $headline );
3979 # Remove link placeholders by the link text.
3980 # <!--LINK number-->
3982 # link text with suffix
3983 $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3985 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3986 $tocline = preg_replace(
3987 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3988 array( '', '<$1>' ),
3991 $tocline = trim( $tocline );
3993 # For the anchor, strip out HTML-y stuff period
3994 $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3995 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
3997 # Save headline for section edit hint before it's escaped
3998 $headlineHint = $safeHeadline;
4000 if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4001 # For reverse compatibility, provide an id that's
4002 # HTML4-compatible, like we used to.
4004 # It may be worth noting, academically, that it's possible for
4005 # the legacy anchor to conflict with a non-legacy headline
4006 # anchor on the page. In this case likely the "correct" thing
4007 # would be to either drop the legacy anchors or make sure
4008 # they're numbered first. However, this would require people
4009 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4010 # manually, so let's not bother worrying about it.
4011 $legacyHeadline = Sanitizer
::escapeId( $safeHeadline,
4012 array( 'noninitial', 'legacy' ) );
4013 $safeHeadline = Sanitizer
::escapeId( $safeHeadline );
4015 if ( $legacyHeadline == $safeHeadline ) {
4016 # No reason to have both (in fact, we can't)
4017 $legacyHeadline = false;
4020 $legacyHeadline = false;
4021 $safeHeadline = Sanitizer
::escapeId( $safeHeadline,
4025 # HTML names must be case-insensitively unique (bug 10721).
4026 # This does not apply to Unicode characters per
4027 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4028 # FIXME: We may be changing them depending on the current locale.
4029 $arrayKey = strtolower( $safeHeadline );
4030 if ( $legacyHeadline === false ) {
4031 $legacyArrayKey = false;
4033 $legacyArrayKey = strtolower( $legacyHeadline );
4036 # count how many in assoc. array so we can track dupes in anchors
4037 if ( isset( $refers[$arrayKey] ) ) {
4038 $refers[$arrayKey]++
;
4040 $refers[$arrayKey] = 1;
4042 if ( isset( $refers[$legacyArrayKey] ) ) {
4043 $refers[$legacyArrayKey]++
;
4045 $refers[$legacyArrayKey] = 1;
4048 # Don't number the heading if it is the only one (looks silly)
4049 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4050 # the two are different if the line contains a link
4051 $headline = $numbering . ' ' . $headline;
4054 # Create the anchor for linking from the TOC to the section
4055 $anchor = $safeHeadline;
4056 $legacyAnchor = $legacyHeadline;
4057 if ( $refers[$arrayKey] > 1 ) {
4058 $anchor .= '_' . $refers[$arrayKey];
4060 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4061 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4063 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) ||
$toclevel < $wgMaxTocLevel ) ) {
4064 $toc .= Linker
::tocLine( $anchor, $tocline,
4065 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4068 # Add the section to the section tree
4069 # Find the DOM node for this header
4070 while ( $node && !$isTemplate ) {
4071 if ( $node->getName() === 'h' ) {
4072 $bits = $node->splitHeading();
4073 if ( $bits['i'] == $sectionIndex ) {
4077 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4078 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4079 $node = $node->getNextSibling();
4082 'toclevel' => $toclevel,
4085 'number' => $numbering,
4086 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4087 'fromtitle' => $titleText,
4088 'byteoffset' => ( $isTemplate ?
null : $byteOffset ),
4089 'anchor' => $anchor,
4092 # give headline the correct <h#> tag
4093 if ( $showEditLink && $sectionIndex !== false ) {
4094 // Output edit section links as markers with styles that can be customized by skins
4095 if ( $isTemplate ) {
4096 # Put a T flag in the section identifier, to indicate to extractSections()
4097 # that sections inside <includeonly> should be counted.
4098 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4100 $editlinkArgs = array( $this->mTitle
->getPrefixedText(), $sectionIndex, $headlineHint );
4102 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4103 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4104 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4105 // so we don't have to worry about a user trying to input one of these markers directly.
4106 // We use a page and section attribute to stop the language converter from converting these important bits
4107 // of data, but put the headline hint inside a content block because the language converter is supposed to
4108 // be able to convert that piece of data.
4109 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4110 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4111 if ( isset($editlinkArgs[2]) ) {
4112 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4119 $head[$headlineCount] = Linker
::makeHeadline( $level,
4120 $matches['attrib'][$headlineCount], $anchor, $headline,
4121 $editlink, $legacyAnchor );
4126 $this->setOutputType( $oldType );
4128 # Never ever show TOC if no headers
4129 if ( $numVisible < 1 ) {
4134 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4135 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4137 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLang() );
4138 $this->mOutput
->setTOCHTML( $toc );
4142 $this->mOutput
->setSections( $tocraw );
4145 # split up and insert constructed headlines
4147 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4150 foreach ( $blocks as $block ) {
4151 if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4152 # This is the [edit] link that appears for the top block of text when
4153 # section editing is enabled
4155 # Disabled because it broke block formatting
4156 # For example, a bullet point in the top line
4157 # $full .= $sk->editSectionLink(0);
4160 if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition
) {
4161 # Top anchor now in skin
4165 if ( !empty( $head[$i] ) ) {
4170 if ( $this->mForceTocPosition
) {
4171 return str_replace( '<!--MWTOC-->', $toc, $full );
4178 * Transform wiki markup when saving a page by doing \r\n -> \n
4179 * conversion, substitting signatures, {{subst:}} templates, etc.
4181 * @param $text String: the text to transform
4182 * @param $title Title: the Title object for the current article
4183 * @param $user User: the User object describing the current user
4184 * @param $options ParserOptions: parsing options
4185 * @param $clearState Boolean: whether to clear the parser state first
4186 * @return String: the altered wiki markup
4188 public function preSaveTransform( $text, Title
$title, User
$user, ParserOptions
$options, $clearState = true ) {
4189 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4190 $this->setUser( $user );
4195 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4196 if( $options->getPreSaveTransform() ) {
4197 $text = $this->pstPass2( $text, $user );
4199 $text = $this->mStripState
->unstripBoth( $text );
4201 $this->setUser( null ); #Reset
4207 * Pre-save transform helper function
4210 function pstPass2( $text, $user ) {
4211 global $wgContLang, $wgLocaltimezone;
4213 # Note: This is the timestamp saved as hardcoded wikitext to
4214 # the database, we use $wgContLang here in order to give
4215 # everyone the same signature and use the default one rather
4216 # than the one selected in each user's preferences.
4217 # (see also bug 12815)
4218 $ts = $this->mOptions
->getTimestamp();
4219 if ( isset( $wgLocaltimezone ) ) {
4220 $tz = $wgLocaltimezone;
4222 $tz = date_default_timezone_get();
4225 $unixts = wfTimestamp( TS_UNIX
, $ts );
4226 $oldtz = date_default_timezone_get();
4227 date_default_timezone_set( $tz );
4228 $ts = date( 'YmdHis', $unixts );
4229 $tzMsg = date( 'T', $unixts ); # might vary on DST changeover!
4231 # Allow translation of timezones through wiki. date() can return
4232 # whatever crap the system uses, localised or not, so we cannot
4233 # ship premade translations.
4234 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4235 $msg = wfMessage( $key )->inContentLanguage();
4236 if ( $msg->exists() ) {
4237 $tzMsg = $msg->text();
4240 date_default_timezone_set( $oldtz );
4242 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4244 # Variable replacement
4245 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4246 $text = $this->replaceVariables( $text );
4248 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4249 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4252 $sigText = $this->getUserSig( $user );
4253 $text = strtr( $text, array(
4255 '~~~~' => "$sigText $d",
4259 # Context links: [[|name]] and [[name (context)|]]
4260 global $wgLegalTitleChars;
4261 $tc = "[$wgLegalTitleChars]";
4262 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4264 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
4265 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(($tc+))\\|]]/"; # [[ns:page(context)|]]
4266 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4267 $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
4269 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4270 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4271 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4272 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4274 $t = $this->mTitle
->getText();
4276 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4277 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4278 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4279 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4281 # if there's no context, don't bother duplicating the title
4282 $text = preg_replace( $p2, '[[\\1]]', $text );
4285 # Trim trailing whitespace
4286 $text = rtrim( $text );
4292 * Fetch the user's signature text, if any, and normalize to
4293 * validated, ready-to-insert wikitext.
4294 * If you have pre-fetched the nickname or the fancySig option, you can
4295 * specify them here to save a database query.
4296 * Do not reuse this parser instance after calling getUserSig(),
4297 * as it may have changed if it's the $wgParser.
4300 * @param $nickname String: nickname to use or false to use user's default nickname
4301 * @param $fancySig Boolean: whether the nicknname is the complete signature
4302 * or null to use default value
4305 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4306 global $wgMaxSigChars;
4308 $username = $user->getName();
4310 # If not given, retrieve from the user object.
4311 if ( $nickname === false )
4312 $nickname = $user->getOption( 'nickname' );
4314 if ( is_null( $fancySig ) ) {
4315 $fancySig = $user->getBoolOption( 'fancysig' );
4318 $nickname = $nickname == null ?
$username : $nickname;
4320 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4321 $nickname = $username;
4322 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4323 } elseif ( $fancySig !== false ) {
4324 # Sig. might contain markup; validate this
4325 if ( $this->validateSig( $nickname ) !== false ) {
4326 # Validated; clean up (if needed) and return it
4327 return $this->cleanSig( $nickname, true );
4329 # Failed to validate; fall back to the default
4330 $nickname = $username;
4331 wfDebug( __METHOD__
.": $username has bad XML tags in signature.\n" );
4335 # Make sure nickname doesnt get a sig in a sig
4336 $nickname = $this->cleanSigInSig( $nickname );
4338 # If we're still here, make it a link to the user page
4339 $userText = wfEscapeWikiText( $username );
4340 $nickText = wfEscapeWikiText( $nickname );
4341 if ( $user->isAnon() ) {
4342 return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4344 return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4349 * Check that the user's signature contains no bad XML
4351 * @param $text String
4352 * @return mixed An expanded string, or false if invalid.
4354 function validateSig( $text ) {
4355 return( Xml
::isWellFormedXmlFragment( $text ) ?
$text : false );
4359 * Clean up signature text
4361 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4362 * 2) Substitute all transclusions
4364 * @param $text String
4365 * @param $parsing Whether we're cleaning (preferences save) or parsing
4366 * @return String: signature text
4368 function cleanSig( $text, $parsing = false ) {
4371 $this->mOptions
= new ParserOptions
;
4372 $this->clearState();
4373 $this->setTitle( $wgTitle );
4374 $this->setOutputType
= self
::OT_PREPROCESS
;
4377 # Option to disable this feature
4378 if ( !$this->mOptions
->getCleanSignatures() ) {
4382 # FIXME: regex doesn't respect extension tags or nowiki
4383 # => Move this logic to braceSubstitution()
4384 $substWord = MagicWord
::get( 'subst' );
4385 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4386 $substText = '{{' . $substWord->getSynonym( 0 );
4388 $text = preg_replace( $substRegex, $substText, $text );
4389 $text = $this->cleanSigInSig( $text );
4390 $dom = $this->preprocessToDom( $text );
4391 $frame = $this->getPreprocessor()->newFrame();
4392 $text = $frame->expand( $dom );
4395 $text = $this->mStripState
->unstripBoth( $text );
4402 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4404 * @param $text String
4405 * @return String: signature text with /~{3,5}/ removed
4407 function cleanSigInSig( $text ) {
4408 $text = preg_replace( '/~{3,5}/', '', $text );
4413 * Set up some variables which are usually set up in parse()
4414 * so that an external function can call some class members with confidence
4416 public function startExternalParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4417 $this->startParse( $title, $options, $outputType, $clearState );
4420 private function startParse( Title
$title = null, ParserOptions
$options, $outputType, $clearState = true ) {
4421 $this->setTitle( $title );
4422 $this->mOptions
= $options;
4423 $this->setOutputType( $outputType );
4424 if ( $clearState ) {
4425 $this->clearState();
4430 * Wrapper for preprocess()
4432 * @param $text String: the text to preprocess
4433 * @param $options ParserOptions: options
4434 * @param $title Title object or null to use $wgTitle
4437 public function transformMsg( $text, $options, $title = null ) {
4438 static $executing = false;
4440 # Guard against infinite recursion
4446 wfProfileIn( __METHOD__
);
4452 # It's not uncommon having a null $wgTitle in scripts. See r80898
4453 # Create a ghost title in such case
4454 $title = Title
::newFromText( 'Dwimmerlaik' );
4456 $text = $this->preprocess( $text, $title, $options );
4459 wfProfileOut( __METHOD__
);
4464 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4465 * The callback should have the following form:
4466 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4468 * Transform and return $text. Use $parser for any required context, e.g. use
4469 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4471 * Hooks may return extended information by returning an array, of which the
4472 * first numbered element (index 0) must be the return string, and all other
4473 * entries are extracted into local variables within an internal function
4474 * in the Parser class.
4476 * This interface (introduced r61913) appears to be undocumented, but
4477 * 'markerName' is used by some core tag hooks to override which strip
4478 * array their results are placed in. **Use great caution if attempting
4479 * this interface, as it is not documented and injudicious use could smash
4480 * private variables.**
4482 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4483 * @param $callback Mixed: the callback function (and object) to use for the tag
4484 * @return The old value of the mTagHooks array associated with the hook
4486 public function setHook( $tag, $callback ) {
4487 $tag = strtolower( $tag );
4488 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4489 $oldVal = isset( $this->mTagHooks
[$tag] ) ?
$this->mTagHooks
[$tag] : null;
4490 $this->mTagHooks
[$tag] = $callback;
4491 if ( !in_array( $tag, $this->mStripList
) ) {
4492 $this->mStripList
[] = $tag;
4499 * As setHook(), but letting the contents be parsed.
4501 * Transparent tag hooks are like regular XML-style tag hooks, except they
4502 * operate late in the transformation sequence, on HTML instead of wikitext.
4504 * This is probably obsoleted by things dealing with parser frames?
4505 * The only extension currently using it is geoserver.
4508 * @todo better document or deprecate this
4510 * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4511 * @param $callback Mixed: the callback function (and object) to use for the tag
4512 * @return The old value of the mTagHooks array associated with the hook
4514 function setTransparentTagHook( $tag, $callback ) {
4515 $tag = strtolower( $tag );
4516 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4517 $oldVal = isset( $this->mTransparentTagHooks
[$tag] ) ?
$this->mTransparentTagHooks
[$tag] : null;
4518 $this->mTransparentTagHooks
[$tag] = $callback;
4524 * Remove all tag hooks
4526 function clearTagHooks() {
4527 $this->mTagHooks
= array();
4528 $this->mStripList
= $this->mDefaultStripList
;
4532 * Create a function, e.g. {{sum:1|2|3}}
4533 * The callback function should have the form:
4534 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4536 * Or with SFH_OBJECT_ARGS:
4537 * function myParserFunction( $parser, $frame, $args ) { ... }
4539 * The callback may either return the text result of the function, or an array with the text
4540 * in element 0, and a number of flags in the other elements. The names of the flags are
4541 * specified in the keys. Valid flags are:
4542 * found The text returned is valid, stop processing the template. This
4544 * nowiki Wiki markup in the return value should be escaped
4545 * isHTML The returned text is HTML, armour it against wikitext transformation
4547 * @param $id String: The magic word ID
4548 * @param $callback Mixed: the callback function (and object) to use
4549 * @param $flags Integer: a combination of the following flags:
4550 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4552 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4553 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4554 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4555 * the arguments, and to control the way they are expanded.
4557 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4558 * arguments, for instance:
4559 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4561 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4562 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4563 * working if/when this is changed.
4565 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4568 * Please read the documentation in includes/parser/Preprocessor.php for more information
4569 * about the methods available in PPFrame and PPNode.
4571 * @return The old callback function for this name, if any
4573 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4576 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4577 $this->mFunctionHooks
[$id] = array( $callback, $flags );
4579 # Add to function cache
4580 $mw = MagicWord
::get( $id );
4582 throw new MWException( __METHOD__
.'() expecting a magic word identifier.' );
4584 $synonyms = $mw->getSynonyms();
4585 $sensitive = intval( $mw->isCaseSensitive() );
4587 foreach ( $synonyms as $syn ) {
4589 if ( !$sensitive ) {
4590 $syn = $wgContLang->lc( $syn );
4593 if ( !( $flags & SFH_NO_HASH
) ) {
4596 # Remove trailing colon
4597 if ( substr( $syn, -1, 1 ) === ':' ) {
4598 $syn = substr( $syn, 0, -1 );
4600 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4606 * Get all registered function hook identifiers
4610 function getFunctionHooks() {
4611 return array_keys( $this->mFunctionHooks
);
4615 * Create a tag function, e.g. <test>some stuff</test>.
4616 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4617 * Unlike parser functions, their content is not preprocessed.
4619 function setFunctionTagHook( $tag, $callback, $flags ) {
4620 $tag = strtolower( $tag );
4621 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4622 $old = isset( $this->mFunctionTagHooks
[$tag] ) ?
4623 $this->mFunctionTagHooks
[$tag] : null;
4624 $this->mFunctionTagHooks
[$tag] = array( $callback, $flags );
4626 if ( !in_array( $tag, $this->mStripList
) ) {
4627 $this->mStripList
[] = $tag;
4634 * FIXME: update documentation. makeLinkObj() is deprecated.
4635 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4636 * Placeholders created in Skin::makeLinkObj()
4637 * Returns an array of link CSS classes, indexed by PDBK.
4639 function replaceLinkHolders( &$text, $options = 0 ) {
4640 return $this->mLinkHolders
->replace( $text );
4644 * Replace <!--LINK--> link placeholders with plain text of links
4645 * (not HTML-formatted).
4647 * @param $text String
4650 function replaceLinkHoldersText( $text ) {
4651 return $this->mLinkHolders
->replaceText( $text );
4655 * Renders an image gallery from a text with one line per image.
4656 * text labels may be given by using |-style alternative text. E.g.
4657 * Image:one.jpg|The number "1"
4658 * Image:tree.jpg|A tree
4659 * given as text will return the HTML of a gallery with two images,
4660 * labeled 'The number "1"' and
4663 * @param string $text
4664 * @param array $param
4665 * @return string HTML
4667 function renderImageGallery( $text, $params ) {
4668 $ig = new ImageGallery();
4669 $ig->setContextTitle( $this->mTitle
);
4670 $ig->setShowBytes( false );
4671 $ig->setShowFilename( false );
4672 $ig->setParser( $this );
4673 $ig->setHideBadImages();
4674 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'table' ) );
4676 if ( isset( $params['showfilename'] ) ) {
4677 $ig->setShowFilename( true );
4679 $ig->setShowFilename( false );
4681 if ( isset( $params['caption'] ) ) {
4682 $caption = $params['caption'];
4683 $caption = htmlspecialchars( $caption );
4684 $caption = $this->replaceInternalLinks( $caption );
4685 $ig->setCaptionHtml( $caption );
4687 if ( isset( $params['perrow'] ) ) {
4688 $ig->setPerRow( $params['perrow'] );
4690 if ( isset( $params['widths'] ) ) {
4691 $ig->setWidths( $params['widths'] );
4693 if ( isset( $params['heights'] ) ) {
4694 $ig->setHeights( $params['heights'] );
4697 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4699 $lines = StringUtils
::explode( "\n", $text );
4700 foreach ( $lines as $line ) {
4701 # match lines like these:
4702 # Image:someimage.jpg|This is some image
4704 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4706 if ( count( $matches ) == 0 ) {
4710 if ( strpos( $matches[0], '%' ) !== false ) {
4711 $matches[1] = rawurldecode( $matches[1] );
4713 $title = Title
::newFromText( $matches[1], NS_FILE
);
4714 if ( is_null( $title ) ) {
4715 # Bogus title. Ignore these so we don't bomb out later.
4721 if ( isset( $matches[3] ) ) {
4722 // look for an |alt= definition while trying not to break existing
4723 // captions with multiple pipes (|) in it, until a more sensible grammar
4724 // is defined for images in galleries
4726 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4727 $altmatches = StringUtils
::explode('|', $matches[3]);
4728 $magicWordAlt = MagicWord
::get( 'img_alt' );
4730 foreach ( $altmatches as $altmatch ) {
4731 $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4733 $alt = $this->stripAltText( $match, false );
4736 // concatenate all other pipes
4737 $label .= '|' . $altmatch;
4740 // remove the first pipe
4741 $label = substr( $label, 1 );
4744 $ig->add( $title, $label, $alt );
4746 return $ig->toHTML();
4749 function getImageParams( $handler ) {
4751 $handlerClass = get_class( $handler );
4755 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
4756 # Initialise static lists
4757 static $internalParamNames = array(
4758 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4759 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4760 'bottom', 'text-bottom' ),
4761 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4762 'upright', 'border', 'link', 'alt' ),
4764 static $internalParamMap;
4765 if ( !$internalParamMap ) {
4766 $internalParamMap = array();
4767 foreach ( $internalParamNames as $type => $names ) {
4768 foreach ( $names as $name ) {
4769 $magicName = str_replace( '-', '_', "img_$name" );
4770 $internalParamMap[$magicName] = array( $type, $name );
4775 # Add handler params
4776 $paramMap = $internalParamMap;
4778 $handlerParamMap = $handler->getParamMap();
4779 foreach ( $handlerParamMap as $magic => $paramName ) {
4780 $paramMap[$magic] = array( 'handler', $paramName );
4783 $this->mImageParams
[$handlerClass] = $paramMap;
4784 $this->mImageParamsMagicArray
[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4786 return array( $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] );
4790 * Parse image options text and use it to make an image
4792 * @param $title Title
4793 * @param $options String
4794 * @param $holders LinkHolderArray
4795 * @return string HTML
4797 function makeImage( $title, $options, $holders = false ) {
4798 # Check if the options text is of the form "options|alt text"
4800 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4801 # * left no resizing, just left align. label is used for alt= only
4802 # * right same, but right aligned
4803 # * none same, but not aligned
4804 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4805 # * center center the image
4806 # * frame Keep original image size, no magnify-button.
4807 # * framed Same as "frame"
4808 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4809 # * upright reduce width for upright images, rounded to full __0 px
4810 # * border draw a 1px border around the image
4811 # * alt Text for HTML alt attribute (defaults to empty)
4812 # * link Set the target of the image link. Can be external, interwiki, or local
4813 # vertical-align values (no % or length right now):
4823 $parts = StringUtils
::explode( "|", $options );
4825 # Give extensions a chance to select the file revision for us
4826 $time = $sha1 = $descQuery = false;
4827 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4828 array( $this, $title, &$time, &$sha1, &$descQuery ) );
4829 # Fetch and register the file (file title may be different via hooks)
4830 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4833 $handler = $file ?
$file->getHandler() : false;
4835 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4838 $this->addTrackingCategory( 'broken-file-category' );
4841 # Process the input parameters
4843 $params = array( 'frame' => array(), 'handler' => array(),
4844 'horizAlign' => array(), 'vertAlign' => array() );
4845 foreach ( $parts as $part ) {
4846 $part = trim( $part );
4847 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4849 if ( isset( $paramMap[$magicName] ) ) {
4850 list( $type, $paramName ) = $paramMap[$magicName];
4852 # Special case; width and height come in one variable together
4853 if ( $type === 'handler' && $paramName === 'width' ) {
4855 # (bug 13500) In both cases (width/height and width only),
4856 # permit trailing "px" for backward compatibility.
4857 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4858 $width = intval( $m[1] );
4859 $height = intval( $m[2] );
4860 if ( $handler->validateParam( 'width', $width ) ) {
4861 $params[$type]['width'] = $width;
4864 if ( $handler->validateParam( 'height', $height ) ) {
4865 $params[$type]['height'] = $height;
4868 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4869 $width = intval( $value );
4870 if ( $handler->validateParam( 'width', $width ) ) {
4871 $params[$type]['width'] = $width;
4874 } # else no validation -- bug 13436
4876 if ( $type === 'handler' ) {
4877 # Validate handler parameter
4878 $validated = $handler->validateParam( $paramName, $value );
4880 # Validate internal parameters
4881 switch( $paramName ) {
4884 # @todo Fixme: possibly check validity here for
4885 # manualthumb? downstream behavior seems odd with
4886 # missing manual thumbs.
4888 $value = $this->stripAltText( $value, $holders );
4891 $chars = self
::EXT_LINK_URL_CLASS
;
4892 $prots = $this->mUrlProtocols
;
4893 if ( $value === '' ) {
4894 $paramName = 'no-link';
4897 } elseif ( preg_match( "/^$prots/", $value ) ) {
4898 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4899 $paramName = 'link-url';
4900 $this->mOutput
->addExternalLink( $value );
4901 if ( $this->mOptions
->getExternalLinkTarget() ) {
4902 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
4907 $linkTitle = Title
::newFromText( $value );
4909 $paramName = 'link-title';
4910 $value = $linkTitle;
4911 $this->mOutput
->addLink( $linkTitle );
4917 # Most other things appear to be empty or numeric...
4918 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
4923 $params[$type][$paramName] = $value;
4927 if ( !$validated ) {
4932 # Process alignment parameters
4933 if ( $params['horizAlign'] ) {
4934 $params['frame']['align'] = key( $params['horizAlign'] );
4936 if ( $params['vertAlign'] ) {
4937 $params['frame']['valign'] = key( $params['vertAlign'] );
4940 $params['frame']['caption'] = $caption;
4942 # Will the image be presented in a frame, with the caption below?
4943 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4944 isset( $params['frame']['framed'] ) ||
4945 isset( $params['frame']['thumbnail'] ) ||
4946 isset( $params['frame']['manualthumb'] );
4948 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
4949 # came to also set the caption, ordinary text after the image -- which
4950 # makes no sense, because that just repeats the text multiple times in
4951 # screen readers. It *also* came to set the title attribute.
4953 # Now that we have an alt attribute, we should not set the alt text to
4954 # equal the caption: that's worse than useless, it just repeats the
4955 # text. This is the framed/thumbnail case. If there's no caption, we
4956 # use the unnamed parameter for alt text as well, just for the time be-
4957 # ing, if the unnamed param is set and the alt param is not.
4959 # For the future, we need to figure out if we want to tweak this more,
4960 # e.g., introducing a title= parameter for the title; ignoring the un-
4961 # named parameter entirely for images without a caption; adding an ex-
4962 # plicit caption= parameter and preserving the old magic unnamed para-
4964 if ( $imageIsFramed ) { # Framed image
4965 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4966 # No caption or alt text, add the filename as the alt text so
4967 # that screen readers at least get some description of the image
4968 $params['frame']['alt'] = $title->getText();
4970 # Do not set $params['frame']['title'] because tooltips don't make sense
4972 } else { # Inline image
4973 if ( !isset( $params['frame']['alt'] ) ) {
4974 # No alt text, use the "caption" for the alt text
4975 if ( $caption !== '') {
4976 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4978 # No caption, fall back to using the filename for the
4980 $params['frame']['alt'] = $title->getText();
4983 # Use the "caption" for the tooltip text
4984 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4987 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4989 # Linker does the rest
4990 $ret = Linker
::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
4991 $time, $descQuery, $this->mOptions
->getThumbSize() );
4993 # Give the handler a chance to modify the parser object
4995 $handler->parserTransformHook( $this, $file );
5001 protected function stripAltText( $caption, $holders ) {
5002 # Strip bad stuff out of the title (tooltip). We can't just use
5003 # replaceLinkHoldersText() here, because if this function is called
5004 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5006 $tooltip = $holders->replaceText( $caption );
5008 $tooltip = $this->replaceLinkHoldersText( $caption );
5011 # make sure there are no placeholders in thumbnail attributes
5012 # that are later expanded to html- so expand them now and
5014 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5015 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5021 * Set a flag in the output object indicating that the content is dynamic and
5022 * shouldn't be cached.
5024 function disableCache() {
5025 wfDebug( "Parser output marked as uncacheable.\n" );
5026 $this->mOutput
->setCacheTime( -1 ); // old style, for compatibility
5027 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5031 * Callback from the Sanitizer for expanding items found in HTML attribute
5032 * values, so they can be safely tested and escaped.
5034 * @param $text String
5035 * @param $frame PPFrame
5038 function attributeStripCallback( &$text, $frame = false ) {
5039 $text = $this->replaceVariables( $text, $frame );
5040 $text = $this->mStripState
->unstripBoth( $text );
5049 function getTags() {
5050 return array_merge( array_keys( $this->mTransparentTagHooks
), array_keys( $this->mTagHooks
) );
5054 * Replace transparent tags in $text with the values given by the callbacks.
5056 * Transparent tag hooks are like regular XML-style tag hooks, except they
5057 * operate late in the transformation sequence, on HTML instead of wikitext.
5059 function replaceTransparentTags( $text ) {
5061 $elements = array_keys( $this->mTransparentTagHooks
);
5062 $text = self
::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix
);
5064 foreach ( $matches as $marker => $data ) {
5065 list( $element, $content, $params, $tag ) = $data;
5066 $tagName = strtolower( $element );
5067 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5068 $output = call_user_func_array( $this->mTransparentTagHooks
[$tagName], array( $content, $params, $this ) );
5072 $this->mStripState
->addGeneral( $marker, $output );
5078 * Break wikitext input into sections, and either pull or replace
5079 * some particular section's text.
5081 * External callers should use the getSection and replaceSection methods.
5083 * @param $text String: Page wikitext
5084 * @param $section String: a section identifier string of the form:
5085 * <flag1> - <flag2> - ... - <section number>
5087 * Currently the only recognised flag is "T", which means the target section number
5088 * was derived during a template inclusion parse, in other words this is a template
5089 * section edit link. If no flags are given, it was an ordinary section edit link.
5090 * This flag is required to avoid a section numbering mismatch when a section is
5091 * enclosed by <includeonly> (bug 6563).
5093 * The section number 0 pulls the text before the first heading; other numbers will
5094 * pull the given section along with its lower-level subsections. If the section is
5095 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5097 * @param $mode String: one of "get" or "replace"
5098 * @param $newText String: replacement text for section data.
5099 * @return String: for "get", the extracted section text.
5100 * for "replace", the whole page with the section replaced.
5102 private function extractSections( $text, $section, $mode, $newText='' ) {
5103 global $wgTitle; # not generally used but removes an ugly failure mode
5104 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5106 $frame = $this->getPreprocessor()->newFrame();
5108 # Process section extraction flags
5110 $sectionParts = explode( '-', $section );
5111 $sectionIndex = array_pop( $sectionParts );
5112 foreach ( $sectionParts as $part ) {
5113 if ( $part === 'T' ) {
5114 $flags |
= self
::PTD_FOR_INCLUSION
;
5117 # Preprocess the text
5118 $root = $this->preprocessToDom( $text, $flags );
5120 # <h> nodes indicate section breaks
5121 # They can only occur at the top level, so we can find them by iterating the root's children
5122 $node = $root->getFirstChild();
5124 # Find the target section
5125 if ( $sectionIndex == 0 ) {
5126 # Section zero doesn't nest, level=big
5127 $targetLevel = 1000;
5129 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5134 if ( $node->getName() === 'h' ) {
5135 $bits = $node->splitHeading();
5136 if ( $bits['i'] == $sectionIndex ) {
5137 $targetLevel = $bits['level'];
5141 if ( $mode === 'replace' ) {
5142 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5144 $node = $node->getNextSibling();
5150 if ( $mode === 'get' ) {
5157 # Find the end of the section, including nested sections
5159 if ( $node->getName() === 'h' ) {
5160 $bits = $node->splitHeading();
5161 $curLevel = $bits['level'];
5162 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5166 if ( $mode === 'get' ) {
5167 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5169 $node = $node->getNextSibling();
5172 # Write out the remainder (in replace mode only)
5173 if ( $mode === 'replace' ) {
5174 # Output the replacement text
5175 # Add two newlines on -- trailing whitespace in $newText is conventionally
5176 # stripped by the editor, so we need both newlines to restore the paragraph gap
5177 # Only add trailing whitespace if there is newText
5178 if ( $newText != "" ) {
5179 $outText .= $newText . "\n\n";
5183 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5184 $node = $node->getNextSibling();
5188 if ( is_string( $outText ) ) {
5189 # Re-insert stripped tags
5190 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5197 * This function returns the text of a section, specified by a number ($section).
5198 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5199 * the first section before any such heading (section 0).
5201 * If a section contains subsections, these are also returned.
5203 * @param $text String: text to look in
5204 * @param $section String: section identifier
5205 * @param $deftext String: default to return if section is not found
5206 * @return string text of the requested section
5208 public function getSection( $text, $section, $deftext='' ) {
5209 return $this->extractSections( $text, $section, "get", $deftext );
5213 * This function returns $oldtext after the content of the section
5214 * specified by $section has been replaced with $text.
5216 * @param $oldtext String: former text of the article
5217 * @param $section Numeric: section identifier
5218 * @param $text String: replacing text
5219 * @return String: modified text
5221 public function replaceSection( $oldtext, $section, $text ) {
5222 return $this->extractSections( $oldtext, $section, "replace", $text );
5226 * Get the ID of the revision we are parsing
5228 * @return Mixed: integer or null
5230 function getRevisionId() {
5231 return $this->mRevisionId
;
5235 * Get the revision object for $this->mRevisionId
5237 * @return Revision|null either a Revision object or null
5239 protected function getRevisionObject() {
5240 if ( !is_null( $this->mRevisionObject
) ) {
5241 return $this->mRevisionObject
;
5243 if ( is_null( $this->mRevisionId
) ) {
5247 $this->mRevisionObject
= Revision
::newFromId( $this->mRevisionId
);
5248 return $this->mRevisionObject
;
5252 * Get the timestamp associated with the current revision, adjusted for
5253 * the default server-local timestamp
5255 function getRevisionTimestamp() {
5256 if ( is_null( $this->mRevisionTimestamp
) ) {
5257 wfProfileIn( __METHOD__
);
5259 $revObject = $this->getRevisionObject();
5260 $timestamp = $revObject ?
$revObject->getTimestamp() : false;
5262 if( $timestamp !== false ) {
5265 # The cryptic '' timezone parameter tells to use the site-default
5266 # timezone offset instead of the user settings.
5268 # Since this value will be saved into the parser cache, served
5269 # to other users, and potentially even used inside links and such,
5270 # it needs to be consistent for all visitors.
5271 $this->mRevisionTimestamp
= $wgContLang->userAdjust( $timestamp, '' );
5274 wfProfileOut( __METHOD__
);
5276 return $this->mRevisionTimestamp
;
5280 * Get the name of the user that edited the last revision
5282 * @return String: user name
5284 function getRevisionUser() {
5285 if( is_null( $this->mRevisionUser
) ) {
5286 $revObject = $this->getRevisionObject();
5288 # if this template is subst: the revision id will be blank,
5289 # so just use the current user's name
5291 $this->mRevisionUser
= $revObject->getUserText();
5292 } elseif( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5293 $this->mRevisionUser
= $this->getUser()->getName();
5296 return $this->mRevisionUser
;
5300 * Mutator for $mDefaultSort
5302 * @param $sort New value
5304 public function setDefaultSort( $sort ) {
5305 $this->mDefaultSort
= $sort;
5306 $this->mOutput
->setProperty( 'defaultsort', $sort );
5310 * Accessor for $mDefaultSort
5311 * Will use the empty string if none is set.
5313 * This value is treated as a prefix, so the
5314 * empty string is equivalent to sorting by
5319 public function getDefaultSort() {
5320 if ( $this->mDefaultSort
!== false ) {
5321 return $this->mDefaultSort
;
5328 * Accessor for $mDefaultSort
5329 * Unlike getDefaultSort(), will return false if none is set
5331 * @return string or false
5333 public function getCustomDefaultSort() {
5334 return $this->mDefaultSort
;
5338 * Try to guess the section anchor name based on a wikitext fragment
5339 * presumably extracted from a heading, for example "Header" from
5342 public function guessSectionNameFromWikiText( $text ) {
5343 # Strip out wikitext links(they break the anchor)
5344 $text = $this->stripSectionName( $text );
5345 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5346 return '#' . Sanitizer
::escapeId( $text, 'noninitial' );
5350 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5351 * instead. For use in redirects, since IE6 interprets Redirect: headers
5352 * as something other than UTF-8 (apparently?), resulting in breakage.
5354 * @param $text String: The section name
5355 * @return string An anchor
5357 public function guessLegacySectionNameFromWikiText( $text ) {
5358 # Strip out wikitext links(they break the anchor)
5359 $text = $this->stripSectionName( $text );
5360 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5361 return '#' . Sanitizer
::escapeId( $text, array( 'noninitial', 'legacy' ) );
5365 * Strips a text string of wikitext for use in a section anchor
5367 * Accepts a text string and then removes all wikitext from the
5368 * string and leaves only the resultant text (i.e. the result of
5369 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5370 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5371 * to create valid section anchors by mimicing the output of the
5372 * parser when headings are parsed.
5374 * @param $text String: text string to be stripped of wikitext
5375 * for use in a Section anchor
5376 * @return Filtered text string
5378 public function stripSectionName( $text ) {
5379 # Strip internal link markup
5380 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5381 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5383 # Strip external link markup (FIXME: Not Tolerant to blank link text
5384 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5385 # on how many empty links there are on the page - need to figure that out.
5386 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5388 # Parse wikitext quotes (italics & bold)
5389 $text = $this->doQuotes( $text );
5392 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
5397 * strip/replaceVariables/unstrip for preprocessor regression testing
5401 function testSrvus( $text, $title, ParserOptions
$options, $outputType = self
::OT_HTML
) {
5402 if ( !$title instanceof Title
) {
5403 $title = Title
::newFromText( $title );
5405 $this->startParse( $title, $options, $outputType, true );
5407 $text = $this->replaceVariables( $text );
5408 $text = $this->mStripState
->unstripBoth( $text );
5409 $text = Sanitizer
::removeHTMLtags( $text );
5413 function testPst( $text, $title, $options ) {
5415 if ( !$title instanceof Title
) {
5416 $title = Title
::newFromText( $title );
5418 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5421 function testPreprocess( $text, $title, $options ) {
5422 if ( !$title instanceof Title
) {
5423 $title = Title
::newFromText( $title );
5425 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
5429 * Call a callback function on all regions of the given text that are not
5430 * inside strip markers, and replace those regions with the return value
5431 * of the callback. For example, with input:
5435 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5436 * two strings will be replaced with the value returned by the callback in
5441 function markerSkipCallback( $s, $callback ) {
5444 while ( $i < strlen( $s ) ) {
5445 $markerStart = strpos( $s, $this->mUniqPrefix
, $i );
5446 if ( $markerStart === false ) {
5447 $out .= call_user_func( $callback, substr( $s, $i ) );
5450 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5451 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
5452 if ( $markerEnd === false ) {
5453 $out .= substr( $s, $markerStart );
5456 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
5457 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5466 * Save the parser state required to convert the given half-parsed text to
5467 * HTML. "Half-parsed" in this context means the output of
5468 * recursiveTagParse() or internalParse(). This output has strip markers
5469 * from replaceVariables (extensionSubstitution() etc.), and link
5470 * placeholders from replaceLinkHolders().
5472 * Returns an array which can be serialized and stored persistently. This
5473 * array can later be loaded into another parser instance with
5474 * unserializeHalfParsedText(). The text can then be safely incorporated into
5475 * the return value of a parser hook.
5479 function serializeHalfParsedText( $text ) {
5480 wfProfileIn( __METHOD__
);
5483 'version' => self
::HALF_PARSED_VERSION
,
5484 'stripState' => $this->mStripState
->getSubState( $text ),
5485 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
5487 wfProfileOut( __METHOD__
);
5492 * Load the parser state given in the $data array, which is assumed to
5493 * have been generated by serializeHalfParsedText(). The text contents is
5494 * extracted from the array, and its markers are transformed into markers
5495 * appropriate for the current Parser instance. This transformed text is
5496 * returned, and can be safely included in the return value of a parser
5499 * If the $data array has been stored persistently, the caller should first
5500 * check whether it is still valid, by calling isValidHalfParsedText().
5502 * @param $data Serialized data
5505 function unserializeHalfParsedText( $data ) {
5506 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
5507 throw new MWException( __METHOD__
.': invalid version' );
5510 # First, extract the strip state.
5511 $texts = array( $data['text'] );
5512 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
5514 # Now renumber links
5515 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
5517 # Should be good to go.
5522 * Returns true if the given array, presumed to be generated by
5523 * serializeHalfParsedText(), is compatible with the current version of the
5526 * @param $data Array
5530 function isValidHalfParsedText( $data ) {
5531 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;