3 * PHP parser that converts wiki markup to HTML.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
23 use MediaWiki\Linker\LinkRenderer
;
24 use MediaWiki\Linker\LinkRendererFactory
;
25 use MediaWiki\MediaWikiServices
;
26 use MediaWiki\Special\SpecialPageFactory
;
27 use Wikimedia\ScopedCallback
;
30 * @defgroup Parser Parser
34 * PHP Parser - Processes wiki markup (which uses a more user-friendly
35 * syntax, such as "[[link]]" for making links), and provides a one-way
36 * transformation of that wiki markup it into (X)HTML output / markup
37 * (which in turn the browser understands, and can display).
39 * There are seven main entry points into the Parser class:
42 * produces HTML output
43 * - Parser::preSaveTransform()
44 * produces altered wiki markup
45 * - Parser::preprocess()
46 * removes HTML comments and expands templates
47 * - Parser::cleanSig() and Parser::cleanSigInSig()
48 * cleans a signature before saving it to preferences
49 * - Parser::getSection()
50 * return the content of a section from an article for section editing
51 * - Parser::replaceSection()
52 * replaces a section by number inside an article
53 * - Parser::getPreloadText()
54 * removes <noinclude> sections and <includeonly> tags
56 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
59 * $wgNamespacesWithSubpages
61 * @par Settings only within ParserOptions:
62 * $wgAllowExternalImages
63 * $wgAllowSpecialInclusion
71 * Update this version number when the ParserOutput format
72 * changes in an incompatible way, so the parser cache
73 * can automatically discard old data.
75 const VERSION
= '1.6.4';
78 * Update this version number when the output of serialiseHalfParsedText()
79 * changes in an incompatible way
81 const HALF_PARSED_VERSION
= 2;
83 # Flags for Parser::setFunctionHook
84 const SFH_NO_HASH
= 1;
85 const SFH_OBJECT_ARGS
= 2;
87 # Constants needed for external link processing
88 # Everything except bracket, space, or control characters
89 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
90 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
91 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
92 # uses to replace invalid HTML characters.
93 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
94 # Simplified expression to match an IPv4 or IPv6 address, or
95 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
96 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
97 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
98 // phpcs:ignore Generic.Files.LineLength
99 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
100 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102 # Regular expression for a non-newline space
103 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105 # Flags for preprocessToDom
106 const PTD_FOR_INCLUSION
= 1;
108 # Allowed values for $this->mOutputType
109 # Parameter to startExternalParse().
110 const OT_HTML
= 1; # like parse()
111 const OT_WIKI
= 2; # like preSaveTransform()
112 const OT_PREPROCESS
= 3; # like preprocess()
114 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
117 * @var string Prefix and suffix for temporary replacement strings
118 * for the multipass parser.
120 * \x7f should never appear in input as it's disallowed in XML.
121 * Using it at the front also gives us a little extra robustness
122 * since it shouldn't match when butted up against identifier-like
125 * Must not consist of all title characters, or else it will change
126 * the behavior of <nowiki> in a link.
128 * Must have a character that needs escaping in attributes, otherwise
129 * someone could put a strip marker in an attribute, to get around
130 * escaping quote marks, and break out of the attribute. Thus we add
133 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
134 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
136 # Markers used for wrapping the table of contents
137 const TOC_START
= '<mw:toc>';
138 const TOC_END
= '</mw:toc>';
140 /** @var int Assume that no output will later be saved this many seconds after parsing */
144 public $mTagHooks = [];
145 public $mTransparentTagHooks = [];
146 public $mFunctionHooks = [];
147 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
148 public $mFunctionTagHooks = [];
149 public $mStripList = [];
150 public $mDefaultStripList = [];
151 public $mVarCache = [];
152 public $mImageParams = [];
153 public $mImageParamsMagicArray = [];
154 public $mMarkerIndex = 0;
156 * @var bool Whether firstCallInit still needs to be called
158 public $mFirstCall = true;
160 # Initialised by initialiseVariables()
163 * @var MagicWordArray
168 * @var MagicWordArray
171 # Initialised in constructor
172 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
174 # Initialized in getPreprocessor()
175 /** @var Preprocessor */
176 public $mPreprocessor;
178 # Cleared with clearState():
190 public $mIncludeCount;
192 * @var LinkHolderArray
194 public $mLinkHolders;
197 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
198 public $mDefaultSort;
199 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
200 public $mExpensiveFunctionCount; # number of expensive parser function calls
201 public $mShowToc, $mForceTocPosition;
206 public $mUser; # User object; only used when doing pre-save transform
209 # These are variables reset at least once per parse regardless of $clearState
219 public $mTitle; # Title context, used for self-link rendering and similar things
220 public $mOutputType; # Output type, one of the OT_xxx constants
221 public $ot; # Shortcut alias, see setOutputType()
222 public $mRevisionObject; # The revision object of the specified revision ID
223 public $mRevisionId; # ID to display in {{REVISIONID}} tags
224 public $mRevisionTimestamp; # The timestamp of the specified revision ID
225 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
226 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
227 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
228 public $mInputSize = false; # For {{PAGESIZE}} on current page.
231 * @var string Deprecated accessor for the strip marker prefix.
232 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
234 public $mUniqPrefix = self
::MARKER_PREFIX
;
237 * @var array Array with the language name of each language link (i.e. the
238 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
239 * duplicate language links to the ParserOutput.
241 public $mLangLinkLanguages;
244 * @var MapCacheLRU|null
247 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
249 public $currentRevisionCache;
252 * @var bool|string Recursive call protection.
253 * This variable should be treated as if it were private.
255 public $mInParse = false;
257 /** @var SectionProfiler */
258 protected $mProfiler;
263 protected $mLinkRenderer;
265 /** @var MagicWordFactory */
266 private $magicWordFactory;
271 /** @var ParserFactory */
274 /** @var SpecialPageFactory */
275 private $specialPageFactory;
280 /** @var LinkRendererFactory */
281 private $linkRendererFactory;
284 * @param array $parserConf See $wgParserConf documentation
285 * @param MagicWordFactory|null $magicWordFactory
286 * @param Language|null $contLang Content language
287 * @param ParserFactory|null $factory
288 * @param string|null $urlProtocols As returned from wfUrlProtocols()
289 * @param SpecialPageFactory|null $spFactory
290 * @param Config|null $siteConfig
291 * @param LinkRendererFactory|null $linkRendererFactory
293 public function __construct(
294 array $parserConf = [], MagicWordFactory
$magicWordFactory = null,
295 Language
$contLang = null, ParserFactory
$factory = null, $urlProtocols = null,
296 SpecialPageFactory
$spFactory = null, Config
$siteConfig = null,
297 LinkRendererFactory
$linkRendererFactory = null
299 $this->mConf
= $parserConf;
300 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
301 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
302 self
::EXT_LINK_ADDR
.
303 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
304 if ( isset( $parserConf['preprocessorClass'] ) ) {
305 $this->mPreprocessorClass
= $parserConf['preprocessorClass'];
306 } elseif ( wfIsHHVM() ) {
307 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
308 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
309 } elseif ( extension_loaded( 'domxml' ) ) {
310 # PECL extension that conflicts with the core DOM extension (T15770)
311 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
312 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
313 } elseif ( extension_loaded( 'dom' ) ) {
314 $this->mPreprocessorClass
= Preprocessor_DOM
::class;
316 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
318 wfDebug( __CLASS__
. ": using preprocessor: {$this->mPreprocessorClass}\n" );
320 $services = MediaWikiServices
::getInstance();
321 $this->magicWordFactory
= $magicWordFactory ??
322 $services->getMagicWordFactory();
324 $this->contLang
= $contLang ??
$services->getContentLanguage();
326 $this->factory
= $factory ??
$services->getParserFactory();
327 $this->specialPageFactory
= $spFactory ??
$services->getSpecialPageFactory();
328 $this->siteConfig
= $siteConfig ?? MediaWikiServices
::getInstance()->getMainConfig();
330 $this->linkRendererFactory
=
331 $linkRendererFactory ?? MediaWikiServices
::getInstance()->getLinkRendererFactory();
335 * Reduce memory usage to reduce the impact of circular references
337 public function __destruct() {
338 if ( isset( $this->mLinkHolders
) ) {
339 unset( $this->mLinkHolders
);
341 foreach ( $this as $name => $value ) {
342 unset( $this->$name );
347 * Allow extensions to clean up when the parser is cloned
349 public function __clone() {
350 $this->mInParse
= false;
352 // T58226: When you create a reference "to" an object field, that
353 // makes the object field itself be a reference too (until the other
354 // reference goes out of scope). When cloning, any field that's a
355 // reference is copied as a reference in the new object. Both of these
356 // are defined PHP5 behaviors, as inconvenient as it is for us when old
357 // hooks from PHP4 days are passing fields by reference.
358 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
359 // Make a non-reference copy of the field, then rebind the field to
360 // reference the new copy.
366 Hooks
::run( 'ParserCloned', [ $this ] );
370 * Do various kinds of initialisation on the first call of the parser
372 public function firstCallInit() {
373 if ( !$this->mFirstCall
) {
376 $this->mFirstCall
= false;
378 CoreParserFunctions
::register( $this );
379 CoreTagHooks
::register( $this );
380 $this->initialiseVariables();
382 // Avoid PHP 7.1 warning from passing $this by reference
384 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
392 public function clearState() {
393 $this->firstCallInit();
394 $this->mOutput
= new ParserOutput
;
395 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
396 $this->mAutonumber
= 0;
397 $this->mIncludeCount
= [];
398 $this->mLinkHolders
= new LinkHolderArray( $this );
400 $this->mRevisionObject
= $this->mRevisionTimestamp
=
401 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
402 $this->mVarCache
= [];
404 $this->mLangLinkLanguages
= [];
405 $this->currentRevisionCache
= null;
407 $this->mStripState
= new StripState( $this );
409 # Clear these on every parse, T6549
410 $this->mTplRedirCache
= $this->mTplDomCache
= [];
412 $this->mShowToc
= true;
413 $this->mForceTocPosition
= false;
414 $this->mIncludeSizes
= [
418 $this->mPPNodeCount
= 0;
419 $this->mGeneratedPPNodeCount
= 0;
420 $this->mHighestExpansionDepth
= 0;
421 $this->mDefaultSort
= false;
422 $this->mHeadings
= [];
423 $this->mDoubleUnderscores
= [];
424 $this->mExpensiveFunctionCount
= 0;
427 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
428 $this->mPreprocessor
= null;
431 $this->mProfiler
= new SectionProfiler();
433 // Avoid PHP 7.1 warning from passing $this by reference
435 Hooks
::run( 'ParserClearState', [ &$parser ] );
439 * Convert wikitext to HTML
440 * Do not call this function recursively.
442 * @param string $text Text we want to parse
443 * @param-taint $text escapes_htmlnoent
444 * @param Title $title
445 * @param ParserOptions $options
446 * @param bool $linestart
447 * @param bool $clearState
448 * @param int|null $revid Number to pass in {{REVISIONID}}
449 * @return ParserOutput A ParserOutput
450 * @return-taint escaped
452 public function parse(
453 $text, Title
$title, ParserOptions
$options,
454 $linestart = true, $clearState = true, $revid = null
457 // We use U+007F DELETE to construct strip markers, so we have to make
458 // sure that this character does not occur in the input text.
459 $text = strtr( $text, "\x7f", "?" );
460 $magicScopeVariable = $this->lock();
462 // Strip U+0000 NULL (T159174)
463 $text = str_replace( "\000", '', $text );
465 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
467 $this->currentRevisionCache
= null;
468 $this->mInputSize
= strlen( $text );
469 if ( $this->mOptions
->getEnableLimitReport() ) {
470 $this->mOutput
->resetParseStartTime();
473 $oldRevisionId = $this->mRevisionId
;
474 $oldRevisionObject = $this->mRevisionObject
;
475 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
476 $oldRevisionUser = $this->mRevisionUser
;
477 $oldRevisionSize = $this->mRevisionSize
;
478 if ( $revid !== null ) {
479 $this->mRevisionId
= $revid;
480 $this->mRevisionObject
= null;
481 $this->mRevisionTimestamp
= null;
482 $this->mRevisionUser
= null;
483 $this->mRevisionSize
= null;
486 // Avoid PHP 7.1 warning from passing $this by reference
488 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
490 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
491 $text = $this->internalParse( $text );
492 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
494 $text = $this->internalParseHalfParsed( $text, true, $linestart );
497 * A converted title will be provided in the output object if title and
498 * content conversion are enabled, the article text does not contain
499 * a conversion-suppressing double-underscore tag, and no
500 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
501 * automatic link conversion.
503 if ( !( $options->getDisableTitleConversion()
504 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
505 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
506 ||
$this->mOutput
->getDisplayTitle() !== false )
508 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
509 if ( $convruletitle ) {
510 $this->mOutput
->setTitleText( $convruletitle );
512 $titleText = $this->getTargetLanguage()->convertTitle( $title );
513 $this->mOutput
->setTitleText( $titleText );
517 # Compute runtime adaptive expiry if set
518 $this->mOutput
->finalizeAdaptiveCacheExpiry();
520 # Warn if too many heavyweight parser functions were used
521 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
522 $this->limitationWarn( 'expensive-parserfunction',
523 $this->mExpensiveFunctionCount
,
524 $this->mOptions
->getExpensiveParserFunctionLimit()
528 # Information on limits, for the benefit of users who try to skirt them
529 if ( $this->mOptions
->getEnableLimitReport() ) {
530 $text .= $this->makeLimitReport();
533 # Wrap non-interface parser output in a <div> so it can be targeted
535 $class = $this->mOptions
->getWrapOutputClass();
536 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
537 $this->mOutput
->addWrapperDivClass( $class );
540 $this->mOutput
->setText( $text );
542 $this->mRevisionId
= $oldRevisionId;
543 $this->mRevisionObject
= $oldRevisionObject;
544 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
545 $this->mRevisionUser
= $oldRevisionUser;
546 $this->mRevisionSize
= $oldRevisionSize;
547 $this->mInputSize
= false;
548 $this->currentRevisionCache
= null;
550 return $this->mOutput
;
554 * Set the limit report data in the current ParserOutput, and return the
555 * limit report HTML comment.
559 protected function makeLimitReport() {
560 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
562 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
563 if ( $cpuTime !== null ) {
564 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
565 sprintf( "%.3f", $cpuTime )
569 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
570 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
571 sprintf( "%.3f", $wallTime )
574 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
575 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
577 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
578 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
580 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
581 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
583 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
584 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
586 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
587 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
589 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
590 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
593 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
594 $this->mOutput
->setLimitReportData( $key, $value );
597 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
599 $limitReport = "NewPP limit report\n";
600 if ( $this->siteConfig
->get( 'ShowHostnames' ) ) {
601 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
603 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
604 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
605 $limitReport .= 'Dynamic content: ' .
606 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
609 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
610 if ( Hooks
::run( 'ParserLimitReportFormat',
611 [ $key, &$value, &$limitReport, false, false ]
613 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
614 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
615 ->inLanguage( 'en' )->useDatabase( false );
616 if ( !$valueMsg->exists() ) {
617 $valueMsg = new RawMessage( '$1' );
619 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
620 $valueMsg->params( $value );
621 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
625 // Since we're not really outputting HTML, decode the entities and
626 // then re-encode the things that need hiding inside HTML comments.
627 $limitReport = htmlspecialchars_decode( $limitReport );
629 // Sanitize for comment. Note '‐' in the replacement is U+2010,
630 // which looks much like the problematic '-'.
631 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
632 $text = "\n<!-- \n$limitReport-->\n";
634 // Add on template profiling data in human/machine readable way
635 $dataByFunc = $this->mProfiler
->getFunctionStats();
636 uasort( $dataByFunc, function ( $a, $b ) {
637 return $b['real'] <=> $a['real']; // descending order
640 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
641 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
642 $item['%real'], $item['real'], $item['calls'],
643 htmlspecialchars( $item['name'] ) );
645 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
646 $text .= implode( "\n", $profileReport ) . "\n-->\n";
648 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
650 // Add other cache related metadata
651 if ( $this->siteConfig
->get( 'ShowHostnames' ) ) {
652 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
654 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
655 $this->mOutput
->getCacheTime() );
656 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
657 $this->mOutput
->getCacheExpiry() );
658 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
659 $this->mOutput
->hasDynamicContent() );
661 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
662 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
663 $this->mTitle
->getPrefixedDBkey() );
669 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
670 * can be called from an extension tag hook.
672 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
673 * instead, which means that lists and links have not been fully parsed yet,
674 * and strip markers are still present.
676 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
678 * Use this function if you're a parser tag hook and you want to parse
679 * wikitext before or after applying additional transformations, and you
680 * intend to *return the result as hook output*, which will cause it to go
681 * through the rest of parsing process automatically.
683 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
684 * $text are not expanded
686 * @param string $text Text extension wants to have parsed
687 * @param-taint $text escapes_htmlnoent
688 * @param bool|PPFrame $frame The frame to use for expanding any template variables
689 * @return string UNSAFE half-parsed HTML
690 * @return-taint escaped
692 public function recursiveTagParse( $text, $frame = false ) {
693 // Avoid PHP 7.1 warning from passing $this by reference
695 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
696 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
697 $text = $this->internalParse( $text, false, $frame );
702 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
703 * point can be called from an extension tag hook.
705 * The output of this function is fully-parsed HTML that is safe for output.
706 * If you're a parser tag hook, you might want to use recursiveTagParse()
709 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
710 * $text are not expanded
714 * @param string $text Text extension wants to have parsed
715 * @param-taint $text escapes_htmlnoent
716 * @param bool|PPFrame $frame The frame to use for expanding any template variables
717 * @return string Fully parsed HTML
718 * @return-taint escaped
720 public function recursiveTagParseFully( $text, $frame = false ) {
721 $text = $this->recursiveTagParse( $text, $frame );
722 $text = $this->internalParseHalfParsed( $text, false );
727 * Expand templates and variables in the text, producing valid, static wikitext.
728 * Also removes comments.
729 * Do not call this function recursively.
730 * @param string $text
731 * @param Title|null $title
732 * @param ParserOptions $options
733 * @param int|null $revid
734 * @param bool|PPFrame $frame
735 * @return mixed|string
737 public function preprocess( $text, Title
$title = null,
738 ParserOptions
$options, $revid = null, $frame = false
740 $magicScopeVariable = $this->lock();
741 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
742 if ( $revid !== null ) {
743 $this->mRevisionId
= $revid;
745 // Avoid PHP 7.1 warning from passing $this by reference
747 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
748 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
749 $text = $this->replaceVariables( $text, $frame );
750 $text = $this->mStripState
->unstripBoth( $text );
755 * Recursive parser entry point that can be called from an extension tag
758 * @param string $text Text to be expanded
759 * @param bool|PPFrame $frame The frame to use for expanding any template variables
763 public function recursivePreprocess( $text, $frame = false ) {
764 $text = $this->replaceVariables( $text, $frame );
765 $text = $this->mStripState
->unstripBoth( $text );
770 * Process the wikitext for the "?preload=" feature. (T7210)
772 * "<noinclude>", "<includeonly>" etc. are parsed as for template
773 * transclusion, comments, templates, arguments, tags hooks and parser
774 * functions are untouched.
776 * @param string $text
777 * @param Title $title
778 * @param ParserOptions $options
779 * @param array $params
782 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
783 $msg = new RawMessage( $text );
784 $text = $msg->params( $params )->plain();
786 # Parser (re)initialisation
787 $magicScopeVariable = $this->lock();
788 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
790 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
791 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
792 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
793 $text = $this->mStripState
->unstripBoth( $text );
798 * Set the current user.
799 * Should only be used when doing pre-save transform.
801 * @param User|null $user User object or null (to reset)
803 public function setUser( $user ) {
804 $this->mUser
= $user;
808 * Set the context title
812 public function setTitle( $t ) {
814 $t = Title
::newFromText( 'NO TITLE' );
817 if ( $t->hasFragment() ) {
818 # Strip the fragment to avoid various odd effects
819 $this->mTitle
= $t->createFragmentTarget( '' );
826 * Accessor for the Title object
830 public function getTitle() {
831 return $this->mTitle
;
835 * Accessor/mutator for the Title object
837 * @param Title|null $x Title object or null to just get the current one
840 public function Title( $x = null ) {
841 return wfSetVar( $this->mTitle
, $x );
845 * Set the output type
847 * @param int $ot New value
849 public function setOutputType( $ot ) {
850 $this->mOutputType
= $ot;
853 'html' => $ot == self
::OT_HTML
,
854 'wiki' => $ot == self
::OT_WIKI
,
855 'pre' => $ot == self
::OT_PREPROCESS
,
856 'plain' => $ot == self
::OT_PLAIN
,
861 * Accessor/mutator for the output type
863 * @param int|null $x New value or null to just get the current one
866 public function OutputType( $x = null ) {
867 return wfSetVar( $this->mOutputType
, $x );
871 * Get the ParserOutput object
873 * @return ParserOutput
875 public function getOutput() {
876 return $this->mOutput
;
880 * Get the ParserOptions object
882 * @return ParserOptions
884 public function getOptions() {
885 return $this->mOptions
;
889 * Accessor/mutator for the ParserOptions object
891 * @param ParserOptions|null $x New value or null to just get the current one
892 * @return ParserOptions Current ParserOptions object
894 public function Options( $x = null ) {
895 return wfSetVar( $this->mOptions
, $x );
901 public function nextLinkID() {
902 return $this->mLinkID++
;
908 public function setLinkID( $id ) {
909 $this->mLinkID
= $id;
913 * Get a language object for use in parser functions such as {{FORMATNUM:}}
916 public function getFunctionLang() {
917 return $this->getTargetLanguage();
921 * Get the target language for the content being parsed. This is usually the
922 * language that the content is in.
926 * @throws MWException
929 public function getTargetLanguage() {
930 $target = $this->mOptions
->getTargetLanguage();
932 if ( $target !== null ) {
934 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
935 return $this->mOptions
->getUserLangObj();
936 } elseif ( is_null( $this->mTitle
) ) {
937 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
940 return $this->mTitle
->getPageLanguage();
944 * Get the language object for language conversion
945 * @deprecated since 1.32, just use getTargetLanguage()
946 * @return Language|null
948 public function getConverterLanguage() {
949 return $this->getTargetLanguage();
953 * Get a User object either from $this->mUser, if set, or from the
954 * ParserOptions object otherwise
958 public function getUser() {
959 if ( !is_null( $this->mUser
) ) {
962 return $this->mOptions
->getUser();
966 * Get a preprocessor object
968 * @return Preprocessor
970 public function getPreprocessor() {
971 if ( !isset( $this->mPreprocessor
) ) {
972 $class = $this->mPreprocessorClass
;
973 $this->mPreprocessor
= new $class( $this );
975 return $this->mPreprocessor
;
979 * Get a LinkRenderer instance to make links with
982 * @return LinkRenderer
984 public function getLinkRenderer() {
985 // XXX We make the LinkRenderer with current options and then cache it forever
986 if ( !$this->mLinkRenderer
) {
987 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
988 $this->mLinkRenderer
->setStubThreshold(
989 $this->getOptions()->getStubThreshold()
993 return $this->mLinkRenderer
;
997 * Get the MagicWordFactory that this Parser is using
1000 * @return MagicWordFactory
1002 public function getMagicWordFactory() {
1003 return $this->magicWordFactory
;
1007 * Get the content language that this Parser is using
1012 public function getContentLanguage() {
1013 return $this->contLang
;
1017 * Replaces all occurrences of HTML-style comments and the given tags
1018 * in the text with a random marker and returns the next text. The output
1019 * parameter $matches will be an associative array filled with data in
1026 * [ 'param' => 'x' ],
1027 * '<element param="x">tag content</element>' ]
1030 * @param array $elements List of element names. Comments are always extracted.
1031 * @param string $text Source text string.
1032 * @param array &$matches Out parameter, Array: extracted tags
1033 * @return string Stripped text
1035 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1040 $taglist = implode( '|', $elements );
1041 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1043 while ( $text != '' ) {
1044 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1046 if ( count( $p ) < 5 ) {
1049 if ( count( $p ) > 5 ) {
1057 list( , $element, $attributes, $close, $inside ) = $p;
1060 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1061 $stripped .= $marker;
1063 if ( $close === '/>' ) {
1064 # Empty element tag, <tag />
1069 if ( $element === '!--' ) {
1072 $end = "/(<\\/$element\\s*>)/i";
1074 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1076 if ( count( $q ) < 3 ) {
1077 # No end tag -- let it run out to the end of the text.
1081 list( , $tail, $text ) = $q;
1085 $matches[$marker] = [ $element,
1087 Sanitizer
::decodeTagAttributes( $attributes ),
1088 "<$element$attributes$close$content$tail" ];
1094 * Get a list of strippable XML-like elements
1098 public function getStripList() {
1099 return $this->mStripList
;
1103 * Add an item to the strip state
1104 * Returns the unique tag which must be inserted into the stripped text
1105 * The tag will be replaced with the original text in unstrip()
1107 * @param string $text
1111 public function insertStripItem( $text ) {
1112 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1113 $this->mMarkerIndex++
;
1114 $this->mStripState
->addGeneral( $marker, $text );
1119 * parse the wiki syntax used to render tables
1122 * @param string $text
1125 public function doTableStuff( $text ) {
1126 $lines = StringUtils
::explode( "\n", $text );
1128 $td_history = []; # Is currently a td tag open?
1129 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1130 $tr_history = []; # Is currently a tr tag open?
1131 $tr_attributes = []; # history of tr attributes
1132 $has_opened_tr = []; # Did this table open a <tr> element?
1133 $indent_level = 0; # indent level of the table
1135 foreach ( $lines as $outLine ) {
1136 $line = trim( $outLine );
1138 if ( $line === '' ) { # empty line, go to next line
1139 $out .= $outLine . "\n";
1143 $first_character = $line[0];
1144 $first_two = substr( $line, 0, 2 );
1147 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1148 # First check if we are starting a new table
1149 $indent_level = strlen( $matches[1] );
1151 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1152 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1154 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1155 array_push( $td_history, false );
1156 array_push( $last_tag_history, '' );
1157 array_push( $tr_history, false );
1158 array_push( $tr_attributes, '' );
1159 array_push( $has_opened_tr, false );
1160 } elseif ( count( $td_history ) == 0 ) {
1161 # Don't do any of the following
1162 $out .= $outLine . "\n";
1164 } elseif ( $first_two === '|}' ) {
1165 # We are ending a table
1166 $line = '</table>' . substr( $line, 2 );
1167 $last_tag = array_pop( $last_tag_history );
1169 if ( !array_pop( $has_opened_tr ) ) {
1170 $line = "<tr><td></td></tr>{$line}";
1173 if ( array_pop( $tr_history ) ) {
1174 $line = "</tr>{$line}";
1177 if ( array_pop( $td_history ) ) {
1178 $line = "</{$last_tag}>{$line}";
1180 array_pop( $tr_attributes );
1181 if ( $indent_level > 0 ) {
1182 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1186 } elseif ( $first_two === '|-' ) {
1187 # Now we have a table row
1188 $line = preg_replace( '#^\|-+#', '', $line );
1190 # Whats after the tag is now only attributes
1191 $attributes = $this->mStripState
->unstripBoth( $line );
1192 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1193 array_pop( $tr_attributes );
1194 array_push( $tr_attributes, $attributes );
1197 $last_tag = array_pop( $last_tag_history );
1198 array_pop( $has_opened_tr );
1199 array_push( $has_opened_tr, true );
1201 if ( array_pop( $tr_history ) ) {
1205 if ( array_pop( $td_history ) ) {
1206 $line = "</{$last_tag}>{$line}";
1210 array_push( $tr_history, false );
1211 array_push( $td_history, false );
1212 array_push( $last_tag_history, '' );
1213 } elseif ( $first_character === '|'
1214 ||
$first_character === '!'
1215 ||
$first_two === '|+'
1217 # This might be cell elements, td, th or captions
1218 if ( $first_two === '|+' ) {
1219 $first_character = '+';
1220 $line = substr( $line, 2 );
1222 $line = substr( $line, 1 );
1225 // Implies both are valid for table headings.
1226 if ( $first_character === '!' ) {
1227 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1230 # Split up multiple cells on the same line.
1231 # FIXME : This can result in improper nesting of tags processed
1232 # by earlier parser steps.
1233 $cells = explode( '||', $line );
1237 # Loop through each table cell
1238 foreach ( $cells as $cell ) {
1240 if ( $first_character !== '+' ) {
1241 $tr_after = array_pop( $tr_attributes );
1242 if ( !array_pop( $tr_history ) ) {
1243 $previous = "<tr{$tr_after}>\n";
1245 array_push( $tr_history, true );
1246 array_push( $tr_attributes, '' );
1247 array_pop( $has_opened_tr );
1248 array_push( $has_opened_tr, true );
1251 $last_tag = array_pop( $last_tag_history );
1253 if ( array_pop( $td_history ) ) {
1254 $previous = "</{$last_tag}>\n{$previous}";
1257 if ( $first_character === '|' ) {
1259 } elseif ( $first_character === '!' ) {
1261 } elseif ( $first_character === '+' ) {
1262 $last_tag = 'caption';
1267 array_push( $last_tag_history, $last_tag );
1269 # A cell could contain both parameters and data
1270 $cell_data = explode( '|', $cell, 2 );
1272 # T2553: Note that a '|' inside an invalid link should not
1273 # be mistaken as delimiting cell parameters
1274 # Bug T153140: Neither should language converter markup.
1275 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1276 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1277 } elseif ( count( $cell_data ) == 1 ) {
1278 // Whitespace in cells is trimmed
1279 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1281 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1282 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1283 // Whitespace in cells is trimmed
1284 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1288 array_push( $td_history, true );
1291 $out .= $outLine . "\n";
1294 # Closing open td, tr && table
1295 while ( count( $td_history ) > 0 ) {
1296 if ( array_pop( $td_history ) ) {
1299 if ( array_pop( $tr_history ) ) {
1302 if ( !array_pop( $has_opened_tr ) ) {
1303 $out .= "<tr><td></td></tr>\n";
1306 $out .= "</table>\n";
1309 # Remove trailing line-ending (b/c)
1310 if ( substr( $out, -1 ) === "\n" ) {
1311 $out = substr( $out, 0, -1 );
1314 # special case: don't return empty table
1315 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1323 * Helper function for parse() that transforms wiki markup into half-parsed
1324 * HTML. Only called for $mOutputType == self::OT_HTML.
1328 * @param string $text The text to parse
1329 * @param-taint $text escapes_html
1330 * @param bool $isMain Whether this is being called from the main parse() function
1331 * @param PPFrame|bool $frame A pre-processor frame
1335 public function internalParse( $text, $isMain = true, $frame = false ) {
1338 // Avoid PHP 7.1 warning from passing $this by reference
1341 # Hook to suspend the parser in this state
1342 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1346 # if $frame is provided, then use $frame for replacing any variables
1348 # use frame depth to infer how include/noinclude tags should be handled
1349 # depth=0 means this is the top-level document; otherwise it's an included document
1350 if ( !$frame->depth
) {
1353 $flag = self
::PTD_FOR_INCLUSION
;
1355 $dom = $this->preprocessToDom( $text, $flag );
1356 $text = $frame->expand( $dom );
1358 # if $frame is not provided, then use old-style replaceVariables
1359 $text = $this->replaceVariables( $text );
1362 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1363 $text = Sanitizer
::removeHTMLtags(
1365 [ $this, 'attributeStripCallback' ],
1367 array_keys( $this->mTransparentTagHooks
),
1369 [ $this, 'addTrackingCategory' ]
1371 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1373 # Tables need to come after variable replacement for things to work
1374 # properly; putting them before other transformations should keep
1375 # exciting things like link expansions from showing up in surprising
1377 $text = $this->doTableStuff( $text );
1379 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1381 $text = $this->doDoubleUnderscore( $text );
1383 $text = $this->doHeadings( $text );
1384 $text = $this->replaceInternalLinks( $text );
1385 $text = $this->doAllQuotes( $text );
1386 $text = $this->replaceExternalLinks( $text );
1388 # replaceInternalLinks may sometimes leave behind
1389 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1390 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1392 $text = $this->doMagicLinks( $text );
1393 $text = $this->formatHeadings( $text, $origText, $isMain );
1399 * Helper function for parse() that transforms half-parsed HTML into fully
1402 * @param string $text
1403 * @param bool $isMain
1404 * @param bool $linestart
1407 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1408 $text = $this->mStripState
->unstripGeneral( $text );
1410 // Avoid PHP 7.1 warning from passing $this by reference
1414 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1417 # Clean up special characters, only run once, next-to-last before doBlockLevels
1418 $text = Sanitizer
::armorFrenchSpaces( $text );
1420 $text = $this->doBlockLevels( $text, $linestart );
1422 $this->replaceLinkHolders( $text );
1425 * The input doesn't get language converted if
1427 * b) Content isn't converted
1428 * c) It's a conversion table
1429 * d) it is an interface message (which is in the user language)
1431 if ( !( $this->mOptions
->getDisableContentConversion()
1432 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1433 && !$this->mOptions
->getInterfaceMessage()
1435 # The position of the convert() call should not be changed. it
1436 # assumes that the links are all replaced and the only thing left
1437 # is the <nowiki> mark.
1438 $text = $this->getTargetLanguage()->convert( $text );
1441 $text = $this->mStripState
->unstripNoWiki( $text );
1444 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1447 $text = $this->replaceTransparentTags( $text );
1448 $text = $this->mStripState
->unstripGeneral( $text );
1450 $text = Sanitizer
::normalizeCharReferences( $text );
1452 if ( MWTidy
::isEnabled() ) {
1453 if ( $this->mOptions
->getTidy() ) {
1454 $text = MWTidy
::tidy( $text );
1457 # attempt to sanitize at least some nesting problems
1458 # (T4702 and quite a few others)
1459 # This code path is buggy and deprecated!
1460 wfDeprecated( 'disabling tidy', '1.33' );
1462 # ''Something [http://www.cool.com cool''] -->
1463 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1464 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1465 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1466 # fix up an anchor inside another anchor, only
1467 # at least for a single single nested link (T5695)
1468 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1469 '\\1\\2</a>\\3</a>\\1\\4</a>',
1470 # fix div inside inline elements- doBlockLevels won't wrap a line which
1471 # contains a div, so fix it up here; replace
1472 # div with escaped text
1473 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1474 '\\1\\3<div\\5>\\6</div>\\8\\9',
1475 # remove empty italic or bold tag pairs, some
1476 # introduced by rules above
1477 '/<([bi])><\/\\1>/' => '',
1480 $text = preg_replace(
1481 array_keys( $tidyregs ),
1482 array_values( $tidyregs ),
1487 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1494 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1495 * magic external links.
1500 * @param string $text
1504 public function doMagicLinks( $text ) {
1505 $prots = wfUrlProtocolsWithoutProtRel();
1506 $urlChar = self
::EXT_LINK_URL_CLASS
;
1507 $addr = self
::EXT_LINK_ADDR
;
1508 $space = self
::SPACE_NOT_NL
; # non-newline space
1509 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1510 $spaces = "$space++"; # possessive match of 1 or more spaces
1511 $text = preg_replace_callback(
1513 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1514 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1515 (\b # m[3]: Free external links
1517 ($addr$urlChar*) # m[4]: Post-protocol path
1519 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1521 \bISBN $spaces ( # m[6]: ISBN, capture number
1522 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1523 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1524 [0-9Xx] # check digit
1526 )!xu", [ $this, 'magicLinkCallback' ], $text );
1531 * @throws MWException
1533 * @return string HTML
1535 public function magicLinkCallback( $m ) {
1536 if ( isset( $m[1] ) && $m[1] !== '' ) {
1539 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1542 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1543 # Free external link
1544 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1545 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1547 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1548 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1553 $cssClass = 'mw-magiclink-rfc';
1554 $trackingCat = 'magiclink-tracking-rfc';
1556 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1557 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1561 $urlmsg = 'pubmedurl';
1562 $cssClass = 'mw-magiclink-pmid';
1563 $trackingCat = 'magiclink-tracking-pmid';
1566 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1567 substr( $m[0], 0, 20 ) . '"' );
1569 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1570 $this->addTrackingCategory( $trackingCat );
1571 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1572 } elseif ( isset( $m[6] ) && $m[6] !== ''
1573 && $this->mOptions
->getMagicISBNLinks()
1577 $space = self
::SPACE_NOT_NL
; # non-newline space
1578 $isbn = preg_replace( "/$space/", ' ', $isbn );
1579 $num = strtr( $isbn, [
1584 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1585 return $this->getLinkRenderer()->makeKnownLink(
1586 SpecialPage
::getTitleFor( 'Booksources', $num ),
1589 'class' => 'internal mw-magiclink-isbn',
1590 'title' => false // suppress title attribute
1599 * Make a free external link, given a user-supplied URL
1601 * @param string $url
1602 * @param int $numPostProto
1603 * The number of characters after the protocol.
1604 * @return string HTML
1607 public function makeFreeExternalLink( $url, $numPostProto ) {
1610 # The characters '<' and '>' (which were escaped by
1611 # removeHTMLtags()) should not be included in
1612 # URLs, per RFC 2396.
1613 # Make terminate a URL as well (bug T84937)
1616 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1621 $trail = substr( $url, $m2[0][1] ) . $trail;
1622 $url = substr( $url, 0, $m2[0][1] );
1625 # Move trailing punctuation to $trail
1627 # If there is no left bracket, then consider right brackets fair game too
1628 if ( strpos( $url, '(' ) === false ) {
1632 $urlRev = strrev( $url );
1633 $numSepChars = strspn( $urlRev, $sep );
1634 # Don't break a trailing HTML entity by moving the ; into $trail
1635 # This is in hot code, so use substr_compare to avoid having to
1636 # create a new string object for the comparison
1637 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1638 # more optimization: instead of running preg_match with a $
1639 # anchor, which can be slow, do the match on the reversed
1640 # string starting at the desired offset.
1641 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1642 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1646 if ( $numSepChars ) {
1647 $trail = substr( $url, -$numSepChars ) . $trail;
1648 $url = substr( $url, 0, -$numSepChars );
1651 # Verify that we still have a real URL after trail removal, and
1652 # not just lone protocol
1653 if ( strlen( $trail ) >= $numPostProto ) {
1654 return $url . $trail;
1657 $url = Sanitizer
::cleanUrl( $url );
1659 # Is this an external image?
1660 $text = $this->maybeMakeExternalImage( $url );
1661 if ( $text === false ) {
1662 # Not an image, make a link
1663 $text = Linker
::makeExternalLink( $url,
1664 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1666 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1667 # Register it in the output object...
1668 $this->mOutput
->addExternalLink( $url );
1670 return $text . $trail;
1674 * Parse headers and return html
1678 * @param string $text
1682 public function doHeadings( $text ) {
1683 for ( $i = 6; $i >= 1; --$i ) {
1684 $h = str_repeat( '=', $i );
1685 // Trim non-newline whitespace from headings
1686 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1687 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1693 * Replace single quotes with HTML markup
1696 * @param string $text
1698 * @return string The altered text
1700 public function doAllQuotes( $text ) {
1702 $lines = StringUtils
::explode( "\n", $text );
1703 foreach ( $lines as $line ) {
1704 $outtext .= $this->doQuotes( $line ) . "\n";
1706 $outtext = substr( $outtext, 0, -1 );
1711 * Helper function for doAllQuotes()
1713 * @param string $text
1717 public function doQuotes( $text ) {
1718 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1719 $countarr = count( $arr );
1720 if ( $countarr == 1 ) {
1724 // First, do some preliminary work. This may shift some apostrophes from
1725 // being mark-up to being text. It also counts the number of occurrences
1726 // of bold and italics mark-ups.
1729 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1730 $thislen = strlen( $arr[$i] );
1731 // If there are ever four apostrophes, assume the first is supposed to
1732 // be text, and the remaining three constitute mark-up for bold text.
1733 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1734 if ( $thislen == 4 ) {
1735 $arr[$i - 1] .= "'";
1738 } elseif ( $thislen > 5 ) {
1739 // If there are more than 5 apostrophes in a row, assume they're all
1740 // text except for the last 5.
1741 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1742 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1746 // Count the number of occurrences of bold and italics mark-ups.
1747 if ( $thislen == 2 ) {
1749 } elseif ( $thislen == 3 ) {
1751 } elseif ( $thislen == 5 ) {
1757 // If there is an odd number of both bold and italics, it is likely
1758 // that one of the bold ones was meant to be an apostrophe followed
1759 // by italics. Which one we cannot know for certain, but it is more
1760 // likely to be one that has a single-letter word before it.
1761 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1762 $firstsingleletterword = -1;
1763 $firstmultiletterword = -1;
1765 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1766 if ( strlen( $arr[$i] ) == 3 ) {
1767 $x1 = substr( $arr[$i - 1], -1 );
1768 $x2 = substr( $arr[$i - 1], -2, 1 );
1769 if ( $x1 === ' ' ) {
1770 if ( $firstspace == -1 ) {
1773 } elseif ( $x2 === ' ' ) {
1774 $firstsingleletterword = $i;
1775 // if $firstsingleletterword is set, we don't
1776 // look at the other options, so we can bail early.
1778 } elseif ( $firstmultiletterword == -1 ) {
1779 $firstmultiletterword = $i;
1784 // If there is a single-letter word, use it!
1785 if ( $firstsingleletterword > -1 ) {
1786 $arr[$firstsingleletterword] = "''";
1787 $arr[$firstsingleletterword - 1] .= "'";
1788 } elseif ( $firstmultiletterword > -1 ) {
1789 // If not, but there's a multi-letter word, use that one.
1790 $arr[$firstmultiletterword] = "''";
1791 $arr[$firstmultiletterword - 1] .= "'";
1792 } elseif ( $firstspace > -1 ) {
1793 // ... otherwise use the first one that has neither.
1794 // (notice that it is possible for all three to be -1 if, for example,
1795 // there is only one pentuple-apostrophe in the line)
1796 $arr[$firstspace] = "''";
1797 $arr[$firstspace - 1] .= "'";
1801 // Now let's actually convert our apostrophic mush to HTML!
1806 foreach ( $arr as $r ) {
1807 if ( ( $i %
2 ) == 0 ) {
1808 if ( $state === 'both' ) {
1814 $thislen = strlen( $r );
1815 if ( $thislen == 2 ) {
1816 if ( $state === 'i' ) {
1819 } elseif ( $state === 'bi' ) {
1822 } elseif ( $state === 'ib' ) {
1823 $output .= '</b></i><b>';
1825 } elseif ( $state === 'both' ) {
1826 $output .= '<b><i>' . $buffer . '</i>';
1828 } else { // $state can be 'b' or ''
1832 } elseif ( $thislen == 3 ) {
1833 if ( $state === 'b' ) {
1836 } elseif ( $state === 'bi' ) {
1837 $output .= '</i></b><i>';
1839 } elseif ( $state === 'ib' ) {
1842 } elseif ( $state === 'both' ) {
1843 $output .= '<i><b>' . $buffer . '</b>';
1845 } else { // $state can be 'i' or ''
1849 } elseif ( $thislen == 5 ) {
1850 if ( $state === 'b' ) {
1851 $output .= '</b><i>';
1853 } elseif ( $state === 'i' ) {
1854 $output .= '</i><b>';
1856 } elseif ( $state === 'bi' ) {
1857 $output .= '</i></b>';
1859 } elseif ( $state === 'ib' ) {
1860 $output .= '</b></i>';
1862 } elseif ( $state === 'both' ) {
1863 $output .= '<i><b>' . $buffer . '</b></i>';
1865 } else { // ($state == '')
1873 // Now close all remaining tags. Notice that the order is important.
1874 if ( $state === 'b' ||
$state === 'ib' ) {
1877 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1880 if ( $state === 'bi' ) {
1883 // There might be lonely ''''', so make sure we have a buffer
1884 if ( $state === 'both' && $buffer ) {
1885 $output .= '<b><i>' . $buffer . '</i></b>';
1891 * Replace external links (REL)
1893 * Note: this is all very hackish and the order of execution matters a lot.
1894 * Make sure to run tests/parser/parserTests.php if you change this code.
1898 * @param string $text
1900 * @throws MWException
1903 public function replaceExternalLinks( $text ) {
1904 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1905 if ( $bits === false ) {
1906 throw new MWException( "PCRE needs to be compiled with "
1907 . "--enable-unicode-properties in order for MediaWiki to function" );
1909 $s = array_shift( $bits );
1912 while ( $i < count( $bits ) ) {
1915 $text = $bits[$i++
];
1916 $trail = $bits[$i++
];
1918 # The characters '<' and '>' (which were escaped by
1919 # removeHTMLtags()) should not be included in
1920 # URLs, per RFC 2396.
1922 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1923 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1924 $url = substr( $url, 0, $m2[0][1] );
1927 # If the link text is an image URL, replace it with an <img> tag
1928 # This happened by accident in the original parser, but some people used it extensively
1929 $img = $this->maybeMakeExternalImage( $text );
1930 if ( $img !== false ) {
1936 # Set linktype for CSS
1939 # No link text, e.g. [http://domain.tld/some.link]
1940 if ( $text == '' ) {
1942 $langObj = $this->getTargetLanguage();
1943 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1944 $linktype = 'autonumber';
1946 # Have link text, e.g. [http://domain.tld/some.link text]s
1948 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1951 // Excluding protocol-relative URLs may avoid many false positives.
1952 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1953 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1956 $url = Sanitizer
::cleanUrl( $url );
1958 # Use the encoded URL
1959 # This means that users can paste URLs directly into the text
1960 # Funny characters like ö aren't valid in URLs anyway
1961 # This was changed in August 2004
1962 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
1963 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
1965 # Register link in the output object.
1966 $this->mOutput
->addExternalLink( $url );
1973 * Get the rel attribute for a particular external link.
1976 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1977 * nofollow if appropriate
1978 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1979 * @return string|null Rel attribute for $url
1981 public static function getExternalLinkRel( $url = false, $title = null ) {
1982 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1983 $ns = $title ?
$title->getNamespace() : false;
1984 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1985 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1993 * Get an associative array of additional HTML attributes appropriate for a
1994 * particular external link. This currently may include rel => nofollow
1995 * (depending on configuration, namespace, and the URL's domain) and/or a
1996 * target attribute (depending on configuration).
1998 * @param string $url URL to extract the domain from for rel =>
1999 * nofollow if appropriate
2000 * @return array Associative array of HTML attributes
2002 public function getExternalLinkAttribs( $url ) {
2004 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2006 $target = $this->mOptions
->getExternalLinkTarget();
2008 $attribs['target'] = $target;
2009 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2010 // T133507. New windows can navigate parent cross-origin.
2011 // Including noreferrer due to lacking browser
2012 // support of noopener. Eventually noreferrer should be removed.
2013 if ( $rel !== '' ) {
2016 $rel .= 'noreferrer noopener';
2019 $attribs['rel'] = $rel;
2024 * Replace unusual escape codes in a URL with their equivalent characters
2026 * This generally follows the syntax defined in RFC 3986, with special
2027 * consideration for HTTP query strings.
2029 * @param string $url
2032 public static function normalizeLinkUrl( $url ) {
2033 # Test for RFC 3986 IPv6 syntax
2034 $scheme = '[a-z][a-z0-9+.-]*:';
2035 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2036 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2037 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2038 IP
::isValid( rawurldecode( $m[1] ) )
2040 $isIPv6 = rawurldecode( $m[1] );
2045 # Make sure unsafe characters are encoded
2046 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2048 return rawurlencode( $m[0] );
2054 $end = strlen( $url );
2056 # Fragment part - 'fragment'
2057 $start = strpos( $url, '#' );
2058 if ( $start !== false && $start < $end ) {
2059 $ret = self
::normalizeUrlComponent(
2060 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2064 # Query part - 'query' minus &=+;
2065 $start = strpos( $url, '?' );
2066 if ( $start !== false && $start < $end ) {
2067 $ret = self
::normalizeUrlComponent(
2068 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2072 # Scheme and path part - 'pchar'
2073 # (we assume no userinfo or encoded colons in the host)
2074 $ret = self
::normalizeUrlComponent(
2075 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2078 if ( $isIPv6 !== false ) {
2079 $ipv6Host = "%5B({$isIPv6})%5D";
2080 $ret = preg_replace(
2081 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2090 private static function normalizeUrlComponent( $component, $unsafe ) {
2091 $callback = function ( $matches ) use ( $unsafe ) {
2092 $char = urldecode( $matches[0] );
2093 $ord = ord( $char );
2094 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2098 # Leave it escaped, but use uppercase for a-f
2099 return strtoupper( $matches[0] );
2102 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2106 * make an image if it's allowed, either through the global
2107 * option, through the exception, or through the on-wiki whitelist
2109 * @param string $url
2113 private function maybeMakeExternalImage( $url ) {
2114 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2115 $imagesexception = !empty( $imagesfrom );
2117 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2118 if ( $imagesexception && is_array( $imagesfrom ) ) {
2119 $imagematch = false;
2120 foreach ( $imagesfrom as $match ) {
2121 if ( strpos( $url, $match ) === 0 ) {
2126 } elseif ( $imagesexception ) {
2127 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2129 $imagematch = false;
2132 if ( $this->mOptions
->getAllowExternalImages()
2133 ||
( $imagesexception && $imagematch )
2135 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2137 $text = Linker
::makeExternalImage( $url );
2140 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2141 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2143 $whitelist = explode(
2145 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2148 foreach ( $whitelist as $entry ) {
2149 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2150 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2153 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2154 # Image matches a whitelist entry
2155 $text = Linker
::makeExternalImage( $url );
2164 * Process [[ ]] wikilinks
2168 * @return string Processed text
2172 public function replaceInternalLinks( $s ) {
2173 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2178 * Process [[ ]] wikilinks (RIL)
2180 * @throws MWException
2181 * @return LinkHolderArray
2185 public function replaceInternalLinks2( &$s ) {
2186 static $tc = false, $e1, $e1_img;
2187 # the % is needed to support urlencoded titles as well
2189 $tc = Title
::legalChars() . '#%';
2190 # Match a link having the form [[namespace:link|alternate]]trail
2191 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2192 # Match cases where there is no "]]", which might still be images
2193 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2196 $holders = new LinkHolderArray( $this );
2198 # split the entire text string on occurrences of [[
2199 $a = StringUtils
::explode( '[[', ' ' . $s );
2200 # get the first element (all text up to first [[), and remove the space we added
2203 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2204 $s = substr( $s, 1 );
2206 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2208 if ( $useLinkPrefixExtension ) {
2209 # Match the end of a line for a word that's not followed by whitespace,
2210 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2211 $charset = $this->contLang
->linkPrefixCharset();
2212 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2215 if ( is_null( $this->mTitle
) ) {
2216 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2218 $nottalk = !$this->mTitle
->isTalkPage();
2220 if ( $useLinkPrefixExtension ) {
2222 if ( preg_match( $e2, $s, $m ) ) {
2223 $first_prefix = $m[2];
2225 $first_prefix = false;
2231 $useSubpages = $this->areSubpagesAllowed();
2233 # Loop for each link
2234 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2235 # Check for excessive memory usage
2236 if ( $holders->isBig() ) {
2238 # Do the existence check, replace the link holders and clear the array
2239 $holders->replace( $s );
2243 if ( $useLinkPrefixExtension ) {
2244 if ( preg_match( $e2, $s, $m ) ) {
2245 list( , $s, $prefix ) = $m;
2250 if ( $first_prefix ) {
2251 $prefix = $first_prefix;
2252 $first_prefix = false;
2256 $might_be_img = false;
2258 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2260 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2261 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2262 # the real problem is with the $e1 regex
2264 # Still some problems for cases where the ] is meant to be outside punctuation,
2265 # and no image is in sight. See T4095.
2267 && substr( $m[3], 0, 1 ) === ']'
2268 && strpos( $text, '[' ) !== false
2270 $text .= ']'; # so that replaceExternalLinks($text) works later
2271 $m[3] = substr( $m[3], 1 );
2273 # fix up urlencoded title texts
2274 if ( strpos( $m[1], '%' ) !== false ) {
2275 # Should anchors '#' also be rejected?
2276 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2279 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2280 # Invalid, but might be an image with a link in its caption
2281 $might_be_img = true;
2283 if ( strpos( $m[1], '%' ) !== false ) {
2284 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2287 } else { # Invalid form; output directly
2288 $s .= $prefix . '[[' . $line;
2292 $origLink = ltrim( $m[1], ' ' );
2294 # Don't allow internal links to pages containing
2295 # PROTO: where PROTO is a valid URL protocol; these
2296 # should be external links.
2297 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2298 $s .= $prefix . '[[' . $line;
2302 # Make subpage if necessary
2303 if ( $useSubpages ) {
2304 $link = $this->maybeDoSubpageLink( $origLink, $text );
2309 // \x7f isn't a default legal title char, so most likely strip
2310 // markers will force us into the "invalid form" path above. But,
2311 // just in case, let's assert that xmlish tags aren't valid in
2312 // the title position.
2313 $unstrip = $this->mStripState
->killMarkers( $link );
2314 $noMarkers = ( $unstrip === $link );
2316 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2317 if ( $nt === null ) {
2318 $s .= $prefix . '[[' . $line;
2322 $ns = $nt->getNamespace();
2323 $iw = $nt->getInterwiki();
2325 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2327 if ( $might_be_img ) { # if this is actually an invalid link
2328 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2331 # look at the next 'line' to see if we can close it there
2333 $next_line = $a->current();
2334 if ( $next_line === false ||
$next_line === null ) {
2337 $m = explode( ']]', $next_line, 3 );
2338 if ( count( $m ) == 3 ) {
2339 # the first ]] closes the inner link, the second the image
2341 $text .= "[[{$m[0]}]]{$m[1]}";
2344 } elseif ( count( $m ) == 2 ) {
2345 # if there's exactly one ]] that's fine, we'll keep looking
2346 $text .= "[[{$m[0]}]]{$m[1]}";
2348 # if $next_line is invalid too, we need look no further
2349 $text .= '[[' . $next_line;
2354 # we couldn't find the end of this imageLink, so output it raw
2355 # but don't ignore what might be perfectly normal links in the text we've examined
2356 $holders->merge( $this->replaceInternalLinks2( $text ) );
2357 $s .= "{$prefix}[[$link|$text";
2358 # note: no $trail, because without an end, there *is* no trail
2361 } else { # it's not an image, so output it raw
2362 $s .= "{$prefix}[[$link|$text";
2363 # note: no $trail, because without an end, there *is* no trail
2368 $wasblank = ( $text == '' );
2372 # Strip off leading ':'
2373 $text = substr( $text, 1 );
2376 # T6598 madness. Handle the quotes only if they come from the alternate part
2377 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2378 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2379 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2380 $text = $this->doQuotes( $text );
2383 # Link not escaped by : , create the various objects
2384 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2387 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2388 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2389 in_array( $iw, $this->siteConfig
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2392 # T26502: filter duplicates
2393 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2394 $this->mLangLinkLanguages
[$iw] = true;
2395 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2399 * Strip the whitespace interwiki links produce, see T10897
2401 $s = rtrim( $s . $prefix ) . $trail; # T175416
2405 if ( $ns == NS_FILE
) {
2406 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2408 # if no parameters were passed, $text
2409 # becomes something like "File:Foo.png",
2410 # which we don't want to pass on to the
2414 # recursively parse links inside the image caption
2415 # actually, this will parse them in any other parameters, too,
2416 # but it might be hard to fix that, and it doesn't matter ATM
2417 $text = $this->replaceExternalLinks( $text );
2418 $holders->merge( $this->replaceInternalLinks2( $text ) );
2420 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2421 $s .= $prefix . $this->armorLinks(
2422 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2425 } elseif ( $ns == NS_CATEGORY
) {
2427 * Strip the whitespace Category links produce, see T2087
2429 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2432 $sortkey = $this->getDefaultSort();
2436 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2437 $sortkey = str_replace( "\n", '', $sortkey );
2438 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2439 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2445 # Self-link checking. For some languages, variants of the title are checked in
2446 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2447 # for linking to a different variant.
2448 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2449 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2453 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2454 # @todo FIXME: Should do batch file existence checks, see comment below
2455 if ( $ns == NS_MEDIA
) {
2456 # Give extensions a chance to select the file revision for us
2459 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2460 [ $this, $nt, &$options, &$descQuery ] );
2461 # Fetch and register the file (file title may be different via hooks)
2462 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2463 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2464 $s .= $prefix . $this->armorLinks(
2465 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2469 # Some titles, such as valid special pages or files in foreign repos, should
2470 # be shown as bluelinks even though they're not included in the page table
2471 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2472 # batch file existence checks for NS_FILE and NS_MEDIA
2473 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2474 $this->mOutput
->addLink( $nt );
2475 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2477 # Links will be added to the output link list after checking
2478 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2485 * Render a forced-blue link inline; protect against double expansion of
2486 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2487 * Since this little disaster has to split off the trail text to avoid
2488 * breaking URLs in the following text without breaking trails on the
2489 * wiki links, it's been made into a horrible function.
2492 * @param string $text
2493 * @param string $trail
2494 * @param string $prefix
2495 * @return string HTML-wikitext mix oh yuck
2497 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2498 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2500 if ( $text == '' ) {
2501 $text = htmlspecialchars( $nt->getPrefixedText() );
2504 $link = $this->getLinkRenderer()->makeKnownLink(
2505 $nt, new HtmlArmor( "$prefix$text$inside" )
2508 return $this->armorLinks( $link ) . $trail;
2512 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2513 * going to go through further parsing steps before inline URL expansion.
2515 * Not needed quite as much as it used to be since free links are a bit
2516 * more sensible these days. But bracketed links are still an issue.
2518 * @param string $text More-or-less HTML
2519 * @return string Less-or-more HTML with NOPARSE bits
2521 public function armorLinks( $text ) {
2522 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2523 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2527 * Return true if subpage links should be expanded on this page.
2530 public function areSubpagesAllowed() {
2531 # Some namespaces don't allow subpages
2532 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2536 * Handle link to subpage if necessary
2538 * @param string $target The source of the link
2539 * @param string &$text The link text, modified as necessary
2540 * @return string The full name of the link
2543 public function maybeDoSubpageLink( $target, &$text ) {
2544 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2548 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2550 * @param string $text
2551 * @param bool $linestart Whether or not this is at the start of a line.
2553 * @return string The lists rendered as HTML
2555 public function doBlockLevels( $text, $linestart ) {
2556 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2560 * Return value of a magic variable (like PAGENAME)
2564 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2565 * @param bool|PPFrame $frame
2567 * @throws MWException
2570 public function getVariableValue( $index, $frame = false ) {
2571 if ( is_null( $this->mTitle
) ) {
2572 // If no title set, bad things are going to happen
2573 // later. Title should always be set since this
2574 // should only be called in the middle of a parse
2575 // operation (but the unit-tests do funky stuff)
2576 throw new MWException( __METHOD__
. ' Should only be '
2577 . ' called while parsing (no title set)' );
2580 // Avoid PHP 7.1 warning from passing $this by reference
2584 * Some of these require message or data lookups and can be
2585 * expensive to check many times.
2587 if ( Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] )
2588 && isset( $this->mVarCache
[$index] )
2590 return $this->mVarCache
[$index];
2593 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2594 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2596 // In miser mode, disable words that always cause double-parses on page save (T137900)
2597 static $slowRevWords = [ 'revisionid' => true ]; // @TODO: 'revisiontimestamp'
2599 isset( $slowRevWords[$index] ) &&
2600 $this->siteConfig
->get( 'MiserMode' ) &&
2601 !$this->mOptions
->getInterfaceMessage() &&
2602 // @TODO: disallow this word on all namespaces
2603 MWNamespace
::isContent( $this->mTitle
->getNamespace() )
2605 return $this->mRevisionId ?
'-' : '';
2608 $pageLang = $this->getFunctionLang();
2614 case 'currentmonth':
2615 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2617 case 'currentmonth1':
2618 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2620 case 'currentmonthname':
2621 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2623 case 'currentmonthnamegen':
2624 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2626 case 'currentmonthabbrev':
2627 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2630 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2633 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2636 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2639 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2641 case 'localmonthname':
2642 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2644 case 'localmonthnamegen':
2645 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2647 case 'localmonthabbrev':
2648 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2651 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2654 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2657 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2660 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2662 case 'fullpagename':
2663 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2665 case 'fullpagenamee':
2666 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2669 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2671 case 'subpagenamee':
2672 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2674 case 'rootpagename':
2675 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2677 case 'rootpagenamee':
2678 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2681 $this->mTitle
->getRootText()
2684 case 'basepagename':
2685 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2687 case 'basepagenamee':
2688 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2691 $this->mTitle
->getBaseText()
2694 case 'talkpagename':
2695 if ( $this->mTitle
->canHaveTalkPage() ) {
2696 $talkPage = $this->mTitle
->getTalkPage();
2697 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2702 case 'talkpagenamee':
2703 if ( $this->mTitle
->canHaveTalkPage() ) {
2704 $talkPage = $this->mTitle
->getTalkPage();
2705 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2710 case 'subjectpagename':
2711 $subjPage = $this->mTitle
->getSubjectPage();
2712 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2714 case 'subjectpagenamee':
2715 $subjPage = $this->mTitle
->getSubjectPage();
2716 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2718 case 'pageid': // requested in T25427
2719 $pageid = $this->getTitle()->getArticleID();
2720 if ( $pageid == 0 ) {
2721 # 0 means the page doesn't exist in the database,
2722 # which means the user is previewing a new page.
2723 # The vary-revision flag must be set, because the magic word
2724 # will have a different value once the page is saved.
2725 $this->mOutput
->setFlag( 'vary-revision' );
2726 wfDebug( __METHOD__
. ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2728 $value = $pageid ?
: null;
2731 # Let the edit saving system know we should parse the page
2732 # *after* a revision ID has been assigned.
2733 $this->mOutput
->setFlag( 'vary-revision-id' );
2734 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2735 $value = $this->mRevisionId
;
2738 $rev = $this->getRevisionObject();
2740 $value = $rev->getId();
2745 $value = $this->mOptions
->getSpeculativeRevId();
2747 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2752 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2754 case 'revisionday2':
2755 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2757 case 'revisionmonth':
2758 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2760 case 'revisionmonth1':
2761 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2763 case 'revisionyear':
2764 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2766 case 'revisiontimestamp':
2767 # Let the edit saving system know we should parse the page
2768 # *after* a revision ID has been assigned. This is for null edits.
2769 $this->mOutput
->setFlag( 'vary-revision' );
2770 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2771 $value = $this->getRevisionTimestamp();
2773 case 'revisionuser':
2774 # Let the edit saving system know we should parse the page
2775 # *after* a revision ID has been assigned for null edits.
2776 $this->mOutput
->setFlag( 'vary-user' );
2777 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-user...\n" );
2778 $value = $this->getRevisionUser();
2780 case 'revisionsize':
2781 $value = $this->getRevisionSize();
2784 $value = str_replace( '_', ' ',
2785 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2788 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2790 case 'namespacenumber':
2791 $value = $this->mTitle
->getNamespace();
2794 $value = $this->mTitle
->canHaveTalkPage()
2795 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2799 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2801 case 'subjectspace':
2802 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2804 case 'subjectspacee':
2805 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2807 case 'currentdayname':
2808 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2811 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2814 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2817 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2820 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2821 # int to remove the padding
2822 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2825 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2827 case 'localdayname':
2828 $value = $pageLang->getWeekdayName(
2829 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2833 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2836 $value = $pageLang->time(
2837 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2843 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2846 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2847 # int to remove the padding
2848 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2851 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2853 case 'numberofarticles':
2854 $value = $pageLang->formatNum( SiteStats
::articles() );
2856 case 'numberoffiles':
2857 $value = $pageLang->formatNum( SiteStats
::images() );
2859 case 'numberofusers':
2860 $value = $pageLang->formatNum( SiteStats
::users() );
2862 case 'numberofactiveusers':
2863 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2865 case 'numberofpages':
2866 $value = $pageLang->formatNum( SiteStats
::pages() );
2868 case 'numberofadmins':
2869 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2871 case 'numberofedits':
2872 $value = $pageLang->formatNum( SiteStats
::edits() );
2874 case 'currenttimestamp':
2875 $value = wfTimestamp( TS_MW
, $ts );
2877 case 'localtimestamp':
2878 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2880 case 'currentversion':
2881 $value = SpecialVersion
::getVersion();
2884 return $this->siteConfig
->get( 'ArticlePath' );
2886 return $this->siteConfig
->get( 'Sitename' );
2888 return $this->siteConfig
->get( 'Server' );
2890 return $this->siteConfig
->get( 'ServerName' );
2892 return $this->siteConfig
->get( 'ScriptPath' );
2894 return $this->siteConfig
->get( 'StylePath' );
2895 case 'directionmark':
2896 return $pageLang->getDirMark();
2897 case 'contentlanguage':
2898 return $this->siteConfig
->get( 'LanguageCode' );
2899 case 'pagelanguage':
2900 $value = $pageLang->getCode();
2902 case 'cascadingsources':
2903 $value = CoreParserFunctions
::cascadingsources( $this );
2908 'ParserGetVariableValueSwitch',
2909 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
2916 $this->mVarCache
[$index] = $value;
2925 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2926 * @param string $variable Parser variable name
2929 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2930 # Get the timezone-adjusted timestamp to be used for this revision
2931 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2932 # Possibly set vary-revision if there is not yet an associated revision
2933 if ( !$this->getRevisionObject() ) {
2934 # Get the timezone-adjusted timestamp $mtts seconds in the future
2936 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
2941 if ( $resNow !== $resThen ) {
2942 # Let the edit saving system know we should parse the page
2943 # *after* a revision ID has been assigned. This is for null edits.
2944 $this->mOutput
->setFlag( 'vary-revision' );
2945 wfDebug( __METHOD__
. ": $variable used, setting vary-revision...\n" );
2953 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2957 public function initialiseVariables() {
2958 $variableIDs = $this->magicWordFactory
->getVariableIDs();
2959 $substIDs = $this->magicWordFactory
->getSubstIDs();
2961 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
2962 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
2966 * Preprocess some wikitext and return the document tree.
2967 * This is the ghost of replace_variables().
2969 * @param string $text The text to parse
2970 * @param int $flags Bitwise combination of:
2971 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2972 * included. Default is to assume a direct page view.
2974 * The generated DOM tree must depend only on the input text and the flags.
2975 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2977 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2978 * change in the DOM tree for a given text, must be passed through the section identifier
2979 * in the section edit link and thus back to extractSections().
2981 * The output of this function is currently only cached in process memory, but a persistent
2982 * cache may be implemented at a later date which takes further advantage of these strict
2983 * dependency requirements.
2987 public function preprocessToDom( $text, $flags = 0 ) {
2988 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2993 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2999 public static function splitWhitespace( $s ) {
3000 $ltrimmed = ltrim( $s );
3001 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3002 $trimmed = rtrim( $ltrimmed );
3003 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3005 $w2 = substr( $ltrimmed, -$diff );
3009 return [ $w1, $trimmed, $w2 ];
3013 * Replace magic variables, templates, and template arguments
3014 * with the appropriate text. Templates are substituted recursively,
3015 * taking care to avoid infinite loops.
3017 * Note that the substitution depends on value of $mOutputType:
3018 * self::OT_WIKI: only {{subst:}} templates
3019 * self::OT_PREPROCESS: templates but not extension tags
3020 * self::OT_HTML: all templates and extension tags
3022 * @param string $text The text to transform
3023 * @param bool|PPFrame $frame Object describing the arguments passed to the
3024 * template. Arguments may also be provided as an associative array, as
3025 * was the usual case before MW1.12. Providing arguments this way may be
3026 * useful for extensions wishing to perform variable replacement
3028 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3029 * double-brace expansion.
3032 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3033 # Is there any text? Also, Prevent too big inclusions!
3034 $textSize = strlen( $text );
3035 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3039 if ( $frame === false ) {
3040 $frame = $this->getPreprocessor()->newFrame();
3041 } elseif ( !( $frame instanceof PPFrame
) ) {
3042 wfDebug( __METHOD__
. " called using plain parameters instead of "
3043 . "a PPFrame instance. Creating custom frame.\n" );
3044 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3047 $dom = $this->preprocessToDom( $text );
3048 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3049 $text = $frame->expand( $dom, $flags );
3055 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3057 * @param array $args
3061 public static function createAssocArgs( $args ) {
3064 foreach ( $args as $arg ) {
3065 $eqpos = strpos( $arg, '=' );
3066 if ( $eqpos === false ) {
3067 $assocArgs[$index++
] = $arg;
3069 $name = trim( substr( $arg, 0, $eqpos ) );
3070 $value = trim( substr( $arg, $eqpos +
1 ) );
3071 if ( $value === false ) {
3074 if ( $name !== false ) {
3075 $assocArgs[$name] = $value;
3084 * Warn the user when a parser limitation is reached
3085 * Will warn at most once the user per limitation type
3087 * The results are shown during preview and run through the Parser (See EditPage.php)
3089 * @param string $limitationType Should be one of:
3090 * 'expensive-parserfunction' (corresponding messages:
3091 * 'expensive-parserfunction-warning',
3092 * 'expensive-parserfunction-category')
3093 * 'post-expand-template-argument' (corresponding messages:
3094 * 'post-expand-template-argument-warning',
3095 * 'post-expand-template-argument-category')
3096 * 'post-expand-template-inclusion' (corresponding messages:
3097 * 'post-expand-template-inclusion-warning',
3098 * 'post-expand-template-inclusion-category')
3099 * 'node-count-exceeded' (corresponding messages:
3100 * 'node-count-exceeded-warning',
3101 * 'node-count-exceeded-category')
3102 * 'expansion-depth-exceeded' (corresponding messages:
3103 * 'expansion-depth-exceeded-warning',
3104 * 'expansion-depth-exceeded-category')
3105 * @param string|int|null $current Current value
3106 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3107 * exceeded, provide the values (optional)
3109 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3110 # does no harm if $current and $max are present but are unnecessary for the message
3111 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3112 # only during preview, and that would split the parser cache unnecessarily.
3113 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3115 $this->mOutput
->addWarning( $warning );
3116 $this->addTrackingCategory( "$limitationType-category" );
3120 * Return the text of a template, after recursively
3121 * replacing any variables or templates within the template.
3123 * @param array $piece The parts of the template
3124 * $piece['title']: the title, i.e. the part before the |
3125 * $piece['parts']: the parameter array
3126 * $piece['lineStart']: whether the brace was at the start of a line
3127 * @param PPFrame $frame The current frame, contains template arguments
3129 * @return string The text of the template
3131 public function braceSubstitution( $piece, $frame ) {
3134 // $text has been filled
3136 // wiki markup in $text should be escaped
3138 // $text is HTML, armour it against wikitext transformation
3140 // Force interwiki transclusion to be done in raw mode not rendered
3141 $forceRawInterwiki = false;
3142 // $text is a DOM node needing expansion in a child frame
3143 $isChildObj = false;
3144 // $text is a DOM node needing expansion in the current frame
3145 $isLocalObj = false;
3147 # Title object, where $text came from
3150 # $part1 is the bit before the first |, and must contain only title characters.
3151 # Various prefixes will be stripped from it later.
3152 $titleWithSpaces = $frame->expand( $piece['title'] );
3153 $part1 = trim( $titleWithSpaces );
3156 # Original title text preserved for various purposes
3157 $originalTitle = $part1;
3159 # $args is a list of argument nodes, starting from index 0, not including $part1
3160 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3161 # below won't work b/c this $args isn't an object
3162 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3164 $profileSection = null; // profile templates
3168 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3170 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3171 # Decide whether to expand template or keep wikitext as-is.
3172 if ( $this->ot
['wiki'] ) {
3173 if ( $substMatch === false ) {
3174 $literal = true; # literal when in PST with no prefix
3176 $literal = false; # expand when in PST with subst: or safesubst:
3179 if ( $substMatch == 'subst' ) {
3180 $literal = true; # literal when not in PST with plain subst:
3182 $literal = false; # expand when not in PST with safesubst: or no prefix
3186 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3193 if ( !$found && $args->getLength() == 0 ) {
3194 $id = $this->mVariables
->matchStartToEnd( $part1 );
3195 if ( $id !== false ) {
3196 $text = $this->getVariableValue( $id, $frame );
3197 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3198 $this->mOutput
->updateCacheExpiry(
3199 $this->magicWordFactory
->getCacheTTL( $id ) );
3205 # MSG, MSGNW and RAW
3208 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3209 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3212 # Remove obsolete MSG:
3213 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3214 $mwMsg->matchStartAndRemove( $part1 );
3218 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3219 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3220 $forceRawInterwiki = true;
3226 $colonPos = strpos( $part1, ':' );
3227 if ( $colonPos !== false ) {
3228 $func = substr( $part1, 0, $colonPos );
3229 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3230 $argsLength = $args->getLength();
3231 for ( $i = 0; $i < $argsLength; $i++
) {
3232 $funcArgs[] = $args->item( $i );
3235 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3237 // Extract any forwarded flags
3238 if ( isset( $result['title'] ) ) {
3239 $title = $result['title'];
3241 if ( isset( $result['found'] ) ) {
3242 $found = $result['found'];
3244 if ( array_key_exists( 'text', $result ) ) {
3246 $text = $result['text'];
3248 if ( isset( $result['nowiki'] ) ) {
3249 $nowiki = $result['nowiki'];
3251 if ( isset( $result['isHTML'] ) ) {
3252 $isHTML = $result['isHTML'];
3254 if ( isset( $result['forceRawInterwiki'] ) ) {
3255 $forceRawInterwiki = $result['forceRawInterwiki'];
3257 if ( isset( $result['isChildObj'] ) ) {
3258 $isChildObj = $result['isChildObj'];
3260 if ( isset( $result['isLocalObj'] ) ) {
3261 $isLocalObj = $result['isLocalObj'];
3266 # Finish mangling title and then check for loops.
3267 # Set $title to a Title object and $titleText to the PDBK
3270 # Split the title into page and subpage
3272 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3273 if ( $part1 !== $relative ) {
3275 $ns = $this->mTitle
->getNamespace();
3277 $title = Title
::newFromText( $part1, $ns );
3279 $titleText = $title->getPrefixedText();
3280 # Check for language variants if the template is not found
3281 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3282 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3284 # Do recursion depth check
3285 $limit = $this->mOptions
->getMaxTemplateDepth();
3286 if ( $frame->depth
>= $limit ) {
3288 $text = '<span class="error">'
3289 . wfMessage( 'parser-template-recursion-depth-warning' )
3290 ->numParams( $limit )->inContentLanguage()->text()
3296 # Load from database
3297 if ( !$found && $title ) {
3298 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3299 if ( !$title->isExternal() ) {
3300 if ( $title->isSpecialPage()
3301 && $this->mOptions
->getAllowSpecialInclusion()
3302 && $this->ot
['html']
3304 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3305 // Pass the template arguments as URL parameters.
3306 // "uselang" will have no effect since the Language object
3307 // is forced to the one defined in ParserOptions.
3309 $argsLength = $args->getLength();
3310 for ( $i = 0; $i < $argsLength; $i++
) {
3311 $bits = $args->item( $i )->splitArg();
3312 if ( strval( $bits['index'] ) === '' ) {
3313 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3314 $value = trim( $frame->expand( $bits['value'] ) );
3315 $pageArgs[$name] = $value;
3319 // Create a new context to execute the special page
3320 $context = new RequestContext
;
3321 $context->setTitle( $title );
3322 $context->setRequest( new FauxRequest( $pageArgs ) );
3323 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3324 $context->setUser( $this->getUser() );
3326 // If this page is cached, then we better not be per user.
3327 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3329 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3330 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3332 $text = $context->getOutput()->getHTML();
3333 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3336 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3337 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3338 $specialPage->maxIncludeCacheTime()
3342 } elseif ( MWNamespace
::isNonincludable( $title->getNamespace() ) ) {
3343 $found = false; # access denied
3344 wfDebug( __METHOD__
. ": template inclusion denied for " .
3345 $title->getPrefixedDBkey() . "\n" );
3347 list( $text, $title ) = $this->getTemplateDom( $title );
3348 if ( $text !== false ) {
3354 # If the title is valid but undisplayable, make a link to it
3355 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3356 $text = "[[:$titleText]]";
3359 } elseif ( $title->isTrans() ) {
3360 # Interwiki transclusion
3361 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3362 $text = $this->interwikiTransclude( $title, 'render' );
3365 $text = $this->interwikiTransclude( $title, 'raw' );
3366 # Preprocess it like a template
3367 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3373 # Do infinite loop check
3374 # This has to be done after redirect resolution to avoid infinite loops via redirects
3375 if ( !$frame->loopCheck( $title ) ) {
3377 $text = '<span class="error">'
3378 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3380 $this->addTrackingCategory( 'template-loop-category' );
3381 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3382 wfEscapeWikiText( $titleText ) )->text() );
3383 wfDebug( __METHOD__
. ": template loop broken at '$titleText'\n" );
3387 # If we haven't found text to substitute by now, we're done
3388 # Recover the source wikitext and return it
3390 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3391 if ( $profileSection ) {
3392 $this->mProfiler
->scopedProfileOut( $profileSection );
3394 return [ 'object' => $text ];
3397 # Expand DOM-style return values in a child frame
3398 if ( $isChildObj ) {
3399 # Clean up argument array
3400 $newFrame = $frame->newChild( $args, $title );
3403 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3404 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3405 # Expansion is eligible for the empty-frame cache
3406 $text = $newFrame->cachedExpand( $titleText, $text );
3408 # Uncached expansion
3409 $text = $newFrame->expand( $text );
3412 if ( $isLocalObj && $nowiki ) {
3413 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3414 $isLocalObj = false;
3417 if ( $profileSection ) {
3418 $this->mProfiler
->scopedProfileOut( $profileSection );
3421 # Replace raw HTML by a placeholder
3423 $text = $this->insertStripItem( $text );
3424 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3425 # Escape nowiki-style return values
3426 $text = wfEscapeWikiText( $text );
3427 } elseif ( is_string( $text )
3428 && !$piece['lineStart']
3429 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3431 # T2529: if the template begins with a table or block-level
3432 # element, it should be treated as beginning a new line.
3433 # This behavior is somewhat controversial.
3434 $text = "\n" . $text;
3437 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3438 # Error, oversize inclusion
3439 if ( $titleText !== false ) {
3440 # Make a working, properly escaped link if possible (T25588)
3441 $text = "[[:$titleText]]";
3443 # This will probably not be a working link, but at least it may
3444 # provide some hint of where the problem is
3445 preg_replace( '/^:/', '', $originalTitle );
3446 $text = "[[:$originalTitle]]";
3448 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3449 . 'post-expand include size too large -->' );
3450 $this->limitationWarn( 'post-expand-template-inclusion' );
3453 if ( $isLocalObj ) {
3454 $ret = [ 'object' => $text ];
3456 $ret = [ 'text' => $text ];
3463 * Call a parser function and return an array with text and flags.
3465 * The returned array will always contain a boolean 'found', indicating
3466 * whether the parser function was found or not. It may also contain the
3468 * text: string|object, resulting wikitext or PP DOM object
3469 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3470 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3471 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3472 * nowiki: bool, wiki markup in $text should be escaped
3475 * @param PPFrame $frame The current frame, contains template arguments
3476 * @param string $function Function name
3477 * @param array $args Arguments to the function
3478 * @throws MWException
3481 public function callParserFunction( $frame, $function, array $args = [] ) {
3482 # Case sensitive functions
3483 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3484 $function = $this->mFunctionSynonyms
[1][$function];
3486 # Case insensitive functions
3487 $function = $this->contLang
->lc( $function );
3488 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3489 $function = $this->mFunctionSynonyms
[0][$function];
3491 return [ 'found' => false ];
3495 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3497 // Avoid PHP 7.1 warning from passing $this by reference
3500 $allArgs = [ &$parser ];
3501 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3502 # Convert arguments to PPNodes and collect for appending to $allArgs
3504 foreach ( $args as $k => $v ) {
3505 if ( $v instanceof PPNode ||
$k === 0 ) {
3508 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3512 # Add a frame parameter, and pass the arguments as an array
3513 $allArgs[] = $frame;
3514 $allArgs[] = $funcArgs;
3516 # Convert arguments to plain text and append to $allArgs
3517 foreach ( $args as $k => $v ) {
3518 if ( $v instanceof PPNode
) {
3519 $allArgs[] = trim( $frame->expand( $v ) );
3520 } elseif ( is_int( $k ) && $k >= 0 ) {
3521 $allArgs[] = trim( $v );
3523 $allArgs[] = trim( "$k=$v" );
3528 $result = $callback( ...$allArgs );
3530 # The interface for function hooks allows them to return a wikitext
3531 # string or an array containing the string and any flags. This mungs
3532 # things around to match what this method should return.
3533 if ( !is_array( $result ) ) {
3539 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3540 $result['text'] = $result[0];
3542 unset( $result[0] );
3549 $preprocessFlags = 0;
3550 if ( isset( $result['noparse'] ) ) {
3551 $noparse = $result['noparse'];
3553 if ( isset( $result['preprocessFlags'] ) ) {
3554 $preprocessFlags = $result['preprocessFlags'];
3558 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3559 $result['isChildObj'] = true;
3566 * Get the semi-parsed DOM representation of a template with a given title,
3567 * and its redirect destination title. Cached.
3569 * @param Title $title
3573 public function getTemplateDom( $title ) {
3574 $cacheTitle = $title;
3575 $titleText = $title->getPrefixedDBkey();
3577 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3578 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3579 $title = Title
::makeTitle( $ns, $dbk );
3580 $titleText = $title->getPrefixedDBkey();
3582 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3583 return [ $this->mTplDomCache
[$titleText], $title ];
3586 # Cache miss, go to the database
3587 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3589 if ( $text === false ) {
3590 $this->mTplDomCache
[$titleText] = false;
3591 return [ false, $title ];
3594 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3595 $this->mTplDomCache
[$titleText] = $dom;
3597 if ( !$title->equals( $cacheTitle ) ) {
3598 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3599 [ $title->getNamespace(), $title->getDBkey() ];
3602 return [ $dom, $title ];
3606 * Fetch the current revision of a given title. Note that the revision
3607 * (and even the title) may not exist in the database, so everything
3608 * contributing to the output of the parser should use this method
3609 * where possible, rather than getting the revisions themselves. This
3610 * method also caches its results, so using it benefits performance.
3613 * @param Title $title
3616 public function fetchCurrentRevisionOfTitle( $title ) {
3617 $cacheKey = $title->getPrefixedDBkey();
3618 if ( !$this->currentRevisionCache
) {
3619 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3621 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3622 $this->currentRevisionCache
->set( $cacheKey,
3623 // Defaults to Parser::statelessFetchRevision()
3624 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3627 return $this->currentRevisionCache
->get( $cacheKey );
3631 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3632 * without passing them on to it.
3635 * @param Title $title
3636 * @param Parser|bool $parser
3637 * @return Revision|bool False if missing
3639 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3640 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3646 * Fetch the unparsed text of a template and register a reference to it.
3647 * @param Title $title
3648 * @return array ( string or false, Title )
3650 public function fetchTemplateAndTitle( $title ) {
3651 // Defaults to Parser::statelessFetchTemplate()
3652 $templateCb = $this->mOptions
->getTemplateCallback();
3653 $stuff = call_user_func( $templateCb, $title, $this );
3654 // We use U+007F DELETE to distinguish strip markers from regular text.
3655 $text = $stuff['text'];
3656 if ( is_string( $stuff['text'] ) ) {
3657 $text = strtr( $text, "\x7f", "?" );
3659 $finalTitle = $stuff['finalTitle'] ??
$title;
3660 if ( isset( $stuff['deps'] ) ) {
3661 foreach ( $stuff['deps'] as $dep ) {
3662 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3663 if ( $dep['title']->equals( $this->getTitle() ) ) {
3664 // If we transclude ourselves, the final result
3665 // will change based on the new version of the page
3666 $this->mOutput
->setFlag( 'vary-revision' );
3670 return [ $text, $finalTitle ];
3674 * Fetch the unparsed text of a template and register a reference to it.
3675 * @param Title $title
3676 * @return string|bool
3678 public function fetchTemplate( $title ) {
3679 return $this->fetchTemplateAndTitle( $title )[0];
3683 * Static function to get a template
3684 * Can be overridden via ParserOptions::setTemplateCallback().
3686 * @param Title $title
3687 * @param bool|Parser $parser
3691 public static function statelessFetchTemplate( $title, $parser = false ) {
3692 $text = $skip = false;
3693 $finalTitle = $title;
3696 # Loop to fetch the article, with up to 1 redirect
3697 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3698 # Give extensions a chance to select the revision instead
3699 $id = false; # Assume current
3700 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3701 [ $parser, $title, &$skip, &$id ] );
3707 'page_id' => $title->getArticleID(),
3714 $rev = Revision
::newFromId( $id );
3715 } elseif ( $parser ) {
3716 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3718 $rev = Revision
::newFromTitle( $title );
3720 $rev_id = $rev ?
$rev->getId() : 0;
3721 # If there is no current revision, there is no page
3722 if ( $id === false && !$rev ) {
3723 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3724 $linkCache->addBadLinkObj( $title );
3729 'page_id' => $title->getArticleID(),
3730 'rev_id' => $rev_id ];
3731 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3732 # We fetched a rev from a different title; register it too...
3734 'title' => $rev->getTitle(),
3735 'page_id' => $rev->getPage(),
3736 'rev_id' => $rev_id ];
3740 $content = $rev->getContent();
3741 $text = $content ?
$content->getWikitextForTransclusion() : null;
3743 Hooks
::run( 'ParserFetchTemplate',
3744 [ $parser, $title, $rev, &$text, &$deps ] );
3746 if ( $text === false ||
$text === null ) {
3750 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3751 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3752 lcfirst( $title->getText() ) )->inContentLanguage();
3753 if ( !$message->exists() ) {
3757 $content = $message->content();
3758 $text = $message->plain();
3766 $finalTitle = $title;
3767 $title = $content->getRedirectTarget();
3771 'finalTitle' => $finalTitle,
3776 * Fetch a file and its title and register a reference to it.
3777 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3778 * @param Title $title
3779 * @param array $options Array of options to RepoGroup::findFile
3781 * @deprecated since 1.32, use fetchFileAndTitle instead
3783 public function fetchFile( $title, $options = [] ) {
3784 wfDeprecated( __METHOD__
, '1.32' );
3785 return $this->fetchFileAndTitle( $title, $options )[0];
3789 * Fetch a file and its title and register a reference to it.
3790 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3791 * @param Title $title
3792 * @param array $options Array of options to RepoGroup::findFile
3793 * @return array ( File or false, Title of file )
3795 public function fetchFileAndTitle( $title, $options = [] ) {
3796 $file = $this->fetchFileNoRegister( $title, $options );
3798 $time = $file ?
$file->getTimestamp() : false;
3799 $sha1 = $file ?
$file->getSha1() : false;
3800 # Register the file as a dependency...
3801 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3802 if ( $file && !$title->equals( $file->getTitle() ) ) {
3803 # Update fetched file title
3804 $title = $file->getTitle();
3805 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3807 return [ $file, $title ];
3811 * Helper function for fetchFileAndTitle.
3813 * Also useful if you need to fetch a file but not use it yet,
3814 * for example to get the file's handler.
3816 * @param Title $title
3817 * @param array $options Array of options to RepoGroup::findFile
3820 protected function fetchFileNoRegister( $title, $options = [] ) {
3821 if ( isset( $options['broken'] ) ) {
3822 $file = false; // broken thumbnail forced by hook
3823 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3824 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3825 } else { // get by (name,timestamp)
3826 $file = wfFindFile( $title, $options );
3832 * Transclude an interwiki link.
3834 * @param Title $title
3835 * @param string $action Usually one of (raw, render)
3839 public function interwikiTransclude( $title, $action ) {
3840 if ( !$this->siteConfig
->get( 'EnableScaryTranscluding' ) ) {
3841 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3844 $url = $title->getFullURL( [ 'action' => $action ] );
3845 if ( strlen( $url ) > 1024 ) {
3846 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3849 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3851 $fname = __METHOD__
;
3852 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3854 $data = $cache->getWithSetCallback(
3855 $cache->makeGlobalKey(
3856 'interwiki-transclude',
3857 ( $wikiId !== false ) ?
$wikiId : 'external',
3860 $this->siteConfig
->get( 'TranscludeCacheExpiry' ),
3861 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3862 $req = MWHttpRequest
::factory( $url, [], $fname );
3864 $status = $req->execute(); // Status object
3865 if ( !$status->isOK() ) {
3866 $ttl = $cache::TTL_UNCACHEABLE
;
3867 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3868 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3872 'text' => $status->isOK() ?
$req->getContent() : null,
3873 'code' => $req->getStatus()
3877 'checkKeys' => ( $wikiId !== false )
3878 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3880 'pcGroup' => 'interwiki-transclude:5',
3881 'pcTTL' => $cache::TTL_PROC_LONG
3885 if ( is_string( $data['text'] ) ) {
3886 $text = $data['text'];
3887 } elseif ( $data['code'] != 200 ) {
3888 // Though we failed to fetch the content, this status is useless.
3889 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3890 ->params( $url, $data['code'] )->inContentLanguage()->text();
3892 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3899 * Triple brace replacement -- used for template arguments
3902 * @param array $piece
3903 * @param PPFrame $frame
3907 public function argSubstitution( $piece, $frame ) {
3909 $parts = $piece['parts'];
3910 $nameWithSpaces = $frame->expand( $piece['title'] );
3911 $argName = trim( $nameWithSpaces );
3913 $text = $frame->getArgument( $argName );
3914 if ( $text === false && $parts->getLength() > 0
3915 && ( $this->ot
['html']
3917 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3920 # No match in frame, use the supplied default
3921 $object = $parts->item( 0 )->getChildren();
3923 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3924 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3925 $this->limitationWarn( 'post-expand-template-argument' );
3928 if ( $text === false && $object === false ) {
3930 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3932 if ( $error !== false ) {
3935 if ( $object !== false ) {
3936 $ret = [ 'object' => $object ];
3938 $ret = [ 'text' => $text ];
3945 * Return the text to be used for a given extension tag.
3946 * This is the ghost of strip().
3948 * @param array $params Associative array of parameters:
3949 * name PPNode for the tag name
3950 * attr PPNode for unparsed text where tag attributes are thought to be
3951 * attributes Optional associative array of parsed attributes
3952 * inner Contents of extension element
3953 * noClose Original text did not have a close tag
3954 * @param PPFrame $frame
3956 * @throws MWException
3959 public function extensionSubstitution( $params, $frame ) {
3960 static $errorStr = '<span class="error">';
3961 static $errorLen = 20;
3963 $name = $frame->expand( $params['name'] );
3964 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3965 // Probably expansion depth or node count exceeded. Just punt the
3970 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3971 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3976 // We can't safely check if the expansion for $content resulted in an
3977 // error, because the content could happen to be the error string
3979 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3981 $marker = self
::MARKER_PREFIX
. "-$name-"
3982 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3984 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
3985 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3986 if ( $isFunctionTag ) {
3987 $markerType = 'none';
3989 $markerType = 'general';
3991 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3992 $name = strtolower( $name );
3993 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3994 if ( isset( $params['attributes'] ) ) {
3995 $attributes +
= $params['attributes'];
3998 if ( isset( $this->mTagHooks
[$name] ) ) {
3999 $output = call_user_func_array( $this->mTagHooks
[$name],
4000 [ $content, $attributes, $this, $frame ] );
4001 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4002 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4004 // Avoid PHP 7.1 warning from passing $this by reference
4006 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4008 $output = '<span class="error">Invalid tag extension name: ' .
4009 htmlspecialchars( $name ) . '</span>';
4012 if ( is_array( $output ) ) {
4015 $output = $flags[0];
4016 if ( isset( $flags['markerType'] ) ) {
4017 $markerType = $flags['markerType'];
4021 if ( is_null( $attrText ) ) {
4024 if ( isset( $params['attributes'] ) ) {
4025 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4026 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4027 htmlspecialchars( $attrValue ) . '"';
4030 if ( $content === null ) {
4031 $output = "<$name$attrText/>";
4033 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4034 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4038 $output = "<$name$attrText>$content$close";
4042 if ( $markerType === 'none' ) {
4044 } elseif ( $markerType === 'nowiki' ) {
4045 $this->mStripState
->addNoWiki( $marker, $output );
4046 } elseif ( $markerType === 'general' ) {
4047 $this->mStripState
->addGeneral( $marker, $output );
4049 throw new MWException( __METHOD__
. ': invalid marker type' );
4055 * Increment an include size counter
4057 * @param string $type The type of expansion
4058 * @param int $size The size of the text
4059 * @return bool False if this inclusion would take it over the maximum, true otherwise
4061 public function incrementIncludeSize( $type, $size ) {
4062 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4065 $this->mIncludeSizes
[$type] +
= $size;
4071 * Increment the expensive function count
4073 * @return bool False if the limit has been exceeded
4075 public function incrementExpensiveFunctionCount() {
4076 $this->mExpensiveFunctionCount++
;
4077 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4081 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4082 * Fills $this->mDoubleUnderscores, returns the modified text
4084 * @param string $text
4088 public function doDoubleUnderscore( $text ) {
4089 # The position of __TOC__ needs to be recorded
4090 $mw = $this->magicWordFactory
->get( 'toc' );
4091 if ( $mw->match( $text ) ) {
4092 $this->mShowToc
= true;
4093 $this->mForceTocPosition
= true;
4095 # Set a placeholder. At the end we'll fill it in with the TOC.
4096 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4098 # Only keep the first one.
4099 $text = $mw->replace( '', $text );
4102 # Now match and remove the rest of them
4103 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4104 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4106 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4107 $this->mOutput
->mNoGallery
= true;
4109 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4110 $this->mShowToc
= false;
4112 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4113 && $this->mTitle
->getNamespace() == NS_CATEGORY
4115 $this->addTrackingCategory( 'hidden-category-category' );
4117 # (T10068) Allow control over whether robots index a page.
4118 # __INDEX__ always overrides __NOINDEX__, see T16899
4119 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4120 $this->mOutput
->setIndexPolicy( 'noindex' );
4121 $this->addTrackingCategory( 'noindex-category' );
4123 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4124 $this->mOutput
->setIndexPolicy( 'index' );
4125 $this->addTrackingCategory( 'index-category' );
4128 # Cache all double underscores in the database
4129 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4130 $this->mOutput
->setProperty( $key, '' );
4137 * @see ParserOutput::addTrackingCategory()
4138 * @param string $msg Message key
4139 * @return bool Whether the addition was successful
4141 public function addTrackingCategory( $msg ) {
4142 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4146 * This function accomplishes several tasks:
4147 * 1) Auto-number headings if that option is enabled
4148 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4149 * 3) Add a Table of contents on the top for users who have enabled the option
4150 * 4) Auto-anchor headings
4152 * It loops through all headlines, collects the necessary data, then splits up the
4153 * string and re-inserts the newly formatted headlines.
4155 * @param string $text
4156 * @param string $origText Original, untouched wikitext
4157 * @param bool $isMain
4158 * @return mixed|string
4161 public function formatHeadings( $text, $origText, $isMain = true ) {
4162 # Inhibit editsection links if requested in the page
4163 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4164 $maybeShowEditLink = false;
4166 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4169 # Get all headlines for numbering them and adding funky stuff like [edit]
4170 # links - this is for later, but we need the number of headlines right now
4171 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4172 # be trimmed here since whitespace in HTML headings is significant.
4174 $numMatches = preg_match_all(
4175 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4180 # if there are fewer than 4 headlines in the article, do not show TOC
4181 # unless it's been explicitly enabled.
4182 $enoughToc = $this->mShowToc
&&
4183 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4185 # Allow user to stipulate that a page should have a "new section"
4186 # link added via __NEWSECTIONLINK__
4187 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4188 $this->mOutput
->setNewSection( true );
4191 # Allow user to remove the "new section"
4192 # link via __NONEWSECTIONLINK__
4193 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4194 $this->mOutput
->hideNewSection( true );
4197 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4198 # override above conditions and always show TOC above first header
4199 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4200 $this->mShowToc
= true;
4208 # Ugh .. the TOC should have neat indentation levels which can be
4209 # passed to the skin functions. These are determined here
4213 $sublevelCount = [];
4219 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4220 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4221 $oldType = $this->mOutputType
;
4222 $this->setOutputType( self
::OT_WIKI
);
4223 $frame = $this->getPreprocessor()->newFrame();
4224 $root = $this->preprocessToDom( $origText );
4225 $node = $root->getFirstChild();
4230 $headlines = $numMatches !== false ?
$matches[3] : [];
4232 $maxTocLevel = $this->siteConfig
->get( 'MaxTocLevel' );
4233 foreach ( $headlines as $headline ) {
4234 $isTemplate = false;
4236 $sectionIndex = false;
4238 $markerMatches = [];
4239 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4240 $serial = $markerMatches[1];
4241 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4242 $isTemplate = ( $titleText != $baseTitleText );
4243 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4247 $prevlevel = $level;
4249 $level = $matches[1][$headlineCount];
4251 if ( $level > $prevlevel ) {
4252 # Increase TOC level
4254 $sublevelCount[$toclevel] = 0;
4255 if ( $toclevel < $maxTocLevel ) {
4256 $prevtoclevel = $toclevel;
4257 $toc .= Linker
::tocIndent();
4260 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4261 # Decrease TOC level, find level to jump to
4263 for ( $i = $toclevel; $i > 0; $i-- ) {
4264 if ( $levelCount[$i] == $level ) {
4265 # Found last matching level
4268 } elseif ( $levelCount[$i] < $level ) {
4269 # Found first matching level below current level
4277 if ( $toclevel < $maxTocLevel ) {
4278 if ( $prevtoclevel < $maxTocLevel ) {
4279 # Unindent only if the previous toc level was shown :p
4280 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4281 $prevtoclevel = $toclevel;
4283 $toc .= Linker
::tocLineEnd();
4287 # No change in level, end TOC line
4288 if ( $toclevel < $maxTocLevel ) {
4289 $toc .= Linker
::tocLineEnd();
4293 $levelCount[$toclevel] = $level;
4295 # count number of headlines for each level
4296 $sublevelCount[$toclevel]++
;
4298 for ( $i = 1; $i <= $toclevel; $i++
) {
4299 if ( !empty( $sublevelCount[$i] ) ) {
4303 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4308 # The safe header is a version of the header text safe to use for links
4310 # Remove link placeholders by the link text.
4311 # <!--LINK number-->
4313 # link text with suffix
4314 # Do this before unstrip since link text can contain strip markers
4315 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4317 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4318 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4320 # Remove any <style> or <script> tags (T198618)
4321 $safeHeadline = preg_replace(
4322 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4327 # Strip out HTML (first regex removes any tag not allowed)
4329 # * <sup> and <sub> (T10393)
4333 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4334 # * <s> and <strike> (T35715)
4335 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4336 # to allow setting directionality in toc items.
4337 $tocline = preg_replace(
4339 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4340 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4346 # Strip '<span></span>', which is the result from the above if
4347 # <span id="foo"></span> is used to produce an additional anchor
4349 $tocline = str_replace( '<span></span>', '', $tocline );
4351 $tocline = trim( $tocline );
4353 # For the anchor, strip out HTML-y stuff period
4354 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4355 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4357 # Save headline for section edit hint before it's escaped
4358 $headlineHint = $safeHeadline;
4360 # Decode HTML entities
4361 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4363 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4365 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4366 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4367 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4368 if ( $fallbackHeadline === $safeHeadline ) {
4369 # No reason to have both (in fact, we can't)
4370 $fallbackHeadline = false;
4373 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4374 # @todo FIXME: We may be changing them depending on the current locale.
4375 $arrayKey = strtolower( $safeHeadline );
4376 if ( $fallbackHeadline === false ) {
4377 $fallbackArrayKey = false;
4379 $fallbackArrayKey = strtolower( $fallbackHeadline );
4382 # Create the anchor for linking from the TOC to the section
4383 $anchor = $safeHeadline;
4384 $fallbackAnchor = $fallbackHeadline;
4385 if ( isset( $refers[$arrayKey] ) ) {
4386 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4387 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4389 $linkAnchor .= "_$i";
4390 $refers["${arrayKey}_$i"] = true;
4392 $refers[$arrayKey] = true;
4394 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4395 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4396 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4397 $fallbackAnchor .= "_$i";
4398 $refers["${fallbackArrayKey}_$i"] = true;
4400 $refers[$fallbackArrayKey] = true;
4403 # Don't number the heading if it is the only one (looks silly)
4404 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4405 # the two are different if the line contains a link
4406 $headline = Html
::element(
4408 [ 'class' => 'mw-headline-number' ],
4410 ) . ' ' . $headline;
4413 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4414 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4415 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4418 # Add the section to the section tree
4419 # Find the DOM node for this header
4420 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4421 while ( $node && !$noOffset ) {
4422 if ( $node->getName() === 'h' ) {
4423 $bits = $node->splitHeading();
4424 if ( $bits['i'] == $sectionIndex ) {
4428 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4429 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4430 $node = $node->getNextSibling();
4433 'toclevel' => $toclevel,
4436 'number' => $numbering,
4437 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4438 'fromtitle' => $titleText,
4439 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4440 'anchor' => $anchor,
4443 # give headline the correct <h#> tag
4444 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4445 // Output edit section links as markers with styles that can be customized by skins
4446 if ( $isTemplate ) {
4447 # Put a T flag in the section identifier, to indicate to extractSections()
4448 # that sections inside <includeonly> should be counted.
4449 $editsectionPage = $titleText;
4450 $editsectionSection = "T-$sectionIndex";
4451 $editsectionContent = null;
4453 $editsectionPage = $this->mTitle
->getPrefixedText();
4454 $editsectionSection = $sectionIndex;
4455 $editsectionContent = $headlineHint;
4457 // We use a bit of pesudo-xml for editsection markers. The
4458 // language converter is run later on. Using a UNIQ style marker
4459 // leads to the converter screwing up the tokens when it
4460 // converts stuff. And trying to insert strip tags fails too. At
4461 // this point all real inputted tags have already been escaped,
4462 // so we don't have to worry about a user trying to input one of
4463 // these markers directly. We use a page and section attribute
4464 // to stop the language converter from converting these
4465 // important bits of data, but put the headline hint inside a
4466 // content block because the language converter is supposed to
4467 // be able to convert that piece of data.
4468 // Gets replaced with html in ParserOutput::getText
4469 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4470 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4471 if ( $editsectionContent !== null ) {
4472 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4479 $head[$headlineCount] = Linker
::makeHeadline( $level,
4480 $matches['attrib'][$headlineCount], $anchor, $headline,
4481 $editlink, $fallbackAnchor );
4486 $this->setOutputType( $oldType );
4488 # Never ever show TOC if no headers
4489 if ( $numVisible < 1 ) {
4494 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4495 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4497 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4498 $this->mOutput
->setTOCHTML( $toc );
4499 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4503 $this->mOutput
->setSections( $tocraw );
4506 # split up and insert constructed headlines
4507 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4510 // build an array of document sections
4512 foreach ( $blocks as $block ) {
4513 // $head is zero-based, sections aren't.
4514 if ( empty( $head[$i - 1] ) ) {
4515 $sections[$i] = $block;
4517 $sections[$i] = $head[$i - 1] . $block;
4521 * Send a hook, one per section.
4522 * The idea here is to be able to make section-level DIVs, but to do so in a
4523 * lower-impact, more correct way than r50769
4526 * $section : the section number
4527 * &$sectionContent : ref to the content of the section
4528 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4530 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4535 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4536 // append the TOC at the beginning
4537 // Top anchor now in skin
4538 $sections[0] .= $toc . "\n";
4541 $full .= implode( '', $sections );
4543 if ( $this->mForceTocPosition
) {
4544 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4551 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4552 * conversion, substituting signatures, {{subst:}} templates, etc.
4554 * @param string $text The text to transform
4555 * @param Title $title The Title object for the current article
4556 * @param User $user The User object describing the current user
4557 * @param ParserOptions $options Parsing options
4558 * @param bool $clearState Whether to clear the parser state first
4559 * @return string The altered wiki markup
4561 public function preSaveTransform( $text, Title
$title, User
$user,
4562 ParserOptions
$options, $clearState = true
4564 if ( $clearState ) {
4565 $magicScopeVariable = $this->lock();
4567 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4568 $this->setUser( $user );
4570 // Strip U+0000 NULL (T159174)
4571 $text = str_replace( "\000", '', $text );
4573 // We still normalize line endings for backwards-compatibility
4574 // with other code that just calls PST, but this should already
4575 // be handled in TextContent subclasses
4576 $text = TextContent
::normalizeLineEndings( $text );
4578 if ( $options->getPreSaveTransform() ) {
4579 $text = $this->pstPass2( $text, $user );
4581 $text = $this->mStripState
->unstripBoth( $text );
4583 $this->setUser( null ); # Reset
4589 * Pre-save transform helper function
4591 * @param string $text
4596 private function pstPass2( $text, $user ) {
4597 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4598 # $this->contLang here in order to give everyone the same signature and use the default one
4599 # rather than the one selected in each user's preferences. (see also T14815)
4600 $ts = $this->mOptions
->getTimestamp();
4601 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4602 $ts = $timestamp->format( 'YmdHis' );
4603 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4605 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4607 # Variable replacement
4608 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4609 $text = $this->replaceVariables( $text );
4611 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4612 # which may corrupt this parser instance via its wfMessage()->text() call-
4615 if ( strpos( $text, '~~~' ) !== false ) {
4616 $sigText = $this->getUserSig( $user );
4617 $text = strtr( $text, [
4619 '~~~~' => "$sigText $d",
4622 # The main two signature forms used above are time-sensitive
4623 $this->mOutput
->setFlag( 'user-signature' );
4626 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4627 $tc = '[' . Title
::legalChars() . ']';
4628 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4630 // [[ns:page (context)|]]
4631 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4632 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4633 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4634 // [[ns:page (context), context|]] (using either single or double-width comma)
4635 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4636 // [[|page]] (reverse pipe trick: add context from page title)
4637 $p2 = "/\[\[\\|($tc+)]]/";
4639 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4640 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4641 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4642 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4644 $t = $this->mTitle
->getText();
4646 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4647 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4648 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4649 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4651 # if there's no context, don't bother duplicating the title
4652 $text = preg_replace( $p2, '[[\\1]]', $text );
4659 * Fetch the user's signature text, if any, and normalize to
4660 * validated, ready-to-insert wikitext.
4661 * If you have pre-fetched the nickname or the fancySig option, you can
4662 * specify them here to save a database query.
4663 * Do not reuse this parser instance after calling getUserSig(),
4664 * as it may have changed if it's the $wgParser.
4666 * @param User &$user
4667 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4668 * @param bool|null $fancySig whether the nicknname is the complete signature
4669 * or null to use default value
4672 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4673 $username = $user->getName();
4675 # If not given, retrieve from the user object.
4676 if ( $nickname === false ) {
4677 $nickname = $user->getOption( 'nickname' );
4680 if ( is_null( $fancySig ) ) {
4681 $fancySig = $user->getBoolOption( 'fancysig' );
4684 $nickname = $nickname == null ?
$username : $nickname;
4686 if ( mb_strlen( $nickname ) > $this->siteConfig
->get( 'MaxSigChars' ) ) {
4687 $nickname = $username;
4688 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4689 } elseif ( $fancySig !== false ) {
4690 # Sig. might contain markup; validate this
4691 if ( $this->validateSig( $nickname ) !== false ) {
4692 # Validated; clean up (if needed) and return it
4693 return $this->cleanSig( $nickname, true );
4695 # Failed to validate; fall back to the default
4696 $nickname = $username;
4697 wfDebug( __METHOD__
. ": $username has bad XML tags in signature.\n" );
4701 # Make sure nickname doesnt get a sig in a sig
4702 $nickname = self
::cleanSigInSig( $nickname );
4704 # If we're still here, make it a link to the user page
4705 $userText = wfEscapeWikiText( $username );
4706 $nickText = wfEscapeWikiText( $nickname );
4707 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4709 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4710 ->title( $this->getTitle() )->text();
4714 * Check that the user's signature contains no bad XML
4716 * @param string $text
4717 * @return string|bool An expanded string, or false if invalid.
4719 public function validateSig( $text ) {
4720 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4724 * Clean up signature text
4726 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4727 * 2) Substitute all transclusions
4729 * @param string $text
4730 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4731 * @return string Signature text
4733 public function cleanSig( $text, $parsing = false ) {
4736 $magicScopeVariable = $this->lock();
4737 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4740 # Option to disable this feature
4741 if ( !$this->mOptions
->getCleanSignatures() ) {
4745 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4746 # => Move this logic to braceSubstitution()
4747 $substWord = $this->magicWordFactory
->get( 'subst' );
4748 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4749 $substText = '{{' . $substWord->getSynonym( 0 );
4751 $text = preg_replace( $substRegex, $substText, $text );
4752 $text = self
::cleanSigInSig( $text );
4753 $dom = $this->preprocessToDom( $text );
4754 $frame = $this->getPreprocessor()->newFrame();
4755 $text = $frame->expand( $dom );
4758 $text = $this->mStripState
->unstripBoth( $text );
4765 * Strip 3, 4 or 5 tildes out of signatures.
4767 * @param string $text
4768 * @return string Signature text with /~{3,5}/ removed
4770 public static function cleanSigInSig( $text ) {
4771 $text = preg_replace( '/~{3,5}/', '', $text );
4776 * Set up some variables which are usually set up in parse()
4777 * so that an external function can call some class members with confidence
4779 * @param Title|null $title
4780 * @param ParserOptions $options
4781 * @param int $outputType
4782 * @param bool $clearState
4784 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4785 $outputType, $clearState = true
4787 $this->startParse( $title, $options, $outputType, $clearState );
4791 * @param Title|null $title
4792 * @param ParserOptions $options
4793 * @param int $outputType
4794 * @param bool $clearState
4796 private function startParse( Title
$title = null, ParserOptions
$options,
4797 $outputType, $clearState = true
4799 $this->setTitle( $title );
4800 $this->mOptions
= $options;
4801 $this->setOutputType( $outputType );
4802 if ( $clearState ) {
4803 $this->clearState();
4808 * Wrapper for preprocess()
4810 * @param string $text The text to preprocess
4811 * @param ParserOptions $options
4812 * @param Title|null $title Title object or null to use $wgTitle
4815 public function transformMsg( $text, $options, $title = null ) {
4816 static $executing = false;
4818 # Guard against infinite recursion
4829 $text = $this->preprocess( $text, $title, $options );
4836 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4837 * The callback should have the following form:
4838 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4840 * Transform and return $text. Use $parser for any required context, e.g. use
4841 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4843 * Hooks may return extended information by returning an array, of which the
4844 * first numbered element (index 0) must be the return string, and all other
4845 * entries are extracted into local variables within an internal function
4846 * in the Parser class.
4848 * This interface (introduced r61913) appears to be undocumented, but
4849 * 'markerType' is used by some core tag hooks to override which strip
4850 * array their results are placed in. **Use great caution if attempting
4851 * this interface, as it is not documented and injudicious use could smash
4852 * private variables.**
4854 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4855 * @param callable $callback The callback function (and object) to use for the tag
4856 * @throws MWException
4857 * @return callable|null The old value of the mTagHooks array associated with the hook
4859 public function setHook( $tag, callable
$callback ) {
4860 $tag = strtolower( $tag );
4861 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4862 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4864 $oldVal = $this->mTagHooks
[$tag] ??
null;
4865 $this->mTagHooks
[$tag] = $callback;
4866 if ( !in_array( $tag, $this->mStripList
) ) {
4867 $this->mStripList
[] = $tag;
4874 * As setHook(), but letting the contents be parsed.
4876 * Transparent tag hooks are like regular XML-style tag hooks, except they
4877 * operate late in the transformation sequence, on HTML instead of wikitext.
4879 * This is probably obsoleted by things dealing with parser frames?
4880 * The only extension currently using it is geoserver.
4883 * @todo better document or deprecate this
4885 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4886 * @param callable $callback The callback function (and object) to use for the tag
4887 * @throws MWException
4888 * @return callable|null The old value of the mTagHooks array associated with the hook
4890 public function setTransparentTagHook( $tag, callable
$callback ) {
4891 $tag = strtolower( $tag );
4892 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4893 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4895 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4896 $this->mTransparentTagHooks
[$tag] = $callback;
4902 * Remove all tag hooks
4904 public function clearTagHooks() {
4905 $this->mTagHooks
= [];
4906 $this->mFunctionTagHooks
= [];
4907 $this->mStripList
= $this->mDefaultStripList
;
4911 * Create a function, e.g. {{sum:1|2|3}}
4912 * The callback function should have the form:
4913 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4915 * Or with Parser::SFH_OBJECT_ARGS:
4916 * function myParserFunction( $parser, $frame, $args ) { ... }
4918 * The callback may either return the text result of the function, or an array with the text
4919 * in element 0, and a number of flags in the other elements. The names of the flags are
4920 * specified in the keys. Valid flags are:
4921 * found The text returned is valid, stop processing the template. This
4923 * nowiki Wiki markup in the return value should be escaped
4924 * isHTML The returned text is HTML, armour it against wikitext transformation
4926 * @param string $id The magic word ID
4927 * @param callable $callback The callback function (and object) to use
4928 * @param int $flags A combination of the following flags:
4929 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4931 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4932 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4933 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4934 * the arguments, and to control the way they are expanded.
4936 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4937 * arguments, for instance:
4938 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4940 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4941 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4942 * working if/when this is changed.
4944 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4947 * Please read the documentation in includes/parser/Preprocessor.php for more information
4948 * about the methods available in PPFrame and PPNode.
4950 * @throws MWException
4951 * @return string|callable The old callback function for this name, if any
4953 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
4954 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4955 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
4957 # Add to function cache
4958 $mw = $this->magicWordFactory
->get( $id );
4960 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
4963 $synonyms = $mw->getSynonyms();
4964 $sensitive = intval( $mw->isCaseSensitive() );
4966 foreach ( $synonyms as $syn ) {
4968 if ( !$sensitive ) {
4969 $syn = $this->contLang
->lc( $syn );
4972 if ( !( $flags & self
::SFH_NO_HASH
) ) {
4975 # Remove trailing colon
4976 if ( substr( $syn, -1, 1 ) === ':' ) {
4977 $syn = substr( $syn, 0, -1 );
4979 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4985 * Get all registered function hook identifiers
4989 public function getFunctionHooks() {
4990 $this->firstCallInit();
4991 return array_keys( $this->mFunctionHooks
);
4995 * Create a tag function, e.g. "<test>some stuff</test>".
4996 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4997 * Unlike parser functions, their content is not preprocessed.
4998 * @param string $tag
4999 * @param callable $callback
5001 * @throws MWException
5004 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5005 $tag = strtolower( $tag );
5006 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5007 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5009 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5010 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5012 if ( !in_array( $tag, $this->mStripList
) ) {
5013 $this->mStripList
[] = $tag;
5020 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5021 * Placeholders created in Linker::link()
5023 * @param string &$text
5024 * @param int $options
5026 public function replaceLinkHolders( &$text, $options = 0 ) {
5027 $this->mLinkHolders
->replace( $text );
5031 * Replace "<!--LINK-->" link placeholders with plain text of links
5032 * (not HTML-formatted).
5034 * @param string $text
5037 public function replaceLinkHoldersText( $text ) {
5038 return $this->mLinkHolders
->replaceText( $text );
5042 * Renders an image gallery from a text with one line per image.
5043 * text labels may be given by using |-style alternative text. E.g.
5044 * Image:one.jpg|The number "1"
5045 * Image:tree.jpg|A tree
5046 * given as text will return the HTML of a gallery with two images,
5047 * labeled 'The number "1"' and
5050 * @param string $text
5051 * @param array $params
5052 * @return string HTML
5054 public function renderImageGallery( $text, $params ) {
5056 if ( isset( $params['mode'] ) ) {
5057 $mode = $params['mode'];
5061 $ig = ImageGalleryBase
::factory( $mode );
5062 } catch ( Exception
$e ) {
5063 // If invalid type set, fallback to default.
5064 $ig = ImageGalleryBase
::factory( false );
5067 $ig->setContextTitle( $this->mTitle
);
5068 $ig->setShowBytes( false );
5069 $ig->setShowDimensions( false );
5070 $ig->setShowFilename( false );
5071 $ig->setParser( $this );
5072 $ig->setHideBadImages();
5073 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5075 if ( isset( $params['showfilename'] ) ) {
5076 $ig->setShowFilename( true );
5078 $ig->setShowFilename( false );
5080 if ( isset( $params['caption'] ) ) {
5081 // NOTE: We aren't passing a frame here or below. Frame info
5082 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5083 // See T107332#4030581
5084 $caption = $this->recursiveTagParse( $params['caption'] );
5085 $ig->setCaptionHtml( $caption );
5087 if ( isset( $params['perrow'] ) ) {
5088 $ig->setPerRow( $params['perrow'] );
5090 if ( isset( $params['widths'] ) ) {
5091 $ig->setWidths( $params['widths'] );
5093 if ( isset( $params['heights'] ) ) {
5094 $ig->setHeights( $params['heights'] );
5096 $ig->setAdditionalOptions( $params );
5098 // Avoid PHP 7.1 warning from passing $this by reference
5100 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5102 $lines = StringUtils
::explode( "\n", $text );
5103 foreach ( $lines as $line ) {
5104 # match lines like these:
5105 # Image:someimage.jpg|This is some image
5107 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5109 if ( count( $matches ) == 0 ) {
5113 if ( strpos( $matches[0], '%' ) !== false ) {
5114 $matches[1] = rawurldecode( $matches[1] );
5116 $title = Title
::newFromText( $matches[1], NS_FILE
);
5117 if ( is_null( $title ) ) {
5118 # Bogus title. Ignore these so we don't bomb out later.
5122 # We need to get what handler the file uses, to figure out parameters.
5123 # Note, a hook can overide the file name, and chose an entirely different
5124 # file (which potentially could be of a different type and have different handler).
5127 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5128 [ $this, $title, &$options, &$descQuery ] );
5129 # Don't register it now, as TraditionalImageGallery does that later.
5130 $file = $this->fetchFileNoRegister( $title, $options );
5131 $handler = $file ?
$file->getHandler() : false;
5134 'img_alt' => 'gallery-internal-alt',
5135 'img_link' => 'gallery-internal-link',
5138 $paramMap +
= $handler->getParamMap();
5139 // We don't want people to specify per-image widths.
5140 // Additionally the width parameter would need special casing anyhow.
5141 unset( $paramMap['img_width'] );
5144 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5149 $handlerOptions = [];
5150 if ( isset( $matches[3] ) ) {
5151 // look for an |alt= definition while trying not to break existing
5152 // captions with multiple pipes (|) in it, until a more sensible grammar
5153 // is defined for images in galleries
5155 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5156 // splitting on '|' is a bit odd, and different from makeImage.
5157 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5158 // Protect LanguageConverter markup
5159 $parameterMatches = StringUtils
::delimiterExplode(
5160 '-{', '}-', '|', $matches[3], true /* nested */
5163 foreach ( $parameterMatches as $parameterMatch ) {
5164 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5166 $paramName = $paramMap[$magicName];
5168 switch ( $paramName ) {
5169 case 'gallery-internal-alt':
5170 $alt = $this->stripAltText( $match, false );
5172 case 'gallery-internal-link':
5173 $linkValue = $this->stripAltText( $match, false );
5174 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5175 // Result of LanguageConverter::markNoConversion
5176 // invoked on an external link.
5177 $linkValue = substr( $linkValue, 4, -2 );
5179 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5180 if ( $type === 'link-url' ) {
5182 $this->mOutput
->addExternalLink( $target );
5183 } elseif ( $type === 'link-title' ) {
5184 $link = $target->getLinkURL();
5185 $this->mOutput
->addLink( $target );
5189 // Must be a handler specific parameter.
5190 if ( $handler->validateParam( $paramName, $match ) ) {
5191 $handlerOptions[$paramName] = $match;
5193 // Guess not, consider it as caption.
5194 wfDebug( "$parameterMatch failed parameter validation\n" );
5195 $label = $parameterMatch;
5201 $label = $parameterMatch;
5206 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5208 $html = $ig->toHTML();
5209 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5214 * @param MediaHandler $handler
5217 public function getImageParams( $handler ) {
5219 $handlerClass = get_class( $handler );
5223 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5224 # Initialise static lists
5225 static $internalParamNames = [
5226 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5227 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5228 'bottom', 'text-bottom' ],
5229 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5230 'upright', 'border', 'link', 'alt', 'class' ],
5232 static $internalParamMap;
5233 if ( !$internalParamMap ) {
5234 $internalParamMap = [];
5235 foreach ( $internalParamNames as $type => $names ) {
5236 foreach ( $names as $name ) {
5237 // For grep: img_left, img_right, img_center, img_none,
5238 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5239 // img_bottom, img_text_bottom,
5240 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5241 // img_border, img_link, img_alt, img_class
5242 $magicName = str_replace( '-', '_', "img_$name" );
5243 $internalParamMap[$magicName] = [ $type, $name ];
5248 # Add handler params
5249 $paramMap = $internalParamMap;
5251 $handlerParamMap = $handler->getParamMap();
5252 foreach ( $handlerParamMap as $magic => $paramName ) {
5253 $paramMap[$magic] = [ 'handler', $paramName ];
5256 $this->mImageParams
[$handlerClass] = $paramMap;
5257 $this->mImageParamsMagicArray
[$handlerClass] =
5258 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5260 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5264 * Parse image options text and use it to make an image
5266 * @param Title $title
5267 * @param string $options
5268 * @param LinkHolderArray|bool $holders
5269 * @return string HTML
5271 public function makeImage( $title, $options, $holders = false ) {
5272 # Check if the options text is of the form "options|alt text"
5274 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5275 # * left no resizing, just left align. label is used for alt= only
5276 # * right same, but right aligned
5277 # * none same, but not aligned
5278 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5279 # * center center the image
5280 # * frame Keep original image size, no magnify-button.
5281 # * framed Same as "frame"
5282 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5283 # * upright reduce width for upright images, rounded to full __0 px
5284 # * border draw a 1px border around the image
5285 # * alt Text for HTML alt attribute (defaults to empty)
5286 # * class Set a class for img node
5287 # * link Set the target of the image link. Can be external, interwiki, or local
5288 # vertical-align values (no % or length right now):
5298 # Protect LanguageConverter markup when splitting into parts
5299 $parts = StringUtils
::delimiterExplode(
5300 '-{', '}-', '|', $options, true /* allow nesting */
5303 # Give extensions a chance to select the file revision for us
5306 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5307 [ $this, $title, &$options, &$descQuery ] );
5308 # Fetch and register the file (file title may be different via hooks)
5309 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5312 $handler = $file ?
$file->getHandler() : false;
5314 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5317 $this->addTrackingCategory( 'broken-file-category' );
5320 # Process the input parameters
5322 $params = [ 'frame' => [], 'handler' => [],
5323 'horizAlign' => [], 'vertAlign' => [] ];
5324 $seenformat = false;
5325 foreach ( $parts as $part ) {
5326 $part = trim( $part );
5327 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5329 if ( isset( $paramMap[$magicName] ) ) {
5330 list( $type, $paramName ) = $paramMap[$magicName];
5332 # Special case; width and height come in one variable together
5333 if ( $type === 'handler' && $paramName === 'width' ) {
5334 $parsedWidthParam = self
::parseWidthParam( $value );
5335 if ( isset( $parsedWidthParam['width'] ) ) {
5336 $width = $parsedWidthParam['width'];
5337 if ( $handler->validateParam( 'width', $width ) ) {
5338 $params[$type]['width'] = $width;
5342 if ( isset( $parsedWidthParam['height'] ) ) {
5343 $height = $parsedWidthParam['height'];
5344 if ( $handler->validateParam( 'height', $height ) ) {
5345 $params[$type]['height'] = $height;
5349 # else no validation -- T15436
5351 if ( $type === 'handler' ) {
5352 # Validate handler parameter
5353 $validated = $handler->validateParam( $paramName, $value );
5355 # Validate internal parameters
5356 switch ( $paramName ) {
5360 # @todo FIXME: Possibly check validity here for
5361 # manualthumb? downstream behavior seems odd with
5362 # missing manual thumbs.
5364 $value = $this->stripAltText( $value, $holders );
5367 list( $paramName, $value ) =
5368 $this->parseLinkParameter(
5369 $this->stripAltText( $value, $holders )
5373 if ( $paramName === 'no-link' ) {
5376 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5377 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5384 // use first appearing option, discard others.
5385 $validated = !$seenformat;
5389 # Most other things appear to be empty or numeric...
5390 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5395 $params[$type][$paramName] = $value;
5399 if ( !$validated ) {
5404 # Process alignment parameters
5405 if ( $params['horizAlign'] ) {
5406 $params['frame']['align'] = key( $params['horizAlign'] );
5408 if ( $params['vertAlign'] ) {
5409 $params['frame']['valign'] = key( $params['vertAlign'] );
5412 $params['frame']['caption'] = $caption;
5414 # Will the image be presented in a frame, with the caption below?
5415 $imageIsFramed = isset( $params['frame']['frame'] )
5416 ||
isset( $params['frame']['framed'] )
5417 ||
isset( $params['frame']['thumbnail'] )
5418 ||
isset( $params['frame']['manualthumb'] );
5420 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5421 # came to also set the caption, ordinary text after the image -- which
5422 # makes no sense, because that just repeats the text multiple times in
5423 # screen readers. It *also* came to set the title attribute.
5424 # Now that we have an alt attribute, we should not set the alt text to
5425 # equal the caption: that's worse than useless, it just repeats the
5426 # text. This is the framed/thumbnail case. If there's no caption, we
5427 # use the unnamed parameter for alt text as well, just for the time be-
5428 # ing, if the unnamed param is set and the alt param is not.
5429 # For the future, we need to figure out if we want to tweak this more,
5430 # e.g., introducing a title= parameter for the title; ignoring the un-
5431 # named parameter entirely for images without a caption; adding an ex-
5432 # plicit caption= parameter and preserving the old magic unnamed para-
5434 if ( $imageIsFramed ) { # Framed image
5435 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5436 # No caption or alt text, add the filename as the alt text so
5437 # that screen readers at least get some description of the image
5438 $params['frame']['alt'] = $title->getText();
5440 # Do not set $params['frame']['title'] because tooltips don't make sense
5442 } else { # Inline image
5443 if ( !isset( $params['frame']['alt'] ) ) {
5444 # No alt text, use the "caption" for the alt text
5445 if ( $caption !== '' ) {
5446 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5448 # No caption, fall back to using the filename for the
5450 $params['frame']['alt'] = $title->getText();
5453 # Use the "caption" for the tooltip text
5454 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5456 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5458 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5460 # Linker does the rest
5461 $time = $options['time'] ??
false;
5462 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5463 $time, $descQuery, $this->mOptions
->getThumbSize() );
5465 # Give the handler a chance to modify the parser object
5467 $handler->parserTransformHook( $this, $file );
5474 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5476 * Adds an entry to appropriate link tables.
5479 * @param string $value
5480 * @return array of `[ type, target ]`, where:
5481 * - `type` is one of:
5482 * - `null`: Given value is not a valid link target, use default
5483 * - `'no-link'`: Given value is empty, do not generate a link
5484 * - `'link-url'`: Given value is a valid external link
5485 * - `'link-title'`: Given value is a valid internal link
5487 * - When `type` is `null` or `'no-link'`: `false`
5488 * - When `type` is `'link-url'`: URL string corresponding to given value
5489 * - When `type` is `'link-title'`: Title object corresponding to given value
5491 public function parseLinkParameter( $value ) {
5492 $chars = self
::EXT_LINK_URL_CLASS
;
5493 $addr = self
::EXT_LINK_ADDR
;
5494 $prots = $this->mUrlProtocols
;
5497 if ( $value === '' ) {
5499 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5500 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5501 $this->mOutput
->addExternalLink( $value );
5506 $linkTitle = Title
::newFromText( $value );
5508 $this->mOutput
->addLink( $linkTitle );
5509 $type = 'link-title';
5510 $target = $linkTitle;
5513 return [ $type, $target ];
5517 * @param string $caption
5518 * @param LinkHolderArray|bool $holders
5519 * @return mixed|string
5521 protected function stripAltText( $caption, $holders ) {
5522 # Strip bad stuff out of the title (tooltip). We can't just use
5523 # replaceLinkHoldersText() here, because if this function is called
5524 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5526 $tooltip = $holders->replaceText( $caption );
5528 $tooltip = $this->replaceLinkHoldersText( $caption );
5531 # make sure there are no placeholders in thumbnail attributes
5532 # that are later expanded to html- so expand them now and
5534 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5535 # Compatibility hack! In HTML certain entity references not terminated
5536 # by a semicolon are decoded (but not if we're in an attribute; that's
5537 # how link URLs get away without properly escaping & in queries).
5538 # But wikitext has always required semicolon-termination of entities,
5539 # so encode & where needed to avoid decode of semicolon-less entities.
5541 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5542 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5543 $tooltip = preg_replace( "/
5544 & # 1. entity prefix
5545 (?= # 2. followed by:
5546 (?: # a. one of the legacy semicolon-less named entities
5547 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5548 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5549 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5550 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5551 U(?:acute|circ|grave|uml)|Yacute|
5552 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5553 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5554 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5555 frac(?:1(?:2|4)|34)|
5556 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5557 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5558 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5559 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5560 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5561 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5562 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5563 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5564 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5566 (?:[^;]|$)) # b. and not followed by a semicolon
5567 # S = study, for efficiency
5568 /Sx", '&', $tooltip );
5569 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5575 * Set a flag in the output object indicating that the content is dynamic and
5576 * shouldn't be cached.
5577 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5579 public function disableCache() {
5580 wfDebug( "Parser output marked as uncacheable.\n" );
5581 if ( !$this->mOutput
) {
5582 throw new MWException( __METHOD__
.
5583 " can only be called when actually parsing something" );
5585 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5589 * Callback from the Sanitizer for expanding items found in HTML attribute
5590 * values, so they can be safely tested and escaped.
5592 * @param string &$text
5593 * @param bool|PPFrame $frame
5596 public function attributeStripCallback( &$text, $frame = false ) {
5597 $text = $this->replaceVariables( $text, $frame );
5598 $text = $this->mStripState
->unstripBoth( $text );
5607 public function getTags() {
5608 $this->firstCallInit();
5610 array_keys( $this->mTransparentTagHooks
),
5611 array_keys( $this->mTagHooks
),
5612 array_keys( $this->mFunctionTagHooks
)
5620 public function getFunctionSynonyms() {
5621 $this->firstCallInit();
5622 return $this->mFunctionSynonyms
;
5629 public function getUrlProtocols() {
5630 return $this->mUrlProtocols
;
5634 * Replace transparent tags in $text with the values given by the callbacks.
5636 * Transparent tag hooks are like regular XML-style tag hooks, except they
5637 * operate late in the transformation sequence, on HTML instead of wikitext.
5639 * @param string $text
5643 public function replaceTransparentTags( $text ) {
5645 $elements = array_keys( $this->mTransparentTagHooks
);
5646 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5649 foreach ( $matches as $marker => $data ) {
5650 list( $element, $content, $params, $tag ) = $data;
5651 $tagName = strtolower( $element );
5652 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5653 $output = call_user_func_array(
5654 $this->mTransparentTagHooks
[$tagName],
5655 [ $content, $params, $this ]
5660 $replacements[$marker] = $output;
5662 return strtr( $text, $replacements );
5666 * Break wikitext input into sections, and either pull or replace
5667 * some particular section's text.
5669 * External callers should use the getSection and replaceSection methods.
5671 * @param string $text Page wikitext
5672 * @param string|int $sectionId A section identifier string of the form:
5673 * "<flag1> - <flag2> - ... - <section number>"
5675 * Currently the only recognised flag is "T", which means the target section number
5676 * was derived during a template inclusion parse, in other words this is a template
5677 * section edit link. If no flags are given, it was an ordinary section edit link.
5678 * This flag is required to avoid a section numbering mismatch when a section is
5679 * enclosed by "<includeonly>" (T8563).
5681 * The section number 0 pulls the text before the first heading; other numbers will
5682 * pull the given section along with its lower-level subsections. If the section is
5683 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5685 * Section 0 is always considered to exist, even if it only contains the empty
5686 * string. If $text is the empty string and section 0 is replaced, $newText is
5689 * @param string $mode One of "get" or "replace"
5690 * @param string $newText Replacement text for section data.
5691 * @return string For "get", the extracted section text.
5692 * for "replace", the whole page with the section replaced.
5694 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5695 global $wgTitle; # not generally used but removes an ugly failure mode
5697 $magicScopeVariable = $this->lock();
5698 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5700 $frame = $this->getPreprocessor()->newFrame();
5702 # Process section extraction flags
5704 $sectionParts = explode( '-', $sectionId );
5705 $sectionIndex = array_pop( $sectionParts );
5706 foreach ( $sectionParts as $part ) {
5707 if ( $part === 'T' ) {
5708 $flags |
= self
::PTD_FOR_INCLUSION
;
5712 # Check for empty input
5713 if ( strval( $text ) === '' ) {
5714 # Only sections 0 and T-0 exist in an empty document
5715 if ( $sectionIndex == 0 ) {
5716 if ( $mode === 'get' ) {
5722 if ( $mode === 'get' ) {
5730 # Preprocess the text
5731 $root = $this->preprocessToDom( $text, $flags );
5733 # <h> nodes indicate section breaks
5734 # They can only occur at the top level, so we can find them by iterating the root's children
5735 $node = $root->getFirstChild();
5737 # Find the target section
5738 if ( $sectionIndex == 0 ) {
5739 # Section zero doesn't nest, level=big
5740 $targetLevel = 1000;
5743 if ( $node->getName() === 'h' ) {
5744 $bits = $node->splitHeading();
5745 if ( $bits['i'] == $sectionIndex ) {
5746 $targetLevel = $bits['level'];
5750 if ( $mode === 'replace' ) {
5751 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5753 $node = $node->getNextSibling();
5759 if ( $mode === 'get' ) {
5766 # Find the end of the section, including nested sections
5768 if ( $node->getName() === 'h' ) {
5769 $bits = $node->splitHeading();
5770 $curLevel = $bits['level'];
5771 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5775 if ( $mode === 'get' ) {
5776 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5778 $node = $node->getNextSibling();
5781 # Write out the remainder (in replace mode only)
5782 if ( $mode === 'replace' ) {
5783 # Output the replacement text
5784 # Add two newlines on -- trailing whitespace in $newText is conventionally
5785 # stripped by the editor, so we need both newlines to restore the paragraph gap
5786 # Only add trailing whitespace if there is newText
5787 if ( $newText != "" ) {
5788 $outText .= $newText . "\n\n";
5792 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5793 $node = $node->getNextSibling();
5797 if ( is_string( $outText ) ) {
5798 # Re-insert stripped tags
5799 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5806 * This function returns the text of a section, specified by a number ($section).
5807 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5808 * the first section before any such heading (section 0).
5810 * If a section contains subsections, these are also returned.
5812 * @param string $text Text to look in
5813 * @param string|int $sectionId Section identifier as a number or string
5814 * (e.g. 0, 1 or 'T-1').
5815 * @param string $defaultText Default to return if section is not found
5817 * @return string Text of the requested section
5819 public function getSection( $text, $sectionId, $defaultText = '' ) {
5820 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5824 * This function returns $oldtext after the content of the section
5825 * specified by $section has been replaced with $text. If the target
5826 * section does not exist, $oldtext is returned unchanged.
5828 * @param string $oldText Former text of the article
5829 * @param string|int $sectionId Section identifier as a number or string
5830 * (e.g. 0, 1 or 'T-1').
5831 * @param string $newText Replacing text
5833 * @return string Modified text
5835 public function replaceSection( $oldText, $sectionId, $newText ) {
5836 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5840 * Get the ID of the revision we are parsing
5844 public function getRevisionId() {
5845 return $this->mRevisionId
;
5849 * Get the revision object for $this->mRevisionId
5851 * @return Revision|null Either a Revision object or null
5852 * @since 1.23 (public since 1.23)
5854 public function getRevisionObject() {
5855 if ( !is_null( $this->mRevisionObject
) ) {
5856 return $this->mRevisionObject
;
5859 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5860 // This is useful when parsing revision that has not yet been saved.
5861 // However, if we get back a saved revision even though we are in
5862 // preview mode, we'll have to ignore it, see below.
5863 // NOTE: This callback may be used to inject an OLD revision that was
5864 // already loaded, so "current" is a bit of a misnomer. We can't just
5865 // skip it if mRevisionId is set.
5866 $rev = call_user_func(
5867 $this->mOptions
->getCurrentRevisionCallback(), $this->getTitle(), $this
5870 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5871 // We are in preview mode (mRevisionId is null), and the current revision callback
5872 // returned an existing revision. Ignore it and return null, it's probably the page's
5873 // current revision, which is not what we want here. Note that we do want to call the
5874 // callback to allow the unsaved revision to be injected here, e.g. for
5875 // self-transclusion previews.
5879 // If the parse is for a new revision, then the callback should have
5880 // already been set to force the object and should match mRevisionId.
5881 // If not, try to fetch by mRevisionId for sanity.
5882 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5883 $rev = Revision
::newFromId( $this->mRevisionId
);
5886 $this->mRevisionObject
= $rev;
5888 return $this->mRevisionObject
;
5892 * Get the timestamp associated with the current revision, adjusted for
5893 * the default server-local timestamp
5896 public function getRevisionTimestamp() {
5897 if ( is_null( $this->mRevisionTimestamp
) ) {
5898 $revObject = $this->getRevisionObject();
5899 $timestamp = $revObject ?
$revObject->getTimestamp() : wfTimestampNow();
5901 # The cryptic '' timezone parameter tells to use the site-default
5902 # timezone offset instead of the user settings.
5903 # Since this value will be saved into the parser cache, served
5904 # to other users, and potentially even used inside links and such,
5905 # it needs to be consistent for all visitors.
5906 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
5908 return $this->mRevisionTimestamp
;
5912 * Get the name of the user that edited the last revision
5914 * @return string User name
5916 public function getRevisionUser() {
5917 if ( is_null( $this->mRevisionUser
) ) {
5918 $revObject = $this->getRevisionObject();
5920 # if this template is subst: the revision id will be blank,
5921 # so just use the current user's name
5923 $this->mRevisionUser
= $revObject->getUserText();
5924 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5925 $this->mRevisionUser
= $this->getUser()->getName();
5928 return $this->mRevisionUser
;
5932 * Get the size of the revision
5934 * @return int|null Revision size
5936 public function getRevisionSize() {
5937 if ( is_null( $this->mRevisionSize
) ) {
5938 $revObject = $this->getRevisionObject();
5940 # if this variable is subst: the revision id will be blank,
5941 # so just use the parser input size, because the own substituation
5942 # will change the size.
5944 $this->mRevisionSize
= $revObject->getSize();
5946 $this->mRevisionSize
= $this->mInputSize
;
5949 return $this->mRevisionSize
;
5953 * Mutator for $mDefaultSort
5955 * @param string $sort New value
5957 public function setDefaultSort( $sort ) {
5958 $this->mDefaultSort
= $sort;
5959 $this->mOutput
->setProperty( 'defaultsort', $sort );
5963 * Accessor for $mDefaultSort
5964 * Will use the empty string if none is set.
5966 * This value is treated as a prefix, so the
5967 * empty string is equivalent to sorting by
5972 public function getDefaultSort() {
5973 if ( $this->mDefaultSort
!== false ) {
5974 return $this->mDefaultSort
;
5981 * Accessor for $mDefaultSort
5982 * Unlike getDefaultSort(), will return false if none is set
5984 * @return string|bool
5986 public function getCustomDefaultSort() {
5987 return $this->mDefaultSort
;
5990 private static function getSectionNameFromStrippedText( $text ) {
5991 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5992 $text = Sanitizer
::decodeCharReferences( $text );
5993 $text = self
::normalizeSectionName( $text );
5997 private static function makeAnchor( $sectionName ) {
5998 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6001 private function makeLegacyAnchor( $sectionName ) {
6002 $fragmentMode = $this->siteConfig
->get( 'FragmentMode' );
6003 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6004 // ForAttribute() and ForLink() are the same for legacy encoding
6005 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6007 $id = Sanitizer
::escapeIdForLink( $sectionName );
6014 * Try to guess the section anchor name based on a wikitext fragment
6015 * presumably extracted from a heading, for example "Header" from
6018 * @param string $text
6019 * @return string Anchor (starting with '#')
6021 public function guessSectionNameFromWikiText( $text ) {
6022 # Strip out wikitext links(they break the anchor)
6023 $text = $this->stripSectionName( $text );
6024 $sectionName = self
::getSectionNameFromStrippedText( $text );
6025 return self
::makeAnchor( $sectionName );
6029 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6030 * instead, if possible. For use in redirects, since various versions
6031 * of Microsoft browsers interpret Location: headers as something other
6032 * than UTF-8, resulting in breakage.
6034 * @param string $text The section name
6035 * @return string Anchor (starting with '#')
6037 public function guessLegacySectionNameFromWikiText( $text ) {
6038 # Strip out wikitext links(they break the anchor)
6039 $text = $this->stripSectionName( $text );
6040 $sectionName = self
::getSectionNameFromStrippedText( $text );
6041 return $this->makeLegacyAnchor( $sectionName );
6045 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6046 * @param string $text Section name (plain text)
6047 * @return string Anchor (starting with '#')
6049 public static function guessSectionNameFromStrippedText( $text ) {
6050 $sectionName = self
::getSectionNameFromStrippedText( $text );
6051 return self
::makeAnchor( $sectionName );
6055 * Apply the same normalization as code making links to this section would
6057 * @param string $text
6060 private static function normalizeSectionName( $text ) {
6061 # T90902: ensure the same normalization is applied for IDs as to links
6062 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6065 $parts = $titleParser->splitTitleString( "#$text" );
6066 } catch ( MalformedTitleException
$ex ) {
6069 return $parts['fragment'];
6073 * Strips a text string of wikitext for use in a section anchor
6075 * Accepts a text string and then removes all wikitext from the
6076 * string and leaves only the resultant text (i.e. the result of
6077 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6078 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6079 * to create valid section anchors by mimicing the output of the
6080 * parser when headings are parsed.
6082 * @param string $text Text string to be stripped of wikitext
6083 * for use in a Section anchor
6084 * @return string Filtered text string
6086 public function stripSectionName( $text ) {
6087 # Strip internal link markup
6088 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6089 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6091 # Strip external link markup
6092 # @todo FIXME: Not tolerant to blank link text
6093 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6094 # on how many empty links there are on the page - need to figure that out.
6095 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6097 # Parse wikitext quotes (italics & bold)
6098 $text = $this->doQuotes( $text );
6101 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6106 * strip/replaceVariables/unstrip for preprocessor regression testing
6108 * @param string $text
6109 * @param Title $title
6110 * @param ParserOptions $options
6111 * @param int $outputType
6115 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6116 $outputType = self
::OT_HTML
6118 $magicScopeVariable = $this->lock();
6119 $this->startParse( $title, $options, $outputType, true );
6121 $text = $this->replaceVariables( $text );
6122 $text = $this->mStripState
->unstripBoth( $text );
6123 $text = Sanitizer
::removeHTMLtags( $text );
6128 * @param string $text
6129 * @param Title $title
6130 * @param ParserOptions $options
6133 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6134 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6138 * @param string $text
6139 * @param Title $title
6140 * @param ParserOptions $options
6143 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6144 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6148 * Call a callback function on all regions of the given text that are not
6149 * inside strip markers, and replace those regions with the return value
6150 * of the callback. For example, with input:
6154 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6155 * two strings will be replaced with the value returned by the callback in
6159 * @param callable $callback
6163 public function markerSkipCallback( $s, $callback ) {
6166 while ( $i < strlen( $s ) ) {
6167 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6168 if ( $markerStart === false ) {
6169 $out .= call_user_func( $callback, substr( $s, $i ) );
6172 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6173 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6174 if ( $markerEnd === false ) {
6175 $out .= substr( $s, $markerStart );
6178 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6179 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6188 * Remove any strip markers found in the given text.
6190 * @param string $text
6193 public function killMarkers( $text ) {
6194 return $this->mStripState
->killMarkers( $text );
6198 * Save the parser state required to convert the given half-parsed text to
6199 * HTML. "Half-parsed" in this context means the output of
6200 * recursiveTagParse() or internalParse(). This output has strip markers
6201 * from replaceVariables (extensionSubstitution() etc.), and link
6202 * placeholders from replaceLinkHolders().
6204 * Returns an array which can be serialized and stored persistently. This
6205 * array can later be loaded into another parser instance with
6206 * unserializeHalfParsedText(). The text can then be safely incorporated into
6207 * the return value of a parser hook.
6209 * @deprecated since 1.31
6210 * @param string $text
6214 public function serializeHalfParsedText( $text ) {
6215 wfDeprecated( __METHOD__
, '1.31' );
6218 'version' => self
::HALF_PARSED_VERSION
,
6219 'stripState' => $this->mStripState
->getSubState( $text ),
6220 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6226 * Load the parser state given in the $data array, which is assumed to
6227 * have been generated by serializeHalfParsedText(). The text contents is
6228 * extracted from the array, and its markers are transformed into markers
6229 * appropriate for the current Parser instance. This transformed text is
6230 * returned, and can be safely included in the return value of a parser
6233 * If the $data array has been stored persistently, the caller should first
6234 * check whether it is still valid, by calling isValidHalfParsedText().
6236 * @deprecated since 1.31
6237 * @param array $data Serialized data
6238 * @throws MWException
6241 public function unserializeHalfParsedText( $data ) {
6242 wfDeprecated( __METHOD__
, '1.31' );
6243 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6244 throw new MWException( __METHOD__
. ': invalid version' );
6247 # First, extract the strip state.
6248 $texts = [ $data['text'] ];
6249 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6251 # Now renumber links
6252 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6254 # Should be good to go.
6259 * Returns true if the given array, presumed to be generated by
6260 * serializeHalfParsedText(), is compatible with the current version of the
6263 * @deprecated since 1.31
6264 * @param array $data
6268 public function isValidHalfParsedText( $data ) {
6269 wfDeprecated( __METHOD__
, '1.31' );
6270 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6274 * Parsed a width param of imagelink like 300px or 200x300px
6276 * @param string $value
6277 * @param bool $parseHeight
6282 public static function parseWidthParam( $value, $parseHeight = true ) {
6283 $parsedWidthParam = [];
6284 if ( $value === '' ) {
6285 return $parsedWidthParam;
6288 # (T15500) In both cases (width/height and width only),
6289 # permit trailing "px" for backward compatibility.
6290 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6291 $width = intval( $m[1] );
6292 $height = intval( $m[2] );
6293 $parsedWidthParam['width'] = $width;
6294 $parsedWidthParam['height'] = $height;
6295 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6296 $width = intval( $value );
6297 $parsedWidthParam['width'] = $width;
6299 return $parsedWidthParam;
6303 * Lock the current instance of the parser.
6305 * This is meant to stop someone from calling the parser
6306 * recursively and messing up all the strip state.
6308 * @throws MWException If parser is in a parse
6309 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6311 protected function lock() {
6312 if ( $this->mInParse
) {
6313 throw new MWException( "Parser state cleared while parsing. "
6314 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6317 // Save the backtrace when locking, so that if some code tries locking again,
6318 // we can print the lock owner's backtrace for easier debugging
6320 $this->mInParse
= $e->getTraceAsString();
6322 $recursiveCheck = new ScopedCallback( function () {
6323 $this->mInParse
= false;
6326 return $recursiveCheck;
6330 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6332 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6333 * or if there is more than one <p/> tag in the input HTML.
6335 * @param string $html
6339 public static function stripOuterParagraph( $html ) {
6341 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6349 * Return this parser if it is not doing anything, otherwise
6350 * get a fresh parser. You can use this method by doing
6351 * $myParser = $wgParser->getFreshParser(), or more simply
6352 * $wgParser->getFreshParser()->parse( ... );
6353 * if you're unsure if $wgParser is safe to use.
6356 * @return Parser A parser object that is not parsing anything
6358 public function getFreshParser() {
6359 if ( $this->mInParse
) {
6360 return $this->factory
->create();
6367 * Set's up the PHP implementation of OOUI for use in this request
6368 * and instructs OutputPage to enable OOUI for itself.
6372 public function enableOOUI() {
6373 OutputPage
::setupOOUI();
6374 $this->mOutput
->setEnableOOUI( true );