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\Config\ServiceOptions
;
24 use MediaWiki\Linker\LinkRenderer
;
25 use MediaWiki\Linker\LinkRendererFactory
;
26 use MediaWiki\Linker\LinkTarget
;
27 use MediaWiki\MediaWikiServices
;
28 use MediaWiki\Special\SpecialPageFactory
;
29 use Wikimedia\ScopedCallback
;
32 * @defgroup Parser Parser
36 * PHP Parser - Processes wiki markup (which uses a more user-friendly
37 * syntax, such as "[[link]]" for making links), and provides a one-way
38 * transformation of that wiki markup it into (X)HTML output / markup
39 * (which in turn the browser understands, and can display).
41 * There are seven main entry points into the Parser class:
44 * produces HTML output
45 * - Parser::preSaveTransform()
46 * produces altered wiki markup
47 * - Parser::preprocess()
48 * removes HTML comments and expands templates
49 * - Parser::cleanSig() and Parser::cleanSigInSig()
50 * cleans a signature before saving it to preferences
51 * - Parser::getSection()
52 * return the content of a section from an article for section editing
53 * - Parser::replaceSection()
54 * replaces a section by number inside an article
55 * - Parser::getPreloadText()
56 * removes <noinclude> sections and <includeonly> tags
58 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
61 * $wgNamespacesWithSubpages
63 * @par Settings only within ParserOptions:
64 * $wgAllowExternalImages
65 * $wgAllowSpecialInclusion
73 * Update this version number when the ParserOutput format
74 * changes in an incompatible way, so the parser cache
75 * can automatically discard old data.
77 const VERSION
= '1.6.4';
80 * Update this version number when the output of serialiseHalfParsedText()
81 * changes in an incompatible way
83 const HALF_PARSED_VERSION
= 2;
85 # Flags for Parser::setFunctionHook
86 const SFH_NO_HASH
= 1;
87 const SFH_OBJECT_ARGS
= 2;
89 # Constants needed for external link processing
90 # Everything except bracket, space, or control characters
91 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
92 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
93 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
94 # uses to replace invalid HTML characters.
95 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
96 # Simplified expression to match an IPv4 or IPv6 address, or
97 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
98 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
99 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
100 // phpcs:ignore Generic.Files.LineLength
101 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
102 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION
= 1;
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML
= 1; # like parse()
113 const OT_WIKI
= 2; # like preSaveTransform()
114 const OT_PREPROCESS
= 3; # like preprocess()
116 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
135 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
136 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
138 # Markers used for wrapping the table of contents
139 const TOC_START
= '<mw:toc>';
140 const TOC_END
= '</mw:toc>';
142 /** @var int Assume that no output will later be saved this many seconds after parsing */
146 public $mTagHooks = [];
147 public $mTransparentTagHooks = [];
148 public $mFunctionHooks = [];
149 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
150 public $mFunctionTagHooks = [];
151 public $mStripList = [];
152 public $mDefaultStripList = [];
153 public $mVarCache = [];
154 public $mImageParams = [];
155 public $mImageParamsMagicArray = [];
156 public $mMarkerIndex = 0;
158 * @var bool Whether firstCallInit still needs to be called
160 public $mFirstCall = true;
162 # Initialised by initialiseVariables()
165 * @var MagicWordArray
170 * @var MagicWordArray
175 * @deprecated since 1.34, there should be no need to use this
180 # Initialised in constructor
181 public $mExtLinkBracketedRegex, $mUrlProtocols;
183 # Initialized in getPreprocessor()
184 /** @var Preprocessor */
185 public $mPreprocessor;
187 # Cleared with clearState():
199 public $mIncludeCount;
201 * @var LinkHolderArray
203 public $mLinkHolders;
206 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
207 public $mDefaultSort;
208 public $mTplRedirCache, $mHeadings, $mDoubleUnderscores;
209 public $mExpensiveFunctionCount; # number of expensive parser function calls
210 public $mShowToc, $mForceTocPosition;
212 public $mTplDomCache;
217 public $mUser; # User object; only used when doing pre-save transform
220 # These are variables reset at least once per parse regardless of $clearState
230 public $mTitle; # Title context, used for self-link rendering and similar things
231 public $mOutputType; # Output type, one of the OT_xxx constants
232 public $ot; # Shortcut alias, see setOutputType()
233 public $mRevisionObject; # The revision object of the specified revision ID
234 public $mRevisionId; # ID to display in {{REVISIONID}} tags
235 public $mRevisionTimestamp; # The timestamp of the specified revision ID
236 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
237 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
238 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
239 public $mInputSize = false; # For {{PAGESIZE}} on current page.
242 * @var array Array with the language name of each language link (i.e. the
243 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
244 * duplicate language links to the ParserOutput.
246 public $mLangLinkLanguages;
249 * @var MapCacheLRU|null
252 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
254 public $currentRevisionCache;
257 * @var bool|string Recursive call protection.
258 * This variable should be treated as if it were private.
260 public $mInParse = false;
262 /** @var SectionProfiler */
263 protected $mProfiler;
268 protected $mLinkRenderer;
270 /** @var MagicWordFactory */
271 private $magicWordFactory;
276 /** @var ParserFactory */
279 /** @var SpecialPageFactory */
280 private $specialPageFactory;
283 * This is called $svcOptions instead of $options like elsewhere to avoid confusion with
284 * $mOptions, which is public and widely used, and also with the local variable $options used
285 * for ParserOptions throughout this file.
287 * @var ServiceOptions
291 /** @var LinkRendererFactory */
292 private $linkRendererFactory;
294 /** @var NamespaceInfo */
298 * TODO Make this a const when HHVM support is dropped (T192166)
303 public static $constructorOptions = [
304 // See $wgParserConf documentation
307 // See documentation for the corresponding config options
309 'EnableScaryTranscluding',
310 'ExtraInterlanguageLinkPrefixes',
322 'TranscludeCacheExpiry',
326 * Constructing parsers directly is deprecated! Use a ParserFactory.
328 * @param ServiceOptions|null $svcOptions
329 * @param MagicWordFactory|null $magicWordFactory
330 * @param Language|null $contLang Content language
331 * @param ParserFactory|null $factory
332 * @param string|null $urlProtocols As returned from wfUrlProtocols()
333 * @param SpecialPageFactory|null $spFactory
334 * @param LinkRendererFactory|null $linkRendererFactory
335 * @param NamespaceInfo|null $nsInfo
337 public function __construct(
338 $svcOptions = null, MagicWordFactory
$magicWordFactory = null,
339 Language
$contLang = null, ParserFactory
$factory = null, $urlProtocols = null,
340 SpecialPageFactory
$spFactory = null, $linkRendererFactory = null, $nsInfo = null
342 $services = MediaWikiServices
::getInstance();
343 if ( !$svcOptions ||
is_array( $svcOptions ) ) {
344 // Pre-1.34 calling convention is the first parameter is just ParserConf, the seventh is
345 // Config, and the eighth is LinkRendererFactory.
346 $this->mConf
= (array)$svcOptions;
347 if ( empty( $this->mConf
['class'] ) ) {
348 $this->mConf
['class'] = self
::class;
350 if ( empty( $this->mConf
['preprocessorClass'] ) ) {
351 $this->mConf
['preprocessorClass'] = self
::getDefaultPreprocessorClass();
353 $this->svcOptions
= new ServiceOptions( self
::$constructorOptions,
355 func_num_args() > 6 ?
func_get_arg( 6 ) : $services->getMainConfig()
357 $linkRendererFactory = func_num_args() > 7 ?
func_get_arg( 7 ) : null;
358 $nsInfo = func_num_args() > 8 ?
func_get_arg( 8 ) : null;
360 // New calling convention
361 $svcOptions->assertRequiredOptions( self
::$constructorOptions );
362 // $this->mConf is public, so we'll keep those two options there as well for
363 // compatibility until it's removed
365 'class' => $svcOptions->get( 'class' ),
366 'preprocessorClass' => $svcOptions->get( 'preprocessorClass' ),
368 $this->svcOptions
= $svcOptions;
371 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
372 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
373 self
::EXT_LINK_ADDR
.
374 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
376 $this->magicWordFactory
= $magicWordFactory ??
377 $services->getMagicWordFactory();
379 $this->contLang
= $contLang ??
$services->getContentLanguage();
381 $this->factory
= $factory ??
$services->getParserFactory();
382 $this->specialPageFactory
= $spFactory ??
$services->getSpecialPageFactory();
383 $this->linkRendererFactory
= $linkRendererFactory ??
$services->getLinkRendererFactory();
384 $this->nsInfo
= $nsInfo ??
$services->getNamespaceInfo();
388 * Reduce memory usage to reduce the impact of circular references
390 public function __destruct() {
391 if ( isset( $this->mLinkHolders
) ) {
392 unset( $this->mLinkHolders
);
394 foreach ( $this as $name => $value ) {
395 unset( $this->$name );
400 * Allow extensions to clean up when the parser is cloned
402 public function __clone() {
403 $this->mInParse
= false;
405 // T58226: When you create a reference "to" an object field, that
406 // makes the object field itself be a reference too (until the other
407 // reference goes out of scope). When cloning, any field that's a
408 // reference is copied as a reference in the new object. Both of these
409 // are defined PHP5 behaviors, as inconvenient as it is for us when old
410 // hooks from PHP4 days are passing fields by reference.
411 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
412 // Make a non-reference copy of the field, then rebind the field to
413 // reference the new copy.
419 Hooks
::run( 'ParserCloned', [ $this ] );
423 * Which class should we use for the preprocessor if not otherwise specified?
426 * @deprecated since 1.34, removing configurability of preprocessor
429 public static function getDefaultPreprocessorClass() {
430 return Preprocessor_Hash
::class;
434 * Do various kinds of initialisation on the first call of the parser
436 public function firstCallInit() {
437 if ( !$this->mFirstCall
) {
440 $this->mFirstCall
= false;
442 CoreParserFunctions
::register( $this );
443 CoreTagHooks
::register( $this );
444 $this->initialiseVariables();
446 // Avoid PHP 7.1 warning from passing $this by reference
448 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
456 public function clearState() {
457 $this->firstCallInit();
458 $this->mOutput
= new ParserOutput
;
459 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
460 $this->mAutonumber
= 0;
461 $this->mIncludeCount
= [];
462 $this->mLinkHolders
= new LinkHolderArray( $this );
464 $this->mRevisionObject
= $this->mRevisionTimestamp
=
465 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
466 $this->mVarCache
= [];
468 $this->mLangLinkLanguages
= [];
469 $this->currentRevisionCache
= null;
471 $this->mStripState
= new StripState( $this );
473 # Clear these on every parse, T6549
474 $this->mTplRedirCache
= $this->mTplDomCache
= [];
476 $this->mShowToc
= true;
477 $this->mForceTocPosition
= false;
478 $this->mIncludeSizes
= [
482 $this->mPPNodeCount
= 0;
483 $this->mGeneratedPPNodeCount
= 0;
484 $this->mHighestExpansionDepth
= 0;
485 $this->mDefaultSort
= false;
486 $this->mHeadings
= [];
487 $this->mDoubleUnderscores
= [];
488 $this->mExpensiveFunctionCount
= 0;
491 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
492 $this->mPreprocessor
= null;
495 $this->mProfiler
= new SectionProfiler();
497 // Avoid PHP 7.1 warning from passing $this by reference
499 Hooks
::run( 'ParserClearState', [ &$parser ] );
503 * Convert wikitext to HTML
504 * Do not call this function recursively.
506 * @param string $text Text we want to parse
507 * @param-taint $text escapes_htmlnoent
508 * @param Title $title
509 * @param ParserOptions $options
510 * @param bool $linestart
511 * @param bool $clearState
512 * @param int|null $revid Number to pass in {{REVISIONID}}
513 * @return ParserOutput A ParserOutput
514 * @return-taint escaped
516 public function parse(
517 $text, Title
$title, ParserOptions
$options,
518 $linestart = true, $clearState = true, $revid = null
521 // We use U+007F DELETE to construct strip markers, so we have to make
522 // sure that this character does not occur in the input text.
523 $text = strtr( $text, "\x7f", "?" );
524 $magicScopeVariable = $this->lock();
526 // Strip U+0000 NULL (T159174)
527 $text = str_replace( "\000", '', $text );
529 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
531 $this->currentRevisionCache
= null;
532 $this->mInputSize
= strlen( $text );
533 if ( $this->mOptions
->getEnableLimitReport() ) {
534 $this->mOutput
->resetParseStartTime();
537 $oldRevisionId = $this->mRevisionId
;
538 $oldRevisionObject = $this->mRevisionObject
;
539 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
540 $oldRevisionUser = $this->mRevisionUser
;
541 $oldRevisionSize = $this->mRevisionSize
;
542 if ( $revid !== null ) {
543 $this->mRevisionId
= $revid;
544 $this->mRevisionObject
= null;
545 $this->mRevisionTimestamp
= null;
546 $this->mRevisionUser
= null;
547 $this->mRevisionSize
= null;
550 // Avoid PHP 7.1 warning from passing $this by reference
552 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
554 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
555 $text = $this->internalParse( $text );
556 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
558 $text = $this->internalParseHalfParsed( $text, true, $linestart );
561 * A converted title will be provided in the output object if title and
562 * content conversion are enabled, the article text does not contain
563 * a conversion-suppressing double-underscore tag, and no
564 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
565 * automatic link conversion.
567 if ( !( $options->getDisableTitleConversion()
568 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
569 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
570 ||
$this->mOutput
->getDisplayTitle() !== false )
572 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
573 if ( $convruletitle ) {
574 $this->mOutput
->setTitleText( $convruletitle );
576 $titleText = $this->getTargetLanguage()->convertTitle( $title );
577 $this->mOutput
->setTitleText( $titleText );
581 # Compute runtime adaptive expiry if set
582 $this->mOutput
->finalizeAdaptiveCacheExpiry();
584 # Warn if too many heavyweight parser functions were used
585 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
586 $this->limitationWarn( 'expensive-parserfunction',
587 $this->mExpensiveFunctionCount
,
588 $this->mOptions
->getExpensiveParserFunctionLimit()
592 # Information on limits, for the benefit of users who try to skirt them
593 if ( $this->mOptions
->getEnableLimitReport() ) {
594 $text .= $this->makeLimitReport();
597 # Wrap non-interface parser output in a <div> so it can be targeted
599 $class = $this->mOptions
->getWrapOutputClass();
600 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
601 $this->mOutput
->addWrapperDivClass( $class );
604 $this->mOutput
->setText( $text );
606 $this->mRevisionId
= $oldRevisionId;
607 $this->mRevisionObject
= $oldRevisionObject;
608 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
609 $this->mRevisionUser
= $oldRevisionUser;
610 $this->mRevisionSize
= $oldRevisionSize;
611 $this->mInputSize
= false;
612 $this->currentRevisionCache
= null;
614 return $this->mOutput
;
618 * Set the limit report data in the current ParserOutput, and return the
619 * limit report HTML comment.
623 protected function makeLimitReport() {
624 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
626 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
627 if ( $cpuTime !== null ) {
628 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
629 sprintf( "%.3f", $cpuTime )
633 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
634 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
635 sprintf( "%.3f", $wallTime )
638 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
639 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
641 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
642 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
644 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
645 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
647 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
648 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
650 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
651 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
653 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
654 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
657 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
658 $this->mOutput
->setLimitReportData( $key, $value );
661 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
663 $limitReport = "NewPP limit report\n";
664 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
665 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
667 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
668 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
669 $limitReport .= 'Dynamic content: ' .
670 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
672 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput
->getAllFlags() ) . "]\n";
674 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
675 if ( Hooks
::run( 'ParserLimitReportFormat',
676 [ $key, &$value, &$limitReport, false, false ]
678 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
679 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
680 ->inLanguage( 'en' )->useDatabase( false );
681 if ( !$valueMsg->exists() ) {
682 $valueMsg = new RawMessage( '$1' );
684 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
685 $valueMsg->params( $value );
686 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
690 // Since we're not really outputting HTML, decode the entities and
691 // then re-encode the things that need hiding inside HTML comments.
692 $limitReport = htmlspecialchars_decode( $limitReport );
694 // Sanitize for comment. Note '‐' in the replacement is U+2010,
695 // which looks much like the problematic '-'.
696 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
697 $text = "\n<!-- \n$limitReport-->\n";
699 // Add on template profiling data in human/machine readable way
700 $dataByFunc = $this->mProfiler
->getFunctionStats();
701 uasort( $dataByFunc, function ( $a, $b ) {
702 return $b['real'] <=> $a['real']; // descending order
705 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
706 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
707 $item['%real'], $item['real'], $item['calls'],
708 htmlspecialchars( $item['name'] ) );
710 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
711 $text .= implode( "\n", $profileReport ) . "\n-->\n";
713 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
715 // Add other cache related metadata
716 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
717 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
719 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
720 $this->mOutput
->getCacheTime() );
721 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
722 $this->mOutput
->getCacheExpiry() );
723 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
724 $this->mOutput
->hasDynamicContent() );
726 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
727 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
728 $this->mTitle
->getPrefixedDBkey() );
734 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
735 * can be called from an extension tag hook.
737 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
738 * instead, which means that lists and links have not been fully parsed yet,
739 * and strip markers are still present.
741 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
743 * Use this function if you're a parser tag hook and you want to parse
744 * wikitext before or after applying additional transformations, and you
745 * intend to *return the result as hook output*, which will cause it to go
746 * through the rest of parsing process automatically.
748 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
749 * $text are not expanded
751 * @param string $text Text extension wants to have parsed
752 * @param-taint $text escapes_htmlnoent
753 * @param bool|PPFrame $frame The frame to use for expanding any template variables
754 * @return string UNSAFE half-parsed HTML
755 * @return-taint escaped
757 public function recursiveTagParse( $text, $frame = false ) {
758 // Avoid PHP 7.1 warning from passing $this by reference
760 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
761 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
762 $text = $this->internalParse( $text, false, $frame );
767 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
768 * point can be called from an extension tag hook.
770 * The output of this function is fully-parsed HTML that is safe for output.
771 * If you're a parser tag hook, you might want to use recursiveTagParse()
774 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
775 * $text are not expanded
779 * @param string $text Text extension wants to have parsed
780 * @param-taint $text escapes_htmlnoent
781 * @param bool|PPFrame $frame The frame to use for expanding any template variables
782 * @return string Fully parsed HTML
783 * @return-taint escaped
785 public function recursiveTagParseFully( $text, $frame = false ) {
786 $text = $this->recursiveTagParse( $text, $frame );
787 $text = $this->internalParseHalfParsed( $text, false );
792 * Expand templates and variables in the text, producing valid, static wikitext.
793 * Also removes comments.
794 * Do not call this function recursively.
795 * @param string $text
796 * @param Title|null $title
797 * @param ParserOptions $options
798 * @param int|null $revid
799 * @param bool|PPFrame $frame
800 * @return mixed|string
802 public function preprocess( $text, Title
$title = null,
803 ParserOptions
$options, $revid = null, $frame = false
805 $magicScopeVariable = $this->lock();
806 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
807 if ( $revid !== null ) {
808 $this->mRevisionId
= $revid;
810 // Avoid PHP 7.1 warning from passing $this by reference
812 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
813 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
814 $text = $this->replaceVariables( $text, $frame );
815 $text = $this->mStripState
->unstripBoth( $text );
820 * Recursive parser entry point that can be called from an extension tag
823 * @param string $text Text to be expanded
824 * @param bool|PPFrame $frame The frame to use for expanding any template variables
828 public function recursivePreprocess( $text, $frame = false ) {
829 $text = $this->replaceVariables( $text, $frame );
830 $text = $this->mStripState
->unstripBoth( $text );
835 * Process the wikitext for the "?preload=" feature. (T7210)
837 * "<noinclude>", "<includeonly>" etc. are parsed as for template
838 * transclusion, comments, templates, arguments, tags hooks and parser
839 * functions are untouched.
841 * @param string $text
842 * @param Title $title
843 * @param ParserOptions $options
844 * @param array $params
847 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
848 $msg = new RawMessage( $text );
849 $text = $msg->params( $params )->plain();
851 # Parser (re)initialisation
852 $magicScopeVariable = $this->lock();
853 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
855 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
856 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
857 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
858 $text = $this->mStripState
->unstripBoth( $text );
863 * Set the current user.
864 * Should only be used when doing pre-save transform.
866 * @param User|null $user User object or null (to reset)
868 public function setUser( $user ) {
869 $this->mUser
= $user;
873 * Set the context title
877 public function setTitle( $t ) {
879 $t = Title
::newFromText( 'NO TITLE' );
882 if ( $t->hasFragment() ) {
883 # Strip the fragment to avoid various odd effects
884 $this->mTitle
= $t->createFragmentTarget( '' );
891 * Accessor for the Title object
895 public function getTitle() {
896 return $this->mTitle
;
900 * Accessor/mutator for the Title object
902 * @param Title|null $x Title object or null to just get the current one
905 public function Title( $x = null ) {
906 return wfSetVar( $this->mTitle
, $x );
910 * Set the output type
912 * @param int $ot New value
914 public function setOutputType( $ot ) {
915 $this->mOutputType
= $ot;
918 'html' => $ot == self
::OT_HTML
,
919 'wiki' => $ot == self
::OT_WIKI
,
920 'pre' => $ot == self
::OT_PREPROCESS
,
921 'plain' => $ot == self
::OT_PLAIN
,
926 * Accessor/mutator for the output type
928 * @param int|null $x New value or null to just get the current one
931 public function OutputType( $x = null ) {
932 return wfSetVar( $this->mOutputType
, $x );
936 * Get the ParserOutput object
938 * @return ParserOutput
940 public function getOutput() {
941 return $this->mOutput
;
945 * Get the ParserOptions object
947 * @return ParserOptions
949 public function getOptions() {
950 return $this->mOptions
;
954 * Accessor/mutator for the ParserOptions object
956 * @param ParserOptions|null $x New value or null to just get the current one
957 * @return ParserOptions Current ParserOptions object
959 public function Options( $x = null ) {
960 return wfSetVar( $this->mOptions
, $x );
966 public function nextLinkID() {
967 return $this->mLinkID++
;
973 public function setLinkID( $id ) {
974 $this->mLinkID
= $id;
978 * Get a language object for use in parser functions such as {{FORMATNUM:}}
981 public function getFunctionLang() {
982 return $this->getTargetLanguage();
986 * Get the target language for the content being parsed. This is usually the
987 * language that the content is in.
991 * @throws MWException
994 public function getTargetLanguage() {
995 $target = $this->mOptions
->getTargetLanguage();
997 if ( $target !== null ) {
999 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
1000 return $this->mOptions
->getUserLangObj();
1001 } elseif ( is_null( $this->mTitle
) ) {
1002 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
1005 return $this->mTitle
->getPageLanguage();
1009 * Get the language object for language conversion
1010 * @deprecated since 1.32, just use getTargetLanguage()
1011 * @return Language|null
1013 public function getConverterLanguage() {
1014 return $this->getTargetLanguage();
1018 * Get a User object either from $this->mUser, if set, or from the
1019 * ParserOptions object otherwise
1023 public function getUser() {
1024 if ( !is_null( $this->mUser
) ) {
1025 return $this->mUser
;
1027 return $this->mOptions
->getUser();
1031 * Get a preprocessor object
1033 * @return Preprocessor
1035 public function getPreprocessor() {
1036 if ( !isset( $this->mPreprocessor
) ) {
1037 $class = $this->svcOptions
->get( 'preprocessorClass' );
1038 $this->mPreprocessor
= new $class( $this );
1040 return $this->mPreprocessor
;
1044 * Get a LinkRenderer instance to make links with
1047 * @return LinkRenderer
1049 public function getLinkRenderer() {
1050 // XXX We make the LinkRenderer with current options and then cache it forever
1051 if ( !$this->mLinkRenderer
) {
1052 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
1053 $this->mLinkRenderer
->setStubThreshold(
1054 $this->getOptions()->getStubThreshold()
1058 return $this->mLinkRenderer
;
1062 * Get the MagicWordFactory that this Parser is using
1065 * @return MagicWordFactory
1067 public function getMagicWordFactory() {
1068 return $this->magicWordFactory
;
1072 * Get the content language that this Parser is using
1077 public function getContentLanguage() {
1078 return $this->contLang
;
1082 * Replaces all occurrences of HTML-style comments and the given tags
1083 * in the text with a random marker and returns the next text. The output
1084 * parameter $matches will be an associative array filled with data in
1091 * [ 'param' => 'x' ],
1092 * '<element param="x">tag content</element>' ]
1095 * @param array $elements List of element names. Comments are always extracted.
1096 * @param string $text Source text string.
1097 * @param array &$matches Out parameter, Array: extracted tags
1098 * @return string Stripped text
1100 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1105 $taglist = implode( '|', $elements );
1106 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1108 while ( $text != '' ) {
1109 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1111 if ( count( $p ) < 5 ) {
1114 if ( count( $p ) > 5 ) {
1122 list( , $element, $attributes, $close, $inside ) = $p;
1125 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1126 $stripped .= $marker;
1128 if ( $close === '/>' ) {
1129 # Empty element tag, <tag />
1134 if ( $element === '!--' ) {
1137 $end = "/(<\\/$element\\s*>)/i";
1139 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1141 if ( count( $q ) < 3 ) {
1142 # No end tag -- let it run out to the end of the text.
1146 list( , $tail, $text ) = $q;
1150 $matches[$marker] = [ $element,
1152 Sanitizer
::decodeTagAttributes( $attributes ),
1153 "<$element$attributes$close$content$tail" ];
1159 * Get a list of strippable XML-like elements
1163 public function getStripList() {
1164 return $this->mStripList
;
1168 * Add an item to the strip state
1169 * Returns the unique tag which must be inserted into the stripped text
1170 * The tag will be replaced with the original text in unstrip()
1172 * @param string $text
1176 public function insertStripItem( $text ) {
1177 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1178 $this->mMarkerIndex++
;
1179 $this->mStripState
->addGeneral( $marker, $text );
1184 * parse the wiki syntax used to render tables
1187 * @param string $text
1190 public function doTableStuff( $text ) {
1191 $lines = StringUtils
::explode( "\n", $text );
1193 $td_history = []; # Is currently a td tag open?
1194 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1195 $tr_history = []; # Is currently a tr tag open?
1196 $tr_attributes = []; # history of tr attributes
1197 $has_opened_tr = []; # Did this table open a <tr> element?
1198 $indent_level = 0; # indent level of the table
1200 foreach ( $lines as $outLine ) {
1201 $line = trim( $outLine );
1203 if ( $line === '' ) { # empty line, go to next line
1204 $out .= $outLine . "\n";
1208 $first_character = $line[0];
1209 $first_two = substr( $line, 0, 2 );
1212 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1213 # First check if we are starting a new table
1214 $indent_level = strlen( $matches[1] );
1216 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1217 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1219 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1220 array_push( $td_history, false );
1221 array_push( $last_tag_history, '' );
1222 array_push( $tr_history, false );
1223 array_push( $tr_attributes, '' );
1224 array_push( $has_opened_tr, false );
1225 } elseif ( count( $td_history ) == 0 ) {
1226 # Don't do any of the following
1227 $out .= $outLine . "\n";
1229 } elseif ( $first_two === '|}' ) {
1230 # We are ending a table
1231 $line = '</table>' . substr( $line, 2 );
1232 $last_tag = array_pop( $last_tag_history );
1234 if ( !array_pop( $has_opened_tr ) ) {
1235 $line = "<tr><td></td></tr>{$line}";
1238 if ( array_pop( $tr_history ) ) {
1239 $line = "</tr>{$line}";
1242 if ( array_pop( $td_history ) ) {
1243 $line = "</{$last_tag}>{$line}";
1245 array_pop( $tr_attributes );
1246 if ( $indent_level > 0 ) {
1247 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1251 } elseif ( $first_two === '|-' ) {
1252 # Now we have a table row
1253 $line = preg_replace( '#^\|-+#', '', $line );
1255 # Whats after the tag is now only attributes
1256 $attributes = $this->mStripState
->unstripBoth( $line );
1257 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1258 array_pop( $tr_attributes );
1259 array_push( $tr_attributes, $attributes );
1262 $last_tag = array_pop( $last_tag_history );
1263 array_pop( $has_opened_tr );
1264 array_push( $has_opened_tr, true );
1266 if ( array_pop( $tr_history ) ) {
1270 if ( array_pop( $td_history ) ) {
1271 $line = "</{$last_tag}>{$line}";
1275 array_push( $tr_history, false );
1276 array_push( $td_history, false );
1277 array_push( $last_tag_history, '' );
1278 } elseif ( $first_character === '|'
1279 ||
$first_character === '!'
1280 ||
$first_two === '|+'
1282 # This might be cell elements, td, th or captions
1283 if ( $first_two === '|+' ) {
1284 $first_character = '+';
1285 $line = substr( $line, 2 );
1287 $line = substr( $line, 1 );
1290 // Implies both are valid for table headings.
1291 if ( $first_character === '!' ) {
1292 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1295 # Split up multiple cells on the same line.
1296 # FIXME : This can result in improper nesting of tags processed
1297 # by earlier parser steps.
1298 $cells = explode( '||', $line );
1302 # Loop through each table cell
1303 foreach ( $cells as $cell ) {
1305 if ( $first_character !== '+' ) {
1306 $tr_after = array_pop( $tr_attributes );
1307 if ( !array_pop( $tr_history ) ) {
1308 $previous = "<tr{$tr_after}>\n";
1310 array_push( $tr_history, true );
1311 array_push( $tr_attributes, '' );
1312 array_pop( $has_opened_tr );
1313 array_push( $has_opened_tr, true );
1316 $last_tag = array_pop( $last_tag_history );
1318 if ( array_pop( $td_history ) ) {
1319 $previous = "</{$last_tag}>\n{$previous}";
1322 if ( $first_character === '|' ) {
1324 } elseif ( $first_character === '!' ) {
1326 } elseif ( $first_character === '+' ) {
1327 $last_tag = 'caption';
1332 array_push( $last_tag_history, $last_tag );
1334 # A cell could contain both parameters and data
1335 $cell_data = explode( '|', $cell, 2 );
1337 # T2553: Note that a '|' inside an invalid link should not
1338 # be mistaken as delimiting cell parameters
1339 # Bug T153140: Neither should language converter markup.
1340 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1341 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1342 } elseif ( count( $cell_data ) == 1 ) {
1343 // Whitespace in cells is trimmed
1344 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1346 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1347 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1348 // Whitespace in cells is trimmed
1349 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1353 array_push( $td_history, true );
1356 $out .= $outLine . "\n";
1359 # Closing open td, tr && table
1360 while ( count( $td_history ) > 0 ) {
1361 if ( array_pop( $td_history ) ) {
1364 if ( array_pop( $tr_history ) ) {
1367 if ( !array_pop( $has_opened_tr ) ) {
1368 $out .= "<tr><td></td></tr>\n";
1371 $out .= "</table>\n";
1374 # Remove trailing line-ending (b/c)
1375 if ( substr( $out, -1 ) === "\n" ) {
1376 $out = substr( $out, 0, -1 );
1379 # special case: don't return empty table
1380 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1388 * Helper function for parse() that transforms wiki markup into half-parsed
1389 * HTML. Only called for $mOutputType == self::OT_HTML.
1393 * @param string $text The text to parse
1394 * @param-taint $text escapes_html
1395 * @param bool $isMain Whether this is being called from the main parse() function
1396 * @param PPFrame|bool $frame A pre-processor frame
1400 public function internalParse( $text, $isMain = true, $frame = false ) {
1403 // Avoid PHP 7.1 warning from passing $this by reference
1406 # Hook to suspend the parser in this state
1407 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1411 # if $frame is provided, then use $frame for replacing any variables
1413 # use frame depth to infer how include/noinclude tags should be handled
1414 # depth=0 means this is the top-level document; otherwise it's an included document
1415 if ( !$frame->depth
) {
1418 $flag = self
::PTD_FOR_INCLUSION
;
1420 $dom = $this->preprocessToDom( $text, $flag );
1421 $text = $frame->expand( $dom );
1423 # if $frame is not provided, then use old-style replaceVariables
1424 $text = $this->replaceVariables( $text );
1427 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1428 $text = Sanitizer
::removeHTMLtags(
1430 [ $this, 'attributeStripCallback' ],
1432 array_keys( $this->mTransparentTagHooks
),
1434 [ $this, 'addTrackingCategory' ]
1436 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1438 # Tables need to come after variable replacement for things to work
1439 # properly; putting them before other transformations should keep
1440 # exciting things like link expansions from showing up in surprising
1442 $text = $this->doTableStuff( $text );
1444 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1446 $text = $this->doDoubleUnderscore( $text );
1448 $text = $this->doHeadings( $text );
1449 $text = $this->replaceInternalLinks( $text );
1450 $text = $this->doAllQuotes( $text );
1451 $text = $this->replaceExternalLinks( $text );
1453 # replaceInternalLinks may sometimes leave behind
1454 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1455 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1457 $text = $this->doMagicLinks( $text );
1458 $text = $this->formatHeadings( $text, $origText, $isMain );
1464 * Helper function for parse() that transforms half-parsed HTML into fully
1467 * @param string $text
1468 * @param bool $isMain
1469 * @param bool $linestart
1472 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1473 $text = $this->mStripState
->unstripGeneral( $text );
1475 // Avoid PHP 7.1 warning from passing $this by reference
1479 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1482 # Clean up special characters, only run once, next-to-last before doBlockLevels
1483 $text = Sanitizer
::armorFrenchSpaces( $text );
1485 $text = $this->doBlockLevels( $text, $linestart );
1487 $this->replaceLinkHolders( $text );
1490 * The input doesn't get language converted if
1492 * b) Content isn't converted
1493 * c) It's a conversion table
1494 * d) it is an interface message (which is in the user language)
1496 if ( !( $this->mOptions
->getDisableContentConversion()
1497 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1498 && !$this->mOptions
->getInterfaceMessage()
1500 # The position of the convert() call should not be changed. it
1501 # assumes that the links are all replaced and the only thing left
1502 # is the <nowiki> mark.
1503 $text = $this->getTargetLanguage()->convert( $text );
1506 $text = $this->mStripState
->unstripNoWiki( $text );
1509 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1512 $text = $this->replaceTransparentTags( $text );
1513 $text = $this->mStripState
->unstripGeneral( $text );
1515 $text = Sanitizer
::normalizeCharReferences( $text );
1517 if ( MWTidy
::isEnabled() ) {
1518 if ( $this->mOptions
->getTidy() ) {
1519 $text = MWTidy
::tidy( $text );
1522 # attempt to sanitize at least some nesting problems
1523 # (T4702 and quite a few others)
1524 # This code path is buggy and deprecated!
1525 wfDeprecated( 'disabling tidy', '1.33' );
1527 # ''Something [http://www.cool.com cool''] -->
1528 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1529 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1530 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1531 # fix up an anchor inside another anchor, only
1532 # at least for a single single nested link (T5695)
1533 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1534 '\\1\\2</a>\\3</a>\\1\\4</a>',
1535 # fix div inside inline elements- doBlockLevels won't wrap a line which
1536 # contains a div, so fix it up here; replace
1537 # div with escaped text
1538 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1539 '\\1\\3<div\\5>\\6</div>\\8\\9',
1540 # remove empty italic or bold tag pairs, some
1541 # introduced by rules above
1542 '/<([bi])><\/\\1>/' => '',
1545 $text = preg_replace(
1546 array_keys( $tidyregs ),
1547 array_values( $tidyregs ),
1552 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1559 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1560 * magic external links.
1565 * @param string $text
1569 public function doMagicLinks( $text ) {
1570 $prots = wfUrlProtocolsWithoutProtRel();
1571 $urlChar = self
::EXT_LINK_URL_CLASS
;
1572 $addr = self
::EXT_LINK_ADDR
;
1573 $space = self
::SPACE_NOT_NL
; # non-newline space
1574 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1575 $spaces = "$space++"; # possessive match of 1 or more spaces
1576 $text = preg_replace_callback(
1578 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1579 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1580 (\b # m[3]: Free external links
1582 ($addr$urlChar*) # m[4]: Post-protocol path
1584 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1586 \bISBN $spaces ( # m[6]: ISBN, capture number
1587 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1588 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1589 [0-9Xx] # check digit
1591 )!xu", [ $this, 'magicLinkCallback' ], $text );
1596 * @throws MWException
1598 * @return string HTML
1600 public function magicLinkCallback( $m ) {
1601 if ( isset( $m[1] ) && $m[1] !== '' ) {
1604 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1607 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1608 # Free external link
1609 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1610 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1612 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1613 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1618 $cssClass = 'mw-magiclink-rfc';
1619 $trackingCat = 'magiclink-tracking-rfc';
1621 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1622 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1626 $urlmsg = 'pubmedurl';
1627 $cssClass = 'mw-magiclink-pmid';
1628 $trackingCat = 'magiclink-tracking-pmid';
1631 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1632 substr( $m[0], 0, 20 ) . '"' );
1634 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1635 $this->addTrackingCategory( $trackingCat );
1636 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1637 } elseif ( isset( $m[6] ) && $m[6] !== ''
1638 && $this->mOptions
->getMagicISBNLinks()
1642 $space = self
::SPACE_NOT_NL
; # non-newline space
1643 $isbn = preg_replace( "/$space/", ' ', $isbn );
1644 $num = strtr( $isbn, [
1649 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1650 return $this->getLinkRenderer()->makeKnownLink(
1651 SpecialPage
::getTitleFor( 'Booksources', $num ),
1654 'class' => 'internal mw-magiclink-isbn',
1655 'title' => false // suppress title attribute
1664 * Make a free external link, given a user-supplied URL
1666 * @param string $url
1667 * @param int $numPostProto
1668 * The number of characters after the protocol.
1669 * @return string HTML
1672 public function makeFreeExternalLink( $url, $numPostProto ) {
1675 # The characters '<' and '>' (which were escaped by
1676 # removeHTMLtags()) should not be included in
1677 # URLs, per RFC 2396.
1678 # Make terminate a URL as well (bug T84937)
1681 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1686 $trail = substr( $url, $m2[0][1] ) . $trail;
1687 $url = substr( $url, 0, $m2[0][1] );
1690 # Move trailing punctuation to $trail
1692 # If there is no left bracket, then consider right brackets fair game too
1693 if ( strpos( $url, '(' ) === false ) {
1697 $urlRev = strrev( $url );
1698 $numSepChars = strspn( $urlRev, $sep );
1699 # Don't break a trailing HTML entity by moving the ; into $trail
1700 # This is in hot code, so use substr_compare to avoid having to
1701 # create a new string object for the comparison
1702 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1703 # more optimization: instead of running preg_match with a $
1704 # anchor, which can be slow, do the match on the reversed
1705 # string starting at the desired offset.
1706 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1707 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1711 if ( $numSepChars ) {
1712 $trail = substr( $url, -$numSepChars ) . $trail;
1713 $url = substr( $url, 0, -$numSepChars );
1716 # Verify that we still have a real URL after trail removal, and
1717 # not just lone protocol
1718 if ( strlen( $trail ) >= $numPostProto ) {
1719 return $url . $trail;
1722 $url = Sanitizer
::cleanUrl( $url );
1724 # Is this an external image?
1725 $text = $this->maybeMakeExternalImage( $url );
1726 if ( $text === false ) {
1727 # Not an image, make a link
1728 $text = Linker
::makeExternalLink( $url,
1729 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1731 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1732 # Register it in the output object...
1733 $this->mOutput
->addExternalLink( $url );
1735 return $text . $trail;
1739 * Parse headers and return html
1743 * @param string $text
1747 public function doHeadings( $text ) {
1748 for ( $i = 6; $i >= 1; --$i ) {
1749 $h = str_repeat( '=', $i );
1750 // Trim non-newline whitespace from headings
1751 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1752 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1758 * Replace single quotes with HTML markup
1761 * @param string $text
1763 * @return string The altered text
1765 public function doAllQuotes( $text ) {
1767 $lines = StringUtils
::explode( "\n", $text );
1768 foreach ( $lines as $line ) {
1769 $outtext .= $this->doQuotes( $line ) . "\n";
1771 $outtext = substr( $outtext, 0, -1 );
1776 * Helper function for doAllQuotes()
1778 * @param string $text
1782 public function doQuotes( $text ) {
1783 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1784 $countarr = count( $arr );
1785 if ( $countarr == 1 ) {
1789 // First, do some preliminary work. This may shift some apostrophes from
1790 // being mark-up to being text. It also counts the number of occurrences
1791 // of bold and italics mark-ups.
1794 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1795 $thislen = strlen( $arr[$i] );
1796 // If there are ever four apostrophes, assume the first is supposed to
1797 // be text, and the remaining three constitute mark-up for bold text.
1798 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1799 if ( $thislen == 4 ) {
1800 $arr[$i - 1] .= "'";
1803 } elseif ( $thislen > 5 ) {
1804 // If there are more than 5 apostrophes in a row, assume they're all
1805 // text except for the last 5.
1806 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1807 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1811 // Count the number of occurrences of bold and italics mark-ups.
1812 if ( $thislen == 2 ) {
1814 } elseif ( $thislen == 3 ) {
1816 } elseif ( $thislen == 5 ) {
1822 // If there is an odd number of both bold and italics, it is likely
1823 // that one of the bold ones was meant to be an apostrophe followed
1824 // by italics. Which one we cannot know for certain, but it is more
1825 // likely to be one that has a single-letter word before it.
1826 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1827 $firstsingleletterword = -1;
1828 $firstmultiletterword = -1;
1830 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1831 if ( strlen( $arr[$i] ) == 3 ) {
1832 $x1 = substr( $arr[$i - 1], -1 );
1833 $x2 = substr( $arr[$i - 1], -2, 1 );
1834 if ( $x1 === ' ' ) {
1835 if ( $firstspace == -1 ) {
1838 } elseif ( $x2 === ' ' ) {
1839 $firstsingleletterword = $i;
1840 // if $firstsingleletterword is set, we don't
1841 // look at the other options, so we can bail early.
1843 } elseif ( $firstmultiletterword == -1 ) {
1844 $firstmultiletterword = $i;
1849 // If there is a single-letter word, use it!
1850 if ( $firstsingleletterword > -1 ) {
1851 $arr[$firstsingleletterword] = "''";
1852 $arr[$firstsingleletterword - 1] .= "'";
1853 } elseif ( $firstmultiletterword > -1 ) {
1854 // If not, but there's a multi-letter word, use that one.
1855 $arr[$firstmultiletterword] = "''";
1856 $arr[$firstmultiletterword - 1] .= "'";
1857 } elseif ( $firstspace > -1 ) {
1858 // ... otherwise use the first one that has neither.
1859 // (notice that it is possible for all three to be -1 if, for example,
1860 // there is only one pentuple-apostrophe in the line)
1861 $arr[$firstspace] = "''";
1862 $arr[$firstspace - 1] .= "'";
1866 // Now let's actually convert our apostrophic mush to HTML!
1871 foreach ( $arr as $r ) {
1872 if ( ( $i %
2 ) == 0 ) {
1873 if ( $state === 'both' ) {
1879 $thislen = strlen( $r );
1880 if ( $thislen == 2 ) {
1881 if ( $state === 'i' ) {
1884 } elseif ( $state === 'bi' ) {
1887 } elseif ( $state === 'ib' ) {
1888 $output .= '</b></i><b>';
1890 } elseif ( $state === 'both' ) {
1891 $output .= '<b><i>' . $buffer . '</i>';
1893 } else { // $state can be 'b' or ''
1897 } elseif ( $thislen == 3 ) {
1898 if ( $state === 'b' ) {
1901 } elseif ( $state === 'bi' ) {
1902 $output .= '</i></b><i>';
1904 } elseif ( $state === 'ib' ) {
1907 } elseif ( $state === 'both' ) {
1908 $output .= '<i><b>' . $buffer . '</b>';
1910 } else { // $state can be 'i' or ''
1914 } elseif ( $thislen == 5 ) {
1915 if ( $state === 'b' ) {
1916 $output .= '</b><i>';
1918 } elseif ( $state === 'i' ) {
1919 $output .= '</i><b>';
1921 } elseif ( $state === 'bi' ) {
1922 $output .= '</i></b>';
1924 } elseif ( $state === 'ib' ) {
1925 $output .= '</b></i>';
1927 } elseif ( $state === 'both' ) {
1928 $output .= '<i><b>' . $buffer . '</b></i>';
1930 } else { // ($state == '')
1938 // Now close all remaining tags. Notice that the order is important.
1939 if ( $state === 'b' ||
$state === 'ib' ) {
1942 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1945 if ( $state === 'bi' ) {
1948 // There might be lonely ''''', so make sure we have a buffer
1949 if ( $state === 'both' && $buffer ) {
1950 $output .= '<b><i>' . $buffer . '</i></b>';
1956 * Replace external links (REL)
1958 * Note: this is all very hackish and the order of execution matters a lot.
1959 * Make sure to run tests/parser/parserTests.php if you change this code.
1963 * @param string $text
1965 * @throws MWException
1968 public function replaceExternalLinks( $text ) {
1969 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1970 if ( $bits === false ) {
1971 throw new MWException( "PCRE needs to be compiled with "
1972 . "--enable-unicode-properties in order for MediaWiki to function" );
1974 $s = array_shift( $bits );
1977 while ( $i < count( $bits ) ) {
1980 $text = $bits[$i++
];
1981 $trail = $bits[$i++
];
1983 # The characters '<' and '>' (which were escaped by
1984 # removeHTMLtags()) should not be included in
1985 # URLs, per RFC 2396.
1987 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1988 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1989 $url = substr( $url, 0, $m2[0][1] );
1992 # If the link text is an image URL, replace it with an <img> tag
1993 # This happened by accident in the original parser, but some people used it extensively
1994 $img = $this->maybeMakeExternalImage( $text );
1995 if ( $img !== false ) {
2001 # Set linktype for CSS
2004 # No link text, e.g. [http://domain.tld/some.link]
2005 if ( $text == '' ) {
2007 $langObj = $this->getTargetLanguage();
2008 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
2009 $linktype = 'autonumber';
2011 # Have link text, e.g. [http://domain.tld/some.link text]s
2013 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
2016 // Excluding protocol-relative URLs may avoid many false positives.
2017 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2018 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2021 $url = Sanitizer
::cleanUrl( $url );
2023 # Use the encoded URL
2024 # This means that users can paste URLs directly into the text
2025 # Funny characters like ö aren't valid in URLs anyway
2026 # This was changed in August 2004
2027 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
2028 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
2030 # Register link in the output object.
2031 $this->mOutput
->addExternalLink( $url );
2038 * Get the rel attribute for a particular external link.
2041 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2042 * nofollow if appropriate
2043 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2044 * @return string|null Rel attribute for $url
2046 public static function getExternalLinkRel( $url = false, $title = null ) {
2047 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2048 $ns = $title ?
$title->getNamespace() : false;
2049 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2050 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2058 * Get an associative array of additional HTML attributes appropriate for a
2059 * particular external link. This currently may include rel => nofollow
2060 * (depending on configuration, namespace, and the URL's domain) and/or a
2061 * target attribute (depending on configuration).
2063 * @param string $url URL to extract the domain from for rel =>
2064 * nofollow if appropriate
2065 * @return array Associative array of HTML attributes
2067 public function getExternalLinkAttribs( $url ) {
2069 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2071 $target = $this->mOptions
->getExternalLinkTarget();
2073 $attribs['target'] = $target;
2074 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2075 // T133507. New windows can navigate parent cross-origin.
2076 // Including noreferrer due to lacking browser
2077 // support of noopener. Eventually noreferrer should be removed.
2078 if ( $rel !== '' ) {
2081 $rel .= 'noreferrer noopener';
2084 $attribs['rel'] = $rel;
2089 * Replace unusual escape codes in a URL with their equivalent characters
2091 * This generally follows the syntax defined in RFC 3986, with special
2092 * consideration for HTTP query strings.
2094 * @param string $url
2097 public static function normalizeLinkUrl( $url ) {
2098 # Test for RFC 3986 IPv6 syntax
2099 $scheme = '[a-z][a-z0-9+.-]*:';
2100 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2101 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2102 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2103 IP
::isValid( rawurldecode( $m[1] ) )
2105 $isIPv6 = rawurldecode( $m[1] );
2110 # Make sure unsafe characters are encoded
2111 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2113 return rawurlencode( $m[0] );
2119 $end = strlen( $url );
2121 # Fragment part - 'fragment'
2122 $start = strpos( $url, '#' );
2123 if ( $start !== false && $start < $end ) {
2124 $ret = self
::normalizeUrlComponent(
2125 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2129 # Query part - 'query' minus &=+;
2130 $start = strpos( $url, '?' );
2131 if ( $start !== false && $start < $end ) {
2132 $ret = self
::normalizeUrlComponent(
2133 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2137 # Scheme and path part - 'pchar'
2138 # (we assume no userinfo or encoded colons in the host)
2139 $ret = self
::normalizeUrlComponent(
2140 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2143 if ( $isIPv6 !== false ) {
2144 $ipv6Host = "%5B({$isIPv6})%5D";
2145 $ret = preg_replace(
2146 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2155 private static function normalizeUrlComponent( $component, $unsafe ) {
2156 $callback = function ( $matches ) use ( $unsafe ) {
2157 $char = urldecode( $matches[0] );
2158 $ord = ord( $char );
2159 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2163 # Leave it escaped, but use uppercase for a-f
2164 return strtoupper( $matches[0] );
2167 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2171 * make an image if it's allowed, either through the global
2172 * option, through the exception, or through the on-wiki whitelist
2174 * @param string $url
2178 private function maybeMakeExternalImage( $url ) {
2179 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2180 $imagesexception = !empty( $imagesfrom );
2182 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2183 if ( $imagesexception && is_array( $imagesfrom ) ) {
2184 $imagematch = false;
2185 foreach ( $imagesfrom as $match ) {
2186 if ( strpos( $url, $match ) === 0 ) {
2191 } elseif ( $imagesexception ) {
2192 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2194 $imagematch = false;
2197 if ( $this->mOptions
->getAllowExternalImages()
2198 ||
( $imagesexception && $imagematch )
2200 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2202 $text = Linker
::makeExternalImage( $url );
2205 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2206 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2208 $whitelist = explode(
2210 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2213 foreach ( $whitelist as $entry ) {
2214 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2215 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2218 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2219 # Image matches a whitelist entry
2220 $text = Linker
::makeExternalImage( $url );
2229 * Process [[ ]] wikilinks
2233 * @return string Processed text
2237 public function replaceInternalLinks( $s ) {
2238 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2243 * Process [[ ]] wikilinks (RIL)
2245 * @throws MWException
2246 * @return LinkHolderArray
2250 public function replaceInternalLinks2( &$s ) {
2251 static $tc = false, $e1, $e1_img;
2252 # the % is needed to support urlencoded titles as well
2254 $tc = Title
::legalChars() . '#%';
2255 # Match a link having the form [[namespace:link|alternate]]trail
2256 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2257 # Match cases where there is no "]]", which might still be images
2258 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2261 $holders = new LinkHolderArray( $this );
2263 # split the entire text string on occurrences of [[
2264 $a = StringUtils
::explode( '[[', ' ' . $s );
2265 # get the first element (all text up to first [[), and remove the space we added
2268 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2269 $s = substr( $s, 1 );
2271 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2273 if ( $useLinkPrefixExtension ) {
2274 # Match the end of a line for a word that's not followed by whitespace,
2275 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2276 $charset = $this->contLang
->linkPrefixCharset();
2277 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2280 if ( is_null( $this->mTitle
) ) {
2281 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2283 $nottalk = !$this->mTitle
->isTalkPage();
2285 if ( $useLinkPrefixExtension ) {
2287 if ( preg_match( $e2, $s, $m ) ) {
2288 $first_prefix = $m[2];
2290 $first_prefix = false;
2296 $useSubpages = $this->areSubpagesAllowed();
2298 # Loop for each link
2299 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2300 # Check for excessive memory usage
2301 if ( $holders->isBig() ) {
2303 # Do the existence check, replace the link holders and clear the array
2304 $holders->replace( $s );
2308 if ( $useLinkPrefixExtension ) {
2309 if ( preg_match( $e2, $s, $m ) ) {
2310 list( , $s, $prefix ) = $m;
2315 if ( $first_prefix ) {
2316 $prefix = $first_prefix;
2317 $first_prefix = false;
2321 $might_be_img = false;
2323 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2325 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2326 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2327 # the real problem is with the $e1 regex
2329 # Still some problems for cases where the ] is meant to be outside punctuation,
2330 # and no image is in sight. See T4095.
2332 && substr( $m[3], 0, 1 ) === ']'
2333 && strpos( $text, '[' ) !== false
2335 $text .= ']'; # so that replaceExternalLinks($text) works later
2336 $m[3] = substr( $m[3], 1 );
2338 # fix up urlencoded title texts
2339 if ( strpos( $m[1], '%' ) !== false ) {
2340 # Should anchors '#' also be rejected?
2341 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2344 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2345 # Invalid, but might be an image with a link in its caption
2346 $might_be_img = true;
2348 if ( strpos( $m[1], '%' ) !== false ) {
2349 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2352 } else { # Invalid form; output directly
2353 $s .= $prefix . '[[' . $line;
2357 $origLink = ltrim( $m[1], ' ' );
2359 # Don't allow internal links to pages containing
2360 # PROTO: where PROTO is a valid URL protocol; these
2361 # should be external links.
2362 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2363 $s .= $prefix . '[[' . $line;
2367 # Make subpage if necessary
2368 if ( $useSubpages ) {
2369 $link = $this->maybeDoSubpageLink( $origLink, $text );
2374 // \x7f isn't a default legal title char, so most likely strip
2375 // markers will force us into the "invalid form" path above. But,
2376 // just in case, let's assert that xmlish tags aren't valid in
2377 // the title position.
2378 $unstrip = $this->mStripState
->killMarkers( $link );
2379 $noMarkers = ( $unstrip === $link );
2381 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2382 if ( $nt === null ) {
2383 $s .= $prefix . '[[' . $line;
2387 $ns = $nt->getNamespace();
2388 $iw = $nt->getInterwiki();
2390 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2392 if ( $might_be_img ) { # if this is actually an invalid link
2393 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2396 # look at the next 'line' to see if we can close it there
2398 $next_line = $a->current();
2399 if ( $next_line === false ||
$next_line === null ) {
2402 $m = explode( ']]', $next_line, 3 );
2403 if ( count( $m ) == 3 ) {
2404 # the first ]] closes the inner link, the second the image
2406 $text .= "[[{$m[0]}]]{$m[1]}";
2409 } elseif ( count( $m ) == 2 ) {
2410 # if there's exactly one ]] that's fine, we'll keep looking
2411 $text .= "[[{$m[0]}]]{$m[1]}";
2413 # if $next_line is invalid too, we need look no further
2414 $text .= '[[' . $next_line;
2419 # we couldn't find the end of this imageLink, so output it raw
2420 # but don't ignore what might be perfectly normal links in the text we've examined
2421 $holders->merge( $this->replaceInternalLinks2( $text ) );
2422 $s .= "{$prefix}[[$link|$text";
2423 # note: no $trail, because without an end, there *is* no trail
2426 } else { # it's not an image, so output it raw
2427 $s .= "{$prefix}[[$link|$text";
2428 # note: no $trail, because without an end, there *is* no trail
2433 $wasblank = ( $text == '' );
2437 # Strip off leading ':'
2438 $text = substr( $text, 1 );
2441 # T6598 madness. Handle the quotes only if they come from the alternate part
2442 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2443 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2444 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2445 $text = $this->doQuotes( $text );
2448 # Link not escaped by : , create the various objects
2449 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2452 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2453 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2454 in_array( $iw, $this->svcOptions
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2457 # T26502: filter duplicates
2458 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2459 $this->mLangLinkLanguages
[$iw] = true;
2460 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2464 * Strip the whitespace interwiki links produce, see T10897
2466 $s = rtrim( $s . $prefix ) . $trail; # T175416
2470 if ( $ns == NS_FILE
) {
2471 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2473 # if no parameters were passed, $text
2474 # becomes something like "File:Foo.png",
2475 # which we don't want to pass on to the
2479 # recursively parse links inside the image caption
2480 # actually, this will parse them in any other parameters, too,
2481 # but it might be hard to fix that, and it doesn't matter ATM
2482 $text = $this->replaceExternalLinks( $text );
2483 $holders->merge( $this->replaceInternalLinks2( $text ) );
2485 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2486 $s .= $prefix . $this->armorLinks(
2487 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2490 } elseif ( $ns == NS_CATEGORY
) {
2492 * Strip the whitespace Category links produce, see T2087
2494 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2497 $sortkey = $this->getDefaultSort();
2501 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2502 $sortkey = str_replace( "\n", '', $sortkey );
2503 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2504 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2510 # Self-link checking. For some languages, variants of the title are checked in
2511 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2512 # for linking to a different variant.
2513 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2514 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2518 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2519 # @todo FIXME: Should do batch file existence checks, see comment below
2520 if ( $ns == NS_MEDIA
) {
2521 # Give extensions a chance to select the file revision for us
2524 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2525 [ $this, $nt, &$options, &$descQuery ] );
2526 # Fetch and register the file (file title may be different via hooks)
2527 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2528 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2529 $s .= $prefix . $this->armorLinks(
2530 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2534 # Some titles, such as valid special pages or files in foreign repos, should
2535 # be shown as bluelinks even though they're not included in the page table
2536 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2537 # batch file existence checks for NS_FILE and NS_MEDIA
2538 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2539 $this->mOutput
->addLink( $nt );
2540 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2542 # Links will be added to the output link list after checking
2543 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2550 * Render a forced-blue link inline; protect against double expansion of
2551 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2552 * Since this little disaster has to split off the trail text to avoid
2553 * breaking URLs in the following text without breaking trails on the
2554 * wiki links, it's been made into a horrible function.
2557 * @param string $text
2558 * @param string $trail
2559 * @param string $prefix
2560 * @return string HTML-wikitext mix oh yuck
2562 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2563 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2565 if ( $text == '' ) {
2566 $text = htmlspecialchars( $nt->getPrefixedText() );
2569 $link = $this->getLinkRenderer()->makeKnownLink(
2570 $nt, new HtmlArmor( "$prefix$text$inside" )
2573 return $this->armorLinks( $link ) . $trail;
2577 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2578 * going to go through further parsing steps before inline URL expansion.
2580 * Not needed quite as much as it used to be since free links are a bit
2581 * more sensible these days. But bracketed links are still an issue.
2583 * @param string $text More-or-less HTML
2584 * @return string Less-or-more HTML with NOPARSE bits
2586 public function armorLinks( $text ) {
2587 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2588 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2592 * Return true if subpage links should be expanded on this page.
2595 public function areSubpagesAllowed() {
2596 # Some namespaces don't allow subpages
2597 return $this->nsInfo
->hasSubpages( $this->mTitle
->getNamespace() );
2601 * Handle link to subpage if necessary
2603 * @param string $target The source of the link
2604 * @param string &$text The link text, modified as necessary
2605 * @return string The full name of the link
2608 public function maybeDoSubpageLink( $target, &$text ) {
2609 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2613 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2615 * @param string $text
2616 * @param bool $linestart Whether or not this is at the start of a line.
2618 * @return string The lists rendered as HTML
2620 public function doBlockLevels( $text, $linestart ) {
2621 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2625 * Return value of a magic variable (like PAGENAME)
2629 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2630 * @param bool|PPFrame $frame
2632 * @throws MWException
2635 public function getVariableValue( $index, $frame = false ) {
2636 if ( is_null( $this->mTitle
) ) {
2637 // If no title set, bad things are going to happen
2638 // later. Title should always be set since this
2639 // should only be called in the middle of a parse
2640 // operation (but the unit-tests do funky stuff)
2641 throw new MWException( __METHOD__
. ' Should only be '
2642 . ' called while parsing (no title set)' );
2645 // Avoid PHP 7.1 warning from passing $this by reference
2649 * Some of these require message or data lookups and can be
2650 * expensive to check many times.
2653 Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) &&
2654 isset( $this->mVarCache
[$index] )
2656 return $this->mVarCache
[$index];
2659 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2660 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2662 $pageLang = $this->getFunctionLang();
2668 case 'currentmonth':
2669 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2671 case 'currentmonth1':
2672 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2674 case 'currentmonthname':
2675 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2677 case 'currentmonthnamegen':
2678 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2680 case 'currentmonthabbrev':
2681 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2684 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2687 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2690 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2693 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2695 case 'localmonthname':
2696 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2698 case 'localmonthnamegen':
2699 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2701 case 'localmonthabbrev':
2702 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2705 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2708 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2711 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2714 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2716 case 'fullpagename':
2717 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2719 case 'fullpagenamee':
2720 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2723 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2725 case 'subpagenamee':
2726 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2728 case 'rootpagename':
2729 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2731 case 'rootpagenamee':
2732 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2735 $this->mTitle
->getRootText()
2738 case 'basepagename':
2739 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2741 case 'basepagenamee':
2742 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2745 $this->mTitle
->getBaseText()
2748 case 'talkpagename':
2749 if ( $this->mTitle
->canHaveTalkPage() ) {
2750 $talkPage = $this->mTitle
->getTalkPage();
2751 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2756 case 'talkpagenamee':
2757 if ( $this->mTitle
->canHaveTalkPage() ) {
2758 $talkPage = $this->mTitle
->getTalkPage();
2759 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2764 case 'subjectpagename':
2765 $subjPage = $this->mTitle
->getSubjectPage();
2766 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2768 case 'subjectpagenamee':
2769 $subjPage = $this->mTitle
->getSubjectPage();
2770 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2772 case 'pageid': // requested in T25427
2773 $pageid = $this->getTitle()->getArticleID();
2774 if ( $pageid == 0 ) {
2775 # 0 means the page doesn't exist in the database,
2776 # which means the user is previewing a new page.
2777 # The vary-revision flag must be set, because the magic word
2778 # will have a different value once the page is saved.
2779 $this->mOutput
->setFlag( 'vary-revision' );
2780 wfDebug( __METHOD__
. ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2782 $value = $pageid ?
: null;
2786 $this->svcOptions
->get( 'MiserMode' ) &&
2787 !$this->mOptions
->getInterfaceMessage() &&
2788 // @TODO: disallow this word on all namespaces
2789 $this->nsInfo
->isContent( $this->mTitle
->getNamespace() )
2791 // Use a stub result instead of the actual revision ID in order to avoid
2792 // double parses on page save but still allow preview detection (T137900)
2793 if ( $this->getRevisionId() ||
$this->mOptions
->getSpeculativeRevId() ) {
2796 $this->mOutput
->setFlag( 'vary-revision-exists' );
2800 # Inform the edit saving system that getting the canonical output after
2801 # revision insertion requires another parse using the actual revision ID
2802 $this->mOutput
->setFlag( 'vary-revision-id' );
2803 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2804 $value = $this->getRevisionId();
2805 if ( $value === 0 ) {
2806 $rev = $this->getRevisionObject();
2807 $value = $rev ?
$rev->getId() : $value;
2810 $value = $this->mOptions
->getSpeculativeRevId();
2812 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2818 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2820 case 'revisionday2':
2821 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2823 case 'revisionmonth':
2824 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2826 case 'revisionmonth1':
2827 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2829 case 'revisionyear':
2830 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2832 case 'revisiontimestamp':
2833 # Let the edit saving system know we should parse the page
2834 # *after* a revision ID has been assigned. This is for null edits.
2835 $this->mOutput
->setFlag( 'vary-revision' );
2836 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2837 $value = $this->getRevisionTimestamp();
2839 case 'revisionuser':
2840 # Let the edit saving system know we should parse the page
2841 # *after* a revision ID has been assigned for null edits.
2842 $this->mOutput
->setFlag( 'vary-user' );
2843 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-user...\n" );
2844 $value = $this->getRevisionUser();
2846 case 'revisionsize':
2847 $value = $this->getRevisionSize();
2850 $value = str_replace( '_', ' ',
2851 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2854 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2856 case 'namespacenumber':
2857 $value = $this->mTitle
->getNamespace();
2860 $value = $this->mTitle
->canHaveTalkPage()
2861 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2865 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2867 case 'subjectspace':
2868 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2870 case 'subjectspacee':
2871 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2873 case 'currentdayname':
2874 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2877 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2880 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2883 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2886 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2887 # int to remove the padding
2888 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2891 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2893 case 'localdayname':
2894 $value = $pageLang->getWeekdayName(
2895 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2899 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2902 $value = $pageLang->time(
2903 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2909 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2912 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2913 # int to remove the padding
2914 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2917 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2919 case 'numberofarticles':
2920 $value = $pageLang->formatNum( SiteStats
::articles() );
2922 case 'numberoffiles':
2923 $value = $pageLang->formatNum( SiteStats
::images() );
2925 case 'numberofusers':
2926 $value = $pageLang->formatNum( SiteStats
::users() );
2928 case 'numberofactiveusers':
2929 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2931 case 'numberofpages':
2932 $value = $pageLang->formatNum( SiteStats
::pages() );
2934 case 'numberofadmins':
2935 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2937 case 'numberofedits':
2938 $value = $pageLang->formatNum( SiteStats
::edits() );
2940 case 'currenttimestamp':
2941 $value = wfTimestamp( TS_MW
, $ts );
2943 case 'localtimestamp':
2944 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2946 case 'currentversion':
2947 $value = SpecialVersion
::getVersion();
2950 return $this->svcOptions
->get( 'ArticlePath' );
2952 return $this->svcOptions
->get( 'Sitename' );
2954 return $this->svcOptions
->get( 'Server' );
2956 return $this->svcOptions
->get( 'ServerName' );
2958 return $this->svcOptions
->get( 'ScriptPath' );
2960 return $this->svcOptions
->get( 'StylePath' );
2961 case 'directionmark':
2962 return $pageLang->getDirMark();
2963 case 'contentlanguage':
2964 return $this->svcOptions
->get( 'LanguageCode' );
2965 case 'pagelanguage':
2966 $value = $pageLang->getCode();
2968 case 'cascadingsources':
2969 $value = CoreParserFunctions
::cascadingsources( $this );
2974 'ParserGetVariableValueSwitch',
2975 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
2982 $this->mVarCache
[$index] = $value;
2991 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2992 * @param string $variable Parser variable name
2995 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2996 # Get the timezone-adjusted timestamp to be used for this revision
2997 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2998 # Possibly set vary-revision if there is not yet an associated revision
2999 if ( !$this->getRevisionObject() ) {
3000 # Get the timezone-adjusted timestamp $mtts seconds in the future
3002 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
3007 if ( $resNow !== $resThen ) {
3008 # Let the edit saving system know we should parse the page
3009 # *after* a revision ID has been assigned. This is for null edits.
3010 $this->mOutput
->setFlag( 'vary-revision' );
3011 wfDebug( __METHOD__
. ": $variable used, setting vary-revision...\n" );
3019 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3023 public function initialiseVariables() {
3024 $variableIDs = $this->magicWordFactory
->getVariableIDs();
3025 $substIDs = $this->magicWordFactory
->getSubstIDs();
3027 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
3028 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
3032 * Preprocess some wikitext and return the document tree.
3033 * This is the ghost of replace_variables().
3035 * @param string $text The text to parse
3036 * @param int $flags Bitwise combination of:
3037 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3038 * included. Default is to assume a direct page view.
3040 * The generated DOM tree must depend only on the input text and the flags.
3041 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3043 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3044 * change in the DOM tree for a given text, must be passed through the section identifier
3045 * in the section edit link and thus back to extractSections().
3047 * The output of this function is currently only cached in process memory, but a persistent
3048 * cache may be implemented at a later date which takes further advantage of these strict
3049 * dependency requirements.
3053 public function preprocessToDom( $text, $flags = 0 ) {
3054 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3059 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3065 public static function splitWhitespace( $s ) {
3066 $ltrimmed = ltrim( $s );
3067 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3068 $trimmed = rtrim( $ltrimmed );
3069 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3071 $w2 = substr( $ltrimmed, -$diff );
3075 return [ $w1, $trimmed, $w2 ];
3079 * Replace magic variables, templates, and template arguments
3080 * with the appropriate text. Templates are substituted recursively,
3081 * taking care to avoid infinite loops.
3083 * Note that the substitution depends on value of $mOutputType:
3084 * self::OT_WIKI: only {{subst:}} templates
3085 * self::OT_PREPROCESS: templates but not extension tags
3086 * self::OT_HTML: all templates and extension tags
3088 * @param string $text The text to transform
3089 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3090 * template. Arguments may also be provided as an associative array, as
3091 * was the usual case before MW1.12. Providing arguments this way may be
3092 * useful for extensions wishing to perform variable replacement
3094 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3095 * double-brace expansion.
3098 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3099 # Is there any text? Also, Prevent too big inclusions!
3100 $textSize = strlen( $text );
3101 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3105 if ( $frame === false ) {
3106 $frame = $this->getPreprocessor()->newFrame();
3107 } elseif ( !( $frame instanceof PPFrame
) ) {
3108 wfDebug( __METHOD__
. " called using plain parameters instead of "
3109 . "a PPFrame instance. Creating custom frame.\n" );
3110 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3113 $dom = $this->preprocessToDom( $text );
3114 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3115 $text = $frame->expand( $dom, $flags );
3121 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3123 * @param array $args
3127 public static function createAssocArgs( $args ) {
3130 foreach ( $args as $arg ) {
3131 $eqpos = strpos( $arg, '=' );
3132 if ( $eqpos === false ) {
3133 $assocArgs[$index++
] = $arg;
3135 $name = trim( substr( $arg, 0, $eqpos ) );
3136 $value = trim( substr( $arg, $eqpos +
1 ) );
3137 if ( $value === false ) {
3140 if ( $name !== false ) {
3141 $assocArgs[$name] = $value;
3150 * Warn the user when a parser limitation is reached
3151 * Will warn at most once the user per limitation type
3153 * The results are shown during preview and run through the Parser (See EditPage.php)
3155 * @param string $limitationType Should be one of:
3156 * 'expensive-parserfunction' (corresponding messages:
3157 * 'expensive-parserfunction-warning',
3158 * 'expensive-parserfunction-category')
3159 * 'post-expand-template-argument' (corresponding messages:
3160 * 'post-expand-template-argument-warning',
3161 * 'post-expand-template-argument-category')
3162 * 'post-expand-template-inclusion' (corresponding messages:
3163 * 'post-expand-template-inclusion-warning',
3164 * 'post-expand-template-inclusion-category')
3165 * 'node-count-exceeded' (corresponding messages:
3166 * 'node-count-exceeded-warning',
3167 * 'node-count-exceeded-category')
3168 * 'expansion-depth-exceeded' (corresponding messages:
3169 * 'expansion-depth-exceeded-warning',
3170 * 'expansion-depth-exceeded-category')
3171 * @param string|int|null $current Current value
3172 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3173 * exceeded, provide the values (optional)
3175 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3176 # does no harm if $current and $max are present but are unnecessary for the message
3177 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3178 # only during preview, and that would split the parser cache unnecessarily.
3179 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3181 $this->mOutput
->addWarning( $warning );
3182 $this->addTrackingCategory( "$limitationType-category" );
3186 * Return the text of a template, after recursively
3187 * replacing any variables or templates within the template.
3189 * @param array $piece The parts of the template
3190 * $piece['title']: the title, i.e. the part before the |
3191 * $piece['parts']: the parameter array
3192 * $piece['lineStart']: whether the brace was at the start of a line
3193 * @param PPFrame $frame The current frame, contains template arguments
3195 * @return string|array The text of the template
3197 public function braceSubstitution( $piece, $frame ) {
3200 // $text has been filled
3202 // wiki markup in $text should be escaped
3204 // $text is HTML, armour it against wikitext transformation
3206 // Force interwiki transclusion to be done in raw mode not rendered
3207 $forceRawInterwiki = false;
3208 // $text is a DOM node needing expansion in a child frame
3209 $isChildObj = false;
3210 // $text is a DOM node needing expansion in the current frame
3211 $isLocalObj = false;
3213 # Title object, where $text came from
3216 # $part1 is the bit before the first |, and must contain only title characters.
3217 # Various prefixes will be stripped from it later.
3218 $titleWithSpaces = $frame->expand( $piece['title'] );
3219 $part1 = trim( $titleWithSpaces );
3222 # Original title text preserved for various purposes
3223 $originalTitle = $part1;
3225 # $args is a list of argument nodes, starting from index 0, not including $part1
3226 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3227 # below won't work b/c this $args isn't an object
3228 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3230 $profileSection = null; // profile templates
3234 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3236 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3237 # Decide whether to expand template or keep wikitext as-is.
3238 if ( $this->ot
['wiki'] ) {
3239 if ( $substMatch === false ) {
3240 $literal = true; # literal when in PST with no prefix
3242 $literal = false; # expand when in PST with subst: or safesubst:
3245 if ( $substMatch == 'subst' ) {
3246 $literal = true; # literal when not in PST with plain subst:
3248 $literal = false; # expand when not in PST with safesubst: or no prefix
3252 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3259 if ( !$found && $args->getLength() == 0 ) {
3260 $id = $this->mVariables
->matchStartToEnd( $part1 );
3261 if ( $id !== false ) {
3262 $text = $this->getVariableValue( $id, $frame );
3263 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3264 $this->mOutput
->updateCacheExpiry(
3265 $this->magicWordFactory
->getCacheTTL( $id ) );
3271 # MSG, MSGNW and RAW
3274 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3275 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3278 # Remove obsolete MSG:
3279 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3280 $mwMsg->matchStartAndRemove( $part1 );
3284 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3285 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3286 $forceRawInterwiki = true;
3292 $colonPos = strpos( $part1, ':' );
3293 if ( $colonPos !== false ) {
3294 $func = substr( $part1, 0, $colonPos );
3295 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3296 $argsLength = $args->getLength();
3297 for ( $i = 0; $i < $argsLength; $i++
) {
3298 $funcArgs[] = $args->item( $i );
3301 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3303 // Extract any forwarded flags
3304 if ( isset( $result['title'] ) ) {
3305 $title = $result['title'];
3307 if ( isset( $result['found'] ) ) {
3308 $found = $result['found'];
3310 if ( array_key_exists( 'text', $result ) ) {
3312 $text = $result['text'];
3314 if ( isset( $result['nowiki'] ) ) {
3315 $nowiki = $result['nowiki'];
3317 if ( isset( $result['isHTML'] ) ) {
3318 $isHTML = $result['isHTML'];
3320 if ( isset( $result['forceRawInterwiki'] ) ) {
3321 $forceRawInterwiki = $result['forceRawInterwiki'];
3323 if ( isset( $result['isChildObj'] ) ) {
3324 $isChildObj = $result['isChildObj'];
3326 if ( isset( $result['isLocalObj'] ) ) {
3327 $isLocalObj = $result['isLocalObj'];
3332 # Finish mangling title and then check for loops.
3333 # Set $title to a Title object and $titleText to the PDBK
3336 # Split the title into page and subpage
3338 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3339 if ( $part1 !== $relative ) {
3341 $ns = $this->mTitle
->getNamespace();
3343 $title = Title
::newFromText( $part1, $ns );
3345 $titleText = $title->getPrefixedText();
3346 # Check for language variants if the template is not found
3347 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3348 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3350 # Do recursion depth check
3351 $limit = $this->mOptions
->getMaxTemplateDepth();
3352 if ( $frame->depth
>= $limit ) {
3354 $text = '<span class="error">'
3355 . wfMessage( 'parser-template-recursion-depth-warning' )
3356 ->numParams( $limit )->inContentLanguage()->text()
3362 # Load from database
3363 if ( !$found && $title ) {
3364 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3365 if ( !$title->isExternal() ) {
3366 if ( $title->isSpecialPage()
3367 && $this->mOptions
->getAllowSpecialInclusion()
3368 && $this->ot
['html']
3370 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3371 // Pass the template arguments as URL parameters.
3372 // "uselang" will have no effect since the Language object
3373 // is forced to the one defined in ParserOptions.
3375 $argsLength = $args->getLength();
3376 for ( $i = 0; $i < $argsLength; $i++
) {
3377 $bits = $args->item( $i )->splitArg();
3378 if ( strval( $bits['index'] ) === '' ) {
3379 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3380 $value = trim( $frame->expand( $bits['value'] ) );
3381 $pageArgs[$name] = $value;
3385 // Create a new context to execute the special page
3386 $context = new RequestContext
;
3387 $context->setTitle( $title );
3388 $context->setRequest( new FauxRequest( $pageArgs ) );
3389 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3390 $context->setUser( $this->getUser() );
3392 // If this page is cached, then we better not be per user.
3393 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3395 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3396 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3398 $text = $context->getOutput()->getHTML();
3399 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3402 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3403 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3404 $specialPage->maxIncludeCacheTime()
3408 } elseif ( $this->nsInfo
->isNonincludable( $title->getNamespace() ) ) {
3409 $found = false; # access denied
3410 wfDebug( __METHOD__
. ": template inclusion denied for " .
3411 $title->getPrefixedDBkey() . "\n" );
3413 list( $text, $title ) = $this->getTemplateDom( $title );
3414 if ( $text !== false ) {
3420 # If the title is valid but undisplayable, make a link to it
3421 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3422 $text = "[[:$titleText]]";
3425 } elseif ( $title->isTrans() ) {
3426 # Interwiki transclusion
3427 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3428 $text = $this->interwikiTransclude( $title, 'render' );
3431 $text = $this->interwikiTransclude( $title, 'raw' );
3432 # Preprocess it like a template
3433 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3439 # Do infinite loop check
3440 # This has to be done after redirect resolution to avoid infinite loops via redirects
3441 if ( !$frame->loopCheck( $title ) ) {
3443 $text = '<span class="error">'
3444 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3446 $this->addTrackingCategory( 'template-loop-category' );
3447 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3448 wfEscapeWikiText( $titleText ) )->text() );
3449 wfDebug( __METHOD__
. ": template loop broken at '$titleText'\n" );
3453 # If we haven't found text to substitute by now, we're done
3454 # Recover the source wikitext and return it
3456 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3457 if ( $profileSection ) {
3458 $this->mProfiler
->scopedProfileOut( $profileSection );
3460 return [ 'object' => $text ];
3463 # Expand DOM-style return values in a child frame
3464 if ( $isChildObj ) {
3465 # Clean up argument array
3466 $newFrame = $frame->newChild( $args, $title );
3469 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3470 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3471 # Expansion is eligible for the empty-frame cache
3472 $text = $newFrame->cachedExpand( $titleText, $text );
3474 # Uncached expansion
3475 $text = $newFrame->expand( $text );
3478 if ( $isLocalObj && $nowiki ) {
3479 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3480 $isLocalObj = false;
3483 if ( $profileSection ) {
3484 $this->mProfiler
->scopedProfileOut( $profileSection );
3487 # Replace raw HTML by a placeholder
3489 $text = $this->insertStripItem( $text );
3490 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3491 # Escape nowiki-style return values
3492 $text = wfEscapeWikiText( $text );
3493 } elseif ( is_string( $text )
3494 && !$piece['lineStart']
3495 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3497 # T2529: if the template begins with a table or block-level
3498 # element, it should be treated as beginning a new line.
3499 # This behavior is somewhat controversial.
3500 $text = "\n" . $text;
3503 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3504 # Error, oversize inclusion
3505 if ( $titleText !== false ) {
3506 # Make a working, properly escaped link if possible (T25588)
3507 $text = "[[:$titleText]]";
3509 # This will probably not be a working link, but at least it may
3510 # provide some hint of where the problem is
3511 preg_replace( '/^:/', '', $originalTitle );
3512 $text = "[[:$originalTitle]]";
3514 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3515 . 'post-expand include size too large -->' );
3516 $this->limitationWarn( 'post-expand-template-inclusion' );
3519 if ( $isLocalObj ) {
3520 $ret = [ 'object' => $text ];
3522 $ret = [ 'text' => $text ];
3529 * Call a parser function and return an array with text and flags.
3531 * The returned array will always contain a boolean 'found', indicating
3532 * whether the parser function was found or not. It may also contain the
3534 * text: string|object, resulting wikitext or PP DOM object
3535 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3536 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3537 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3538 * nowiki: bool, wiki markup in $text should be escaped
3541 * @param PPFrame $frame The current frame, contains template arguments
3542 * @param string $function Function name
3543 * @param array $args Arguments to the function
3544 * @throws MWException
3547 public function callParserFunction( $frame, $function, array $args = [] ) {
3548 # Case sensitive functions
3549 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3550 $function = $this->mFunctionSynonyms
[1][$function];
3552 # Case insensitive functions
3553 $function = $this->contLang
->lc( $function );
3554 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3555 $function = $this->mFunctionSynonyms
[0][$function];
3557 return [ 'found' => false ];
3561 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3563 // Avoid PHP 7.1 warning from passing $this by reference
3566 $allArgs = [ &$parser ];
3567 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3568 # Convert arguments to PPNodes and collect for appending to $allArgs
3570 foreach ( $args as $k => $v ) {
3571 if ( $v instanceof PPNode ||
$k === 0 ) {
3574 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3578 # Add a frame parameter, and pass the arguments as an array
3579 $allArgs[] = $frame;
3580 $allArgs[] = $funcArgs;
3582 # Convert arguments to plain text and append to $allArgs
3583 foreach ( $args as $k => $v ) {
3584 if ( $v instanceof PPNode
) {
3585 $allArgs[] = trim( $frame->expand( $v ) );
3586 } elseif ( is_int( $k ) && $k >= 0 ) {
3587 $allArgs[] = trim( $v );
3589 $allArgs[] = trim( "$k=$v" );
3594 $result = $callback( ...$allArgs );
3596 # The interface for function hooks allows them to return a wikitext
3597 # string or an array containing the string and any flags. This mungs
3598 # things around to match what this method should return.
3599 if ( !is_array( $result ) ) {
3605 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3606 $result['text'] = $result[0];
3608 unset( $result[0] );
3615 $preprocessFlags = 0;
3616 if ( isset( $result['noparse'] ) ) {
3617 $noparse = $result['noparse'];
3619 if ( isset( $result['preprocessFlags'] ) ) {
3620 $preprocessFlags = $result['preprocessFlags'];
3624 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3625 $result['isChildObj'] = true;
3632 * Get the semi-parsed DOM representation of a template with a given title,
3633 * and its redirect destination title. Cached.
3635 * @param Title $title
3639 public function getTemplateDom( $title ) {
3640 $cacheTitle = $title;
3641 $titleText = $title->getPrefixedDBkey();
3643 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3644 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3645 $title = Title
::makeTitle( $ns, $dbk );
3646 $titleText = $title->getPrefixedDBkey();
3648 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3649 return [ $this->mTplDomCache
[$titleText], $title ];
3652 # Cache miss, go to the database
3653 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3655 if ( $text === false ) {
3656 $this->mTplDomCache
[$titleText] = false;
3657 return [ false, $title ];
3660 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3661 $this->mTplDomCache
[$titleText] = $dom;
3663 if ( !$title->equals( $cacheTitle ) ) {
3664 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3665 [ $title->getNamespace(), $title->getDBkey() ];
3668 return [ $dom, $title ];
3672 * Fetch the current revision of a given title. Note that the revision
3673 * (and even the title) may not exist in the database, so everything
3674 * contributing to the output of the parser should use this method
3675 * where possible, rather than getting the revisions themselves. This
3676 * method also caches its results, so using it benefits performance.
3679 * @param Title $title
3682 public function fetchCurrentRevisionOfTitle( $title ) {
3683 $cacheKey = $title->getPrefixedDBkey();
3684 if ( !$this->currentRevisionCache
) {
3685 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3687 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3688 $this->currentRevisionCache
->set( $cacheKey,
3689 // Defaults to Parser::statelessFetchRevision()
3690 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3693 return $this->currentRevisionCache
->get( $cacheKey );
3697 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3698 * without passing them on to it.
3701 * @param Title $title
3702 * @param Parser|bool $parser
3703 * @return Revision|bool False if missing
3705 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3706 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3712 * Fetch the unparsed text of a template and register a reference to it.
3713 * @param Title $title
3714 * @return array ( string or false, Title )
3716 public function fetchTemplateAndTitle( $title ) {
3717 // Defaults to Parser::statelessFetchTemplate()
3718 $templateCb = $this->mOptions
->getTemplateCallback();
3719 $stuff = call_user_func( $templateCb, $title, $this );
3720 // We use U+007F DELETE to distinguish strip markers from regular text.
3721 $text = $stuff['text'];
3722 if ( is_string( $stuff['text'] ) ) {
3723 $text = strtr( $text, "\x7f", "?" );
3725 $finalTitle = $stuff['finalTitle'] ??
$title;
3726 if ( isset( $stuff['deps'] ) ) {
3727 foreach ( $stuff['deps'] as $dep ) {
3728 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3729 if ( $dep['title']->equals( $this->getTitle() ) ) {
3730 // If we transclude ourselves, the final result
3731 // will change based on the new version of the page
3732 $this->mOutput
->setFlag( 'vary-revision' );
3736 return [ $text, $finalTitle ];
3740 * Fetch the unparsed text of a template and register a reference to it.
3741 * @param Title $title
3742 * @return string|bool
3744 public function fetchTemplate( $title ) {
3745 return $this->fetchTemplateAndTitle( $title )[0];
3749 * Static function to get a template
3750 * Can be overridden via ParserOptions::setTemplateCallback().
3752 * @param Title $title
3753 * @param bool|Parser $parser
3757 public static function statelessFetchTemplate( $title, $parser = false ) {
3758 $text = $skip = false;
3759 $finalTitle = $title;
3762 # Loop to fetch the article, with up to 1 redirect
3763 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3764 # Give extensions a chance to select the revision instead
3765 $id = false; # Assume current
3766 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3767 [ $parser, $title, &$skip, &$id ] );
3773 'page_id' => $title->getArticleID(),
3780 $rev = Revision
::newFromId( $id );
3781 } elseif ( $parser ) {
3782 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3784 $rev = Revision
::newFromTitle( $title );
3786 $rev_id = $rev ?
$rev->getId() : 0;
3787 # If there is no current revision, there is no page
3788 if ( $id === false && !$rev ) {
3789 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3790 $linkCache->addBadLinkObj( $title );
3795 'page_id' => $title->getArticleID(),
3796 'rev_id' => $rev_id ];
3797 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3798 # We fetched a rev from a different title; register it too...
3800 'title' => $rev->getTitle(),
3801 'page_id' => $rev->getPage(),
3802 'rev_id' => $rev_id ];
3806 $content = $rev->getContent();
3807 $text = $content ?
$content->getWikitextForTransclusion() : null;
3809 Hooks
::run( 'ParserFetchTemplate',
3810 [ $parser, $title, $rev, &$text, &$deps ] );
3812 if ( $text === false ||
$text === null ) {
3816 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3817 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3818 lcfirst( $title->getText() ) )->inContentLanguage();
3819 if ( !$message->exists() ) {
3823 $content = $message->content();
3824 $text = $message->plain();
3832 $finalTitle = $title;
3833 $title = $content->getRedirectTarget();
3837 'finalTitle' => $finalTitle,
3842 * Fetch a file and its title and register a reference to it.
3843 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3844 * @param Title $title
3845 * @param array $options Array of options to RepoGroup::findFile
3847 * @deprecated since 1.32, use fetchFileAndTitle instead
3849 public function fetchFile( $title, $options = [] ) {
3850 wfDeprecated( __METHOD__
, '1.32' );
3851 return $this->fetchFileAndTitle( $title, $options )[0];
3855 * Fetch a file and its title and register a reference to it.
3856 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3857 * @param Title $title
3858 * @param array $options Array of options to RepoGroup::findFile
3859 * @return array ( File or false, Title of file )
3861 public function fetchFileAndTitle( $title, $options = [] ) {
3862 $file = $this->fetchFileNoRegister( $title, $options );
3864 $time = $file ?
$file->getTimestamp() : false;
3865 $sha1 = $file ?
$file->getSha1() : false;
3866 # Register the file as a dependency...
3867 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3868 if ( $file && !$title->equals( $file->getTitle() ) ) {
3869 # Update fetched file title
3870 $title = $file->getTitle();
3871 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3873 return [ $file, $title ];
3877 * Helper function for fetchFileAndTitle.
3879 * Also useful if you need to fetch a file but not use it yet,
3880 * for example to get the file's handler.
3882 * @param Title $title
3883 * @param array $options Array of options to RepoGroup::findFile
3886 protected function fetchFileNoRegister( $title, $options = [] ) {
3887 if ( isset( $options['broken'] ) ) {
3888 $file = false; // broken thumbnail forced by hook
3889 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3890 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3891 } else { // get by (name,timestamp)
3892 $file = MediaWikiServices
::getInstance()->getRepoGroup()->findFile( $title, $options );
3898 * Transclude an interwiki link.
3900 * @param Title $title
3901 * @param string $action Usually one of (raw, render)
3905 public function interwikiTransclude( $title, $action ) {
3906 if ( !$this->svcOptions
->get( 'EnableScaryTranscluding' ) ) {
3907 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3910 $url = $title->getFullURL( [ 'action' => $action ] );
3911 if ( strlen( $url ) > 1024 ) {
3912 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3915 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3917 $fname = __METHOD__
;
3918 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3920 $data = $cache->getWithSetCallback(
3921 $cache->makeGlobalKey(
3922 'interwiki-transclude',
3923 ( $wikiId !== false ) ?
$wikiId : 'external',
3926 $this->svcOptions
->get( 'TranscludeCacheExpiry' ),
3927 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3928 $req = MWHttpRequest
::factory( $url, [], $fname );
3930 $status = $req->execute(); // Status object
3931 if ( !$status->isOK() ) {
3932 $ttl = $cache::TTL_UNCACHEABLE
;
3933 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3934 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3938 'text' => $status->isOK() ?
$req->getContent() : null,
3939 'code' => $req->getStatus()
3943 'checkKeys' => ( $wikiId !== false )
3944 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3946 'pcGroup' => 'interwiki-transclude:5',
3947 'pcTTL' => $cache::TTL_PROC_LONG
3951 if ( is_string( $data['text'] ) ) {
3952 $text = $data['text'];
3953 } elseif ( $data['code'] != 200 ) {
3954 // Though we failed to fetch the content, this status is useless.
3955 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3956 ->params( $url, $data['code'] )->inContentLanguage()->text();
3958 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3965 * Triple brace replacement -- used for template arguments
3968 * @param array $piece
3969 * @param PPFrame $frame
3973 public function argSubstitution( $piece, $frame ) {
3975 $parts = $piece['parts'];
3976 $nameWithSpaces = $frame->expand( $piece['title'] );
3977 $argName = trim( $nameWithSpaces );
3979 $text = $frame->getArgument( $argName );
3980 if ( $text === false && $parts->getLength() > 0
3981 && ( $this->ot
['html']
3983 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3986 # No match in frame, use the supplied default
3987 $object = $parts->item( 0 )->getChildren();
3989 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3990 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3991 $this->limitationWarn( 'post-expand-template-argument' );
3994 if ( $text === false && $object === false ) {
3996 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3998 if ( $error !== false ) {
4001 if ( $object !== false ) {
4002 $ret = [ 'object' => $object ];
4004 $ret = [ 'text' => $text ];
4011 * Return the text to be used for a given extension tag.
4012 * This is the ghost of strip().
4014 * @param array $params Associative array of parameters:
4015 * name PPNode for the tag name
4016 * attr PPNode for unparsed text where tag attributes are thought to be
4017 * attributes Optional associative array of parsed attributes
4018 * inner Contents of extension element
4019 * noClose Original text did not have a close tag
4020 * @param PPFrame $frame
4022 * @throws MWException
4025 public function extensionSubstitution( $params, $frame ) {
4026 static $errorStr = '<span class="error">';
4027 static $errorLen = 20;
4029 $name = $frame->expand( $params['name'] );
4030 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4031 // Probably expansion depth or node count exceeded. Just punt the
4036 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
4037 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4042 // We can't safely check if the expansion for $content resulted in an
4043 // error, because the content could happen to be the error string
4045 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
4047 $marker = self
::MARKER_PREFIX
. "-$name-"
4048 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
4050 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
4051 ( $this->ot
['html'] ||
$this->ot
['pre'] );
4052 if ( $isFunctionTag ) {
4053 $markerType = 'none';
4055 $markerType = 'general';
4057 if ( $this->ot
['html'] ||
$isFunctionTag ) {
4058 $name = strtolower( $name );
4059 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
4060 if ( isset( $params['attributes'] ) ) {
4061 $attributes +
= $params['attributes'];
4064 if ( isset( $this->mTagHooks
[$name] ) ) {
4065 $output = call_user_func_array( $this->mTagHooks
[$name],
4066 [ $content, $attributes, $this, $frame ] );
4067 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4068 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4070 // Avoid PHP 7.1 warning from passing $this by reference
4072 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4074 $output = '<span class="error">Invalid tag extension name: ' .
4075 htmlspecialchars( $name ) . '</span>';
4078 if ( is_array( $output ) ) {
4081 $output = $flags[0];
4082 if ( isset( $flags['markerType'] ) ) {
4083 $markerType = $flags['markerType'];
4087 if ( is_null( $attrText ) ) {
4090 if ( isset( $params['attributes'] ) ) {
4091 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4092 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4093 htmlspecialchars( $attrValue ) . '"';
4096 if ( $content === null ) {
4097 $output = "<$name$attrText/>";
4099 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4100 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4104 $output = "<$name$attrText>$content$close";
4108 if ( $markerType === 'none' ) {
4110 } elseif ( $markerType === 'nowiki' ) {
4111 $this->mStripState
->addNoWiki( $marker, $output );
4112 } elseif ( $markerType === 'general' ) {
4113 $this->mStripState
->addGeneral( $marker, $output );
4115 throw new MWException( __METHOD__
. ': invalid marker type' );
4121 * Increment an include size counter
4123 * @param string $type The type of expansion
4124 * @param int $size The size of the text
4125 * @return bool False if this inclusion would take it over the maximum, true otherwise
4127 public function incrementIncludeSize( $type, $size ) {
4128 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4131 $this->mIncludeSizes
[$type] +
= $size;
4137 * Increment the expensive function count
4139 * @return bool False if the limit has been exceeded
4141 public function incrementExpensiveFunctionCount() {
4142 $this->mExpensiveFunctionCount++
;
4143 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4147 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4148 * Fills $this->mDoubleUnderscores, returns the modified text
4150 * @param string $text
4154 public function doDoubleUnderscore( $text ) {
4155 # The position of __TOC__ needs to be recorded
4156 $mw = $this->magicWordFactory
->get( 'toc' );
4157 if ( $mw->match( $text ) ) {
4158 $this->mShowToc
= true;
4159 $this->mForceTocPosition
= true;
4161 # Set a placeholder. At the end we'll fill it in with the TOC.
4162 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4164 # Only keep the first one.
4165 $text = $mw->replace( '', $text );
4168 # Now match and remove the rest of them
4169 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4170 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4172 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4173 $this->mOutput
->mNoGallery
= true;
4175 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4176 $this->mShowToc
= false;
4178 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4179 && $this->mTitle
->getNamespace() == NS_CATEGORY
4181 $this->addTrackingCategory( 'hidden-category-category' );
4183 # (T10068) Allow control over whether robots index a page.
4184 # __INDEX__ always overrides __NOINDEX__, see T16899
4185 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4186 $this->mOutput
->setIndexPolicy( 'noindex' );
4187 $this->addTrackingCategory( 'noindex-category' );
4189 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4190 $this->mOutput
->setIndexPolicy( 'index' );
4191 $this->addTrackingCategory( 'index-category' );
4194 # Cache all double underscores in the database
4195 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4196 $this->mOutput
->setProperty( $key, '' );
4203 * @see ParserOutput::addTrackingCategory()
4204 * @param string $msg Message key
4205 * @return bool Whether the addition was successful
4207 public function addTrackingCategory( $msg ) {
4208 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4212 * This function accomplishes several tasks:
4213 * 1) Auto-number headings if that option is enabled
4214 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4215 * 3) Add a Table of contents on the top for users who have enabled the option
4216 * 4) Auto-anchor headings
4218 * It loops through all headlines, collects the necessary data, then splits up the
4219 * string and re-inserts the newly formatted headlines.
4221 * @param string $text
4222 * @param string $origText Original, untouched wikitext
4223 * @param bool $isMain
4224 * @return mixed|string
4227 public function formatHeadings( $text, $origText, $isMain = true ) {
4228 # Inhibit editsection links if requested in the page
4229 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4230 $maybeShowEditLink = false;
4232 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4235 # Get all headlines for numbering them and adding funky stuff like [edit]
4236 # links - this is for later, but we need the number of headlines right now
4237 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4238 # be trimmed here since whitespace in HTML headings is significant.
4240 $numMatches = preg_match_all(
4241 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4246 # if there are fewer than 4 headlines in the article, do not show TOC
4247 # unless it's been explicitly enabled.
4248 $enoughToc = $this->mShowToc
&&
4249 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4251 # Allow user to stipulate that a page should have a "new section"
4252 # link added via __NEWSECTIONLINK__
4253 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4254 $this->mOutput
->setNewSection( true );
4257 # Allow user to remove the "new section"
4258 # link via __NONEWSECTIONLINK__
4259 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4260 $this->mOutput
->hideNewSection( true );
4263 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4264 # override above conditions and always show TOC above first header
4265 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4266 $this->mShowToc
= true;
4274 # Ugh .. the TOC should have neat indentation levels which can be
4275 # passed to the skin functions. These are determined here
4279 $sublevelCount = [];
4285 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4286 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4287 $oldType = $this->mOutputType
;
4288 $this->setOutputType( self
::OT_WIKI
);
4289 $frame = $this->getPreprocessor()->newFrame();
4290 $root = $this->preprocessToDom( $origText );
4291 $node = $root->getFirstChild();
4296 $headlines = $numMatches !== false ?
$matches[3] : [];
4298 $maxTocLevel = $this->svcOptions
->get( 'MaxTocLevel' );
4299 foreach ( $headlines as $headline ) {
4300 $isTemplate = false;
4302 $sectionIndex = false;
4304 $markerMatches = [];
4305 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4306 $serial = $markerMatches[1];
4307 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4308 $isTemplate = ( $titleText != $baseTitleText );
4309 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4313 $prevlevel = $level;
4315 $level = $matches[1][$headlineCount];
4317 if ( $level > $prevlevel ) {
4318 # Increase TOC level
4320 $sublevelCount[$toclevel] = 0;
4321 if ( $toclevel < $maxTocLevel ) {
4322 $prevtoclevel = $toclevel;
4323 $toc .= Linker
::tocIndent();
4326 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4327 # Decrease TOC level, find level to jump to
4329 for ( $i = $toclevel; $i > 0; $i-- ) {
4330 if ( $levelCount[$i] == $level ) {
4331 # Found last matching level
4334 } elseif ( $levelCount[$i] < $level ) {
4335 # Found first matching level below current level
4343 if ( $toclevel < $maxTocLevel ) {
4344 if ( $prevtoclevel < $maxTocLevel ) {
4345 # Unindent only if the previous toc level was shown :p
4346 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4347 $prevtoclevel = $toclevel;
4349 $toc .= Linker
::tocLineEnd();
4353 # No change in level, end TOC line
4354 if ( $toclevel < $maxTocLevel ) {
4355 $toc .= Linker
::tocLineEnd();
4359 $levelCount[$toclevel] = $level;
4361 # count number of headlines for each level
4362 $sublevelCount[$toclevel]++
;
4364 for ( $i = 1; $i <= $toclevel; $i++
) {
4365 if ( !empty( $sublevelCount[$i] ) ) {
4369 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4374 # The safe header is a version of the header text safe to use for links
4376 # Remove link placeholders by the link text.
4377 # <!--LINK number-->
4379 # link text with suffix
4380 # Do this before unstrip since link text can contain strip markers
4381 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4383 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4384 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4386 # Remove any <style> or <script> tags (T198618)
4387 $safeHeadline = preg_replace(
4388 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4393 # Strip out HTML (first regex removes any tag not allowed)
4395 # * <sup> and <sub> (T10393)
4399 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4400 # * <s> and <strike> (T35715)
4401 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4402 # to allow setting directionality in toc items.
4403 $tocline = preg_replace(
4405 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4406 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4412 # Strip '<span></span>', which is the result from the above if
4413 # <span id="foo"></span> is used to produce an additional anchor
4415 $tocline = str_replace( '<span></span>', '', $tocline );
4417 $tocline = trim( $tocline );
4419 # For the anchor, strip out HTML-y stuff period
4420 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4421 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4423 # Save headline for section edit hint before it's escaped
4424 $headlineHint = $safeHeadline;
4426 # Decode HTML entities
4427 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4429 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4431 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4432 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4433 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4434 if ( $fallbackHeadline === $safeHeadline ) {
4435 # No reason to have both (in fact, we can't)
4436 $fallbackHeadline = false;
4439 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4440 # @todo FIXME: We may be changing them depending on the current locale.
4441 $arrayKey = strtolower( $safeHeadline );
4442 if ( $fallbackHeadline === false ) {
4443 $fallbackArrayKey = false;
4445 $fallbackArrayKey = strtolower( $fallbackHeadline );
4448 # Create the anchor for linking from the TOC to the section
4449 $anchor = $safeHeadline;
4450 $fallbackAnchor = $fallbackHeadline;
4451 if ( isset( $refers[$arrayKey] ) ) {
4452 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4453 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4455 $linkAnchor .= "_$i";
4456 $refers["${arrayKey}_$i"] = true;
4458 $refers[$arrayKey] = true;
4460 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4461 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4462 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4463 $fallbackAnchor .= "_$i";
4464 $refers["${fallbackArrayKey}_$i"] = true;
4466 $refers[$fallbackArrayKey] = true;
4469 # Don't number the heading if it is the only one (looks silly)
4470 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4471 # the two are different if the line contains a link
4472 $headline = Html
::element(
4474 [ 'class' => 'mw-headline-number' ],
4476 ) . ' ' . $headline;
4479 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4480 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4481 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4484 # Add the section to the section tree
4485 # Find the DOM node for this header
4486 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4487 while ( $node && !$noOffset ) {
4488 if ( $node->getName() === 'h' ) {
4489 $bits = $node->splitHeading();
4490 if ( $bits['i'] == $sectionIndex ) {
4494 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4495 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4496 $node = $node->getNextSibling();
4499 'toclevel' => $toclevel,
4502 'number' => $numbering,
4503 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4504 'fromtitle' => $titleText,
4505 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4506 'anchor' => $anchor,
4509 # give headline the correct <h#> tag
4510 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4511 // Output edit section links as markers with styles that can be customized by skins
4512 if ( $isTemplate ) {
4513 # Put a T flag in the section identifier, to indicate to extractSections()
4514 # that sections inside <includeonly> should be counted.
4515 $editsectionPage = $titleText;
4516 $editsectionSection = "T-$sectionIndex";
4517 $editsectionContent = null;
4519 $editsectionPage = $this->mTitle
->getPrefixedText();
4520 $editsectionSection = $sectionIndex;
4521 $editsectionContent = $headlineHint;
4523 // We use a bit of pesudo-xml for editsection markers. The
4524 // language converter is run later on. Using a UNIQ style marker
4525 // leads to the converter screwing up the tokens when it
4526 // converts stuff. And trying to insert strip tags fails too. At
4527 // this point all real inputted tags have already been escaped,
4528 // so we don't have to worry about a user trying to input one of
4529 // these markers directly. We use a page and section attribute
4530 // to stop the language converter from converting these
4531 // important bits of data, but put the headline hint inside a
4532 // content block because the language converter is supposed to
4533 // be able to convert that piece of data.
4534 // Gets replaced with html in ParserOutput::getText
4535 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4536 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4537 if ( $editsectionContent !== null ) {
4538 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4545 $head[$headlineCount] = Linker
::makeHeadline( $level,
4546 $matches['attrib'][$headlineCount], $anchor, $headline,
4547 $editlink, $fallbackAnchor );
4552 $this->setOutputType( $oldType );
4554 # Never ever show TOC if no headers
4555 if ( $numVisible < 1 ) {
4560 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4561 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4563 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4564 $this->mOutput
->setTOCHTML( $toc );
4565 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4569 $this->mOutput
->setSections( $tocraw );
4572 # split up and insert constructed headlines
4573 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4576 // build an array of document sections
4578 foreach ( $blocks as $block ) {
4579 // $head is zero-based, sections aren't.
4580 if ( empty( $head[$i - 1] ) ) {
4581 $sections[$i] = $block;
4583 $sections[$i] = $head[$i - 1] . $block;
4587 * Send a hook, one per section.
4588 * The idea here is to be able to make section-level DIVs, but to do so in a
4589 * lower-impact, more correct way than r50769
4592 * $section : the section number
4593 * &$sectionContent : ref to the content of the section
4594 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4596 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4601 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4602 // append the TOC at the beginning
4603 // Top anchor now in skin
4604 $sections[0] .= $toc . "\n";
4607 $full .= implode( '', $sections );
4609 if ( $this->mForceTocPosition
) {
4610 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4617 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4618 * conversion, substituting signatures, {{subst:}} templates, etc.
4620 * @param string $text The text to transform
4621 * @param Title $title The Title object for the current article
4622 * @param User $user The User object describing the current user
4623 * @param ParserOptions $options Parsing options
4624 * @param bool $clearState Whether to clear the parser state first
4625 * @return string The altered wiki markup
4627 public function preSaveTransform( $text, Title
$title, User
$user,
4628 ParserOptions
$options, $clearState = true
4630 if ( $clearState ) {
4631 $magicScopeVariable = $this->lock();
4633 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4634 $this->setUser( $user );
4636 // Strip U+0000 NULL (T159174)
4637 $text = str_replace( "\000", '', $text );
4639 // We still normalize line endings for backwards-compatibility
4640 // with other code that just calls PST, but this should already
4641 // be handled in TextContent subclasses
4642 $text = TextContent
::normalizeLineEndings( $text );
4644 if ( $options->getPreSaveTransform() ) {
4645 $text = $this->pstPass2( $text, $user );
4647 $text = $this->mStripState
->unstripBoth( $text );
4649 $this->setUser( null ); # Reset
4655 * Pre-save transform helper function
4657 * @param string $text
4662 private function pstPass2( $text, $user ) {
4663 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4664 # $this->contLang here in order to give everyone the same signature and use the default one
4665 # rather than the one selected in each user's preferences. (see also T14815)
4666 $ts = $this->mOptions
->getTimestamp();
4667 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4668 $ts = $timestamp->format( 'YmdHis' );
4669 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4671 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4673 # Variable replacement
4674 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4675 $text = $this->replaceVariables( $text );
4677 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4678 # which may corrupt this parser instance via its wfMessage()->text() call-
4681 if ( strpos( $text, '~~~' ) !== false ) {
4682 $sigText = $this->getUserSig( $user );
4683 $text = strtr( $text, [
4685 '~~~~' => "$sigText $d",
4688 # The main two signature forms used above are time-sensitive
4689 $this->mOutput
->setFlag( 'user-signature' );
4692 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4693 $tc = '[' . Title
::legalChars() . ']';
4694 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4696 // [[ns:page (context)|]]
4697 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4698 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4699 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4700 // [[ns:page (context), context|]] (using either single or double-width comma)
4701 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4702 // [[|page]] (reverse pipe trick: add context from page title)
4703 $p2 = "/\[\[\\|($tc+)]]/";
4705 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4706 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4707 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4708 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4710 $t = $this->mTitle
->getText();
4712 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4713 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4714 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4715 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4717 # if there's no context, don't bother duplicating the title
4718 $text = preg_replace( $p2, '[[\\1]]', $text );
4725 * Fetch the user's signature text, if any, and normalize to
4726 * validated, ready-to-insert wikitext.
4727 * If you have pre-fetched the nickname or the fancySig option, you can
4728 * specify them here to save a database query.
4729 * Do not reuse this parser instance after calling getUserSig(),
4730 * as it may have changed.
4732 * @param User &$user
4733 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4734 * @param bool|null $fancySig whether the nicknname is the complete signature
4735 * or null to use default value
4738 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4739 $username = $user->getName();
4741 # If not given, retrieve from the user object.
4742 if ( $nickname === false ) {
4743 $nickname = $user->getOption( 'nickname' );
4746 if ( is_null( $fancySig ) ) {
4747 $fancySig = $user->getBoolOption( 'fancysig' );
4750 $nickname = $nickname == null ?
$username : $nickname;
4752 if ( mb_strlen( $nickname ) > $this->svcOptions
->get( 'MaxSigChars' ) ) {
4753 $nickname = $username;
4754 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4755 } elseif ( $fancySig !== false ) {
4756 # Sig. might contain markup; validate this
4757 if ( $this->validateSig( $nickname ) !== false ) {
4758 # Validated; clean up (if needed) and return it
4759 return $this->cleanSig( $nickname, true );
4761 # Failed to validate; fall back to the default
4762 $nickname = $username;
4763 wfDebug( __METHOD__
. ": $username has bad XML tags in signature.\n" );
4767 # Make sure nickname doesnt get a sig in a sig
4768 $nickname = self
::cleanSigInSig( $nickname );
4770 # If we're still here, make it a link to the user page
4771 $userText = wfEscapeWikiText( $username );
4772 $nickText = wfEscapeWikiText( $nickname );
4773 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4775 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4776 ->title( $this->getTitle() )->text();
4780 * Check that the user's signature contains no bad XML
4782 * @param string $text
4783 * @return string|bool An expanded string, or false if invalid.
4785 public function validateSig( $text ) {
4786 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4790 * Clean up signature text
4792 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4793 * 2) Substitute all transclusions
4795 * @param string $text
4796 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4797 * @return string Signature text
4799 public function cleanSig( $text, $parsing = false ) {
4802 $magicScopeVariable = $this->lock();
4803 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4806 # Option to disable this feature
4807 if ( !$this->mOptions
->getCleanSignatures() ) {
4811 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4812 # => Move this logic to braceSubstitution()
4813 $substWord = $this->magicWordFactory
->get( 'subst' );
4814 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4815 $substText = '{{' . $substWord->getSynonym( 0 );
4817 $text = preg_replace( $substRegex, $substText, $text );
4818 $text = self
::cleanSigInSig( $text );
4819 $dom = $this->preprocessToDom( $text );
4820 $frame = $this->getPreprocessor()->newFrame();
4821 $text = $frame->expand( $dom );
4824 $text = $this->mStripState
->unstripBoth( $text );
4831 * Strip 3, 4 or 5 tildes out of signatures.
4833 * @param string $text
4834 * @return string Signature text with /~{3,5}/ removed
4836 public static function cleanSigInSig( $text ) {
4837 $text = preg_replace( '/~{3,5}/', '', $text );
4842 * Set up some variables which are usually set up in parse()
4843 * so that an external function can call some class members with confidence
4845 * @param Title|null $title
4846 * @param ParserOptions $options
4847 * @param int $outputType
4848 * @param bool $clearState
4850 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4851 $outputType, $clearState = true
4853 $this->startParse( $title, $options, $outputType, $clearState );
4857 * @param Title|null $title
4858 * @param ParserOptions $options
4859 * @param int $outputType
4860 * @param bool $clearState
4862 private function startParse( Title
$title = null, ParserOptions
$options,
4863 $outputType, $clearState = true
4865 $this->setTitle( $title );
4866 $this->mOptions
= $options;
4867 $this->setOutputType( $outputType );
4868 if ( $clearState ) {
4869 $this->clearState();
4874 * Wrapper for preprocess()
4876 * @param string $text The text to preprocess
4877 * @param ParserOptions $options
4878 * @param Title|null $title Title object or null to use $wgTitle
4881 public function transformMsg( $text, $options, $title = null ) {
4882 static $executing = false;
4884 # Guard against infinite recursion
4895 $text = $this->preprocess( $text, $title, $options );
4902 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4903 * The callback should have the following form:
4904 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4906 * Transform and return $text. Use $parser for any required context, e.g. use
4907 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4909 * Hooks may return extended information by returning an array, of which the
4910 * first numbered element (index 0) must be the return string, and all other
4911 * entries are extracted into local variables within an internal function
4912 * in the Parser class.
4914 * This interface (introduced r61913) appears to be undocumented, but
4915 * 'markerType' is used by some core tag hooks to override which strip
4916 * array their results are placed in. **Use great caution if attempting
4917 * this interface, as it is not documented and injudicious use could smash
4918 * private variables.**
4920 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4921 * @param callable $callback The callback function (and object) to use for the tag
4922 * @throws MWException
4923 * @return callable|null The old value of the mTagHooks array associated with the hook
4925 public function setHook( $tag, callable
$callback ) {
4926 $tag = strtolower( $tag );
4927 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4928 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4930 $oldVal = $this->mTagHooks
[$tag] ??
null;
4931 $this->mTagHooks
[$tag] = $callback;
4932 if ( !in_array( $tag, $this->mStripList
) ) {
4933 $this->mStripList
[] = $tag;
4940 * As setHook(), but letting the contents be parsed.
4942 * Transparent tag hooks are like regular XML-style tag hooks, except they
4943 * operate late in the transformation sequence, on HTML instead of wikitext.
4945 * This is probably obsoleted by things dealing with parser frames?
4946 * The only extension currently using it is geoserver.
4949 * @todo better document or deprecate this
4951 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4952 * @param callable $callback The callback function (and object) to use for the tag
4953 * @throws MWException
4954 * @return callable|null The old value of the mTagHooks array associated with the hook
4956 public function setTransparentTagHook( $tag, callable
$callback ) {
4957 $tag = strtolower( $tag );
4958 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4959 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4961 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4962 $this->mTransparentTagHooks
[$tag] = $callback;
4968 * Remove all tag hooks
4970 public function clearTagHooks() {
4971 $this->mTagHooks
= [];
4972 $this->mFunctionTagHooks
= [];
4973 $this->mStripList
= $this->mDefaultStripList
;
4977 * Create a function, e.g. {{sum:1|2|3}}
4978 * The callback function should have the form:
4979 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4981 * Or with Parser::SFH_OBJECT_ARGS:
4982 * function myParserFunction( $parser, $frame, $args ) { ... }
4984 * The callback may either return the text result of the function, or an array with the text
4985 * in element 0, and a number of flags in the other elements. The names of the flags are
4986 * specified in the keys. Valid flags are:
4987 * found The text returned is valid, stop processing the template. This
4989 * nowiki Wiki markup in the return value should be escaped
4990 * isHTML The returned text is HTML, armour it against wikitext transformation
4992 * @param string $id The magic word ID
4993 * @param callable $callback The callback function (and object) to use
4994 * @param int $flags A combination of the following flags:
4995 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4997 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4998 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4999 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5000 * the arguments, and to control the way they are expanded.
5002 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5003 * arguments, for instance:
5004 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5006 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5007 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5008 * working if/when this is changed.
5010 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5013 * Please read the documentation in includes/parser/Preprocessor.php for more information
5014 * about the methods available in PPFrame and PPNode.
5016 * @throws MWException
5017 * @return string|callable The old callback function for this name, if any
5019 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
5020 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
5021 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
5023 # Add to function cache
5024 $mw = $this->magicWordFactory
->get( $id );
5026 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
5029 $synonyms = $mw->getSynonyms();
5030 $sensitive = intval( $mw->isCaseSensitive() );
5032 foreach ( $synonyms as $syn ) {
5034 if ( !$sensitive ) {
5035 $syn = $this->contLang
->lc( $syn );
5038 if ( !( $flags & self
::SFH_NO_HASH
) ) {
5041 # Remove trailing colon
5042 if ( substr( $syn, -1, 1 ) === ':' ) {
5043 $syn = substr( $syn, 0, -1 );
5045 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
5051 * Get all registered function hook identifiers
5055 public function getFunctionHooks() {
5056 $this->firstCallInit();
5057 return array_keys( $this->mFunctionHooks
);
5061 * Create a tag function, e.g. "<test>some stuff</test>".
5062 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5063 * Unlike parser functions, their content is not preprocessed.
5064 * @param string $tag
5065 * @param callable $callback
5067 * @throws MWException
5070 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5071 $tag = strtolower( $tag );
5072 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5073 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5075 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5076 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5078 if ( !in_array( $tag, $this->mStripList
) ) {
5079 $this->mStripList
[] = $tag;
5086 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5087 * Placeholders created in Linker::link()
5089 * @param string &$text
5090 * @param int $options
5092 public function replaceLinkHolders( &$text, $options = 0 ) {
5093 $this->mLinkHolders
->replace( $text );
5097 * Replace "<!--LINK-->" link placeholders with plain text of links
5098 * (not HTML-formatted).
5100 * @param string $text
5103 public function replaceLinkHoldersText( $text ) {
5104 return $this->mLinkHolders
->replaceText( $text );
5108 * Renders an image gallery from a text with one line per image.
5109 * text labels may be given by using |-style alternative text. E.g.
5110 * Image:one.jpg|The number "1"
5111 * Image:tree.jpg|A tree
5112 * given as text will return the HTML of a gallery with two images,
5113 * labeled 'The number "1"' and
5116 * @param string $text
5117 * @param array $params
5118 * @return string HTML
5120 public function renderImageGallery( $text, $params ) {
5122 if ( isset( $params['mode'] ) ) {
5123 $mode = $params['mode'];
5127 $ig = ImageGalleryBase
::factory( $mode );
5128 } catch ( Exception
$e ) {
5129 // If invalid type set, fallback to default.
5130 $ig = ImageGalleryBase
::factory( false );
5133 $ig->setContextTitle( $this->mTitle
);
5134 $ig->setShowBytes( false );
5135 $ig->setShowDimensions( false );
5136 $ig->setShowFilename( false );
5137 $ig->setParser( $this );
5138 $ig->setHideBadImages();
5139 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5141 if ( isset( $params['showfilename'] ) ) {
5142 $ig->setShowFilename( true );
5144 $ig->setShowFilename( false );
5146 if ( isset( $params['caption'] ) ) {
5147 // NOTE: We aren't passing a frame here or below. Frame info
5148 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5149 // See T107332#4030581
5150 $caption = $this->recursiveTagParse( $params['caption'] );
5151 $ig->setCaptionHtml( $caption );
5153 if ( isset( $params['perrow'] ) ) {
5154 $ig->setPerRow( $params['perrow'] );
5156 if ( isset( $params['widths'] ) ) {
5157 $ig->setWidths( $params['widths'] );
5159 if ( isset( $params['heights'] ) ) {
5160 $ig->setHeights( $params['heights'] );
5162 $ig->setAdditionalOptions( $params );
5164 // Avoid PHP 7.1 warning from passing $this by reference
5166 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5168 $lines = StringUtils
::explode( "\n", $text );
5169 foreach ( $lines as $line ) {
5170 # match lines like these:
5171 # Image:someimage.jpg|This is some image
5173 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5175 if ( count( $matches ) == 0 ) {
5179 if ( strpos( $matches[0], '%' ) !== false ) {
5180 $matches[1] = rawurldecode( $matches[1] );
5182 $title = Title
::newFromText( $matches[1], NS_FILE
);
5183 if ( is_null( $title ) ) {
5184 # Bogus title. Ignore these so we don't bomb out later.
5188 # We need to get what handler the file uses, to figure out parameters.
5189 # Note, a hook can overide the file name, and chose an entirely different
5190 # file (which potentially could be of a different type and have different handler).
5193 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5194 [ $this, $title, &$options, &$descQuery ] );
5195 # Don't register it now, as TraditionalImageGallery does that later.
5196 $file = $this->fetchFileNoRegister( $title, $options );
5197 $handler = $file ?
$file->getHandler() : false;
5200 'img_alt' => 'gallery-internal-alt',
5201 'img_link' => 'gallery-internal-link',
5204 $paramMap +
= $handler->getParamMap();
5205 // We don't want people to specify per-image widths.
5206 // Additionally the width parameter would need special casing anyhow.
5207 unset( $paramMap['img_width'] );
5210 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5215 $handlerOptions = [];
5216 if ( isset( $matches[3] ) ) {
5217 // look for an |alt= definition while trying not to break existing
5218 // captions with multiple pipes (|) in it, until a more sensible grammar
5219 // is defined for images in galleries
5221 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5222 // splitting on '|' is a bit odd, and different from makeImage.
5223 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5224 // Protect LanguageConverter markup
5225 $parameterMatches = StringUtils
::delimiterExplode(
5226 '-{', '}-', '|', $matches[3], true /* nested */
5229 foreach ( $parameterMatches as $parameterMatch ) {
5230 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5232 $paramName = $paramMap[$magicName];
5234 switch ( $paramName ) {
5235 case 'gallery-internal-alt':
5236 $alt = $this->stripAltText( $match, false );
5238 case 'gallery-internal-link':
5239 $linkValue = $this->stripAltText( $match, false );
5240 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5241 // Result of LanguageConverter::markNoConversion
5242 // invoked on an external link.
5243 $linkValue = substr( $linkValue, 4, -2 );
5245 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5246 if ( $type === 'link-url' ) {
5248 $this->mOutput
->addExternalLink( $target );
5249 } elseif ( $type === 'link-title' ) {
5250 $link = $target->getLinkURL();
5251 $this->mOutput
->addLink( $target );
5255 // Must be a handler specific parameter.
5256 if ( $handler->validateParam( $paramName, $match ) ) {
5257 $handlerOptions[$paramName] = $match;
5259 // Guess not, consider it as caption.
5260 wfDebug( "$parameterMatch failed parameter validation\n" );
5261 $label = $parameterMatch;
5267 $label = $parameterMatch;
5272 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5274 $html = $ig->toHTML();
5275 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5280 * @param MediaHandler $handler
5283 public function getImageParams( $handler ) {
5285 $handlerClass = get_class( $handler );
5289 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5290 # Initialise static lists
5291 static $internalParamNames = [
5292 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5293 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5294 'bottom', 'text-bottom' ],
5295 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5296 'upright', 'border', 'link', 'alt', 'class' ],
5298 static $internalParamMap;
5299 if ( !$internalParamMap ) {
5300 $internalParamMap = [];
5301 foreach ( $internalParamNames as $type => $names ) {
5302 foreach ( $names as $name ) {
5303 // For grep: img_left, img_right, img_center, img_none,
5304 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5305 // img_bottom, img_text_bottom,
5306 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5307 // img_border, img_link, img_alt, img_class
5308 $magicName = str_replace( '-', '_', "img_$name" );
5309 $internalParamMap[$magicName] = [ $type, $name ];
5314 # Add handler params
5315 $paramMap = $internalParamMap;
5317 $handlerParamMap = $handler->getParamMap();
5318 foreach ( $handlerParamMap as $magic => $paramName ) {
5319 $paramMap[$magic] = [ 'handler', $paramName ];
5322 $this->mImageParams
[$handlerClass] = $paramMap;
5323 $this->mImageParamsMagicArray
[$handlerClass] =
5324 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5326 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5330 * Parse image options text and use it to make an image
5332 * @param Title $title
5333 * @param string $options
5334 * @param LinkHolderArray|bool $holders
5335 * @return string HTML
5337 public function makeImage( $title, $options, $holders = false ) {
5338 # Check if the options text is of the form "options|alt text"
5340 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5341 # * left no resizing, just left align. label is used for alt= only
5342 # * right same, but right aligned
5343 # * none same, but not aligned
5344 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5345 # * center center the image
5346 # * frame Keep original image size, no magnify-button.
5347 # * framed Same as "frame"
5348 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5349 # * upright reduce width for upright images, rounded to full __0 px
5350 # * border draw a 1px border around the image
5351 # * alt Text for HTML alt attribute (defaults to empty)
5352 # * class Set a class for img node
5353 # * link Set the target of the image link. Can be external, interwiki, or local
5354 # vertical-align values (no % or length right now):
5364 # Protect LanguageConverter markup when splitting into parts
5365 $parts = StringUtils
::delimiterExplode(
5366 '-{', '}-', '|', $options, true /* allow nesting */
5369 # Give extensions a chance to select the file revision for us
5372 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5373 [ $this, $title, &$options, &$descQuery ] );
5374 # Fetch and register the file (file title may be different via hooks)
5375 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5378 $handler = $file ?
$file->getHandler() : false;
5380 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5383 $this->addTrackingCategory( 'broken-file-category' );
5386 # Process the input parameters
5388 $params = [ 'frame' => [], 'handler' => [],
5389 'horizAlign' => [], 'vertAlign' => [] ];
5390 $seenformat = false;
5391 foreach ( $parts as $part ) {
5392 $part = trim( $part );
5393 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5395 if ( isset( $paramMap[$magicName] ) ) {
5396 list( $type, $paramName ) = $paramMap[$magicName];
5398 # Special case; width and height come in one variable together
5399 if ( $type === 'handler' && $paramName === 'width' ) {
5400 $parsedWidthParam = self
::parseWidthParam( $value );
5401 if ( isset( $parsedWidthParam['width'] ) ) {
5402 $width = $parsedWidthParam['width'];
5403 if ( $handler->validateParam( 'width', $width ) ) {
5404 $params[$type]['width'] = $width;
5408 if ( isset( $parsedWidthParam['height'] ) ) {
5409 $height = $parsedWidthParam['height'];
5410 if ( $handler->validateParam( 'height', $height ) ) {
5411 $params[$type]['height'] = $height;
5415 # else no validation -- T15436
5417 if ( $type === 'handler' ) {
5418 # Validate handler parameter
5419 $validated = $handler->validateParam( $paramName, $value );
5421 # Validate internal parameters
5422 switch ( $paramName ) {
5426 # @todo FIXME: Possibly check validity here for
5427 # manualthumb? downstream behavior seems odd with
5428 # missing manual thumbs.
5430 $value = $this->stripAltText( $value, $holders );
5433 list( $paramName, $value ) =
5434 $this->parseLinkParameter(
5435 $this->stripAltText( $value, $holders )
5439 if ( $paramName === 'no-link' ) {
5442 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5443 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5450 // use first appearing option, discard others.
5451 $validated = !$seenformat;
5455 # Most other things appear to be empty or numeric...
5456 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5461 $params[$type][$paramName] = $value;
5465 if ( !$validated ) {
5470 # Process alignment parameters
5471 if ( $params['horizAlign'] ) {
5472 $params['frame']['align'] = key( $params['horizAlign'] );
5474 if ( $params['vertAlign'] ) {
5475 $params['frame']['valign'] = key( $params['vertAlign'] );
5478 $params['frame']['caption'] = $caption;
5480 # Will the image be presented in a frame, with the caption below?
5481 $imageIsFramed = isset( $params['frame']['frame'] )
5482 ||
isset( $params['frame']['framed'] )
5483 ||
isset( $params['frame']['thumbnail'] )
5484 ||
isset( $params['frame']['manualthumb'] );
5486 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5487 # came to also set the caption, ordinary text after the image -- which
5488 # makes no sense, because that just repeats the text multiple times in
5489 # screen readers. It *also* came to set the title attribute.
5490 # Now that we have an alt attribute, we should not set the alt text to
5491 # equal the caption: that's worse than useless, it just repeats the
5492 # text. This is the framed/thumbnail case. If there's no caption, we
5493 # use the unnamed parameter for alt text as well, just for the time be-
5494 # ing, if the unnamed param is set and the alt param is not.
5495 # For the future, we need to figure out if we want to tweak this more,
5496 # e.g., introducing a title= parameter for the title; ignoring the un-
5497 # named parameter entirely for images without a caption; adding an ex-
5498 # plicit caption= parameter and preserving the old magic unnamed para-
5500 if ( $imageIsFramed ) { # Framed image
5501 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5502 # No caption or alt text, add the filename as the alt text so
5503 # that screen readers at least get some description of the image
5504 $params['frame']['alt'] = $title->getText();
5506 # Do not set $params['frame']['title'] because tooltips don't make sense
5508 } else { # Inline image
5509 if ( !isset( $params['frame']['alt'] ) ) {
5510 # No alt text, use the "caption" for the alt text
5511 if ( $caption !== '' ) {
5512 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5514 # No caption, fall back to using the filename for the
5516 $params['frame']['alt'] = $title->getText();
5519 # Use the "caption" for the tooltip text
5520 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5522 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5524 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5526 # Linker does the rest
5527 $time = $options['time'] ??
false;
5528 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5529 $time, $descQuery, $this->mOptions
->getThumbSize() );
5531 # Give the handler a chance to modify the parser object
5533 $handler->parserTransformHook( $this, $file );
5540 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5542 * Adds an entry to appropriate link tables.
5545 * @param string $value
5546 * @return array of `[ type, target ]`, where:
5547 * - `type` is one of:
5548 * - `null`: Given value is not a valid link target, use default
5549 * - `'no-link'`: Given value is empty, do not generate a link
5550 * - `'link-url'`: Given value is a valid external link
5551 * - `'link-title'`: Given value is a valid internal link
5553 * - When `type` is `null` or `'no-link'`: `false`
5554 * - When `type` is `'link-url'`: URL string corresponding to given value
5555 * - When `type` is `'link-title'`: Title object corresponding to given value
5557 public function parseLinkParameter( $value ) {
5558 $chars = self
::EXT_LINK_URL_CLASS
;
5559 $addr = self
::EXT_LINK_ADDR
;
5560 $prots = $this->mUrlProtocols
;
5563 if ( $value === '' ) {
5565 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5566 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5567 $this->mOutput
->addExternalLink( $value );
5572 $linkTitle = Title
::newFromText( $value );
5574 $this->mOutput
->addLink( $linkTitle );
5575 $type = 'link-title';
5576 $target = $linkTitle;
5579 return [ $type, $target ];
5583 * @param string $caption
5584 * @param LinkHolderArray|bool $holders
5585 * @return mixed|string
5587 protected function stripAltText( $caption, $holders ) {
5588 # Strip bad stuff out of the title (tooltip). We can't just use
5589 # replaceLinkHoldersText() here, because if this function is called
5590 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5592 $tooltip = $holders->replaceText( $caption );
5594 $tooltip = $this->replaceLinkHoldersText( $caption );
5597 # make sure there are no placeholders in thumbnail attributes
5598 # that are later expanded to html- so expand them now and
5600 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5601 # Compatibility hack! In HTML certain entity references not terminated
5602 # by a semicolon are decoded (but not if we're in an attribute; that's
5603 # how link URLs get away without properly escaping & in queries).
5604 # But wikitext has always required semicolon-termination of entities,
5605 # so encode & where needed to avoid decode of semicolon-less entities.
5607 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5608 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5609 $tooltip = preg_replace( "/
5610 & # 1. entity prefix
5611 (?= # 2. followed by:
5612 (?: # a. one of the legacy semicolon-less named entities
5613 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5614 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5615 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5616 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5617 U(?:acute|circ|grave|uml)|Yacute|
5618 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5619 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5620 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5621 frac(?:1(?:2|4)|34)|
5622 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5623 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5624 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5625 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5626 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5627 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5628 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5629 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5630 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5632 (?:[^;]|$)) # b. and not followed by a semicolon
5633 # S = study, for efficiency
5634 /Sx", '&', $tooltip );
5635 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5641 * Set a flag in the output object indicating that the content is dynamic and
5642 * shouldn't be cached.
5643 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5645 public function disableCache() {
5646 wfDebug( "Parser output marked as uncacheable.\n" );
5647 if ( !$this->mOutput
) {
5648 throw new MWException( __METHOD__
.
5649 " can only be called when actually parsing something" );
5651 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5655 * Callback from the Sanitizer for expanding items found in HTML attribute
5656 * values, so they can be safely tested and escaped.
5658 * @param string &$text
5659 * @param bool|PPFrame $frame
5662 public function attributeStripCallback( &$text, $frame = false ) {
5663 $text = $this->replaceVariables( $text, $frame );
5664 $text = $this->mStripState
->unstripBoth( $text );
5673 public function getTags() {
5674 $this->firstCallInit();
5676 array_keys( $this->mTransparentTagHooks
),
5677 array_keys( $this->mTagHooks
),
5678 array_keys( $this->mFunctionTagHooks
)
5686 public function getFunctionSynonyms() {
5687 $this->firstCallInit();
5688 return $this->mFunctionSynonyms
;
5695 public function getUrlProtocols() {
5696 return $this->mUrlProtocols
;
5700 * Replace transparent tags in $text with the values given by the callbacks.
5702 * Transparent tag hooks are like regular XML-style tag hooks, except they
5703 * operate late in the transformation sequence, on HTML instead of wikitext.
5705 * @param string $text
5709 public function replaceTransparentTags( $text ) {
5711 $elements = array_keys( $this->mTransparentTagHooks
);
5712 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5715 foreach ( $matches as $marker => $data ) {
5716 list( $element, $content, $params, $tag ) = $data;
5717 $tagName = strtolower( $element );
5718 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5719 $output = call_user_func_array(
5720 $this->mTransparentTagHooks
[$tagName],
5721 [ $content, $params, $this ]
5726 $replacements[$marker] = $output;
5728 return strtr( $text, $replacements );
5732 * Break wikitext input into sections, and either pull or replace
5733 * some particular section's text.
5735 * External callers should use the getSection and replaceSection methods.
5737 * @param string $text Page wikitext
5738 * @param string|int $sectionId A section identifier string of the form:
5739 * "<flag1> - <flag2> - ... - <section number>"
5741 * Currently the only recognised flag is "T", which means the target section number
5742 * was derived during a template inclusion parse, in other words this is a template
5743 * section edit link. If no flags are given, it was an ordinary section edit link.
5744 * This flag is required to avoid a section numbering mismatch when a section is
5745 * enclosed by "<includeonly>" (T8563).
5747 * The section number 0 pulls the text before the first heading; other numbers will
5748 * pull the given section along with its lower-level subsections. If the section is
5749 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5751 * Section 0 is always considered to exist, even if it only contains the empty
5752 * string. If $text is the empty string and section 0 is replaced, $newText is
5755 * @param string $mode One of "get" or "replace"
5756 * @param string $newText Replacement text for section data.
5757 * @return string For "get", the extracted section text.
5758 * for "replace", the whole page with the section replaced.
5760 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5761 global $wgTitle; # not generally used but removes an ugly failure mode
5763 $magicScopeVariable = $this->lock();
5764 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5766 $frame = $this->getPreprocessor()->newFrame();
5768 # Process section extraction flags
5770 $sectionParts = explode( '-', $sectionId );
5771 $sectionIndex = array_pop( $sectionParts );
5772 foreach ( $sectionParts as $part ) {
5773 if ( $part === 'T' ) {
5774 $flags |
= self
::PTD_FOR_INCLUSION
;
5778 # Check for empty input
5779 if ( strval( $text ) === '' ) {
5780 # Only sections 0 and T-0 exist in an empty document
5781 if ( $sectionIndex == 0 ) {
5782 if ( $mode === 'get' ) {
5788 if ( $mode === 'get' ) {
5796 # Preprocess the text
5797 $root = $this->preprocessToDom( $text, $flags );
5799 # <h> nodes indicate section breaks
5800 # They can only occur at the top level, so we can find them by iterating the root's children
5801 $node = $root->getFirstChild();
5803 # Find the target section
5804 if ( $sectionIndex == 0 ) {
5805 # Section zero doesn't nest, level=big
5806 $targetLevel = 1000;
5809 if ( $node->getName() === 'h' ) {
5810 $bits = $node->splitHeading();
5811 if ( $bits['i'] == $sectionIndex ) {
5812 $targetLevel = $bits['level'];
5816 if ( $mode === 'replace' ) {
5817 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5819 $node = $node->getNextSibling();
5825 if ( $mode === 'get' ) {
5832 # Find the end of the section, including nested sections
5834 if ( $node->getName() === 'h' ) {
5835 $bits = $node->splitHeading();
5836 $curLevel = $bits['level'];
5837 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5841 if ( $mode === 'get' ) {
5842 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5844 $node = $node->getNextSibling();
5847 # Write out the remainder (in replace mode only)
5848 if ( $mode === 'replace' ) {
5849 # Output the replacement text
5850 # Add two newlines on -- trailing whitespace in $newText is conventionally
5851 # stripped by the editor, so we need both newlines to restore the paragraph gap
5852 # Only add trailing whitespace if there is newText
5853 if ( $newText != "" ) {
5854 $outText .= $newText . "\n\n";
5858 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5859 $node = $node->getNextSibling();
5863 if ( is_string( $outText ) ) {
5864 # Re-insert stripped tags
5865 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5872 * This function returns the text of a section, specified by a number ($section).
5873 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5874 * the first section before any such heading (section 0).
5876 * If a section contains subsections, these are also returned.
5878 * @param string $text Text to look in
5879 * @param string|int $sectionId Section identifier as a number or string
5880 * (e.g. 0, 1 or 'T-1').
5881 * @param string $defaultText Default to return if section is not found
5883 * @return string Text of the requested section
5885 public function getSection( $text, $sectionId, $defaultText = '' ) {
5886 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5890 * This function returns $oldtext after the content of the section
5891 * specified by $section has been replaced with $text. If the target
5892 * section does not exist, $oldtext is returned unchanged.
5894 * @param string $oldText Former text of the article
5895 * @param string|int $sectionId Section identifier as a number or string
5896 * (e.g. 0, 1 or 'T-1').
5897 * @param string $newText Replacing text
5899 * @return string Modified text
5901 public function replaceSection( $oldText, $sectionId, $newText ) {
5902 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5906 * Get the ID of the revision we are parsing
5908 * The return value will be either:
5909 * - a) Positive, indicating a specific revision ID (current or old)
5910 * - b) Zero, meaning the revision ID specified by getCurrentRevisionCallback()
5911 * - c) Null, meaning the parse is for preview mode and there is no revision
5915 public function getRevisionId() {
5916 return $this->mRevisionId
;
5920 * Get the revision object for $this->mRevisionId
5922 * @return Revision|null Either a Revision object or null
5923 * @since 1.23 (public since 1.23)
5925 public function getRevisionObject() {
5926 if ( !is_null( $this->mRevisionObject
) ) {
5927 return $this->mRevisionObject
;
5930 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5931 // This is useful when parsing revision that has not yet been saved.
5932 // However, if we get back a saved revision even though we are in
5933 // preview mode, we'll have to ignore it, see below.
5934 // NOTE: This callback may be used to inject an OLD revision that was
5935 // already loaded, so "current" is a bit of a misnomer. We can't just
5936 // skip it if mRevisionId is set.
5937 $rev = call_user_func(
5938 $this->mOptions
->getCurrentRevisionCallback(), $this->getTitle(), $this
5941 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5942 // We are in preview mode (mRevisionId is null), and the current revision callback
5943 // returned an existing revision. Ignore it and return null, it's probably the page's
5944 // current revision, which is not what we want here. Note that we do want to call the
5945 // callback to allow the unsaved revision to be injected here, e.g. for
5946 // self-transclusion previews.
5950 // If the parse is for a new revision, then the callback should have
5951 // already been set to force the object and should match mRevisionId.
5952 // If not, try to fetch by mRevisionId for sanity.
5953 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5954 $rev = Revision
::newFromId( $this->mRevisionId
);
5957 $this->mRevisionObject
= $rev;
5959 return $this->mRevisionObject
;
5963 * Get the timestamp associated with the current revision, adjusted for
5964 * the default server-local timestamp
5967 public function getRevisionTimestamp() {
5968 if ( is_null( $this->mRevisionTimestamp
) ) {
5969 $revObject = $this->getRevisionObject();
5970 $timestamp = $revObject ?
$revObject->getTimestamp() : wfTimestampNow();
5972 # The cryptic '' timezone parameter tells to use the site-default
5973 # timezone offset instead of the user settings.
5974 # Since this value will be saved into the parser cache, served
5975 # to other users, and potentially even used inside links and such,
5976 # it needs to be consistent for all visitors.
5977 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
5979 return $this->mRevisionTimestamp
;
5983 * Get the name of the user that edited the last revision
5985 * @return string User name
5987 public function getRevisionUser() {
5988 if ( is_null( $this->mRevisionUser
) ) {
5989 $revObject = $this->getRevisionObject();
5991 # if this template is subst: the revision id will be blank,
5992 # so just use the current user's name
5994 $this->mRevisionUser
= $revObject->getUserText();
5995 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5996 $this->mRevisionUser
= $this->getUser()->getName();
5999 return $this->mRevisionUser
;
6003 * Get the size of the revision
6005 * @return int|null Revision size
6007 public function getRevisionSize() {
6008 if ( is_null( $this->mRevisionSize
) ) {
6009 $revObject = $this->getRevisionObject();
6011 # if this variable is subst: the revision id will be blank,
6012 # so just use the parser input size, because the own substituation
6013 # will change the size.
6015 $this->mRevisionSize
= $revObject->getSize();
6017 $this->mRevisionSize
= $this->mInputSize
;
6020 return $this->mRevisionSize
;
6024 * Mutator for $mDefaultSort
6026 * @param string $sort New value
6028 public function setDefaultSort( $sort ) {
6029 $this->mDefaultSort
= $sort;
6030 $this->mOutput
->setProperty( 'defaultsort', $sort );
6034 * Accessor for $mDefaultSort
6035 * Will use the empty string if none is set.
6037 * This value is treated as a prefix, so the
6038 * empty string is equivalent to sorting by
6043 public function getDefaultSort() {
6044 if ( $this->mDefaultSort
!== false ) {
6045 return $this->mDefaultSort
;
6052 * Accessor for $mDefaultSort
6053 * Unlike getDefaultSort(), will return false if none is set
6055 * @return string|bool
6057 public function getCustomDefaultSort() {
6058 return $this->mDefaultSort
;
6061 private static function getSectionNameFromStrippedText( $text ) {
6062 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
6063 $text = Sanitizer
::decodeCharReferences( $text );
6064 $text = self
::normalizeSectionName( $text );
6068 private static function makeAnchor( $sectionName ) {
6069 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6072 private function makeLegacyAnchor( $sectionName ) {
6073 $fragmentMode = $this->svcOptions
->get( 'FragmentMode' );
6074 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6075 // ForAttribute() and ForLink() are the same for legacy encoding
6076 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6078 $id = Sanitizer
::escapeIdForLink( $sectionName );
6085 * Try to guess the section anchor name based on a wikitext fragment
6086 * presumably extracted from a heading, for example "Header" from
6089 * @param string $text
6090 * @return string Anchor (starting with '#')
6092 public function guessSectionNameFromWikiText( $text ) {
6093 # Strip out wikitext links(they break the anchor)
6094 $text = $this->stripSectionName( $text );
6095 $sectionName = self
::getSectionNameFromStrippedText( $text );
6096 return self
::makeAnchor( $sectionName );
6100 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6101 * instead, if possible. For use in redirects, since various versions
6102 * of Microsoft browsers interpret Location: headers as something other
6103 * than UTF-8, resulting in breakage.
6105 * @param string $text The section name
6106 * @return string Anchor (starting with '#')
6108 public function guessLegacySectionNameFromWikiText( $text ) {
6109 # Strip out wikitext links(they break the anchor)
6110 $text = $this->stripSectionName( $text );
6111 $sectionName = self
::getSectionNameFromStrippedText( $text );
6112 return $this->makeLegacyAnchor( $sectionName );
6116 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6117 * @param string $text Section name (plain text)
6118 * @return string Anchor (starting with '#')
6120 public static function guessSectionNameFromStrippedText( $text ) {
6121 $sectionName = self
::getSectionNameFromStrippedText( $text );
6122 return self
::makeAnchor( $sectionName );
6126 * Apply the same normalization as code making links to this section would
6128 * @param string $text
6131 private static function normalizeSectionName( $text ) {
6132 # T90902: ensure the same normalization is applied for IDs as to links
6133 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6136 $parts = $titleParser->splitTitleString( "#$text" );
6137 } catch ( MalformedTitleException
$ex ) {
6140 return $parts['fragment'];
6144 * Strips a text string of wikitext for use in a section anchor
6146 * Accepts a text string and then removes all wikitext from the
6147 * string and leaves only the resultant text (i.e. the result of
6148 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6149 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6150 * to create valid section anchors by mimicing the output of the
6151 * parser when headings are parsed.
6153 * @param string $text Text string to be stripped of wikitext
6154 * for use in a Section anchor
6155 * @return string Filtered text string
6157 public function stripSectionName( $text ) {
6158 # Strip internal link markup
6159 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6160 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6162 # Strip external link markup
6163 # @todo FIXME: Not tolerant to blank link text
6164 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6165 # on how many empty links there are on the page - need to figure that out.
6166 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6168 # Parse wikitext quotes (italics & bold)
6169 $text = $this->doQuotes( $text );
6172 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6177 * strip/replaceVariables/unstrip for preprocessor regression testing
6179 * @param string $text
6180 * @param Title $title
6181 * @param ParserOptions $options
6182 * @param int $outputType
6186 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6187 $outputType = self
::OT_HTML
6189 $magicScopeVariable = $this->lock();
6190 $this->startParse( $title, $options, $outputType, true );
6192 $text = $this->replaceVariables( $text );
6193 $text = $this->mStripState
->unstripBoth( $text );
6194 $text = Sanitizer
::removeHTMLtags( $text );
6199 * @param string $text
6200 * @param Title $title
6201 * @param ParserOptions $options
6204 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6205 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6209 * @param string $text
6210 * @param Title $title
6211 * @param ParserOptions $options
6214 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6215 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6219 * Call a callback function on all regions of the given text that are not
6220 * inside strip markers, and replace those regions with the return value
6221 * of the callback. For example, with input:
6225 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6226 * two strings will be replaced with the value returned by the callback in
6230 * @param callable $callback
6234 public function markerSkipCallback( $s, $callback ) {
6237 while ( $i < strlen( $s ) ) {
6238 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6239 if ( $markerStart === false ) {
6240 $out .= call_user_func( $callback, substr( $s, $i ) );
6243 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6244 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6245 if ( $markerEnd === false ) {
6246 $out .= substr( $s, $markerStart );
6249 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6250 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6259 * Remove any strip markers found in the given text.
6261 * @param string $text
6264 public function killMarkers( $text ) {
6265 return $this->mStripState
->killMarkers( $text );
6269 * Save the parser state required to convert the given half-parsed text to
6270 * HTML. "Half-parsed" in this context means the output of
6271 * recursiveTagParse() or internalParse(). This output has strip markers
6272 * from replaceVariables (extensionSubstitution() etc.), and link
6273 * placeholders from replaceLinkHolders().
6275 * Returns an array which can be serialized and stored persistently. This
6276 * array can later be loaded into another parser instance with
6277 * unserializeHalfParsedText(). The text can then be safely incorporated into
6278 * the return value of a parser hook.
6280 * @deprecated since 1.31
6281 * @param string $text
6285 public function serializeHalfParsedText( $text ) {
6286 wfDeprecated( __METHOD__
, '1.31' );
6289 'version' => self
::HALF_PARSED_VERSION
,
6290 'stripState' => $this->mStripState
->getSubState( $text ),
6291 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6297 * Load the parser state given in the $data array, which is assumed to
6298 * have been generated by serializeHalfParsedText(). The text contents is
6299 * extracted from the array, and its markers are transformed into markers
6300 * appropriate for the current Parser instance. This transformed text is
6301 * returned, and can be safely included in the return value of a parser
6304 * If the $data array has been stored persistently, the caller should first
6305 * check whether it is still valid, by calling isValidHalfParsedText().
6307 * @deprecated since 1.31
6308 * @param array $data Serialized data
6309 * @throws MWException
6312 public function unserializeHalfParsedText( $data ) {
6313 wfDeprecated( __METHOD__
, '1.31' );
6314 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6315 throw new MWException( __METHOD__
. ': invalid version' );
6318 # First, extract the strip state.
6319 $texts = [ $data['text'] ];
6320 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6322 # Now renumber links
6323 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6325 # Should be good to go.
6330 * Returns true if the given array, presumed to be generated by
6331 * serializeHalfParsedText(), is compatible with the current version of the
6334 * @deprecated since 1.31
6335 * @param array $data
6339 public function isValidHalfParsedText( $data ) {
6340 wfDeprecated( __METHOD__
, '1.31' );
6341 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6345 * Parsed a width param of imagelink like 300px or 200x300px
6347 * @param string $value
6348 * @param bool $parseHeight
6353 public static function parseWidthParam( $value, $parseHeight = true ) {
6354 $parsedWidthParam = [];
6355 if ( $value === '' ) {
6356 return $parsedWidthParam;
6359 # (T15500) In both cases (width/height and width only),
6360 # permit trailing "px" for backward compatibility.
6361 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6362 $width = intval( $m[1] );
6363 $height = intval( $m[2] );
6364 $parsedWidthParam['width'] = $width;
6365 $parsedWidthParam['height'] = $height;
6366 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6367 $width = intval( $value );
6368 $parsedWidthParam['width'] = $width;
6370 return $parsedWidthParam;
6374 * Lock the current instance of the parser.
6376 * This is meant to stop someone from calling the parser
6377 * recursively and messing up all the strip state.
6379 * @throws MWException If parser is in a parse
6380 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6382 protected function lock() {
6383 if ( $this->mInParse
) {
6384 throw new MWException( "Parser state cleared while parsing. "
6385 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6388 // Save the backtrace when locking, so that if some code tries locking again,
6389 // we can print the lock owner's backtrace for easier debugging
6391 $this->mInParse
= $e->getTraceAsString();
6393 $recursiveCheck = new ScopedCallback( function () {
6394 $this->mInParse
= false;
6397 return $recursiveCheck;
6401 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6403 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6404 * or if there is more than one <p/> tag in the input HTML.
6406 * @param string $html
6410 public static function stripOuterParagraph( $html ) {
6412 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6420 * Return this parser if it is not doing anything, otherwise
6421 * get a fresh parser. You can use this method by doing
6422 * $newParser = $oldParser->getFreshParser(), or more simply
6423 * $oldParser->getFreshParser()->parse( ... );
6424 * if you're unsure if $oldParser is safe to use.
6427 * @return Parser A parser object that is not parsing anything
6429 public function getFreshParser() {
6430 if ( $this->mInParse
) {
6431 return $this->factory
->create();
6438 * Set's up the PHP implementation of OOUI for use in this request
6439 * and instructs OutputPage to enable OOUI for itself.
6443 public function enableOOUI() {
6444 OutputPage
::setupOOUI();
6445 $this->mOutput
->setEnableOOUI( true );