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\BadFileLookup
;
24 use MediaWiki\Config\ServiceOptions
;
25 use MediaWiki\Linker\LinkRenderer
;
26 use MediaWiki\Linker\LinkRendererFactory
;
27 use MediaWiki\Linker\LinkTarget
;
28 use MediaWiki\MediaWikiServices
;
29 use MediaWiki\Special\SpecialPageFactory
;
30 use Psr\Log\NullLogger
;
31 use Wikimedia\ScopedCallback
;
32 use Psr\Log\LoggerInterface
;
35 * @defgroup Parser Parser
39 * PHP Parser - Processes wiki markup (which uses a more user-friendly
40 * syntax, such as "[[link]]" for making links), and provides a one-way
41 * transformation of that wiki markup it into (X)HTML output / markup
42 * (which in turn the browser understands, and can display).
44 * There are seven main entry points into the Parser class:
47 * produces HTML output
48 * - Parser::preSaveTransform()
49 * produces altered wiki markup
50 * - Parser::preprocess()
51 * removes HTML comments and expands templates
52 * - Parser::cleanSig() and Parser::cleanSigInSig()
53 * cleans a signature before saving it to preferences
54 * - Parser::getSection()
55 * return the content of a section from an article for section editing
56 * - Parser::replaceSection()
57 * replaces a section by number inside an article
58 * - Parser::getPreloadText()
59 * removes <noinclude> sections and <includeonly> tags
61 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
64 * $wgNamespacesWithSubpages
66 * @par Settings only within ParserOptions:
67 * $wgAllowExternalImages
68 * $wgAllowSpecialInclusion
76 * Update this version number when the ParserOutput format
77 * changes in an incompatible way, so the parser cache
78 * can automatically discard old data.
80 const VERSION
= '1.6.4';
83 * Update this version number when the output of serialiseHalfParsedText()
84 * changes in an incompatible way
86 const HALF_PARSED_VERSION
= 2;
88 # Flags for Parser::setFunctionHook
89 const SFH_NO_HASH
= 1;
90 const SFH_OBJECT_ARGS
= 2;
92 # Constants needed for external link processing
93 # Everything except bracket, space, or control characters
94 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
95 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
96 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
97 # uses to replace invalid HTML characters.
98 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
99 # Simplified expression to match an IPv4 or IPv6 address, or
100 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
101 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
102 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
103 // phpcs:ignore Generic.Files.LineLength
104 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
105 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
107 # Regular expression for a non-newline space
108 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
110 # Flags for preprocessToDom
111 const PTD_FOR_INCLUSION
= 1;
113 # Allowed values for $this->mOutputType
114 # Parameter to startExternalParse().
115 const OT_HTML
= 1; # like parse()
116 const OT_WIKI
= 2; # like preSaveTransform()
117 const OT_PREPROCESS
= 3; # like preprocess()
119 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
122 * @var string Prefix and suffix for temporary replacement strings
123 * for the multipass parser.
125 * \x7f should never appear in input as it's disallowed in XML.
126 * Using it at the front also gives us a little extra robustness
127 * since it shouldn't match when butted up against identifier-like
130 * Must not consist of all title characters, or else it will change
131 * the behavior of <nowiki> in a link.
133 * Must have a character that needs escaping in attributes, otherwise
134 * someone could put a strip marker in an attribute, to get around
135 * escaping quote marks, and break out of the attribute. Thus we add
138 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
139 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
141 # Markers used for wrapping the table of contents
142 const TOC_START
= '<mw:toc>';
143 const TOC_END
= '</mw:toc>';
145 /** @var int Assume that no output will later be saved this many seconds after parsing */
149 public $mTagHooks = [];
150 public $mTransparentTagHooks = [];
151 public $mFunctionHooks = [];
152 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
153 public $mFunctionTagHooks = [];
154 public $mStripList = [];
155 public $mDefaultStripList = [];
156 public $mVarCache = [];
157 public $mImageParams = [];
158 public $mImageParamsMagicArray = [];
159 public $mMarkerIndex = 0;
161 * @var bool Whether firstCallInit still needs to be called
163 public $mFirstCall = true;
165 # Initialised by initialiseVariables()
168 * @var MagicWordArray
173 * @var MagicWordArray
178 * @deprecated since 1.34, there should be no need to use this
183 # Initialised in constructor
184 public $mExtLinkBracketedRegex, $mUrlProtocols;
186 # Initialized in getPreprocessor()
187 /** @var Preprocessor */
188 public $mPreprocessor;
190 # Cleared with clearState():
202 public $mIncludeCount;
204 * @var LinkHolderArray
206 public $mLinkHolders;
209 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
210 public $mDefaultSort;
211 public $mTplRedirCache, $mHeadings, $mDoubleUnderscores;
212 public $mExpensiveFunctionCount; # number of expensive parser function calls
213 public $mShowToc, $mForceTocPosition;
215 public $mTplDomCache;
220 public $mUser; # User object; only used when doing pre-save transform
223 # These are variables reset at least once per parse regardless of $clearState
233 public $mTitle; # Title context, used for self-link rendering and similar things
234 public $mOutputType; # Output type, one of the OT_xxx constants
235 public $ot; # Shortcut alias, see setOutputType()
236 public $mRevisionObject; # The revision object of the specified revision ID
237 public $mRevisionId; # ID to display in {{REVISIONID}} tags
238 public $mRevisionTimestamp; # The timestamp of the specified revision ID
239 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
240 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
241 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
242 public $mInputSize = false; # For {{PAGESIZE}} on current page.
245 * @var array Array with the language name of each language link (i.e. the
246 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
247 * duplicate language links to the ParserOutput.
249 public $mLangLinkLanguages;
252 * @var MapCacheLRU|null
255 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
257 public $currentRevisionCache;
260 * @var bool|string Recursive call protection.
261 * This variable should be treated as if it were private.
263 public $mInParse = false;
265 /** @var SectionProfiler */
266 protected $mProfiler;
271 protected $mLinkRenderer;
273 /** @var MagicWordFactory */
274 private $magicWordFactory;
279 /** @var ParserFactory */
282 /** @var SpecialPageFactory */
283 private $specialPageFactory;
286 * This is called $svcOptions instead of $options like elsewhere to avoid confusion with
287 * $mOptions, which is public and widely used, and also with the local variable $options used
288 * for ParserOptions throughout this file.
290 * @var ServiceOptions
294 /** @var LinkRendererFactory */
295 private $linkRendererFactory;
297 /** @var NamespaceInfo */
300 /** @var LoggerInterface */
303 /** @var BadFileLookup */
304 private $badFileLookup;
307 * TODO Make this a const when HHVM support is dropped (T192166)
312 public static $constructorOptions = [
313 // See $wgParserConf documentation
316 // See documentation for the corresponding config options
318 'EnableScaryTranscluding',
319 'ExtraInterlanguageLinkPrefixes',
331 'TranscludeCacheExpiry',
335 * Constructing parsers directly is deprecated! Use a ParserFactory.
337 * @param ServiceOptions|null $svcOptions
338 * @param MagicWordFactory|null $magicWordFactory
339 * @param Language|null $contLang Content language
340 * @param ParserFactory|null $factory
341 * @param string|null $urlProtocols As returned from wfUrlProtocols()
342 * @param SpecialPageFactory|null $spFactory
343 * @param LinkRendererFactory|null $linkRendererFactory
344 * @param NamespaceInfo|null $nsInfo
345 * @param LoggerInterface|null $logger
346 * @param BadFileLookup|null $badFileLookup
348 public function __construct(
350 MagicWordFactory
$magicWordFactory = null,
351 Language
$contLang = null,
352 ParserFactory
$factory = null,
353 $urlProtocols = null,
354 SpecialPageFactory
$spFactory = null,
355 $linkRendererFactory = null,
358 BadFileLookup
$badFileLookup = null
360 if ( !$svcOptions ||
is_array( $svcOptions ) ) {
361 // Pre-1.34 calling convention is the first parameter is just ParserConf, the seventh is
362 // Config, and the eighth is LinkRendererFactory.
363 $this->mConf
= (array)$svcOptions;
364 if ( empty( $this->mConf
['class'] ) ) {
365 $this->mConf
['class'] = self
::class;
367 if ( empty( $this->mConf
['preprocessorClass'] ) ) {
368 $this->mConf
['preprocessorClass'] = self
::getDefaultPreprocessorClass();
370 $this->svcOptions
= new ServiceOptions( self
::$constructorOptions,
371 $this->mConf
, func_num_args() > 6
372 ?
func_get_arg( 6 ) : MediaWikiServices
::getInstance()->getMainConfig()
374 $linkRendererFactory = func_num_args() > 7 ?
func_get_arg( 7 ) : null;
375 $nsInfo = func_num_args() > 8 ?
func_get_arg( 8 ) : null;
377 // New calling convention
378 $svcOptions->assertRequiredOptions( self
::$constructorOptions );
379 // $this->mConf is public, so we'll keep those two options there as well for
380 // compatibility until it's removed
382 'class' => $svcOptions->get( 'class' ),
383 'preprocessorClass' => $svcOptions->get( 'preprocessorClass' ),
385 $this->svcOptions
= $svcOptions;
388 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
389 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
390 self
::EXT_LINK_ADDR
.
391 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
393 $this->magicWordFactory
= $magicWordFactory ??
394 MediaWikiServices
::getInstance()->getMagicWordFactory();
396 $this->contLang
= $contLang ?? MediaWikiServices
::getInstance()->getContentLanguage();
398 $this->factory
= $factory ?? MediaWikiServices
::getInstance()->getParserFactory();
399 $this->specialPageFactory
= $spFactory ??
400 MediaWikiServices
::getInstance()->getSpecialPageFactory();
401 $this->linkRendererFactory
= $linkRendererFactory ??
402 MediaWikiServices
::getInstance()->getLinkRendererFactory();
403 $this->nsInfo
= $nsInfo ?? MediaWikiServices
::getInstance()->getNamespaceInfo();
404 $this->logger
= $logger ?
: new NullLogger();
405 $this->badFileLookup
= $badFileLookup ??
406 MediaWikiServices
::getInstance()->getBadFileLookup();
410 * Reduce memory usage to reduce the impact of circular references
412 public function __destruct() {
413 if ( isset( $this->mLinkHolders
) ) {
414 unset( $this->mLinkHolders
);
416 foreach ( $this as $name => $value ) {
417 unset( $this->$name );
422 * Allow extensions to clean up when the parser is cloned
424 public function __clone() {
425 $this->mInParse
= false;
427 // T58226: When you create a reference "to" an object field, that
428 // makes the object field itself be a reference too (until the other
429 // reference goes out of scope). When cloning, any field that's a
430 // reference is copied as a reference in the new object. Both of these
431 // are defined PHP5 behaviors, as inconvenient as it is for us when old
432 // hooks from PHP4 days are passing fields by reference.
433 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
434 // Make a non-reference copy of the field, then rebind the field to
435 // reference the new copy.
441 Hooks
::run( 'ParserCloned', [ $this ] );
445 * Which class should we use for the preprocessor if not otherwise specified?
448 * @deprecated since 1.34, removing configurability of preprocessor
451 public static function getDefaultPreprocessorClass() {
452 return Preprocessor_Hash
::class;
456 * Do various kinds of initialisation on the first call of the parser
458 public function firstCallInit() {
459 if ( !$this->mFirstCall
) {
462 $this->mFirstCall
= false;
464 CoreParserFunctions
::register( $this );
465 CoreTagHooks
::register( $this );
466 $this->initialiseVariables();
468 // Avoid PHP 7.1 warning from passing $this by reference
470 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
478 public function clearState() {
479 $this->firstCallInit();
480 $this->resetOutput();
481 $this->mAutonumber
= 0;
482 $this->mIncludeCount
= [];
483 $this->mLinkHolders
= new LinkHolderArray( $this );
485 $this->mRevisionObject
= $this->mRevisionTimestamp
=
486 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
487 $this->mVarCache
= [];
489 $this->mLangLinkLanguages
= [];
490 $this->currentRevisionCache
= null;
492 $this->mStripState
= new StripState( $this );
494 # Clear these on every parse, T6549
495 $this->mTplRedirCache
= $this->mTplDomCache
= [];
497 $this->mShowToc
= true;
498 $this->mForceTocPosition
= false;
499 $this->mIncludeSizes
= [
503 $this->mPPNodeCount
= 0;
504 $this->mGeneratedPPNodeCount
= 0;
505 $this->mHighestExpansionDepth
= 0;
506 $this->mDefaultSort
= false;
507 $this->mHeadings
= [];
508 $this->mDoubleUnderscores
= [];
509 $this->mExpensiveFunctionCount
= 0;
512 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
513 $this->mPreprocessor
= null;
516 $this->mProfiler
= new SectionProfiler();
518 // Avoid PHP 7.1 warning from passing $this by reference
520 Hooks
::run( 'ParserClearState', [ &$parser ] );
524 * Reset the ParserOutput
526 public function resetOutput() {
527 $this->mOutput
= new ParserOutput
;
528 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
532 * Convert wikitext to HTML
533 * Do not call this function recursively.
535 * @param string $text Text we want to parse
536 * @param-taint $text escapes_htmlnoent
537 * @param Title $title
538 * @param ParserOptions $options
539 * @param bool $linestart
540 * @param bool $clearState
541 * @param int|null $revid ID of the revision being rendered. This is used to render
542 * REVISION* magic words. 0 means that any current revision will be used. Null means
543 * that {{REVISIONID}}/{{REVISIONUSER}} will be empty and {{REVISIONTIMESTAMP}} will
544 * use the current timestamp.
545 * @return ParserOutput A ParserOutput
546 * @return-taint escaped
548 public function parse(
549 $text, Title
$title, ParserOptions
$options,
550 $linestart = true, $clearState = true, $revid = null
553 // We use U+007F DELETE to construct strip markers, so we have to make
554 // sure that this character does not occur in the input text.
555 $text = strtr( $text, "\x7f", "?" );
556 $magicScopeVariable = $this->lock();
558 // Strip U+0000 NULL (T159174)
559 $text = str_replace( "\000", '', $text );
561 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
563 $this->currentRevisionCache
= null;
564 $this->mInputSize
= strlen( $text );
565 if ( $this->mOptions
->getEnableLimitReport() ) {
566 $this->mOutput
->resetParseStartTime();
569 $oldRevisionId = $this->mRevisionId
;
570 $oldRevisionObject = $this->mRevisionObject
;
571 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
572 $oldRevisionUser = $this->mRevisionUser
;
573 $oldRevisionSize = $this->mRevisionSize
;
574 if ( $revid !== null ) {
575 $this->mRevisionId
= $revid;
576 $this->mRevisionObject
= null;
577 $this->mRevisionTimestamp
= null;
578 $this->mRevisionUser
= null;
579 $this->mRevisionSize
= null;
582 // Avoid PHP 7.1 warning from passing $this by reference
584 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
586 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
587 $text = $this->internalParse( $text );
588 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
590 $text = $this->internalParseHalfParsed( $text, true, $linestart );
593 * A converted title will be provided in the output object if title and
594 * content conversion are enabled, the article text does not contain
595 * a conversion-suppressing double-underscore tag, and no
596 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
597 * automatic link conversion.
599 if ( !( $options->getDisableTitleConversion()
600 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
601 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
602 ||
$this->mOutput
->getDisplayTitle() !== false )
604 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
605 if ( $convruletitle ) {
606 $this->mOutput
->setTitleText( $convruletitle );
608 $titleText = $this->getTargetLanguage()->convertTitle( $title );
609 $this->mOutput
->setTitleText( $titleText );
613 # Compute runtime adaptive expiry if set
614 $this->mOutput
->finalizeAdaptiveCacheExpiry();
616 # Warn if too many heavyweight parser functions were used
617 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
618 $this->limitationWarn( 'expensive-parserfunction',
619 $this->mExpensiveFunctionCount
,
620 $this->mOptions
->getExpensiveParserFunctionLimit()
624 # Information on limits, for the benefit of users who try to skirt them
625 if ( $this->mOptions
->getEnableLimitReport() ) {
626 $text .= $this->makeLimitReport();
629 # Wrap non-interface parser output in a <div> so it can be targeted
631 $class = $this->mOptions
->getWrapOutputClass();
632 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
633 $this->mOutput
->addWrapperDivClass( $class );
636 $this->mOutput
->setText( $text );
638 $this->mRevisionId
= $oldRevisionId;
639 $this->mRevisionObject
= $oldRevisionObject;
640 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
641 $this->mRevisionUser
= $oldRevisionUser;
642 $this->mRevisionSize
= $oldRevisionSize;
643 $this->mInputSize
= false;
644 $this->currentRevisionCache
= null;
646 return $this->mOutput
;
650 * Set the limit report data in the current ParserOutput, and return the
651 * limit report HTML comment.
655 protected function makeLimitReport() {
656 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
658 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
659 if ( $cpuTime !== null ) {
660 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
661 sprintf( "%.3f", $cpuTime )
665 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
666 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
667 sprintf( "%.3f", $wallTime )
670 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
671 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
673 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
674 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
676 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
677 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
679 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
680 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
682 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
683 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
685 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
686 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
689 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
690 $this->mOutput
->setLimitReportData( $key, $value );
693 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
695 $limitReport = "NewPP limit report\n";
696 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
697 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
699 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
700 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
701 $limitReport .= 'Dynamic content: ' .
702 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
704 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput
->getAllFlags() ) . "]\n";
706 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
707 if ( Hooks
::run( 'ParserLimitReportFormat',
708 [ $key, &$value, &$limitReport, false, false ]
710 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
711 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
712 ->inLanguage( 'en' )->useDatabase( false );
713 if ( !$valueMsg->exists() ) {
714 $valueMsg = new RawMessage( '$1' );
716 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
717 $valueMsg->params( $value );
718 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
722 // Since we're not really outputting HTML, decode the entities and
723 // then re-encode the things that need hiding inside HTML comments.
724 $limitReport = htmlspecialchars_decode( $limitReport );
726 // Sanitize for comment. Note '‐' in the replacement is U+2010,
727 // which looks much like the problematic '-'.
728 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
729 $text = "\n<!-- \n$limitReport-->\n";
731 // Add on template profiling data in human/machine readable way
732 $dataByFunc = $this->mProfiler
->getFunctionStats();
733 uasort( $dataByFunc, function ( $a, $b ) {
734 return $b['real'] <=> $a['real']; // descending order
737 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
738 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
739 $item['%real'], $item['real'], $item['calls'],
740 htmlspecialchars( $item['name'] ) );
742 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
743 $text .= implode( "\n", $profileReport ) . "\n-->\n";
745 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
747 // Add other cache related metadata
748 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
749 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
751 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
752 $this->mOutput
->getCacheTime() );
753 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
754 $this->mOutput
->getCacheExpiry() );
755 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
756 $this->mOutput
->hasDynamicContent() );
758 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
759 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
760 $this->mTitle
->getPrefixedDBkey() );
766 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
767 * can be called from an extension tag hook.
769 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
770 * instead, which means that lists and links have not been fully parsed yet,
771 * and strip markers are still present.
773 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
775 * Use this function if you're a parser tag hook and you want to parse
776 * wikitext before or after applying additional transformations, and you
777 * intend to *return the result as hook output*, which will cause it to go
778 * through the rest of parsing process automatically.
780 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
781 * $text are not expanded
783 * @param string $text Text extension wants to have parsed
784 * @param-taint $text escapes_htmlnoent
785 * @param bool|PPFrame $frame The frame to use for expanding any template variables
786 * @return string UNSAFE half-parsed HTML
787 * @return-taint escaped
789 public function recursiveTagParse( $text, $frame = false ) {
790 // Avoid PHP 7.1 warning from passing $this by reference
792 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
793 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
794 $text = $this->internalParse( $text, false, $frame );
799 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
800 * point can be called from an extension tag hook.
802 * The output of this function is fully-parsed HTML that is safe for output.
803 * If you're a parser tag hook, you might want to use recursiveTagParse()
806 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
807 * $text are not expanded
811 * @param string $text Text extension wants to have parsed
812 * @param-taint $text escapes_htmlnoent
813 * @param bool|PPFrame $frame The frame to use for expanding any template variables
814 * @return string Fully parsed HTML
815 * @return-taint escaped
817 public function recursiveTagParseFully( $text, $frame = false ) {
818 $text = $this->recursiveTagParse( $text, $frame );
819 $text = $this->internalParseHalfParsed( $text, false );
824 * Expand templates and variables in the text, producing valid, static wikitext.
825 * Also removes comments.
826 * Do not call this function recursively.
827 * @param string $text
828 * @param Title|null $title
829 * @param ParserOptions $options
830 * @param int|null $revid
831 * @param bool|PPFrame $frame
832 * @return mixed|string
834 public function preprocess( $text, Title
$title = null,
835 ParserOptions
$options, $revid = null, $frame = false
837 $magicScopeVariable = $this->lock();
838 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
839 if ( $revid !== null ) {
840 $this->mRevisionId
= $revid;
842 // Avoid PHP 7.1 warning from passing $this by reference
844 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
845 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
846 $text = $this->replaceVariables( $text, $frame );
847 $text = $this->mStripState
->unstripBoth( $text );
852 * Recursive parser entry point that can be called from an extension tag
855 * @param string $text Text to be expanded
856 * @param bool|PPFrame $frame The frame to use for expanding any template variables
860 public function recursivePreprocess( $text, $frame = false ) {
861 $text = $this->replaceVariables( $text, $frame );
862 $text = $this->mStripState
->unstripBoth( $text );
867 * Process the wikitext for the "?preload=" feature. (T7210)
869 * "<noinclude>", "<includeonly>" etc. are parsed as for template
870 * transclusion, comments, templates, arguments, tags hooks and parser
871 * functions are untouched.
873 * @param string $text
874 * @param Title $title
875 * @param ParserOptions $options
876 * @param array $params
879 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
880 $msg = new RawMessage( $text );
881 $text = $msg->params( $params )->plain();
883 # Parser (re)initialisation
884 $magicScopeVariable = $this->lock();
885 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
887 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
888 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
889 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
890 $text = $this->mStripState
->unstripBoth( $text );
895 * Set the current user.
896 * Should only be used when doing pre-save transform.
898 * @param User|null $user User object or null (to reset)
900 public function setUser( $user ) {
901 $this->mUser
= $user;
905 * Set the context title
909 public function setTitle( $t ) {
911 $t = Title
::newFromText( 'NO TITLE' );
914 if ( $t->hasFragment() ) {
915 # Strip the fragment to avoid various odd effects
916 $this->mTitle
= $t->createFragmentTarget( '' );
923 * Accessor for the Title object
927 public function getTitle() {
928 return $this->mTitle
;
932 * Accessor/mutator for the Title object
934 * @param Title|null $x Title object or null to just get the current one
937 public function Title( $x = null ) {
938 return wfSetVar( $this->mTitle
, $x );
942 * Set the output type
944 * @param int $ot New value
946 public function setOutputType( $ot ) {
947 $this->mOutputType
= $ot;
950 'html' => $ot == self
::OT_HTML
,
951 'wiki' => $ot == self
::OT_WIKI
,
952 'pre' => $ot == self
::OT_PREPROCESS
,
953 'plain' => $ot == self
::OT_PLAIN
,
958 * Accessor/mutator for the output type
960 * @param int|null $x New value or null to just get the current one
963 public function OutputType( $x = null ) {
964 return wfSetVar( $this->mOutputType
, $x );
968 * Get the ParserOutput object
970 * @return ParserOutput
972 public function getOutput() {
973 return $this->mOutput
;
977 * Get the ParserOptions object
979 * @return ParserOptions
981 public function getOptions() {
982 return $this->mOptions
;
986 * Accessor/mutator for the ParserOptions object
988 * @param ParserOptions|null $x New value or null to just get the current one
989 * @return ParserOptions Current ParserOptions object
991 public function Options( $x = null ) {
992 return wfSetVar( $this->mOptions
, $x );
998 public function nextLinkID() {
999 return $this->mLinkID++
;
1005 public function setLinkID( $id ) {
1006 $this->mLinkID
= $id;
1010 * Get a language object for use in parser functions such as {{FORMATNUM:}}
1013 public function getFunctionLang() {
1014 return $this->getTargetLanguage();
1018 * Get the target language for the content being parsed. This is usually the
1019 * language that the content is in.
1023 * @throws MWException
1026 public function getTargetLanguage() {
1027 $target = $this->mOptions
->getTargetLanguage();
1029 if ( $target !== null ) {
1031 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
1032 return $this->mOptions
->getUserLangObj();
1033 } elseif ( is_null( $this->mTitle
) ) {
1034 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
1037 return $this->mTitle
->getPageLanguage();
1041 * Get the language object for language conversion
1042 * @deprecated since 1.32, just use getTargetLanguage()
1043 * @return Language|null
1045 public function getConverterLanguage() {
1046 return $this->getTargetLanguage();
1050 * Get a User object either from $this->mUser, if set, or from the
1051 * ParserOptions object otherwise
1055 public function getUser() {
1056 if ( !is_null( $this->mUser
) ) {
1057 return $this->mUser
;
1059 return $this->mOptions
->getUser();
1063 * Get a preprocessor object
1065 * @return Preprocessor
1067 public function getPreprocessor() {
1068 if ( !isset( $this->mPreprocessor
) ) {
1069 $class = $this->svcOptions
->get( 'preprocessorClass' );
1070 $this->mPreprocessor
= new $class( $this );
1072 return $this->mPreprocessor
;
1076 * Get a LinkRenderer instance to make links with
1079 * @return LinkRenderer
1081 public function getLinkRenderer() {
1082 // XXX We make the LinkRenderer with current options and then cache it forever
1083 if ( !$this->mLinkRenderer
) {
1084 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
1085 $this->mLinkRenderer
->setStubThreshold(
1086 $this->getOptions()->getStubThreshold()
1090 return $this->mLinkRenderer
;
1094 * Get the MagicWordFactory that this Parser is using
1097 * @return MagicWordFactory
1099 public function getMagicWordFactory() {
1100 return $this->magicWordFactory
;
1104 * Get the content language that this Parser is using
1109 public function getContentLanguage() {
1110 return $this->contLang
;
1114 * Replaces all occurrences of HTML-style comments and the given tags
1115 * in the text with a random marker and returns the next text. The output
1116 * parameter $matches will be an associative array filled with data in
1123 * [ 'param' => 'x' ],
1124 * '<element param="x">tag content</element>' ]
1127 * @param array $elements List of element names. Comments are always extracted.
1128 * @param string $text Source text string.
1129 * @param array &$matches Out parameter, Array: extracted tags
1130 * @return string Stripped text
1132 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1137 $taglist = implode( '|', $elements );
1138 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1140 while ( $text != '' ) {
1141 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1143 if ( count( $p ) < 5 ) {
1146 if ( count( $p ) > 5 ) {
1154 list( , $element, $attributes, $close, $inside ) = $p;
1157 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1158 $stripped .= $marker;
1160 if ( $close === '/>' ) {
1161 # Empty element tag, <tag />
1166 if ( $element === '!--' ) {
1169 $end = "/(<\\/$element\\s*>)/i";
1171 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1173 if ( count( $q ) < 3 ) {
1174 # No end tag -- let it run out to the end of the text.
1178 list( , $tail, $text ) = $q;
1182 $matches[$marker] = [ $element,
1184 Sanitizer
::decodeTagAttributes( $attributes ),
1185 "<$element$attributes$close$content$tail" ];
1191 * Get a list of strippable XML-like elements
1195 public function getStripList() {
1196 return $this->mStripList
;
1200 * Get the StripState
1202 * @return StripState
1204 public function getStripState() {
1205 return $this->mStripState
;
1209 * Add an item to the strip state
1210 * Returns the unique tag which must be inserted into the stripped text
1211 * The tag will be replaced with the original text in unstrip()
1213 * @param string $text
1217 public function insertStripItem( $text ) {
1218 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1219 $this->mMarkerIndex++
;
1220 $this->mStripState
->addGeneral( $marker, $text );
1225 * parse the wiki syntax used to render tables
1228 * @param string $text
1231 public function doTableStuff( $text ) {
1232 $lines = StringUtils
::explode( "\n", $text );
1234 $td_history = []; # Is currently a td tag open?
1235 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1236 $tr_history = []; # Is currently a tr tag open?
1237 $tr_attributes = []; # history of tr attributes
1238 $has_opened_tr = []; # Did this table open a <tr> element?
1239 $indent_level = 0; # indent level of the table
1241 foreach ( $lines as $outLine ) {
1242 $line = trim( $outLine );
1244 if ( $line === '' ) { # empty line, go to next line
1245 $out .= $outLine . "\n";
1249 $first_character = $line[0];
1250 $first_two = substr( $line, 0, 2 );
1253 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1254 # First check if we are starting a new table
1255 $indent_level = strlen( $matches[1] );
1257 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1258 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1260 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1261 array_push( $td_history, false );
1262 array_push( $last_tag_history, '' );
1263 array_push( $tr_history, false );
1264 array_push( $tr_attributes, '' );
1265 array_push( $has_opened_tr, false );
1266 } elseif ( count( $td_history ) == 0 ) {
1267 # Don't do any of the following
1268 $out .= $outLine . "\n";
1270 } elseif ( $first_two === '|}' ) {
1271 # We are ending a table
1272 $line = '</table>' . substr( $line, 2 );
1273 $last_tag = array_pop( $last_tag_history );
1275 if ( !array_pop( $has_opened_tr ) ) {
1276 $line = "<tr><td></td></tr>{$line}";
1279 if ( array_pop( $tr_history ) ) {
1280 $line = "</tr>{$line}";
1283 if ( array_pop( $td_history ) ) {
1284 $line = "</{$last_tag}>{$line}";
1286 array_pop( $tr_attributes );
1287 if ( $indent_level > 0 ) {
1288 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1292 } elseif ( $first_two === '|-' ) {
1293 # Now we have a table row
1294 $line = preg_replace( '#^\|-+#', '', $line );
1296 # Whats after the tag is now only attributes
1297 $attributes = $this->mStripState
->unstripBoth( $line );
1298 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1299 array_pop( $tr_attributes );
1300 array_push( $tr_attributes, $attributes );
1303 $last_tag = array_pop( $last_tag_history );
1304 array_pop( $has_opened_tr );
1305 array_push( $has_opened_tr, true );
1307 if ( array_pop( $tr_history ) ) {
1311 if ( array_pop( $td_history ) ) {
1312 $line = "</{$last_tag}>{$line}";
1316 array_push( $tr_history, false );
1317 array_push( $td_history, false );
1318 array_push( $last_tag_history, '' );
1319 } elseif ( $first_character === '|'
1320 ||
$first_character === '!'
1321 ||
$first_two === '|+'
1323 # This might be cell elements, td, th or captions
1324 if ( $first_two === '|+' ) {
1325 $first_character = '+';
1326 $line = substr( $line, 2 );
1328 $line = substr( $line, 1 );
1331 // Implies both are valid for table headings.
1332 if ( $first_character === '!' ) {
1333 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1336 # Split up multiple cells on the same line.
1337 # FIXME : This can result in improper nesting of tags processed
1338 # by earlier parser steps.
1339 $cells = explode( '||', $line );
1343 # Loop through each table cell
1344 foreach ( $cells as $cell ) {
1346 if ( $first_character !== '+' ) {
1347 $tr_after = array_pop( $tr_attributes );
1348 if ( !array_pop( $tr_history ) ) {
1349 $previous = "<tr{$tr_after}>\n";
1351 array_push( $tr_history, true );
1352 array_push( $tr_attributes, '' );
1353 array_pop( $has_opened_tr );
1354 array_push( $has_opened_tr, true );
1357 $last_tag = array_pop( $last_tag_history );
1359 if ( array_pop( $td_history ) ) {
1360 $previous = "</{$last_tag}>\n{$previous}";
1363 if ( $first_character === '|' ) {
1365 } elseif ( $first_character === '!' ) {
1367 } elseif ( $first_character === '+' ) {
1368 $last_tag = 'caption';
1373 array_push( $last_tag_history, $last_tag );
1375 # A cell could contain both parameters and data
1376 $cell_data = explode( '|', $cell, 2 );
1378 # T2553: Note that a '|' inside an invalid link should not
1379 # be mistaken as delimiting cell parameters
1380 # Bug T153140: Neither should language converter markup.
1381 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1382 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1383 } elseif ( count( $cell_data ) == 1 ) {
1384 // Whitespace in cells is trimmed
1385 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1387 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1388 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1389 // Whitespace in cells is trimmed
1390 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1394 array_push( $td_history, true );
1397 $out .= $outLine . "\n";
1400 # Closing open td, tr && table
1401 while ( count( $td_history ) > 0 ) {
1402 if ( array_pop( $td_history ) ) {
1405 if ( array_pop( $tr_history ) ) {
1408 if ( !array_pop( $has_opened_tr ) ) {
1409 $out .= "<tr><td></td></tr>\n";
1412 $out .= "</table>\n";
1415 # Remove trailing line-ending (b/c)
1416 if ( substr( $out, -1 ) === "\n" ) {
1417 $out = substr( $out, 0, -1 );
1420 # special case: don't return empty table
1421 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1429 * Helper function for parse() that transforms wiki markup into half-parsed
1430 * HTML. Only called for $mOutputType == self::OT_HTML.
1434 * @param string $text The text to parse
1435 * @param-taint $text escapes_html
1436 * @param bool $isMain Whether this is being called from the main parse() function
1437 * @param PPFrame|bool $frame A pre-processor frame
1441 public function internalParse( $text, $isMain = true, $frame = false ) {
1444 // Avoid PHP 7.1 warning from passing $this by reference
1447 # Hook to suspend the parser in this state
1448 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1452 # if $frame is provided, then use $frame for replacing any variables
1454 # use frame depth to infer how include/noinclude tags should be handled
1455 # depth=0 means this is the top-level document; otherwise it's an included document
1456 if ( !$frame->depth
) {
1459 $flag = self
::PTD_FOR_INCLUSION
;
1461 $dom = $this->preprocessToDom( $text, $flag );
1462 $text = $frame->expand( $dom );
1464 # if $frame is not provided, then use old-style replaceVariables
1465 $text = $this->replaceVariables( $text );
1468 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1469 $text = Sanitizer
::removeHTMLtags(
1471 [ $this, 'attributeStripCallback' ],
1473 array_keys( $this->mTransparentTagHooks
),
1475 [ $this, 'addTrackingCategory' ]
1477 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1479 # Tables need to come after variable replacement for things to work
1480 # properly; putting them before other transformations should keep
1481 # exciting things like link expansions from showing up in surprising
1483 $text = $this->doTableStuff( $text );
1485 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1487 $text = $this->doDoubleUnderscore( $text );
1489 $text = $this->doHeadings( $text );
1490 $text = $this->replaceInternalLinks( $text );
1491 $text = $this->doAllQuotes( $text );
1492 $text = $this->replaceExternalLinks( $text );
1494 # replaceInternalLinks may sometimes leave behind
1495 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1496 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1498 $text = $this->doMagicLinks( $text );
1499 $text = $this->formatHeadings( $text, $origText, $isMain );
1505 * Helper function for parse() that transforms half-parsed HTML into fully
1508 * @param string $text
1509 * @param bool $isMain
1510 * @param bool $linestart
1513 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1514 $text = $this->mStripState
->unstripGeneral( $text );
1516 // Avoid PHP 7.1 warning from passing $this by reference
1520 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1523 # Clean up special characters, only run once, next-to-last before doBlockLevels
1524 $text = Sanitizer
::armorFrenchSpaces( $text );
1526 $text = $this->doBlockLevels( $text, $linestart );
1528 $this->replaceLinkHolders( $text );
1531 * The input doesn't get language converted if
1533 * b) Content isn't converted
1534 * c) It's a conversion table
1535 * d) it is an interface message (which is in the user language)
1537 if ( !( $this->mOptions
->getDisableContentConversion()
1538 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1539 && !$this->mOptions
->getInterfaceMessage()
1541 # The position of the convert() call should not be changed. it
1542 # assumes that the links are all replaced and the only thing left
1543 # is the <nowiki> mark.
1544 $text = $this->getTargetLanguage()->convert( $text );
1547 $text = $this->mStripState
->unstripNoWiki( $text );
1550 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1553 $text = $this->replaceTransparentTags( $text );
1554 $text = $this->mStripState
->unstripGeneral( $text );
1556 $text = Sanitizer
::normalizeCharReferences( $text );
1558 if ( MWTidy
::isEnabled() ) {
1559 if ( $this->mOptions
->getTidy() ) {
1560 $text = MWTidy
::tidy( $text );
1563 # attempt to sanitize at least some nesting problems
1564 # (T4702 and quite a few others)
1565 # This code path is buggy and deprecated!
1566 wfDeprecated( 'disabling tidy', '1.33' );
1568 # ''Something [http://www.cool.com cool''] -->
1569 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1570 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1571 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1572 # fix up an anchor inside another anchor, only
1573 # at least for a single single nested link (T5695)
1574 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1575 '\\1\\2</a>\\3</a>\\1\\4</a>',
1576 # fix div inside inline elements- doBlockLevels won't wrap a line which
1577 # contains a div, so fix it up here; replace
1578 # div with escaped text
1579 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1580 '\\1\\3<div\\5>\\6</div>\\8\\9',
1581 # remove empty italic or bold tag pairs, some
1582 # introduced by rules above
1583 '/<([bi])><\/\\1>/' => '',
1586 $text = preg_replace(
1587 array_keys( $tidyregs ),
1588 array_values( $tidyregs ),
1593 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1600 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1601 * magic external links.
1606 * @param string $text
1610 public function doMagicLinks( $text ) {
1611 $prots = wfUrlProtocolsWithoutProtRel();
1612 $urlChar = self
::EXT_LINK_URL_CLASS
;
1613 $addr = self
::EXT_LINK_ADDR
;
1614 $space = self
::SPACE_NOT_NL
; # non-newline space
1615 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1616 $spaces = "$space++"; # possessive match of 1 or more spaces
1617 $text = preg_replace_callback(
1619 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1620 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1621 (\b # m[3]: Free external links
1623 ($addr$urlChar*) # m[4]: Post-protocol path
1625 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1627 \bISBN $spaces ( # m[6]: ISBN, capture number
1628 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1629 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1630 [0-9Xx] # check digit
1632 )!xu", [ $this, 'magicLinkCallback' ], $text );
1637 * @throws MWException
1639 * @return string HTML
1641 public function magicLinkCallback( $m ) {
1642 if ( isset( $m[1] ) && $m[1] !== '' ) {
1645 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1648 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1649 # Free external link
1650 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1651 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1653 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1654 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1659 $cssClass = 'mw-magiclink-rfc';
1660 $trackingCat = 'magiclink-tracking-rfc';
1662 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1663 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1667 $urlmsg = 'pubmedurl';
1668 $cssClass = 'mw-magiclink-pmid';
1669 $trackingCat = 'magiclink-tracking-pmid';
1672 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1673 substr( $m[0], 0, 20 ) . '"' );
1675 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1676 $this->addTrackingCategory( $trackingCat );
1677 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1678 } elseif ( isset( $m[6] ) && $m[6] !== ''
1679 && $this->mOptions
->getMagicISBNLinks()
1683 $space = self
::SPACE_NOT_NL
; # non-newline space
1684 $isbn = preg_replace( "/$space/", ' ', $isbn );
1685 $num = strtr( $isbn, [
1690 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1691 return $this->getLinkRenderer()->makeKnownLink(
1692 SpecialPage
::getTitleFor( 'Booksources', $num ),
1695 'class' => 'internal mw-magiclink-isbn',
1696 'title' => false // suppress title attribute
1705 * Make a free external link, given a user-supplied URL
1707 * @param string $url
1708 * @param int $numPostProto
1709 * The number of characters after the protocol.
1710 * @return string HTML
1713 public function makeFreeExternalLink( $url, $numPostProto ) {
1716 # The characters '<' and '>' (which were escaped by
1717 # removeHTMLtags()) should not be included in
1718 # URLs, per RFC 2396.
1719 # Make terminate a URL as well (bug T84937)
1722 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1727 $trail = substr( $url, $m2[0][1] ) . $trail;
1728 $url = substr( $url, 0, $m2[0][1] );
1731 # Move trailing punctuation to $trail
1733 # If there is no left bracket, then consider right brackets fair game too
1734 if ( strpos( $url, '(' ) === false ) {
1738 $urlRev = strrev( $url );
1739 $numSepChars = strspn( $urlRev, $sep );
1740 # Don't break a trailing HTML entity by moving the ; into $trail
1741 # This is in hot code, so use substr_compare to avoid having to
1742 # create a new string object for the comparison
1743 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1744 # more optimization: instead of running preg_match with a $
1745 # anchor, which can be slow, do the match on the reversed
1746 # string starting at the desired offset.
1747 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1748 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1752 if ( $numSepChars ) {
1753 $trail = substr( $url, -$numSepChars ) . $trail;
1754 $url = substr( $url, 0, -$numSepChars );
1757 # Verify that we still have a real URL after trail removal, and
1758 # not just lone protocol
1759 if ( strlen( $trail ) >= $numPostProto ) {
1760 return $url . $trail;
1763 $url = Sanitizer
::cleanUrl( $url );
1765 # Is this an external image?
1766 $text = $this->maybeMakeExternalImage( $url );
1767 if ( $text === false ) {
1768 # Not an image, make a link
1769 $text = Linker
::makeExternalLink( $url,
1770 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1772 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1773 # Register it in the output object...
1774 $this->mOutput
->addExternalLink( $url );
1776 return $text . $trail;
1780 * Parse headers and return html
1784 * @param string $text
1788 public function doHeadings( $text ) {
1789 for ( $i = 6; $i >= 1; --$i ) {
1790 $h = str_repeat( '=', $i );
1791 // Trim non-newline whitespace from headings
1792 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1793 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1799 * Replace single quotes with HTML markup
1802 * @param string $text
1804 * @return string The altered text
1806 public function doAllQuotes( $text ) {
1808 $lines = StringUtils
::explode( "\n", $text );
1809 foreach ( $lines as $line ) {
1810 $outtext .= $this->doQuotes( $line ) . "\n";
1812 $outtext = substr( $outtext, 0, -1 );
1817 * Helper function for doAllQuotes()
1819 * @param string $text
1823 public function doQuotes( $text ) {
1824 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1825 $countarr = count( $arr );
1826 if ( $countarr == 1 ) {
1830 // First, do some preliminary work. This may shift some apostrophes from
1831 // being mark-up to being text. It also counts the number of occurrences
1832 // of bold and italics mark-ups.
1835 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1836 $thislen = strlen( $arr[$i] );
1837 // If there are ever four apostrophes, assume the first is supposed to
1838 // be text, and the remaining three constitute mark-up for bold text.
1839 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1840 if ( $thislen == 4 ) {
1841 $arr[$i - 1] .= "'";
1844 } elseif ( $thislen > 5 ) {
1845 // If there are more than 5 apostrophes in a row, assume they're all
1846 // text except for the last 5.
1847 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1848 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1852 // Count the number of occurrences of bold and italics mark-ups.
1853 if ( $thislen == 2 ) {
1855 } elseif ( $thislen == 3 ) {
1857 } elseif ( $thislen == 5 ) {
1863 // If there is an odd number of both bold and italics, it is likely
1864 // that one of the bold ones was meant to be an apostrophe followed
1865 // by italics. Which one we cannot know for certain, but it is more
1866 // likely to be one that has a single-letter word before it.
1867 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1868 $firstsingleletterword = -1;
1869 $firstmultiletterword = -1;
1871 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1872 if ( strlen( $arr[$i] ) == 3 ) {
1873 $x1 = substr( $arr[$i - 1], -1 );
1874 $x2 = substr( $arr[$i - 1], -2, 1 );
1875 if ( $x1 === ' ' ) {
1876 if ( $firstspace == -1 ) {
1879 } elseif ( $x2 === ' ' ) {
1880 $firstsingleletterword = $i;
1881 // if $firstsingleletterword is set, we don't
1882 // look at the other options, so we can bail early.
1884 } elseif ( $firstmultiletterword == -1 ) {
1885 $firstmultiletterword = $i;
1890 // If there is a single-letter word, use it!
1891 if ( $firstsingleletterword > -1 ) {
1892 $arr[$firstsingleletterword] = "''";
1893 $arr[$firstsingleletterword - 1] .= "'";
1894 } elseif ( $firstmultiletterword > -1 ) {
1895 // If not, but there's a multi-letter word, use that one.
1896 $arr[$firstmultiletterword] = "''";
1897 $arr[$firstmultiletterword - 1] .= "'";
1898 } elseif ( $firstspace > -1 ) {
1899 // ... otherwise use the first one that has neither.
1900 // (notice that it is possible for all three to be -1 if, for example,
1901 // there is only one pentuple-apostrophe in the line)
1902 $arr[$firstspace] = "''";
1903 $arr[$firstspace - 1] .= "'";
1907 // Now let's actually convert our apostrophic mush to HTML!
1912 foreach ( $arr as $r ) {
1913 if ( ( $i %
2 ) == 0 ) {
1914 if ( $state === 'both' ) {
1920 $thislen = strlen( $r );
1921 if ( $thislen == 2 ) {
1922 if ( $state === 'i' ) {
1925 } elseif ( $state === 'bi' ) {
1928 } elseif ( $state === 'ib' ) {
1929 $output .= '</b></i><b>';
1931 } elseif ( $state === 'both' ) {
1932 $output .= '<b><i>' . $buffer . '</i>';
1934 } else { // $state can be 'b' or ''
1938 } elseif ( $thislen == 3 ) {
1939 if ( $state === 'b' ) {
1942 } elseif ( $state === 'bi' ) {
1943 $output .= '</i></b><i>';
1945 } elseif ( $state === 'ib' ) {
1948 } elseif ( $state === 'both' ) {
1949 $output .= '<i><b>' . $buffer . '</b>';
1951 } else { // $state can be 'i' or ''
1955 } elseif ( $thislen == 5 ) {
1956 if ( $state === 'b' ) {
1957 $output .= '</b><i>';
1959 } elseif ( $state === 'i' ) {
1960 $output .= '</i><b>';
1962 } elseif ( $state === 'bi' ) {
1963 $output .= '</i></b>';
1965 } elseif ( $state === 'ib' ) {
1966 $output .= '</b></i>';
1968 } elseif ( $state === 'both' ) {
1969 $output .= '<i><b>' . $buffer . '</b></i>';
1971 } else { // ($state == '')
1979 // Now close all remaining tags. Notice that the order is important.
1980 if ( $state === 'b' ||
$state === 'ib' ) {
1983 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1986 if ( $state === 'bi' ) {
1989 // There might be lonely ''''', so make sure we have a buffer
1990 if ( $state === 'both' && $buffer ) {
1991 $output .= '<b><i>' . $buffer . '</i></b>';
1997 * Replace external links (REL)
1999 * Note: this is all very hackish and the order of execution matters a lot.
2000 * Make sure to run tests/parser/parserTests.php if you change this code.
2004 * @param string $text
2006 * @throws MWException
2009 public function replaceExternalLinks( $text ) {
2010 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
2011 if ( $bits === false ) {
2012 throw new MWException( "PCRE needs to be compiled with "
2013 . "--enable-unicode-properties in order for MediaWiki to function" );
2015 $s = array_shift( $bits );
2018 while ( $i < count( $bits ) ) {
2021 $text = $bits[$i++
];
2022 $trail = $bits[$i++
];
2024 # The characters '<' and '>' (which were escaped by
2025 # removeHTMLtags()) should not be included in
2026 # URLs, per RFC 2396.
2028 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
2029 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
2030 $url = substr( $url, 0, $m2[0][1] );
2033 # If the link text is an image URL, replace it with an <img> tag
2034 # This happened by accident in the original parser, but some people used it extensively
2035 $img = $this->maybeMakeExternalImage( $text );
2036 if ( $img !== false ) {
2042 # Set linktype for CSS
2045 # No link text, e.g. [http://domain.tld/some.link]
2046 if ( $text == '' ) {
2048 $langObj = $this->getTargetLanguage();
2049 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
2050 $linktype = 'autonumber';
2052 # Have link text, e.g. [http://domain.tld/some.link text]s
2054 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
2057 // Excluding protocol-relative URLs may avoid many false positives.
2058 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2059 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2062 $url = Sanitizer
::cleanUrl( $url );
2064 # Use the encoded URL
2065 # This means that users can paste URLs directly into the text
2066 # Funny characters like ö aren't valid in URLs anyway
2067 # This was changed in August 2004
2068 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
2069 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
2071 # Register link in the output object.
2072 $this->mOutput
->addExternalLink( $url );
2079 * Get the rel attribute for a particular external link.
2082 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2083 * nofollow if appropriate
2084 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2085 * @return string|null Rel attribute for $url
2087 public static function getExternalLinkRel( $url = false, $title = null ) {
2088 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2089 $ns = $title ?
$title->getNamespace() : false;
2090 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2091 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2099 * Get an associative array of additional HTML attributes appropriate for a
2100 * particular external link. This currently may include rel => nofollow
2101 * (depending on configuration, namespace, and the URL's domain) and/or a
2102 * target attribute (depending on configuration).
2104 * @param string $url URL to extract the domain from for rel =>
2105 * nofollow if appropriate
2106 * @return array Associative array of HTML attributes
2108 public function getExternalLinkAttribs( $url ) {
2110 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2112 $target = $this->mOptions
->getExternalLinkTarget();
2114 $attribs['target'] = $target;
2115 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2116 // T133507. New windows can navigate parent cross-origin.
2117 // Including noreferrer due to lacking browser
2118 // support of noopener. Eventually noreferrer should be removed.
2119 if ( $rel !== '' ) {
2122 $rel .= 'noreferrer noopener';
2125 $attribs['rel'] = $rel;
2130 * Replace unusual escape codes in a URL with their equivalent characters
2132 * This generally follows the syntax defined in RFC 3986, with special
2133 * consideration for HTTP query strings.
2135 * @param string $url
2138 public static function normalizeLinkUrl( $url ) {
2139 # Test for RFC 3986 IPv6 syntax
2140 $scheme = '[a-z][a-z0-9+.-]*:';
2141 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2142 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2143 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2144 IP
::isValid( rawurldecode( $m[1] ) )
2146 $isIPv6 = rawurldecode( $m[1] );
2151 # Make sure unsafe characters are encoded
2152 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2154 return rawurlencode( $m[0] );
2160 $end = strlen( $url );
2162 # Fragment part - 'fragment'
2163 $start = strpos( $url, '#' );
2164 if ( $start !== false && $start < $end ) {
2165 $ret = self
::normalizeUrlComponent(
2166 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2170 # Query part - 'query' minus &=+;
2171 $start = strpos( $url, '?' );
2172 if ( $start !== false && $start < $end ) {
2173 $ret = self
::normalizeUrlComponent(
2174 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2178 # Scheme and path part - 'pchar'
2179 # (we assume no userinfo or encoded colons in the host)
2180 $ret = self
::normalizeUrlComponent(
2181 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2184 if ( $isIPv6 !== false ) {
2185 $ipv6Host = "%5B({$isIPv6})%5D";
2186 $ret = preg_replace(
2187 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2196 private static function normalizeUrlComponent( $component, $unsafe ) {
2197 $callback = function ( $matches ) use ( $unsafe ) {
2198 $char = urldecode( $matches[0] );
2199 $ord = ord( $char );
2200 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2204 # Leave it escaped, but use uppercase for a-f
2205 return strtoupper( $matches[0] );
2208 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2212 * make an image if it's allowed, either through the global
2213 * option, through the exception, or through the on-wiki whitelist
2215 * @param string $url
2219 private function maybeMakeExternalImage( $url ) {
2220 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2221 $imagesexception = !empty( $imagesfrom );
2223 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2224 if ( $imagesexception && is_array( $imagesfrom ) ) {
2225 $imagematch = false;
2226 foreach ( $imagesfrom as $match ) {
2227 if ( strpos( $url, $match ) === 0 ) {
2232 } elseif ( $imagesexception ) {
2233 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2235 $imagematch = false;
2238 if ( $this->mOptions
->getAllowExternalImages()
2239 ||
( $imagesexception && $imagematch )
2241 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2243 $text = Linker
::makeExternalImage( $url );
2246 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2247 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2249 $whitelist = explode(
2251 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2254 foreach ( $whitelist as $entry ) {
2255 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2256 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2259 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2260 # Image matches a whitelist entry
2261 $text = Linker
::makeExternalImage( $url );
2270 * Process [[ ]] wikilinks
2274 * @return string Processed text
2278 public function replaceInternalLinks( $s ) {
2279 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2284 * Process [[ ]] wikilinks (RIL)
2286 * @throws MWException
2287 * @return LinkHolderArray
2291 public function replaceInternalLinks2( &$s ) {
2292 static $tc = false, $e1, $e1_img;
2293 # the % is needed to support urlencoded titles as well
2295 $tc = Title
::legalChars() . '#%';
2296 # Match a link having the form [[namespace:link|alternate]]trail
2297 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2298 # Match cases where there is no "]]", which might still be images
2299 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2302 $holders = new LinkHolderArray( $this );
2304 # split the entire text string on occurrences of [[
2305 $a = StringUtils
::explode( '[[', ' ' . $s );
2306 # get the first element (all text up to first [[), and remove the space we added
2309 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2310 $s = substr( $s, 1 );
2312 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2314 if ( $useLinkPrefixExtension ) {
2315 # Match the end of a line for a word that's not followed by whitespace,
2316 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2317 $charset = $this->contLang
->linkPrefixCharset();
2318 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2321 if ( is_null( $this->mTitle
) ) {
2322 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2324 $nottalk = !$this->mTitle
->isTalkPage();
2326 if ( $useLinkPrefixExtension ) {
2328 if ( preg_match( $e2, $s, $m ) ) {
2329 $first_prefix = $m[2];
2331 $first_prefix = false;
2337 $useSubpages = $this->areSubpagesAllowed();
2339 # Loop for each link
2340 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2341 # Check for excessive memory usage
2342 if ( $holders->isBig() ) {
2344 # Do the existence check, replace the link holders and clear the array
2345 $holders->replace( $s );
2349 if ( $useLinkPrefixExtension ) {
2350 if ( preg_match( $e2, $s, $m ) ) {
2351 list( , $s, $prefix ) = $m;
2356 if ( $first_prefix ) {
2357 $prefix = $first_prefix;
2358 $first_prefix = false;
2362 $might_be_img = false;
2364 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2366 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2367 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2368 # the real problem is with the $e1 regex
2370 # Still some problems for cases where the ] is meant to be outside punctuation,
2371 # and no image is in sight. See T4095.
2373 && substr( $m[3], 0, 1 ) === ']'
2374 && strpos( $text, '[' ) !== false
2376 $text .= ']'; # so that replaceExternalLinks($text) works later
2377 $m[3] = substr( $m[3], 1 );
2379 # fix up urlencoded title texts
2380 if ( strpos( $m[1], '%' ) !== false ) {
2381 # Should anchors '#' also be rejected?
2382 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2385 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2386 # Invalid, but might be an image with a link in its caption
2387 $might_be_img = true;
2389 if ( strpos( $m[1], '%' ) !== false ) {
2390 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2393 } else { # Invalid form; output directly
2394 $s .= $prefix . '[[' . $line;
2398 $origLink = ltrim( $m[1], ' ' );
2400 # Don't allow internal links to pages containing
2401 # PROTO: where PROTO is a valid URL protocol; these
2402 # should be external links.
2403 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2404 $s .= $prefix . '[[' . $line;
2408 # Make subpage if necessary
2409 if ( $useSubpages ) {
2410 $link = $this->maybeDoSubpageLink( $origLink, $text );
2415 // \x7f isn't a default legal title char, so most likely strip
2416 // markers will force us into the "invalid form" path above. But,
2417 // just in case, let's assert that xmlish tags aren't valid in
2418 // the title position.
2419 $unstrip = $this->mStripState
->killMarkers( $link );
2420 $noMarkers = ( $unstrip === $link );
2422 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2423 if ( $nt === null ) {
2424 $s .= $prefix . '[[' . $line;
2428 $ns = $nt->getNamespace();
2429 $iw = $nt->getInterwiki();
2431 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2433 if ( $might_be_img ) { # if this is actually an invalid link
2434 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2437 # look at the next 'line' to see if we can close it there
2439 $next_line = $a->current();
2440 if ( $next_line === false ||
$next_line === null ) {
2443 $m = explode( ']]', $next_line, 3 );
2444 if ( count( $m ) == 3 ) {
2445 # the first ]] closes the inner link, the second the image
2447 $text .= "[[{$m[0]}]]{$m[1]}";
2450 } elseif ( count( $m ) == 2 ) {
2451 # if there's exactly one ]] that's fine, we'll keep looking
2452 $text .= "[[{$m[0]}]]{$m[1]}";
2454 # if $next_line is invalid too, we need look no further
2455 $text .= '[[' . $next_line;
2460 # we couldn't find the end of this imageLink, so output it raw
2461 # but don't ignore what might be perfectly normal links in the text we've examined
2462 $holders->merge( $this->replaceInternalLinks2( $text ) );
2463 $s .= "{$prefix}[[$link|$text";
2464 # note: no $trail, because without an end, there *is* no trail
2467 } else { # it's not an image, so output it raw
2468 $s .= "{$prefix}[[$link|$text";
2469 # note: no $trail, because without an end, there *is* no trail
2474 $wasblank = ( $text == '' );
2478 # Strip off leading ':'
2479 $text = substr( $text, 1 );
2482 # T6598 madness. Handle the quotes only if they come from the alternate part
2483 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2484 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2485 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2486 $text = $this->doQuotes( $text );
2489 # Link not escaped by : , create the various objects
2490 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2493 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2494 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2495 in_array( $iw, $this->svcOptions
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2498 # T26502: filter duplicates
2499 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2500 $this->mLangLinkLanguages
[$iw] = true;
2501 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2505 * Strip the whitespace interwiki links produce, see T10897
2507 $s = rtrim( $s . $prefix ) . $trail; # T175416
2511 if ( $ns == NS_FILE
) {
2512 if ( !$this->badFileLookup
->isBadFile( $nt->getDBkey(), $this->mTitle
) ) {
2514 # if no parameters were passed, $text
2515 # becomes something like "File:Foo.png",
2516 # which we don't want to pass on to the
2520 # recursively parse links inside the image caption
2521 # actually, this will parse them in any other parameters, too,
2522 # but it might be hard to fix that, and it doesn't matter ATM
2523 $text = $this->replaceExternalLinks( $text );
2524 $holders->merge( $this->replaceInternalLinks2( $text ) );
2526 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2527 $s .= $prefix . $this->armorLinks(
2528 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2531 } elseif ( $ns == NS_CATEGORY
) {
2533 * Strip the whitespace Category links produce, see T2087
2535 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2538 $sortkey = $this->getDefaultSort();
2542 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2543 $sortkey = str_replace( "\n", '', $sortkey );
2544 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2545 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2551 # Self-link checking. For some languages, variants of the title are checked in
2552 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2553 # for linking to a different variant.
2554 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2555 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2559 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2560 # @todo FIXME: Should do batch file existence checks, see comment below
2561 if ( $ns == NS_MEDIA
) {
2562 # Give extensions a chance to select the file revision for us
2565 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2566 [ $this, $nt, &$options, &$descQuery ] );
2567 # Fetch and register the file (file title may be different via hooks)
2568 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2569 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2570 $s .= $prefix . $this->armorLinks(
2571 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2575 # Some titles, such as valid special pages or files in foreign repos, should
2576 # be shown as bluelinks even though they're not included in the page table
2577 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2578 # batch file existence checks for NS_FILE and NS_MEDIA
2579 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2580 $this->mOutput
->addLink( $nt );
2581 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2583 # Links will be added to the output link list after checking
2584 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2591 * Render a forced-blue link inline; protect against double expansion of
2592 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2593 * Since this little disaster has to split off the trail text to avoid
2594 * breaking URLs in the following text without breaking trails on the
2595 * wiki links, it's been made into a horrible function.
2598 * @param string $text
2599 * @param string $trail
2600 * @param string $prefix
2601 * @return string HTML-wikitext mix oh yuck
2603 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2604 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2606 if ( $text == '' ) {
2607 $text = htmlspecialchars( $nt->getPrefixedText() );
2610 $link = $this->getLinkRenderer()->makeKnownLink(
2611 $nt, new HtmlArmor( "$prefix$text$inside" )
2614 return $this->armorLinks( $link ) . $trail;
2618 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2619 * going to go through further parsing steps before inline URL expansion.
2621 * Not needed quite as much as it used to be since free links are a bit
2622 * more sensible these days. But bracketed links are still an issue.
2624 * @param string $text More-or-less HTML
2625 * @return string Less-or-more HTML with NOPARSE bits
2627 public function armorLinks( $text ) {
2628 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2629 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2633 * Return true if subpage links should be expanded on this page.
2636 public function areSubpagesAllowed() {
2637 # Some namespaces don't allow subpages
2638 return $this->nsInfo
->hasSubpages( $this->mTitle
->getNamespace() );
2642 * Handle link to subpage if necessary
2644 * @param string $target The source of the link
2645 * @param string &$text The link text, modified as necessary
2646 * @return string The full name of the link
2649 public function maybeDoSubpageLink( $target, &$text ) {
2650 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2654 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2656 * @param string $text
2657 * @param bool $linestart Whether or not this is at the start of a line.
2659 * @return string The lists rendered as HTML
2661 public function doBlockLevels( $text, $linestart ) {
2662 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2666 * Return value of a magic variable (like PAGENAME)
2670 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2671 * @param bool|PPFrame $frame
2673 * @throws MWException
2676 public function getVariableValue( $index, $frame = false ) {
2677 if ( is_null( $this->mTitle
) ) {
2678 // If no title set, bad things are going to happen
2679 // later. Title should always be set since this
2680 // should only be called in the middle of a parse
2681 // operation (but the unit-tests do funky stuff)
2682 throw new MWException( __METHOD__
. ' Should only be '
2683 . ' called while parsing (no title set)' );
2686 // Avoid PHP 7.1 warning from passing $this by reference
2690 * Some of these require message or data lookups and can be
2691 * expensive to check many times.
2694 Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) &&
2695 isset( $this->mVarCache
[$index] )
2697 return $this->mVarCache
[$index];
2700 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2701 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2703 $pageLang = $this->getFunctionLang();
2709 case 'currentmonth':
2710 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2712 case 'currentmonth1':
2713 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2715 case 'currentmonthname':
2716 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2718 case 'currentmonthnamegen':
2719 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2721 case 'currentmonthabbrev':
2722 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2725 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2728 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2731 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2734 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2736 case 'localmonthname':
2737 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2739 case 'localmonthnamegen':
2740 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2742 case 'localmonthabbrev':
2743 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2746 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2749 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2752 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2755 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2757 case 'fullpagename':
2758 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2760 case 'fullpagenamee':
2761 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2764 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2766 case 'subpagenamee':
2767 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2769 case 'rootpagename':
2770 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2772 case 'rootpagenamee':
2773 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2776 $this->mTitle
->getRootText()
2779 case 'basepagename':
2780 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2782 case 'basepagenamee':
2783 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2786 $this->mTitle
->getBaseText()
2789 case 'talkpagename':
2790 if ( $this->mTitle
->canHaveTalkPage() ) {
2791 $talkPage = $this->mTitle
->getTalkPage();
2792 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2797 case 'talkpagenamee':
2798 if ( $this->mTitle
->canHaveTalkPage() ) {
2799 $talkPage = $this->mTitle
->getTalkPage();
2800 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2805 case 'subjectpagename':
2806 $subjPage = $this->mTitle
->getSubjectPage();
2807 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2809 case 'subjectpagenamee':
2810 $subjPage = $this->mTitle
->getSubjectPage();
2811 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2813 case 'pageid': // requested in T25427
2814 # Inform the edit saving system that getting the canonical output
2815 # after page insertion requires a parse that used that exact page ID
2816 $this->setOutputFlag( 'vary-page-id', '{{PAGEID}} used' );
2817 $value = $this->mTitle
->getArticleID();
2819 $value = $this->mOptions
->getSpeculativePageId();
2821 $this->mOutput
->setSpeculativePageIdUsed( $value );
2827 $this->svcOptions
->get( 'MiserMode' ) &&
2828 !$this->mOptions
->getInterfaceMessage() &&
2829 // @TODO: disallow this word on all namespaces
2830 $this->nsInfo
->isContent( $this->mTitle
->getNamespace() )
2832 // Use a stub result instead of the actual revision ID in order to avoid
2833 // double parses on page save but still allow preview detection (T137900)
2834 if ( $this->getRevisionId() ||
$this->mOptions
->getSpeculativeRevId() ) {
2837 $this->setOutputFlag( 'vary-revision-exists', '{{REVISIONID}} used' );
2841 # Inform the edit saving system that getting the canonical output after
2842 # revision insertion requires a parse that used that exact revision ID
2843 $this->setOutputFlag( 'vary-revision-id', '{{REVISIONID}} used' );
2844 $value = $this->getRevisionId();
2845 if ( $value === 0 ) {
2846 $rev = $this->getRevisionObject();
2847 $value = $rev ?
$rev->getId() : $value;
2850 $value = $this->mOptions
->getSpeculativeRevId();
2852 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2858 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2860 case 'revisionday2':
2861 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2863 case 'revisionmonth':
2864 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2866 case 'revisionmonth1':
2867 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2869 case 'revisionyear':
2870 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2872 case 'revisiontimestamp':
2873 $value = $this->getRevisionTimestampSubstring( 0, 14, self
::MAX_TTS
, $index );
2875 case 'revisionuser':
2876 # Inform the edit saving system that getting the canonical output after
2877 # revision insertion requires a parse that used the actual user ID
2878 $this->setOutputFlag( 'vary-user', '{{REVISIONUSER}} used' );
2879 $value = $this->getRevisionUser();
2881 case 'revisionsize':
2882 $value = $this->getRevisionSize();
2885 $value = str_replace( '_', ' ',
2886 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2889 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2891 case 'namespacenumber':
2892 $value = $this->mTitle
->getNamespace();
2895 $value = $this->mTitle
->canHaveTalkPage()
2896 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2900 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2902 case 'subjectspace':
2903 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2905 case 'subjectspacee':
2906 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2908 case 'currentdayname':
2909 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2912 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2915 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2918 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2921 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2922 # int to remove the padding
2923 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2926 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2928 case 'localdayname':
2929 $value = $pageLang->getWeekdayName(
2930 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2934 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2937 $value = $pageLang->time(
2938 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2944 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2947 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2948 # int to remove the padding
2949 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2952 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2954 case 'numberofarticles':
2955 $value = $pageLang->formatNum( SiteStats
::articles() );
2957 case 'numberoffiles':
2958 $value = $pageLang->formatNum( SiteStats
::images() );
2960 case 'numberofusers':
2961 $value = $pageLang->formatNum( SiteStats
::users() );
2963 case 'numberofactiveusers':
2964 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2966 case 'numberofpages':
2967 $value = $pageLang->formatNum( SiteStats
::pages() );
2969 case 'numberofadmins':
2970 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2972 case 'numberofedits':
2973 $value = $pageLang->formatNum( SiteStats
::edits() );
2975 case 'currenttimestamp':
2976 $value = wfTimestamp( TS_MW
, $ts );
2978 case 'localtimestamp':
2979 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2981 case 'currentversion':
2982 $value = SpecialVersion
::getVersion();
2985 return $this->svcOptions
->get( 'ArticlePath' );
2987 return $this->svcOptions
->get( 'Sitename' );
2989 return $this->svcOptions
->get( 'Server' );
2991 return $this->svcOptions
->get( 'ServerName' );
2993 return $this->svcOptions
->get( 'ScriptPath' );
2995 return $this->svcOptions
->get( 'StylePath' );
2996 case 'directionmark':
2997 return $pageLang->getDirMark();
2998 case 'contentlanguage':
2999 return $this->svcOptions
->get( 'LanguageCode' );
3000 case 'pagelanguage':
3001 $value = $pageLang->getCode();
3003 case 'cascadingsources':
3004 $value = CoreParserFunctions
::cascadingsources( $this );
3009 'ParserGetVariableValueSwitch',
3010 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
3017 $this->mVarCache
[$index] = $value;
3026 * @param int $mtts Max time-till-save; sets vary-revision-timestamp if result changes by then
3027 * @param string $variable Parser variable name
3030 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
3031 # Get the timezone-adjusted timestamp to be used for this revision
3032 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
3033 # Possibly set vary-revision if there is not yet an associated revision
3034 if ( !$this->getRevisionObject() ) {
3035 # Get the timezone-adjusted timestamp $mtts seconds in the future.
3036 # This future is relative to the current time and not that of the
3037 # parser options. The rendered timestamp can be compared to that
3038 # of the timestamp specified by the parser options.
3040 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
3045 if ( $resNow !== $resThen ) {
3046 # Inform the edit saving system that getting the canonical output after
3047 # revision insertion requires a parse that used an actual revision timestamp
3048 $this->setOutputFlag( 'vary-revision-timestamp', "$variable used" );
3056 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3060 public function initialiseVariables() {
3061 $variableIDs = $this->magicWordFactory
->getVariableIDs();
3062 $substIDs = $this->magicWordFactory
->getSubstIDs();
3064 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
3065 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
3069 * Preprocess some wikitext and return the document tree.
3070 * This is the ghost of replace_variables().
3072 * @param string $text The text to parse
3073 * @param int $flags Bitwise combination of:
3074 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3075 * included. Default is to assume a direct page view.
3077 * The generated DOM tree must depend only on the input text and the flags.
3078 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3080 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3081 * change in the DOM tree for a given text, must be passed through the section identifier
3082 * in the section edit link and thus back to extractSections().
3084 * The output of this function is currently only cached in process memory, but a persistent
3085 * cache may be implemented at a later date which takes further advantage of these strict
3086 * dependency requirements.
3090 public function preprocessToDom( $text, $flags = 0 ) {
3091 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3096 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3102 public static function splitWhitespace( $s ) {
3103 $ltrimmed = ltrim( $s );
3104 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3105 $trimmed = rtrim( $ltrimmed );
3106 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3108 $w2 = substr( $ltrimmed, -$diff );
3112 return [ $w1, $trimmed, $w2 ];
3116 * Replace magic variables, templates, and template arguments
3117 * with the appropriate text. Templates are substituted recursively,
3118 * taking care to avoid infinite loops.
3120 * Note that the substitution depends on value of $mOutputType:
3121 * self::OT_WIKI: only {{subst:}} templates
3122 * self::OT_PREPROCESS: templates but not extension tags
3123 * self::OT_HTML: all templates and extension tags
3125 * @param string $text The text to transform
3126 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3127 * template. Arguments may also be provided as an associative array, as
3128 * was the usual case before MW1.12. Providing arguments this way may be
3129 * useful for extensions wishing to perform variable replacement
3131 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3132 * double-brace expansion.
3135 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3136 # Is there any text? Also, Prevent too big inclusions!
3137 $textSize = strlen( $text );
3138 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3142 if ( $frame === false ) {
3143 $frame = $this->getPreprocessor()->newFrame();
3144 } elseif ( !( $frame instanceof PPFrame
) ) {
3145 $this->logger
->debug(
3146 __METHOD__
. " called using plain parameters instead of " .
3147 "a PPFrame instance. Creating custom frame."
3149 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3152 $dom = $this->preprocessToDom( $text );
3153 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3154 $text = $frame->expand( $dom, $flags );
3160 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3162 * @param array $args
3166 public static function createAssocArgs( $args ) {
3169 foreach ( $args as $arg ) {
3170 $eqpos = strpos( $arg, '=' );
3171 if ( $eqpos === false ) {
3172 $assocArgs[$index++
] = $arg;
3174 $name = trim( substr( $arg, 0, $eqpos ) );
3175 $value = trim( substr( $arg, $eqpos +
1 ) );
3176 if ( $value === false ) {
3179 if ( $name !== false ) {
3180 $assocArgs[$name] = $value;
3189 * Warn the user when a parser limitation is reached
3190 * Will warn at most once the user per limitation type
3192 * The results are shown during preview and run through the Parser (See EditPage.php)
3194 * @param string $limitationType Should be one of:
3195 * 'expensive-parserfunction' (corresponding messages:
3196 * 'expensive-parserfunction-warning',
3197 * 'expensive-parserfunction-category')
3198 * 'post-expand-template-argument' (corresponding messages:
3199 * 'post-expand-template-argument-warning',
3200 * 'post-expand-template-argument-category')
3201 * 'post-expand-template-inclusion' (corresponding messages:
3202 * 'post-expand-template-inclusion-warning',
3203 * 'post-expand-template-inclusion-category')
3204 * 'node-count-exceeded' (corresponding messages:
3205 * 'node-count-exceeded-warning',
3206 * 'node-count-exceeded-category')
3207 * 'expansion-depth-exceeded' (corresponding messages:
3208 * 'expansion-depth-exceeded-warning',
3209 * 'expansion-depth-exceeded-category')
3210 * @param string|int|null $current Current value
3211 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3212 * exceeded, provide the values (optional)
3214 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3215 # does no harm if $current and $max are present but are unnecessary for the message
3216 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3217 # only during preview, and that would split the parser cache unnecessarily.
3218 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3220 $this->mOutput
->addWarning( $warning );
3221 $this->addTrackingCategory( "$limitationType-category" );
3225 * Return the text of a template, after recursively
3226 * replacing any variables or templates within the template.
3228 * @param array $piece The parts of the template
3229 * $piece['title']: the title, i.e. the part before the |
3230 * $piece['parts']: the parameter array
3231 * $piece['lineStart']: whether the brace was at the start of a line
3232 * @param PPFrame $frame The current frame, contains template arguments
3234 * @return string|array The text of the template
3236 public function braceSubstitution( $piece, $frame ) {
3239 // $text has been filled
3241 // wiki markup in $text should be escaped
3243 // $text is HTML, armour it against wikitext transformation
3245 // Force interwiki transclusion to be done in raw mode not rendered
3246 $forceRawInterwiki = false;
3247 // $text is a DOM node needing expansion in a child frame
3248 $isChildObj = false;
3249 // $text is a DOM node needing expansion in the current frame
3250 $isLocalObj = false;
3252 # Title object, where $text came from
3255 # $part1 is the bit before the first |, and must contain only title characters.
3256 # Various prefixes will be stripped from it later.
3257 $titleWithSpaces = $frame->expand( $piece['title'] );
3258 $part1 = trim( $titleWithSpaces );
3261 # Original title text preserved for various purposes
3262 $originalTitle = $part1;
3264 # $args is a list of argument nodes, starting from index 0, not including $part1
3265 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3266 # below won't work b/c this $args isn't an object
3267 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3269 $profileSection = null; // profile templates
3273 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3275 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3276 # Decide whether to expand template or keep wikitext as-is.
3277 if ( $this->ot
['wiki'] ) {
3278 if ( $substMatch === false ) {
3279 $literal = true; # literal when in PST with no prefix
3281 $literal = false; # expand when in PST with subst: or safesubst:
3284 if ( $substMatch == 'subst' ) {
3285 $literal = true; # literal when not in PST with plain subst:
3287 $literal = false; # expand when not in PST with safesubst: or no prefix
3291 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3298 if ( !$found && $args->getLength() == 0 ) {
3299 $id = $this->mVariables
->matchStartToEnd( $part1 );
3300 if ( $id !== false ) {
3301 $text = $this->getVariableValue( $id, $frame );
3302 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3303 $this->mOutput
->updateCacheExpiry(
3304 $this->magicWordFactory
->getCacheTTL( $id ) );
3310 # MSG, MSGNW and RAW
3313 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3314 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3317 # Remove obsolete MSG:
3318 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3319 $mwMsg->matchStartAndRemove( $part1 );
3323 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3324 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3325 $forceRawInterwiki = true;
3331 $colonPos = strpos( $part1, ':' );
3332 if ( $colonPos !== false ) {
3333 $func = substr( $part1, 0, $colonPos );
3334 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3335 $argsLength = $args->getLength();
3336 for ( $i = 0; $i < $argsLength; $i++
) {
3337 $funcArgs[] = $args->item( $i );
3340 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3342 // Extract any forwarded flags
3343 if ( isset( $result['title'] ) ) {
3344 $title = $result['title'];
3346 if ( isset( $result['found'] ) ) {
3347 $found = $result['found'];
3349 if ( array_key_exists( 'text', $result ) ) {
3351 $text = $result['text'];
3353 if ( isset( $result['nowiki'] ) ) {
3354 $nowiki = $result['nowiki'];
3356 if ( isset( $result['isHTML'] ) ) {
3357 $isHTML = $result['isHTML'];
3359 if ( isset( $result['forceRawInterwiki'] ) ) {
3360 $forceRawInterwiki = $result['forceRawInterwiki'];
3362 if ( isset( $result['isChildObj'] ) ) {
3363 $isChildObj = $result['isChildObj'];
3365 if ( isset( $result['isLocalObj'] ) ) {
3366 $isLocalObj = $result['isLocalObj'];
3371 # Finish mangling title and then check for loops.
3372 # Set $title to a Title object and $titleText to the PDBK
3375 # Split the title into page and subpage
3377 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3378 if ( $part1 !== $relative ) {
3380 $ns = $this->mTitle
->getNamespace();
3382 $title = Title
::newFromText( $part1, $ns );
3384 $titleText = $title->getPrefixedText();
3385 # Check for language variants if the template is not found
3386 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3387 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3389 # Do recursion depth check
3390 $limit = $this->mOptions
->getMaxTemplateDepth();
3391 if ( $frame->depth
>= $limit ) {
3393 $text = '<span class="error">'
3394 . wfMessage( 'parser-template-recursion-depth-warning' )
3395 ->numParams( $limit )->inContentLanguage()->text()
3401 # Load from database
3402 if ( !$found && $title ) {
3403 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3404 if ( !$title->isExternal() ) {
3405 if ( $title->isSpecialPage()
3406 && $this->mOptions
->getAllowSpecialInclusion()
3407 && $this->ot
['html']
3409 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3410 // Pass the template arguments as URL parameters.
3411 // "uselang" will have no effect since the Language object
3412 // is forced to the one defined in ParserOptions.
3414 $argsLength = $args->getLength();
3415 for ( $i = 0; $i < $argsLength; $i++
) {
3416 $bits = $args->item( $i )->splitArg();
3417 if ( strval( $bits['index'] ) === '' ) {
3418 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3419 $value = trim( $frame->expand( $bits['value'] ) );
3420 $pageArgs[$name] = $value;
3424 // Create a new context to execute the special page
3425 $context = new RequestContext
;
3426 $context->setTitle( $title );
3427 $context->setRequest( new FauxRequest( $pageArgs ) );
3428 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3429 $context->setUser( $this->getUser() );
3431 // If this page is cached, then we better not be per user.
3432 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3434 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3435 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3437 $text = $context->getOutput()->getHTML();
3438 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3441 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3442 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3443 $specialPage->maxIncludeCacheTime()
3447 } elseif ( $this->nsInfo
->isNonincludable( $title->getNamespace() ) ) {
3448 $found = false; # access denied
3449 $this->logger
->debug(
3451 ": template inclusion denied for " . $title->getPrefixedDBkey()
3454 list( $text, $title ) = $this->getTemplateDom( $title );
3455 if ( $text !== false ) {
3461 # If the title is valid but undisplayable, make a link to it
3462 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3463 $text = "[[:$titleText]]";
3466 } elseif ( $title->isTrans() ) {
3467 # Interwiki transclusion
3468 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3469 $text = $this->interwikiTransclude( $title, 'render' );
3472 $text = $this->interwikiTransclude( $title, 'raw' );
3473 # Preprocess it like a template
3474 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3480 # Do infinite loop check
3481 # This has to be done after redirect resolution to avoid infinite loops via redirects
3482 if ( !$frame->loopCheck( $title ) ) {
3484 $text = '<span class="error">'
3485 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3487 $this->addTrackingCategory( 'template-loop-category' );
3488 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3489 wfEscapeWikiText( $titleText ) )->text() );
3490 $this->logger
->debug( __METHOD__
. ": template loop broken at '$titleText'" );
3494 # If we haven't found text to substitute by now, we're done
3495 # Recover the source wikitext and return it
3497 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3498 if ( $profileSection ) {
3499 $this->mProfiler
->scopedProfileOut( $profileSection );
3501 return [ 'object' => $text ];
3504 # Expand DOM-style return values in a child frame
3505 if ( $isChildObj ) {
3506 # Clean up argument array
3507 $newFrame = $frame->newChild( $args, $title );
3510 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3511 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3512 # Expansion is eligible for the empty-frame cache
3513 $text = $newFrame->cachedExpand( $titleText, $text );
3515 # Uncached expansion
3516 $text = $newFrame->expand( $text );
3519 if ( $isLocalObj && $nowiki ) {
3520 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3521 $isLocalObj = false;
3524 if ( $profileSection ) {
3525 $this->mProfiler
->scopedProfileOut( $profileSection );
3528 # Replace raw HTML by a placeholder
3530 $text = $this->insertStripItem( $text );
3531 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3532 # Escape nowiki-style return values
3533 $text = wfEscapeWikiText( $text );
3534 } elseif ( is_string( $text )
3535 && !$piece['lineStart']
3536 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3538 # T2529: if the template begins with a table or block-level
3539 # element, it should be treated as beginning a new line.
3540 # This behavior is somewhat controversial.
3541 $text = "\n" . $text;
3544 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3545 # Error, oversize inclusion
3546 if ( $titleText !== false ) {
3547 # Make a working, properly escaped link if possible (T25588)
3548 $text = "[[:$titleText]]";
3550 # This will probably not be a working link, but at least it may
3551 # provide some hint of where the problem is
3552 preg_replace( '/^:/', '', $originalTitle );
3553 $text = "[[:$originalTitle]]";
3555 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3556 . 'post-expand include size too large -->' );
3557 $this->limitationWarn( 'post-expand-template-inclusion' );
3560 if ( $isLocalObj ) {
3561 $ret = [ 'object' => $text ];
3563 $ret = [ 'text' => $text ];
3570 * Call a parser function and return an array with text and flags.
3572 * The returned array will always contain a boolean 'found', indicating
3573 * whether the parser function was found or not. It may also contain the
3575 * text: string|object, resulting wikitext or PP DOM object
3576 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3577 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3578 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3579 * nowiki: bool, wiki markup in $text should be escaped
3582 * @param PPFrame $frame The current frame, contains template arguments
3583 * @param string $function Function name
3584 * @param array $args Arguments to the function
3585 * @throws MWException
3588 public function callParserFunction( $frame, $function, array $args = [] ) {
3589 # Case sensitive functions
3590 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3591 $function = $this->mFunctionSynonyms
[1][$function];
3593 # Case insensitive functions
3594 $function = $this->contLang
->lc( $function );
3595 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3596 $function = $this->mFunctionSynonyms
[0][$function];
3598 return [ 'found' => false ];
3602 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3604 // Avoid PHP 7.1 warning from passing $this by reference
3607 $allArgs = [ &$parser ];
3608 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3609 # Convert arguments to PPNodes and collect for appending to $allArgs
3611 foreach ( $args as $k => $v ) {
3612 if ( $v instanceof PPNode ||
$k === 0 ) {
3615 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3619 # Add a frame parameter, and pass the arguments as an array
3620 $allArgs[] = $frame;
3621 $allArgs[] = $funcArgs;
3623 # Convert arguments to plain text and append to $allArgs
3624 foreach ( $args as $k => $v ) {
3625 if ( $v instanceof PPNode
) {
3626 $allArgs[] = trim( $frame->expand( $v ) );
3627 } elseif ( is_int( $k ) && $k >= 0 ) {
3628 $allArgs[] = trim( $v );
3630 $allArgs[] = trim( "$k=$v" );
3635 $result = $callback( ...$allArgs );
3637 # The interface for function hooks allows them to return a wikitext
3638 # string or an array containing the string and any flags. This mungs
3639 # things around to match what this method should return.
3640 if ( !is_array( $result ) ) {
3646 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3647 $result['text'] = $result[0];
3649 unset( $result[0] );
3656 $preprocessFlags = 0;
3657 if ( isset( $result['noparse'] ) ) {
3658 $noparse = $result['noparse'];
3660 if ( isset( $result['preprocessFlags'] ) ) {
3661 $preprocessFlags = $result['preprocessFlags'];
3665 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3666 $result['isChildObj'] = true;
3673 * Get the semi-parsed DOM representation of a template with a given title,
3674 * and its redirect destination title. Cached.
3676 * @param Title $title
3680 public function getTemplateDom( $title ) {
3681 $cacheTitle = $title;
3682 $titleText = $title->getPrefixedDBkey();
3684 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3685 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3686 $title = Title
::makeTitle( $ns, $dbk );
3687 $titleText = $title->getPrefixedDBkey();
3689 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3690 return [ $this->mTplDomCache
[$titleText], $title ];
3693 # Cache miss, go to the database
3694 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3696 if ( $text === false ) {
3697 $this->mTplDomCache
[$titleText] = false;
3698 return [ false, $title ];
3701 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3702 $this->mTplDomCache
[$titleText] = $dom;
3704 if ( !$title->equals( $cacheTitle ) ) {
3705 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3706 [ $title->getNamespace(), $title->getDBkey() ];
3709 return [ $dom, $title ];
3713 * Fetch the current revision of a given title. Note that the revision
3714 * (and even the title) may not exist in the database, so everything
3715 * contributing to the output of the parser should use this method
3716 * where possible, rather than getting the revisions themselves. This
3717 * method also caches its results, so using it benefits performance.
3720 * @param Title $title
3723 public function fetchCurrentRevisionOfTitle( $title ) {
3724 $cacheKey = $title->getPrefixedDBkey();
3725 if ( !$this->currentRevisionCache
) {
3726 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3728 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3729 $this->currentRevisionCache
->set( $cacheKey,
3730 // Defaults to Parser::statelessFetchRevision()
3731 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3734 return $this->currentRevisionCache
->get( $cacheKey );
3738 * @param Title $title
3742 public function isCurrentRevisionOfTitleCached( $title ) {
3744 $this->currentRevisionCache
&&
3745 $this->currentRevisionCache
->has( $title->getPrefixedText() )
3750 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3751 * without passing them on to it.
3754 * @param Title $title
3755 * @param Parser|bool $parser
3756 * @return Revision|bool False if missing
3758 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3759 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3765 * Fetch the unparsed text of a template and register a reference to it.
3766 * @param Title $title
3767 * @return array ( string or false, Title )
3769 public function fetchTemplateAndTitle( $title ) {
3770 // Defaults to Parser::statelessFetchTemplate()
3771 $templateCb = $this->mOptions
->getTemplateCallback();
3772 $stuff = call_user_func( $templateCb, $title, $this );
3773 $rev = $stuff['revision'] ??
null;
3774 $text = $stuff['text'];
3775 if ( is_string( $stuff['text'] ) ) {
3776 // We use U+007F DELETE to distinguish strip markers from regular text
3777 $text = strtr( $text, "\x7f", "?" );
3779 $finalTitle = $stuff['finalTitle'] ??
$title;
3780 foreach ( ( $stuff['deps'] ??
[] ) as $dep ) {
3781 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3782 if ( $dep['title']->equals( $this->getTitle() ) && $rev instanceof Revision
) {
3783 // Self-transclusion; final result may change based on the new page version
3784 $this->setOutputFlag( 'vary-revision-sha1', 'Self transclusion' );
3785 $this->getOutput()->setRevisionUsedSha1Base36( $rev->getSha1() );
3789 return [ $text, $finalTitle ];
3793 * Fetch the unparsed text of a template and register a reference to it.
3794 * @param Title $title
3795 * @return string|bool
3797 public function fetchTemplate( $title ) {
3798 return $this->fetchTemplateAndTitle( $title )[0];
3802 * Static function to get a template
3803 * Can be overridden via ParserOptions::setTemplateCallback().
3805 * @param Title $title
3806 * @param bool|Parser $parser
3810 public static function statelessFetchTemplate( $title, $parser = false ) {
3811 $text = $skip = false;
3812 $finalTitle = $title;
3816 # Loop to fetch the article, with up to 1 redirect
3817 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3818 # Give extensions a chance to select the revision instead
3819 $id = false; # Assume current
3820 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3821 [ $parser, $title, &$skip, &$id ] );
3827 'page_id' => $title->getArticleID(),
3834 $rev = Revision
::newFromId( $id );
3835 } elseif ( $parser ) {
3836 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3838 $rev = Revision
::newFromTitle( $title );
3840 $rev_id = $rev ?
$rev->getId() : 0;
3841 # If there is no current revision, there is no page
3842 if ( $id === false && !$rev ) {
3843 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3844 $linkCache->addBadLinkObj( $title );
3849 'page_id' => $title->getArticleID(),
3852 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3853 # We fetched a rev from a different title; register it too...
3855 'title' => $rev->getTitle(),
3856 'page_id' => $rev->getPage(),
3862 $content = $rev->getContent();
3863 $text = $content ?
$content->getWikitextForTransclusion() : null;
3865 Hooks
::run( 'ParserFetchTemplate',
3866 [ $parser, $title, $rev, &$text, &$deps ] );
3868 if ( $text === false ||
$text === null ) {
3872 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3873 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3874 lcfirst( $title->getText() ) )->inContentLanguage();
3875 if ( !$message->exists() ) {
3879 $content = $message->content();
3880 $text = $message->plain();
3888 $finalTitle = $title;
3889 $title = $content->getRedirectTarget();
3894 'finalTitle' => $finalTitle,
3900 * Fetch a file and its title and register a reference to it.
3901 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3902 * @param Title $title
3903 * @param array $options Array of options to RepoGroup::findFile
3904 * @return array ( File or false, Title of file )
3906 public function fetchFileAndTitle( $title, $options = [] ) {
3907 $file = $this->fetchFileNoRegister( $title, $options );
3909 $time = $file ?
$file->getTimestamp() : false;
3910 $sha1 = $file ?
$file->getSha1() : false;
3911 # Register the file as a dependency...
3912 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3913 if ( $file && !$title->equals( $file->getTitle() ) ) {
3914 # Update fetched file title
3915 $title = $file->getTitle();
3916 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3918 return [ $file, $title ];
3922 * Helper function for fetchFileAndTitle.
3924 * Also useful if you need to fetch a file but not use it yet,
3925 * for example to get the file's handler.
3927 * @param Title $title
3928 * @param array $options Array of options to RepoGroup::findFile
3931 protected function fetchFileNoRegister( $title, $options = [] ) {
3932 if ( isset( $options['broken'] ) ) {
3933 $file = false; // broken thumbnail forced by hook
3934 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3935 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3936 } else { // get by (name,timestamp)
3937 $file = MediaWikiServices
::getInstance()->getRepoGroup()->findFile( $title, $options );
3943 * Transclude an interwiki link.
3945 * @param Title $title
3946 * @param string $action Usually one of (raw, render)
3950 public function interwikiTransclude( $title, $action ) {
3951 if ( !$this->svcOptions
->get( 'EnableScaryTranscluding' ) ) {
3952 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3955 $url = $title->getFullURL( [ 'action' => $action ] );
3956 if ( strlen( $url ) > 1024 ) {
3957 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3960 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3962 $fname = __METHOD__
;
3963 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3965 $data = $cache->getWithSetCallback(
3966 $cache->makeGlobalKey(
3967 'interwiki-transclude',
3968 ( $wikiId !== false ) ?
$wikiId : 'external',
3971 $this->svcOptions
->get( 'TranscludeCacheExpiry' ),
3972 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3973 $req = MWHttpRequest
::factory( $url, [], $fname );
3975 $status = $req->execute(); // Status object
3976 if ( !$status->isOK() ) {
3977 $ttl = $cache::TTL_UNCACHEABLE
;
3978 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3979 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3983 'text' => $status->isOK() ?
$req->getContent() : null,
3984 'code' => $req->getStatus()
3988 'checkKeys' => ( $wikiId !== false )
3989 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3991 'pcGroup' => 'interwiki-transclude:5',
3992 'pcTTL' => $cache::TTL_PROC_LONG
3996 if ( is_string( $data['text'] ) ) {
3997 $text = $data['text'];
3998 } elseif ( $data['code'] != 200 ) {
3999 // Though we failed to fetch the content, this status is useless.
4000 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
4001 ->params( $url, $data['code'] )->inContentLanguage()->text();
4003 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4010 * Triple brace replacement -- used for template arguments
4013 * @param array $piece
4014 * @param PPFrame $frame
4018 public function argSubstitution( $piece, $frame ) {
4020 $parts = $piece['parts'];
4021 $nameWithSpaces = $frame->expand( $piece['title'] );
4022 $argName = trim( $nameWithSpaces );
4024 $text = $frame->getArgument( $argName );
4025 if ( $text === false && $parts->getLength() > 0
4026 && ( $this->ot
['html']
4028 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
4031 # No match in frame, use the supplied default
4032 $object = $parts->item( 0 )->getChildren();
4034 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4035 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4036 $this->limitationWarn( 'post-expand-template-argument' );
4039 if ( $text === false && $object === false ) {
4041 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4043 if ( $error !== false ) {
4046 if ( $object !== false ) {
4047 $ret = [ 'object' => $object ];
4049 $ret = [ 'text' => $text ];
4056 * Return the text to be used for a given extension tag.
4057 * This is the ghost of strip().
4059 * @param array $params Associative array of parameters:
4060 * name PPNode for the tag name
4061 * attr PPNode for unparsed text where tag attributes are thought to be
4062 * attributes Optional associative array of parsed attributes
4063 * inner Contents of extension element
4064 * noClose Original text did not have a close tag
4065 * @param PPFrame $frame
4067 * @throws MWException
4070 public function extensionSubstitution( $params, $frame ) {
4071 static $errorStr = '<span class="error">';
4072 static $errorLen = 20;
4074 $name = $frame->expand( $params['name'] );
4075 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4076 // Probably expansion depth or node count exceeded. Just punt the
4081 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
4082 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4087 // We can't safely check if the expansion for $content resulted in an
4088 // error, because the content could happen to be the error string
4090 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
4092 $marker = self
::MARKER_PREFIX
. "-$name-"
4093 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
4095 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
4096 ( $this->ot
['html'] ||
$this->ot
['pre'] );
4097 if ( $isFunctionTag ) {
4098 $markerType = 'none';
4100 $markerType = 'general';
4102 if ( $this->ot
['html'] ||
$isFunctionTag ) {
4103 $name = strtolower( $name );
4104 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
4105 if ( isset( $params['attributes'] ) ) {
4106 $attributes +
= $params['attributes'];
4109 if ( isset( $this->mTagHooks
[$name] ) ) {
4110 $output = call_user_func_array( $this->mTagHooks
[$name],
4111 [ $content, $attributes, $this, $frame ] );
4112 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4113 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4115 // Avoid PHP 7.1 warning from passing $this by reference
4117 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4119 $output = '<span class="error">Invalid tag extension name: ' .
4120 htmlspecialchars( $name ) . '</span>';
4123 if ( is_array( $output ) ) {
4126 $output = $flags[0];
4127 if ( isset( $flags['markerType'] ) ) {
4128 $markerType = $flags['markerType'];
4132 if ( is_null( $attrText ) ) {
4135 if ( isset( $params['attributes'] ) ) {
4136 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4137 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4138 htmlspecialchars( $attrValue ) . '"';
4141 if ( $content === null ) {
4142 $output = "<$name$attrText/>";
4144 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4145 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4149 $output = "<$name$attrText>$content$close";
4153 if ( $markerType === 'none' ) {
4155 } elseif ( $markerType === 'nowiki' ) {
4156 $this->mStripState
->addNoWiki( $marker, $output );
4157 } elseif ( $markerType === 'general' ) {
4158 $this->mStripState
->addGeneral( $marker, $output );
4160 throw new MWException( __METHOD__
. ': invalid marker type' );
4166 * Increment an include size counter
4168 * @param string $type The type of expansion
4169 * @param int $size The size of the text
4170 * @return bool False if this inclusion would take it over the maximum, true otherwise
4172 public function incrementIncludeSize( $type, $size ) {
4173 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4176 $this->mIncludeSizes
[$type] +
= $size;
4182 * Increment the expensive function count
4184 * @return bool False if the limit has been exceeded
4186 public function incrementExpensiveFunctionCount() {
4187 $this->mExpensiveFunctionCount++
;
4188 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4192 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4193 * Fills $this->mDoubleUnderscores, returns the modified text
4195 * @param string $text
4199 public function doDoubleUnderscore( $text ) {
4200 # The position of __TOC__ needs to be recorded
4201 $mw = $this->magicWordFactory
->get( 'toc' );
4202 if ( $mw->match( $text ) ) {
4203 $this->mShowToc
= true;
4204 $this->mForceTocPosition
= true;
4206 # Set a placeholder. At the end we'll fill it in with the TOC.
4207 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4209 # Only keep the first one.
4210 $text = $mw->replace( '', $text );
4213 # Now match and remove the rest of them
4214 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4215 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4217 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4218 $this->mOutput
->mNoGallery
= true;
4220 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4221 $this->mShowToc
= false;
4223 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4224 && $this->mTitle
->getNamespace() == NS_CATEGORY
4226 $this->addTrackingCategory( 'hidden-category-category' );
4228 # (T10068) Allow control over whether robots index a page.
4229 # __INDEX__ always overrides __NOINDEX__, see T16899
4230 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4231 $this->mOutput
->setIndexPolicy( 'noindex' );
4232 $this->addTrackingCategory( 'noindex-category' );
4234 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4235 $this->mOutput
->setIndexPolicy( 'index' );
4236 $this->addTrackingCategory( 'index-category' );
4239 # Cache all double underscores in the database
4240 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4241 $this->mOutput
->setProperty( $key, '' );
4248 * @see ParserOutput::addTrackingCategory()
4249 * @param string $msg Message key
4250 * @return bool Whether the addition was successful
4252 public function addTrackingCategory( $msg ) {
4253 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4257 * This function accomplishes several tasks:
4258 * 1) Auto-number headings if that option is enabled
4259 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4260 * 3) Add a Table of contents on the top for users who have enabled the option
4261 * 4) Auto-anchor headings
4263 * It loops through all headlines, collects the necessary data, then splits up the
4264 * string and re-inserts the newly formatted headlines.
4266 * @param string $text
4267 * @param string $origText Original, untouched wikitext
4268 * @param bool $isMain
4269 * @return mixed|string
4272 public function formatHeadings( $text, $origText, $isMain = true ) {
4273 # Inhibit editsection links if requested in the page
4274 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4275 $maybeShowEditLink = false;
4277 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4280 # Get all headlines for numbering them and adding funky stuff like [edit]
4281 # links - this is for later, but we need the number of headlines right now
4282 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4283 # be trimmed here since whitespace in HTML headings is significant.
4285 $numMatches = preg_match_all(
4286 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4291 # if there are fewer than 4 headlines in the article, do not show TOC
4292 # unless it's been explicitly enabled.
4293 $enoughToc = $this->mShowToc
&&
4294 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4296 # Allow user to stipulate that a page should have a "new section"
4297 # link added via __NEWSECTIONLINK__
4298 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4299 $this->mOutput
->setNewSection( true );
4302 # Allow user to remove the "new section"
4303 # link via __NONEWSECTIONLINK__
4304 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4305 $this->mOutput
->hideNewSection( true );
4308 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4309 # override above conditions and always show TOC above first header
4310 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4311 $this->mShowToc
= true;
4319 # Ugh .. the TOC should have neat indentation levels which can be
4320 # passed to the skin functions. These are determined here
4324 $sublevelCount = [];
4330 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4331 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4332 $oldType = $this->mOutputType
;
4333 $this->setOutputType( self
::OT_WIKI
);
4334 $frame = $this->getPreprocessor()->newFrame();
4335 $root = $this->preprocessToDom( $origText );
4336 $node = $root->getFirstChild();
4341 $headlines = $numMatches !== false ?
$matches[3] : [];
4343 $maxTocLevel = $this->svcOptions
->get( 'MaxTocLevel' );
4344 foreach ( $headlines as $headline ) {
4345 $isTemplate = false;
4347 $sectionIndex = false;
4349 $markerMatches = [];
4350 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4351 $serial = $markerMatches[1];
4352 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4353 $isTemplate = ( $titleText != $baseTitleText );
4354 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4358 $prevlevel = $level;
4360 $level = $matches[1][$headlineCount];
4362 if ( $level > $prevlevel ) {
4363 # Increase TOC level
4365 $sublevelCount[$toclevel] = 0;
4366 if ( $toclevel < $maxTocLevel ) {
4367 $prevtoclevel = $toclevel;
4368 $toc .= Linker
::tocIndent();
4371 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4372 # Decrease TOC level, find level to jump to
4374 for ( $i = $toclevel; $i > 0; $i-- ) {
4375 if ( $levelCount[$i] == $level ) {
4376 # Found last matching level
4379 } elseif ( $levelCount[$i] < $level ) {
4380 # Found first matching level below current level
4388 if ( $toclevel < $maxTocLevel ) {
4389 if ( $prevtoclevel < $maxTocLevel ) {
4390 # Unindent only if the previous toc level was shown :p
4391 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4392 $prevtoclevel = $toclevel;
4394 $toc .= Linker
::tocLineEnd();
4398 # No change in level, end TOC line
4399 if ( $toclevel < $maxTocLevel ) {
4400 $toc .= Linker
::tocLineEnd();
4404 $levelCount[$toclevel] = $level;
4406 # count number of headlines for each level
4407 $sublevelCount[$toclevel]++
;
4409 for ( $i = 1; $i <= $toclevel; $i++
) {
4410 if ( !empty( $sublevelCount[$i] ) ) {
4414 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4419 # The safe header is a version of the header text safe to use for links
4421 # Remove link placeholders by the link text.
4422 # <!--LINK number-->
4424 # link text with suffix
4425 # Do this before unstrip since link text can contain strip markers
4426 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4428 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4429 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4431 # Remove any <style> or <script> tags (T198618)
4432 $safeHeadline = preg_replace(
4433 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4438 # Strip out HTML (first regex removes any tag not allowed)
4440 # * <sup> and <sub> (T10393)
4444 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4445 # * <s> and <strike> (T35715)
4446 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4447 # to allow setting directionality in toc items.
4448 $tocline = preg_replace(
4450 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4451 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4457 # Strip '<span></span>', which is the result from the above if
4458 # <span id="foo"></span> is used to produce an additional anchor
4460 $tocline = str_replace( '<span></span>', '', $tocline );
4462 $tocline = trim( $tocline );
4464 # For the anchor, strip out HTML-y stuff period
4465 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4466 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4468 # Save headline for section edit hint before it's escaped
4469 $headlineHint = $safeHeadline;
4471 # Decode HTML entities
4472 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4474 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4476 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4477 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4478 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4479 if ( $fallbackHeadline === $safeHeadline ) {
4480 # No reason to have both (in fact, we can't)
4481 $fallbackHeadline = false;
4484 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4485 # @todo FIXME: We may be changing them depending on the current locale.
4486 $arrayKey = strtolower( $safeHeadline );
4487 if ( $fallbackHeadline === false ) {
4488 $fallbackArrayKey = false;
4490 $fallbackArrayKey = strtolower( $fallbackHeadline );
4493 # Create the anchor for linking from the TOC to the section
4494 $anchor = $safeHeadline;
4495 $fallbackAnchor = $fallbackHeadline;
4496 if ( isset( $refers[$arrayKey] ) ) {
4497 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4498 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4500 $linkAnchor .= "_$i";
4501 $refers["${arrayKey}_$i"] = true;
4503 $refers[$arrayKey] = true;
4505 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4506 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4507 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4508 $fallbackAnchor .= "_$i";
4509 $refers["${fallbackArrayKey}_$i"] = true;
4511 $refers[$fallbackArrayKey] = true;
4514 # Don't number the heading if it is the only one (looks silly)
4515 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4516 # the two are different if the line contains a link
4517 $headline = Html
::element(
4519 [ 'class' => 'mw-headline-number' ],
4521 ) . ' ' . $headline;
4524 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4525 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4526 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4529 # Add the section to the section tree
4530 # Find the DOM node for this header
4531 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4532 while ( $node && !$noOffset ) {
4533 if ( $node->getName() === 'h' ) {
4534 $bits = $node->splitHeading();
4535 if ( $bits['i'] == $sectionIndex ) {
4539 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4540 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4541 $node = $node->getNextSibling();
4544 'toclevel' => $toclevel,
4547 'number' => $numbering,
4548 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4549 'fromtitle' => $titleText,
4550 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4551 'anchor' => $anchor,
4554 # give headline the correct <h#> tag
4555 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4556 // Output edit section links as markers with styles that can be customized by skins
4557 if ( $isTemplate ) {
4558 # Put a T flag in the section identifier, to indicate to extractSections()
4559 # that sections inside <includeonly> should be counted.
4560 $editsectionPage = $titleText;
4561 $editsectionSection = "T-$sectionIndex";
4562 $editsectionContent = null;
4564 $editsectionPage = $this->mTitle
->getPrefixedText();
4565 $editsectionSection = $sectionIndex;
4566 $editsectionContent = $headlineHint;
4568 // We use a bit of pesudo-xml for editsection markers. The
4569 // language converter is run later on. Using a UNIQ style marker
4570 // leads to the converter screwing up the tokens when it
4571 // converts stuff. And trying to insert strip tags fails too. At
4572 // this point all real inputted tags have already been escaped,
4573 // so we don't have to worry about a user trying to input one of
4574 // these markers directly. We use a page and section attribute
4575 // to stop the language converter from converting these
4576 // important bits of data, but put the headline hint inside a
4577 // content block because the language converter is supposed to
4578 // be able to convert that piece of data.
4579 // Gets replaced with html in ParserOutput::getText
4580 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4581 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4582 if ( $editsectionContent !== null ) {
4583 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4590 $head[$headlineCount] = Linker
::makeHeadline( $level,
4591 $matches['attrib'][$headlineCount], $anchor, $headline,
4592 $editlink, $fallbackAnchor );
4597 $this->setOutputType( $oldType );
4599 # Never ever show TOC if no headers
4600 if ( $numVisible < 1 ) {
4605 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4606 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4608 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4609 $this->mOutput
->setTOCHTML( $toc );
4610 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4614 $this->mOutput
->setSections( $tocraw );
4617 # split up and insert constructed headlines
4618 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4621 // build an array of document sections
4623 foreach ( $blocks as $block ) {
4624 // $head is zero-based, sections aren't.
4625 if ( empty( $head[$i - 1] ) ) {
4626 $sections[$i] = $block;
4628 $sections[$i] = $head[$i - 1] . $block;
4632 * Send a hook, one per section.
4633 * The idea here is to be able to make section-level DIVs, but to do so in a
4634 * lower-impact, more correct way than r50769
4637 * $section : the section number
4638 * &$sectionContent : ref to the content of the section
4639 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4641 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4646 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4647 // append the TOC at the beginning
4648 // Top anchor now in skin
4649 $sections[0] .= $toc . "\n";
4652 $full .= implode( '', $sections );
4654 if ( $this->mForceTocPosition
) {
4655 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4662 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4663 * conversion, substituting signatures, {{subst:}} templates, etc.
4665 * @param string $text The text to transform
4666 * @param Title $title The Title object for the current article
4667 * @param User $user The User object describing the current user
4668 * @param ParserOptions $options Parsing options
4669 * @param bool $clearState Whether to clear the parser state first
4670 * @return string The altered wiki markup
4672 public function preSaveTransform( $text, Title
$title, User
$user,
4673 ParserOptions
$options, $clearState = true
4675 if ( $clearState ) {
4676 $magicScopeVariable = $this->lock();
4678 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4679 $this->setUser( $user );
4681 // Strip U+0000 NULL (T159174)
4682 $text = str_replace( "\000", '', $text );
4684 // We still normalize line endings for backwards-compatibility
4685 // with other code that just calls PST, but this should already
4686 // be handled in TextContent subclasses
4687 $text = TextContent
::normalizeLineEndings( $text );
4689 if ( $options->getPreSaveTransform() ) {
4690 $text = $this->pstPass2( $text, $user );
4692 $text = $this->mStripState
->unstripBoth( $text );
4694 $this->setUser( null ); # Reset
4700 * Pre-save transform helper function
4702 * @param string $text
4707 private function pstPass2( $text, $user ) {
4708 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4709 # $this->contLang here in order to give everyone the same signature and use the default one
4710 # rather than the one selected in each user's preferences. (see also T14815)
4711 $ts = $this->mOptions
->getTimestamp();
4712 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4713 $ts = $timestamp->format( 'YmdHis' );
4714 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4716 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4718 # Variable replacement
4719 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4720 $text = $this->replaceVariables( $text );
4722 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4723 # which may corrupt this parser instance via its wfMessage()->text() call-
4726 if ( strpos( $text, '~~~' ) !== false ) {
4727 $sigText = $this->getUserSig( $user );
4728 $text = strtr( $text, [
4730 '~~~~' => "$sigText $d",
4733 # The main two signature forms used above are time-sensitive
4734 $this->setOutputFlag( 'user-signature', 'User signature detected' );
4737 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4738 $tc = '[' . Title
::legalChars() . ']';
4739 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4741 // [[ns:page (context)|]]
4742 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4743 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4744 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4745 // [[ns:page (context), context|]] (using either single or double-width comma)
4746 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4747 // [[|page]] (reverse pipe trick: add context from page title)
4748 $p2 = "/\[\[\\|($tc+)]]/";
4750 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4751 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4752 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4753 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4755 $t = $this->mTitle
->getText();
4757 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4758 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4759 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4760 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4762 # if there's no context, don't bother duplicating the title
4763 $text = preg_replace( $p2, '[[\\1]]', $text );
4770 * Fetch the user's signature text, if any, and normalize to
4771 * validated, ready-to-insert wikitext.
4772 * If you have pre-fetched the nickname or the fancySig option, you can
4773 * specify them here to save a database query.
4774 * Do not reuse this parser instance after calling getUserSig(),
4775 * as it may have changed.
4777 * @param User &$user
4778 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4779 * @param bool|null $fancySig whether the nicknname is the complete signature
4780 * or null to use default value
4783 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4784 $username = $user->getName();
4786 # If not given, retrieve from the user object.
4787 if ( $nickname === false ) {
4788 $nickname = $user->getOption( 'nickname' );
4791 if ( is_null( $fancySig ) ) {
4792 $fancySig = $user->getBoolOption( 'fancysig' );
4795 $nickname = $nickname == null ?
$username : $nickname;
4797 if ( mb_strlen( $nickname ) > $this->svcOptions
->get( 'MaxSigChars' ) ) {
4798 $nickname = $username;
4799 $this->logger
->debug( __METHOD__
. ": $username has overlong signature." );
4800 } elseif ( $fancySig !== false ) {
4801 # Sig. might contain markup; validate this
4802 if ( $this->validateSig( $nickname ) !== false ) {
4803 # Validated; clean up (if needed) and return it
4804 return $this->cleanSig( $nickname, true );
4806 # Failed to validate; fall back to the default
4807 $nickname = $username;
4808 $this->logger
->debug( __METHOD__
. ": $username has bad XML tags in signature." );
4812 # Make sure nickname doesnt get a sig in a sig
4813 $nickname = self
::cleanSigInSig( $nickname );
4815 # If we're still here, make it a link to the user page
4816 $userText = wfEscapeWikiText( $username );
4817 $nickText = wfEscapeWikiText( $nickname );
4818 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4820 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4821 ->title( $this->getTitle() )->text();
4825 * Check that the user's signature contains no bad XML
4827 * @param string $text
4828 * @return string|bool An expanded string, or false if invalid.
4830 public function validateSig( $text ) {
4831 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4835 * Clean up signature text
4837 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4838 * 2) Substitute all transclusions
4840 * @param string $text
4841 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4842 * @return string Signature text
4844 public function cleanSig( $text, $parsing = false ) {
4847 $magicScopeVariable = $this->lock();
4848 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4851 # Option to disable this feature
4852 if ( !$this->mOptions
->getCleanSignatures() ) {
4856 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4857 # => Move this logic to braceSubstitution()
4858 $substWord = $this->magicWordFactory
->get( 'subst' );
4859 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4860 $substText = '{{' . $substWord->getSynonym( 0 );
4862 $text = preg_replace( $substRegex, $substText, $text );
4863 $text = self
::cleanSigInSig( $text );
4864 $dom = $this->preprocessToDom( $text );
4865 $frame = $this->getPreprocessor()->newFrame();
4866 $text = $frame->expand( $dom );
4869 $text = $this->mStripState
->unstripBoth( $text );
4876 * Strip 3, 4 or 5 tildes out of signatures.
4878 * @param string $text
4879 * @return string Signature text with /~{3,5}/ removed
4881 public static function cleanSigInSig( $text ) {
4882 $text = preg_replace( '/~{3,5}/', '', $text );
4887 * Set up some variables which are usually set up in parse()
4888 * so that an external function can call some class members with confidence
4890 * @param Title|null $title
4891 * @param ParserOptions $options
4892 * @param int $outputType
4893 * @param bool $clearState
4894 * @param int|null $revId
4896 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4897 $outputType, $clearState = true, $revId = null
4899 $this->startParse( $title, $options, $outputType, $clearState );
4900 if ( $revId !== null ) {
4901 $this->mRevisionId
= $revId;
4906 * @param Title|null $title
4907 * @param ParserOptions $options
4908 * @param int $outputType
4909 * @param bool $clearState
4911 private function startParse( Title
$title = null, ParserOptions
$options,
4912 $outputType, $clearState = true
4914 $this->setTitle( $title );
4915 $this->mOptions
= $options;
4916 $this->setOutputType( $outputType );
4917 if ( $clearState ) {
4918 $this->clearState();
4923 * Wrapper for preprocess()
4925 * @param string $text The text to preprocess
4926 * @param ParserOptions $options
4927 * @param Title|null $title Title object or null to use $wgTitle
4930 public function transformMsg( $text, $options, $title = null ) {
4931 static $executing = false;
4933 # Guard against infinite recursion
4944 $text = $this->preprocess( $text, $title, $options );
4951 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4952 * The callback should have the following form:
4953 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4955 * Transform and return $text. Use $parser for any required context, e.g. use
4956 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4958 * Hooks may return extended information by returning an array, of which the
4959 * first numbered element (index 0) must be the return string, and all other
4960 * entries are extracted into local variables within an internal function
4961 * in the Parser class.
4963 * This interface (introduced r61913) appears to be undocumented, but
4964 * 'markerType' is used by some core tag hooks to override which strip
4965 * array their results are placed in. **Use great caution if attempting
4966 * this interface, as it is not documented and injudicious use could smash
4967 * private variables.**
4969 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4970 * @param callable $callback The callback function (and object) to use for the tag
4971 * @throws MWException
4972 * @return callable|null The old value of the mTagHooks array associated with the hook
4974 public function setHook( $tag, callable
$callback ) {
4975 $tag = strtolower( $tag );
4976 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4977 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4979 $oldVal = $this->mTagHooks
[$tag] ??
null;
4980 $this->mTagHooks
[$tag] = $callback;
4981 if ( !in_array( $tag, $this->mStripList
) ) {
4982 $this->mStripList
[] = $tag;
4989 * As setHook(), but letting the contents be parsed.
4991 * Transparent tag hooks are like regular XML-style tag hooks, except they
4992 * operate late in the transformation sequence, on HTML instead of wikitext.
4994 * This is probably obsoleted by things dealing with parser frames?
4995 * The only extension currently using it is geoserver.
4998 * @todo better document or deprecate this
5000 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5001 * @param callable $callback The callback function (and object) to use for the tag
5002 * @throws MWException
5003 * @return callable|null The old value of the mTagHooks array associated with the hook
5005 public function setTransparentTagHook( $tag, callable
$callback ) {
5006 $tag = strtolower( $tag );
5007 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5008 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5010 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
5011 $this->mTransparentTagHooks
[$tag] = $callback;
5017 * Remove all tag hooks
5019 public function clearTagHooks() {
5020 $this->mTagHooks
= [];
5021 $this->mFunctionTagHooks
= [];
5022 $this->mStripList
= $this->mDefaultStripList
;
5026 * Create a function, e.g. {{sum:1|2|3}}
5027 * The callback function should have the form:
5028 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5030 * Or with Parser::SFH_OBJECT_ARGS:
5031 * function myParserFunction( $parser, $frame, $args ) { ... }
5033 * The callback may either return the text result of the function, or an array with the text
5034 * in element 0, and a number of flags in the other elements. The names of the flags are
5035 * specified in the keys. Valid flags are:
5036 * found The text returned is valid, stop processing the template. This
5038 * nowiki Wiki markup in the return value should be escaped
5039 * isHTML The returned text is HTML, armour it against wikitext transformation
5041 * @param string $id The magic word ID
5042 * @param callable $callback The callback function (and object) to use
5043 * @param int $flags A combination of the following flags:
5044 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5046 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5047 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5048 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5049 * the arguments, and to control the way they are expanded.
5051 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5052 * arguments, for instance:
5053 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5055 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5056 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5057 * working if/when this is changed.
5059 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5062 * Please read the documentation in includes/parser/Preprocessor.php for more information
5063 * about the methods available in PPFrame and PPNode.
5065 * @throws MWException
5066 * @return string|callable The old callback function for this name, if any
5068 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
5069 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
5070 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
5072 # Add to function cache
5073 $mw = $this->magicWordFactory
->get( $id );
5075 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
5078 $synonyms = $mw->getSynonyms();
5079 $sensitive = intval( $mw->isCaseSensitive() );
5081 foreach ( $synonyms as $syn ) {
5083 if ( !$sensitive ) {
5084 $syn = $this->contLang
->lc( $syn );
5087 if ( !( $flags & self
::SFH_NO_HASH
) ) {
5090 # Remove trailing colon
5091 if ( substr( $syn, -1, 1 ) === ':' ) {
5092 $syn = substr( $syn, 0, -1 );
5094 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
5100 * Get all registered function hook identifiers
5104 public function getFunctionHooks() {
5105 $this->firstCallInit();
5106 return array_keys( $this->mFunctionHooks
);
5110 * Create a tag function, e.g. "<test>some stuff</test>".
5111 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5112 * Unlike parser functions, their content is not preprocessed.
5113 * @param string $tag
5114 * @param callable $callback
5116 * @throws MWException
5119 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5120 $tag = strtolower( $tag );
5121 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5122 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5124 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5125 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5127 if ( !in_array( $tag, $this->mStripList
) ) {
5128 $this->mStripList
[] = $tag;
5135 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5136 * Placeholders created in Linker::link()
5138 * @param string &$text
5139 * @param int $options
5141 public function replaceLinkHolders( &$text, $options = 0 ) {
5142 $this->mLinkHolders
->replace( $text );
5146 * Replace "<!--LINK-->" link placeholders with plain text of links
5147 * (not HTML-formatted).
5149 * @param string $text
5152 public function replaceLinkHoldersText( $text ) {
5153 return $this->mLinkHolders
->replaceText( $text );
5157 * Renders an image gallery from a text with one line per image.
5158 * text labels may be given by using |-style alternative text. E.g.
5159 * Image:one.jpg|The number "1"
5160 * Image:tree.jpg|A tree
5161 * given as text will return the HTML of a gallery with two images,
5162 * labeled 'The number "1"' and
5165 * @param string $text
5166 * @param array $params
5167 * @return string HTML
5169 public function renderImageGallery( $text, $params ) {
5171 if ( isset( $params['mode'] ) ) {
5172 $mode = $params['mode'];
5176 $ig = ImageGalleryBase
::factory( $mode );
5177 } catch ( Exception
$e ) {
5178 // If invalid type set, fallback to default.
5179 $ig = ImageGalleryBase
::factory( false );
5182 $ig->setContextTitle( $this->mTitle
);
5183 $ig->setShowBytes( false );
5184 $ig->setShowDimensions( false );
5185 $ig->setShowFilename( false );
5186 $ig->setParser( $this );
5187 $ig->setHideBadImages();
5188 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5190 if ( isset( $params['showfilename'] ) ) {
5191 $ig->setShowFilename( true );
5193 $ig->setShowFilename( false );
5195 if ( isset( $params['caption'] ) ) {
5196 // NOTE: We aren't passing a frame here or below. Frame info
5197 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5198 // See T107332#4030581
5199 $caption = $this->recursiveTagParse( $params['caption'] );
5200 $ig->setCaptionHtml( $caption );
5202 if ( isset( $params['perrow'] ) ) {
5203 $ig->setPerRow( $params['perrow'] );
5205 if ( isset( $params['widths'] ) ) {
5206 $ig->setWidths( $params['widths'] );
5208 if ( isset( $params['heights'] ) ) {
5209 $ig->setHeights( $params['heights'] );
5211 $ig->setAdditionalOptions( $params );
5213 // Avoid PHP 7.1 warning from passing $this by reference
5215 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5217 $lines = StringUtils
::explode( "\n", $text );
5218 foreach ( $lines as $line ) {
5219 # match lines like these:
5220 # Image:someimage.jpg|This is some image
5222 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5224 if ( count( $matches ) == 0 ) {
5228 if ( strpos( $matches[0], '%' ) !== false ) {
5229 $matches[1] = rawurldecode( $matches[1] );
5231 $title = Title
::newFromText( $matches[1], NS_FILE
);
5232 if ( is_null( $title ) ) {
5233 # Bogus title. Ignore these so we don't bomb out later.
5237 # We need to get what handler the file uses, to figure out parameters.
5238 # Note, a hook can overide the file name, and chose an entirely different
5239 # file (which potentially could be of a different type and have different handler).
5242 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5243 [ $this, $title, &$options, &$descQuery ] );
5244 # Don't register it now, as TraditionalImageGallery does that later.
5245 $file = $this->fetchFileNoRegister( $title, $options );
5246 $handler = $file ?
$file->getHandler() : false;
5249 'img_alt' => 'gallery-internal-alt',
5250 'img_link' => 'gallery-internal-link',
5253 $paramMap +
= $handler->getParamMap();
5254 // We don't want people to specify per-image widths.
5255 // Additionally the width parameter would need special casing anyhow.
5256 unset( $paramMap['img_width'] );
5259 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5264 $handlerOptions = [];
5265 if ( isset( $matches[3] ) ) {
5266 // look for an |alt= definition while trying not to break existing
5267 // captions with multiple pipes (|) in it, until a more sensible grammar
5268 // is defined for images in galleries
5270 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5271 // splitting on '|' is a bit odd, and different from makeImage.
5272 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5273 // Protect LanguageConverter markup
5274 $parameterMatches = StringUtils
::delimiterExplode(
5275 '-{', '}-', '|', $matches[3], true /* nested */
5278 foreach ( $parameterMatches as $parameterMatch ) {
5279 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5281 $paramName = $paramMap[$magicName];
5283 switch ( $paramName ) {
5284 case 'gallery-internal-alt':
5285 $alt = $this->stripAltText( $match, false );
5287 case 'gallery-internal-link':
5288 $linkValue = $this->stripAltText( $match, false );
5289 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5290 // Result of LanguageConverter::markNoConversion
5291 // invoked on an external link.
5292 $linkValue = substr( $linkValue, 4, -2 );
5294 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5295 if ( $type === 'link-url' ) {
5297 $this->mOutput
->addExternalLink( $target );
5298 } elseif ( $type === 'link-title' ) {
5299 $link = $target->getLinkURL();
5300 $this->mOutput
->addLink( $target );
5304 // Must be a handler specific parameter.
5305 if ( $handler->validateParam( $paramName, $match ) ) {
5306 $handlerOptions[$paramName] = $match;
5308 // Guess not, consider it as caption.
5309 $this->logger
->debug(
5310 "$parameterMatch failed parameter validation" );
5311 $label = $parameterMatch;
5317 $label = $parameterMatch;
5322 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5324 $html = $ig->toHTML();
5325 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5330 * @param MediaHandler $handler
5333 public function getImageParams( $handler ) {
5335 $handlerClass = get_class( $handler );
5339 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5340 # Initialise static lists
5341 static $internalParamNames = [
5342 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5343 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5344 'bottom', 'text-bottom' ],
5345 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5346 'upright', 'border', 'link', 'alt', 'class' ],
5348 static $internalParamMap;
5349 if ( !$internalParamMap ) {
5350 $internalParamMap = [];
5351 foreach ( $internalParamNames as $type => $names ) {
5352 foreach ( $names as $name ) {
5353 // For grep: img_left, img_right, img_center, img_none,
5354 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5355 // img_bottom, img_text_bottom,
5356 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5357 // img_border, img_link, img_alt, img_class
5358 $magicName = str_replace( '-', '_', "img_$name" );
5359 $internalParamMap[$magicName] = [ $type, $name ];
5364 # Add handler params
5365 $paramMap = $internalParamMap;
5367 $handlerParamMap = $handler->getParamMap();
5368 foreach ( $handlerParamMap as $magic => $paramName ) {
5369 $paramMap[$magic] = [ 'handler', $paramName ];
5372 $this->mImageParams
[$handlerClass] = $paramMap;
5373 $this->mImageParamsMagicArray
[$handlerClass] =
5374 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5376 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5380 * Parse image options text and use it to make an image
5382 * @param Title $title
5383 * @param string $options
5384 * @param LinkHolderArray|bool $holders
5385 * @return string HTML
5387 public function makeImage( $title, $options, $holders = false ) {
5388 # Check if the options text is of the form "options|alt text"
5390 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5391 # * left no resizing, just left align. label is used for alt= only
5392 # * right same, but right aligned
5393 # * none same, but not aligned
5394 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5395 # * center center the image
5396 # * frame Keep original image size, no magnify-button.
5397 # * framed Same as "frame"
5398 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5399 # * upright reduce width for upright images, rounded to full __0 px
5400 # * border draw a 1px border around the image
5401 # * alt Text for HTML alt attribute (defaults to empty)
5402 # * class Set a class for img node
5403 # * link Set the target of the image link. Can be external, interwiki, or local
5404 # vertical-align values (no % or length right now):
5414 # Protect LanguageConverter markup when splitting into parts
5415 $parts = StringUtils
::delimiterExplode(
5416 '-{', '}-', '|', $options, true /* allow nesting */
5419 # Give extensions a chance to select the file revision for us
5422 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5423 [ $this, $title, &$options, &$descQuery ] );
5424 # Fetch and register the file (file title may be different via hooks)
5425 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5428 $handler = $file ?
$file->getHandler() : false;
5430 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5433 $this->addTrackingCategory( 'broken-file-category' );
5436 # Process the input parameters
5438 $params = [ 'frame' => [], 'handler' => [],
5439 'horizAlign' => [], 'vertAlign' => [] ];
5440 $seenformat = false;
5441 foreach ( $parts as $part ) {
5442 $part = trim( $part );
5443 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5445 if ( isset( $paramMap[$magicName] ) ) {
5446 list( $type, $paramName ) = $paramMap[$magicName];
5448 # Special case; width and height come in one variable together
5449 if ( $type === 'handler' && $paramName === 'width' ) {
5450 $parsedWidthParam = self
::parseWidthParam( $value );
5451 if ( isset( $parsedWidthParam['width'] ) ) {
5452 $width = $parsedWidthParam['width'];
5453 if ( $handler->validateParam( 'width', $width ) ) {
5454 $params[$type]['width'] = $width;
5458 if ( isset( $parsedWidthParam['height'] ) ) {
5459 $height = $parsedWidthParam['height'];
5460 if ( $handler->validateParam( 'height', $height ) ) {
5461 $params[$type]['height'] = $height;
5465 # else no validation -- T15436
5467 if ( $type === 'handler' ) {
5468 # Validate handler parameter
5469 $validated = $handler->validateParam( $paramName, $value );
5471 # Validate internal parameters
5472 switch ( $paramName ) {
5476 # @todo FIXME: Possibly check validity here for
5477 # manualthumb? downstream behavior seems odd with
5478 # missing manual thumbs.
5480 $value = $this->stripAltText( $value, $holders );
5483 list( $paramName, $value ) =
5484 $this->parseLinkParameter(
5485 $this->stripAltText( $value, $holders )
5489 if ( $paramName === 'no-link' ) {
5492 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5493 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5500 // use first appearing option, discard others.
5501 $validated = !$seenformat;
5505 # Most other things appear to be empty or numeric...
5506 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5511 $params[$type][$paramName] = $value;
5515 if ( !$validated ) {
5520 # Process alignment parameters
5521 if ( $params['horizAlign'] ) {
5522 $params['frame']['align'] = key( $params['horizAlign'] );
5524 if ( $params['vertAlign'] ) {
5525 $params['frame']['valign'] = key( $params['vertAlign'] );
5528 $params['frame']['caption'] = $caption;
5530 # Will the image be presented in a frame, with the caption below?
5531 $imageIsFramed = isset( $params['frame']['frame'] )
5532 ||
isset( $params['frame']['framed'] )
5533 ||
isset( $params['frame']['thumbnail'] )
5534 ||
isset( $params['frame']['manualthumb'] );
5536 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5537 # came to also set the caption, ordinary text after the image -- which
5538 # makes no sense, because that just repeats the text multiple times in
5539 # screen readers. It *also* came to set the title attribute.
5540 # Now that we have an alt attribute, we should not set the alt text to
5541 # equal the caption: that's worse than useless, it just repeats the
5542 # text. This is the framed/thumbnail case. If there's no caption, we
5543 # use the unnamed parameter for alt text as well, just for the time be-
5544 # ing, if the unnamed param is set and the alt param is not.
5545 # For the future, we need to figure out if we want to tweak this more,
5546 # e.g., introducing a title= parameter for the title; ignoring the un-
5547 # named parameter entirely for images without a caption; adding an ex-
5548 # plicit caption= parameter and preserving the old magic unnamed para-
5550 if ( $imageIsFramed ) { # Framed image
5551 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5552 # No caption or alt text, add the filename as the alt text so
5553 # that screen readers at least get some description of the image
5554 $params['frame']['alt'] = $title->getText();
5556 # Do not set $params['frame']['title'] because tooltips don't make sense
5558 } else { # Inline image
5559 if ( !isset( $params['frame']['alt'] ) ) {
5560 # No alt text, use the "caption" for the alt text
5561 if ( $caption !== '' ) {
5562 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5564 # No caption, fall back to using the filename for the
5566 $params['frame']['alt'] = $title->getText();
5569 # Use the "caption" for the tooltip text
5570 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5572 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5574 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5576 # Linker does the rest
5577 $time = $options['time'] ??
false;
5578 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5579 $time, $descQuery, $this->mOptions
->getThumbSize() );
5581 # Give the handler a chance to modify the parser object
5583 $handler->parserTransformHook( $this, $file );
5590 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5592 * Adds an entry to appropriate link tables.
5595 * @param string $value
5596 * @return array of `[ type, target ]`, where:
5597 * - `type` is one of:
5598 * - `null`: Given value is not a valid link target, use default
5599 * - `'no-link'`: Given value is empty, do not generate a link
5600 * - `'link-url'`: Given value is a valid external link
5601 * - `'link-title'`: Given value is a valid internal link
5603 * - When `type` is `null` or `'no-link'`: `false`
5604 * - When `type` is `'link-url'`: URL string corresponding to given value
5605 * - When `type` is `'link-title'`: Title object corresponding to given value
5607 public function parseLinkParameter( $value ) {
5608 $chars = self
::EXT_LINK_URL_CLASS
;
5609 $addr = self
::EXT_LINK_ADDR
;
5610 $prots = $this->mUrlProtocols
;
5613 if ( $value === '' ) {
5615 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5616 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5617 $this->mOutput
->addExternalLink( $value );
5622 $linkTitle = Title
::newFromText( $value );
5624 $this->mOutput
->addLink( $linkTitle );
5625 $type = 'link-title';
5626 $target = $linkTitle;
5629 return [ $type, $target ];
5633 * @param string $caption
5634 * @param LinkHolderArray|bool $holders
5635 * @return mixed|string
5637 protected function stripAltText( $caption, $holders ) {
5638 # Strip bad stuff out of the title (tooltip). We can't just use
5639 # replaceLinkHoldersText() here, because if this function is called
5640 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5642 $tooltip = $holders->replaceText( $caption );
5644 $tooltip = $this->replaceLinkHoldersText( $caption );
5647 # make sure there are no placeholders in thumbnail attributes
5648 # that are later expanded to html- so expand them now and
5650 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5651 # Compatibility hack! In HTML certain entity references not terminated
5652 # by a semicolon are decoded (but not if we're in an attribute; that's
5653 # how link URLs get away without properly escaping & in queries).
5654 # But wikitext has always required semicolon-termination of entities,
5655 # so encode & where needed to avoid decode of semicolon-less entities.
5657 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5658 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5659 $tooltip = preg_replace( "/
5660 & # 1. entity prefix
5661 (?= # 2. followed by:
5662 (?: # a. one of the legacy semicolon-less named entities
5663 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5664 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5665 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5666 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5667 U(?:acute|circ|grave|uml)|Yacute|
5668 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5669 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5670 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5671 frac(?:1(?:2|4)|34)|
5672 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5673 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5674 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5675 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5676 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5677 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5678 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5679 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5680 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5682 (?:[^;]|$)) # b. and not followed by a semicolon
5683 # S = study, for efficiency
5684 /Sx", '&', $tooltip );
5685 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5691 * Set a flag in the output object indicating that the content is dynamic and
5692 * shouldn't be cached.
5693 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5695 public function disableCache() {
5696 $this->logger
->debug( "Parser output marked as uncacheable." );
5697 if ( !$this->mOutput
) {
5698 throw new MWException( __METHOD__
.
5699 " can only be called when actually parsing something" );
5701 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5705 * Callback from the Sanitizer for expanding items found in HTML attribute
5706 * values, so they can be safely tested and escaped.
5708 * @param string &$text
5709 * @param bool|PPFrame $frame
5712 public function attributeStripCallback( &$text, $frame = false ) {
5713 $text = $this->replaceVariables( $text, $frame );
5714 $text = $this->mStripState
->unstripBoth( $text );
5723 public function getTags() {
5724 $this->firstCallInit();
5726 array_keys( $this->mTransparentTagHooks
),
5727 array_keys( $this->mTagHooks
),
5728 array_keys( $this->mFunctionTagHooks
)
5736 public function getFunctionSynonyms() {
5737 $this->firstCallInit();
5738 return $this->mFunctionSynonyms
;
5745 public function getUrlProtocols() {
5746 return $this->mUrlProtocols
;
5750 * Replace transparent tags in $text with the values given by the callbacks.
5752 * Transparent tag hooks are like regular XML-style tag hooks, except they
5753 * operate late in the transformation sequence, on HTML instead of wikitext.
5755 * @param string $text
5759 public function replaceTransparentTags( $text ) {
5761 $elements = array_keys( $this->mTransparentTagHooks
);
5762 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5765 foreach ( $matches as $marker => $data ) {
5766 list( $element, $content, $params, $tag ) = $data;
5767 $tagName = strtolower( $element );
5768 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5769 $output = call_user_func_array(
5770 $this->mTransparentTagHooks
[$tagName],
5771 [ $content, $params, $this ]
5776 $replacements[$marker] = $output;
5778 return strtr( $text, $replacements );
5782 * Break wikitext input into sections, and either pull or replace
5783 * some particular section's text.
5785 * External callers should use the getSection and replaceSection methods.
5787 * @param string $text Page wikitext
5788 * @param string|int $sectionId A section identifier string of the form:
5789 * "<flag1> - <flag2> - ... - <section number>"
5791 * Currently the only recognised flag is "T", which means the target section number
5792 * was derived during a template inclusion parse, in other words this is a template
5793 * section edit link. If no flags are given, it was an ordinary section edit link.
5794 * This flag is required to avoid a section numbering mismatch when a section is
5795 * enclosed by "<includeonly>" (T8563).
5797 * The section number 0 pulls the text before the first heading; other numbers will
5798 * pull the given section along with its lower-level subsections. If the section is
5799 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5801 * Section 0 is always considered to exist, even if it only contains the empty
5802 * string. If $text is the empty string and section 0 is replaced, $newText is
5805 * @param string $mode One of "get" or "replace"
5806 * @param string $newText Replacement text for section data.
5807 * @return string For "get", the extracted section text.
5808 * for "replace", the whole page with the section replaced.
5810 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5811 global $wgTitle; # not generally used but removes an ugly failure mode
5813 $magicScopeVariable = $this->lock();
5814 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5816 $frame = $this->getPreprocessor()->newFrame();
5818 # Process section extraction flags
5820 $sectionParts = explode( '-', $sectionId );
5821 $sectionIndex = array_pop( $sectionParts );
5822 foreach ( $sectionParts as $part ) {
5823 if ( $part === 'T' ) {
5824 $flags |
= self
::PTD_FOR_INCLUSION
;
5828 # Check for empty input
5829 if ( strval( $text ) === '' ) {
5830 # Only sections 0 and T-0 exist in an empty document
5831 if ( $sectionIndex == 0 ) {
5832 if ( $mode === 'get' ) {
5838 if ( $mode === 'get' ) {
5846 # Preprocess the text
5847 $root = $this->preprocessToDom( $text, $flags );
5849 # <h> nodes indicate section breaks
5850 # They can only occur at the top level, so we can find them by iterating the root's children
5851 $node = $root->getFirstChild();
5853 # Find the target section
5854 if ( $sectionIndex == 0 ) {
5855 # Section zero doesn't nest, level=big
5856 $targetLevel = 1000;
5859 if ( $node->getName() === 'h' ) {
5860 $bits = $node->splitHeading();
5861 if ( $bits['i'] == $sectionIndex ) {
5862 $targetLevel = $bits['level'];
5866 if ( $mode === 'replace' ) {
5867 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5869 $node = $node->getNextSibling();
5875 if ( $mode === 'get' ) {
5882 # Find the end of the section, including nested sections
5884 if ( $node->getName() === 'h' ) {
5885 $bits = $node->splitHeading();
5886 $curLevel = $bits['level'];
5887 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5891 if ( $mode === 'get' ) {
5892 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5894 $node = $node->getNextSibling();
5897 # Write out the remainder (in replace mode only)
5898 if ( $mode === 'replace' ) {
5899 # Output the replacement text
5900 # Add two newlines on -- trailing whitespace in $newText is conventionally
5901 # stripped by the editor, so we need both newlines to restore the paragraph gap
5902 # Only add trailing whitespace if there is newText
5903 if ( $newText != "" ) {
5904 $outText .= $newText . "\n\n";
5908 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5909 $node = $node->getNextSibling();
5913 if ( is_string( $outText ) ) {
5914 # Re-insert stripped tags
5915 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5922 * This function returns the text of a section, specified by a number ($section).
5923 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5924 * the first section before any such heading (section 0).
5926 * If a section contains subsections, these are also returned.
5928 * @param string $text Text to look in
5929 * @param string|int $sectionId Section identifier as a number or string
5930 * (e.g. 0, 1 or 'T-1').
5931 * @param string $defaultText Default to return if section is not found
5933 * @return string Text of the requested section
5935 public function getSection( $text, $sectionId, $defaultText = '' ) {
5936 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5940 * This function returns $oldtext after the content of the section
5941 * specified by $section has been replaced with $text. If the target
5942 * section does not exist, $oldtext is returned unchanged.
5944 * @param string $oldText Former text of the article
5945 * @param string|int $sectionId Section identifier as a number or string
5946 * (e.g. 0, 1 or 'T-1').
5947 * @param string $newText Replacing text
5949 * @return string Modified text
5951 public function replaceSection( $oldText, $sectionId, $newText ) {
5952 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5956 * Get the ID of the revision we are parsing
5958 * The return value will be either:
5959 * - a) Positive, indicating a specific revision ID (current or old)
5960 * - b) Zero, meaning the revision ID is specified by getCurrentRevisionCallback()
5961 * - c) Null, meaning the parse is for preview mode and there is no revision
5965 public function getRevisionId() {
5966 return $this->mRevisionId
;
5970 * Get the revision object for $this->mRevisionId
5972 * @return Revision|null Either a Revision object or null
5973 * @since 1.23 (public since 1.23)
5975 public function getRevisionObject() {
5976 if ( $this->mRevisionObject
) {
5977 return $this->mRevisionObject
;
5980 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5981 // This is useful when parsing a revision that has not yet been saved.
5982 // However, if we get back a saved revision even though we are in
5983 // preview mode, we'll have to ignore it, see below.
5984 // NOTE: This callback may be used to inject an OLD revision that was
5985 // already loaded, so "current" is a bit of a misnomer. We can't just
5986 // skip it if mRevisionId is set.
5987 $rev = call_user_func(
5988 $this->mOptions
->getCurrentRevisionCallback(),
5993 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5994 // We are in preview mode (mRevisionId is null), and the current revision callback
5995 // returned an existing revision. Ignore it and return null, it's probably the page's
5996 // current revision, which is not what we want here. Note that we do want to call the
5997 // callback to allow the unsaved revision to be injected here, e.g. for
5998 // self-transclusion previews.
6002 // If the parse is for a new revision, then the callback should have
6003 // already been set to force the object and should match mRevisionId.
6004 // If not, try to fetch by mRevisionId for sanity.
6005 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
6006 $rev = Revision
::newFromId( $this->mRevisionId
);
6009 $this->mRevisionObject
= $rev;
6011 return $this->mRevisionObject
;
6015 * Get the timestamp associated with the current revision, adjusted for
6016 * the default server-local timestamp
6017 * @return string TS_MW timestamp
6019 public function getRevisionTimestamp() {
6020 if ( $this->mRevisionTimestamp
!== null ) {
6021 return $this->mRevisionTimestamp
;
6024 # Use specified revision timestamp, falling back to the current timestamp
6025 $revObject = $this->getRevisionObject();
6026 $timestamp = $revObject ?
$revObject->getTimestamp() : $this->mOptions
->getTimestamp();
6027 $this->mOutput
->setRevisionTimestampUsed( $timestamp ); // unadjusted time zone
6029 # The cryptic '' timezone parameter tells to use the site-default
6030 # timezone offset instead of the user settings.
6031 # Since this value will be saved into the parser cache, served
6032 # to other users, and potentially even used inside links and such,
6033 # it needs to be consistent for all visitors.
6034 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
6036 return $this->mRevisionTimestamp
;
6040 * Get the name of the user that edited the last revision
6042 * @return string User name
6044 public function getRevisionUser() {
6045 if ( is_null( $this->mRevisionUser
) ) {
6046 $revObject = $this->getRevisionObject();
6048 # if this template is subst: the revision id will be blank,
6049 # so just use the current user's name
6051 $this->mRevisionUser
= $revObject->getUserText();
6052 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
6053 $this->mRevisionUser
= $this->getUser()->getName();
6056 return $this->mRevisionUser
;
6060 * Get the size of the revision
6062 * @return int|null Revision size
6064 public function getRevisionSize() {
6065 if ( is_null( $this->mRevisionSize
) ) {
6066 $revObject = $this->getRevisionObject();
6068 # if this variable is subst: the revision id will be blank,
6069 # so just use the parser input size, because the own substituation
6070 # will change the size.
6072 $this->mRevisionSize
= $revObject->getSize();
6074 $this->mRevisionSize
= $this->mInputSize
;
6077 return $this->mRevisionSize
;
6081 * Mutator for $mDefaultSort
6083 * @param string $sort New value
6085 public function setDefaultSort( $sort ) {
6086 $this->mDefaultSort
= $sort;
6087 $this->mOutput
->setProperty( 'defaultsort', $sort );
6091 * Accessor for $mDefaultSort
6092 * Will use the empty string if none is set.
6094 * This value is treated as a prefix, so the
6095 * empty string is equivalent to sorting by
6100 public function getDefaultSort() {
6101 if ( $this->mDefaultSort
!== false ) {
6102 return $this->mDefaultSort
;
6109 * Accessor for $mDefaultSort
6110 * Unlike getDefaultSort(), will return false if none is set
6112 * @return string|bool
6114 public function getCustomDefaultSort() {
6115 return $this->mDefaultSort
;
6118 private static function getSectionNameFromStrippedText( $text ) {
6119 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
6120 $text = Sanitizer
::decodeCharReferences( $text );
6121 $text = self
::normalizeSectionName( $text );
6125 private static function makeAnchor( $sectionName ) {
6126 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6129 private function makeLegacyAnchor( $sectionName ) {
6130 $fragmentMode = $this->svcOptions
->get( 'FragmentMode' );
6131 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6132 // ForAttribute() and ForLink() are the same for legacy encoding
6133 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6135 $id = Sanitizer
::escapeIdForLink( $sectionName );
6142 * Try to guess the section anchor name based on a wikitext fragment
6143 * presumably extracted from a heading, for example "Header" from
6146 * @param string $text
6147 * @return string Anchor (starting with '#')
6149 public function guessSectionNameFromWikiText( $text ) {
6150 # Strip out wikitext links(they break the anchor)
6151 $text = $this->stripSectionName( $text );
6152 $sectionName = self
::getSectionNameFromStrippedText( $text );
6153 return self
::makeAnchor( $sectionName );
6157 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6158 * instead, if possible. For use in redirects, since various versions
6159 * of Microsoft browsers interpret Location: headers as something other
6160 * than UTF-8, resulting in breakage.
6162 * @param string $text The section name
6163 * @return string Anchor (starting with '#')
6165 public function guessLegacySectionNameFromWikiText( $text ) {
6166 # Strip out wikitext links(they break the anchor)
6167 $text = $this->stripSectionName( $text );
6168 $sectionName = self
::getSectionNameFromStrippedText( $text );
6169 return $this->makeLegacyAnchor( $sectionName );
6173 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6174 * @param string $text Section name (plain text)
6175 * @return string Anchor (starting with '#')
6177 public static function guessSectionNameFromStrippedText( $text ) {
6178 $sectionName = self
::getSectionNameFromStrippedText( $text );
6179 return self
::makeAnchor( $sectionName );
6183 * Apply the same normalization as code making links to this section would
6185 * @param string $text
6188 private static function normalizeSectionName( $text ) {
6189 # T90902: ensure the same normalization is applied for IDs as to links
6190 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6193 $parts = $titleParser->splitTitleString( "#$text" );
6194 } catch ( MalformedTitleException
$ex ) {
6197 return $parts['fragment'];
6201 * Strips a text string of wikitext for use in a section anchor
6203 * Accepts a text string and then removes all wikitext from the
6204 * string and leaves only the resultant text (i.e. the result of
6205 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6206 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6207 * to create valid section anchors by mimicing the output of the
6208 * parser when headings are parsed.
6210 * @param string $text Text string to be stripped of wikitext
6211 * for use in a Section anchor
6212 * @return string Filtered text string
6214 public function stripSectionName( $text ) {
6215 # Strip internal link markup
6216 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6217 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6219 # Strip external link markup
6220 # @todo FIXME: Not tolerant to blank link text
6221 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6222 # on how many empty links there are on the page - need to figure that out.
6223 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6225 # Parse wikitext quotes (italics & bold)
6226 $text = $this->doQuotes( $text );
6229 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6234 * strip/replaceVariables/unstrip for preprocessor regression testing
6236 * @param string $text
6237 * @param Title $title
6238 * @param ParserOptions $options
6239 * @param int $outputType
6243 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6244 $outputType = self
::OT_HTML
6246 $magicScopeVariable = $this->lock();
6247 $this->startParse( $title, $options, $outputType, true );
6249 $text = $this->replaceVariables( $text );
6250 $text = $this->mStripState
->unstripBoth( $text );
6251 $text = Sanitizer
::removeHTMLtags( $text );
6256 * @param string $text
6257 * @param Title $title
6258 * @param ParserOptions $options
6261 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6262 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6266 * @param string $text
6267 * @param Title $title
6268 * @param ParserOptions $options
6271 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6272 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6276 * Call a callback function on all regions of the given text that are not
6277 * inside strip markers, and replace those regions with the return value
6278 * of the callback. For example, with input:
6282 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6283 * two strings will be replaced with the value returned by the callback in
6287 * @param callable $callback
6291 public function markerSkipCallback( $s, $callback ) {
6294 while ( $i < strlen( $s ) ) {
6295 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6296 if ( $markerStart === false ) {
6297 $out .= call_user_func( $callback, substr( $s, $i ) );
6300 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6301 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6302 if ( $markerEnd === false ) {
6303 $out .= substr( $s, $markerStart );
6306 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6307 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6316 * Remove any strip markers found in the given text.
6318 * @param string $text
6321 public function killMarkers( $text ) {
6322 return $this->mStripState
->killMarkers( $text );
6326 * Save the parser state required to convert the given half-parsed text to
6327 * HTML. "Half-parsed" in this context means the output of
6328 * recursiveTagParse() or internalParse(). This output has strip markers
6329 * from replaceVariables (extensionSubstitution() etc.), and link
6330 * placeholders from replaceLinkHolders().
6332 * Returns an array which can be serialized and stored persistently. This
6333 * array can later be loaded into another parser instance with
6334 * unserializeHalfParsedText(). The text can then be safely incorporated into
6335 * the return value of a parser hook.
6337 * @deprecated since 1.31
6338 * @param string $text
6342 public function serializeHalfParsedText( $text ) {
6343 wfDeprecated( __METHOD__
, '1.31' );
6346 'version' => self
::HALF_PARSED_VERSION
,
6347 'stripState' => $this->mStripState
->getSubState( $text ),
6348 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6354 * Load the parser state given in the $data array, which is assumed to
6355 * have been generated by serializeHalfParsedText(). The text contents is
6356 * extracted from the array, and its markers are transformed into markers
6357 * appropriate for the current Parser instance. This transformed text is
6358 * returned, and can be safely included in the return value of a parser
6361 * If the $data array has been stored persistently, the caller should first
6362 * check whether it is still valid, by calling isValidHalfParsedText().
6364 * @deprecated since 1.31
6365 * @param array $data Serialized data
6366 * @throws MWException
6369 public function unserializeHalfParsedText( $data ) {
6370 wfDeprecated( __METHOD__
, '1.31' );
6371 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6372 throw new MWException( __METHOD__
. ': invalid version' );
6375 # First, extract the strip state.
6376 $texts = [ $data['text'] ];
6377 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6379 # Now renumber links
6380 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6382 # Should be good to go.
6387 * Returns true if the given array, presumed to be generated by
6388 * serializeHalfParsedText(), is compatible with the current version of the
6391 * @deprecated since 1.31
6392 * @param array $data
6396 public function isValidHalfParsedText( $data ) {
6397 wfDeprecated( __METHOD__
, '1.31' );
6398 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6402 * Parsed a width param of imagelink like 300px or 200x300px
6404 * @param string $value
6405 * @param bool $parseHeight
6410 public static function parseWidthParam( $value, $parseHeight = true ) {
6411 $parsedWidthParam = [];
6412 if ( $value === '' ) {
6413 return $parsedWidthParam;
6416 # (T15500) In both cases (width/height and width only),
6417 # permit trailing "px" for backward compatibility.
6418 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6419 $width = intval( $m[1] );
6420 $height = intval( $m[2] );
6421 $parsedWidthParam['width'] = $width;
6422 $parsedWidthParam['height'] = $height;
6423 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6424 $width = intval( $value );
6425 $parsedWidthParam['width'] = $width;
6427 return $parsedWidthParam;
6431 * Lock the current instance of the parser.
6433 * This is meant to stop someone from calling the parser
6434 * recursively and messing up all the strip state.
6436 * @throws MWException If parser is in a parse
6437 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6439 protected function lock() {
6440 if ( $this->mInParse
) {
6441 throw new MWException( "Parser state cleared while parsing. "
6442 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6445 // Save the backtrace when locking, so that if some code tries locking again,
6446 // we can print the lock owner's backtrace for easier debugging
6448 $this->mInParse
= $e->getTraceAsString();
6450 $recursiveCheck = new ScopedCallback( function () {
6451 $this->mInParse
= false;
6454 return $recursiveCheck;
6458 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6460 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6461 * or if there is more than one <p/> tag in the input HTML.
6463 * @param string $html
6467 public static function stripOuterParagraph( $html ) {
6469 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6477 * Return this parser if it is not doing anything, otherwise
6478 * get a fresh parser. You can use this method by doing
6479 * $newParser = $oldParser->getFreshParser(), or more simply
6480 * $oldParser->getFreshParser()->parse( ... );
6481 * if you're unsure if $oldParser is safe to use.
6484 * @return Parser A parser object that is not parsing anything
6486 public function getFreshParser() {
6487 if ( $this->mInParse
) {
6488 return $this->factory
->create();
6495 * Set's up the PHP implementation of OOUI for use in this request
6496 * and instructs OutputPage to enable OOUI for itself.
6500 public function enableOOUI() {
6501 OutputPage
::setupOOUI();
6502 $this->mOutput
->setEnableOOUI( true );
6506 * @param string $flag
6507 * @param string $reason
6509 protected function setOutputFlag( $flag, $reason ) {
6510 $this->mOutput
->setFlag( $flag );
6511 $name = $this->mTitle
->getPrefixedText();
6512 $this->logger
->debug( __METHOD__
. ": set $flag flag on '$name'; $reason" );