3 * PHP parser that converts wiki markup to HTML.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
23 use MediaWiki\Config\ServiceOptions
;
24 use MediaWiki\Linker\LinkRenderer
;
25 use MediaWiki\Linker\LinkRendererFactory
;
26 use MediaWiki\Linker\LinkTarget
;
27 use MediaWiki\MediaWikiServices
;
28 use MediaWiki\Special\SpecialPageFactory
;
29 use Psr\Log\NullLogger
;
30 use Wikimedia\ScopedCallback
;
31 use Psr\Log\LoggerInterface
;
34 * @defgroup Parser Parser
38 * PHP Parser - Processes wiki markup (which uses a more user-friendly
39 * syntax, such as "[[link]]" for making links), and provides a one-way
40 * transformation of that wiki markup it into (X)HTML output / markup
41 * (which in turn the browser understands, and can display).
43 * There are seven main entry points into the Parser class:
46 * produces HTML output
47 * - Parser::preSaveTransform()
48 * produces altered wiki markup
49 * - Parser::preprocess()
50 * removes HTML comments and expands templates
51 * - Parser::cleanSig() and Parser::cleanSigInSig()
52 * cleans a signature before saving it to preferences
53 * - Parser::getSection()
54 * return the content of a section from an article for section editing
55 * - Parser::replaceSection()
56 * replaces a section by number inside an article
57 * - Parser::getPreloadText()
58 * removes <noinclude> sections and <includeonly> tags
60 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
63 * $wgNamespacesWithSubpages
65 * @par Settings only within ParserOptions:
66 * $wgAllowExternalImages
67 * $wgAllowSpecialInclusion
75 * Update this version number when the ParserOutput format
76 * changes in an incompatible way, so the parser cache
77 * can automatically discard old data.
79 const VERSION
= '1.6.4';
82 * Update this version number when the output of serialiseHalfParsedText()
83 * changes in an incompatible way
85 const HALF_PARSED_VERSION
= 2;
87 # Flags for Parser::setFunctionHook
88 const SFH_NO_HASH
= 1;
89 const SFH_OBJECT_ARGS
= 2;
91 # Constants needed for external link processing
92 # Everything except bracket, space, or control characters
93 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
94 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
95 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
96 # uses to replace invalid HTML characters.
97 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
98 # Simplified expression to match an IPv4 or IPv6 address, or
99 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
100 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
101 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
102 // phpcs:ignore Generic.Files.LineLength
103 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
104 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
106 # Regular expression for a non-newline space
107 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
109 # Flags for preprocessToDom
110 const PTD_FOR_INCLUSION
= 1;
112 # Allowed values for $this->mOutputType
113 # Parameter to startExternalParse().
114 const OT_HTML
= 1; # like parse()
115 const OT_WIKI
= 2; # like preSaveTransform()
116 const OT_PREPROCESS
= 3; # like preprocess()
118 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
121 * @var string Prefix and suffix for temporary replacement strings
122 * for the multipass parser.
124 * \x7f should never appear in input as it's disallowed in XML.
125 * Using it at the front also gives us a little extra robustness
126 * since it shouldn't match when butted up against identifier-like
129 * Must not consist of all title characters, or else it will change
130 * the behavior of <nowiki> in a link.
132 * Must have a character that needs escaping in attributes, otherwise
133 * someone could put a strip marker in an attribute, to get around
134 * escaping quote marks, and break out of the attribute. Thus we add
137 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
138 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
140 # Markers used for wrapping the table of contents
141 const TOC_START
= '<mw:toc>';
142 const TOC_END
= '</mw:toc>';
144 /** @var int Assume that no output will later be saved this many seconds after parsing */
148 public $mTagHooks = [];
149 public $mTransparentTagHooks = [];
150 public $mFunctionHooks = [];
151 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
152 public $mFunctionTagHooks = [];
153 public $mStripList = [];
154 public $mDefaultStripList = [];
155 public $mVarCache = [];
156 public $mImageParams = [];
157 public $mImageParamsMagicArray = [];
158 public $mMarkerIndex = 0;
160 * @var bool Whether firstCallInit still needs to be called
162 public $mFirstCall = true;
164 # Initialised by initialiseVariables()
167 * @var MagicWordArray
172 * @var MagicWordArray
177 * @deprecated since 1.34, there should be no need to use this
182 # Initialised in constructor
183 public $mExtLinkBracketedRegex, $mUrlProtocols;
185 # Initialized in getPreprocessor()
186 /** @var Preprocessor */
187 public $mPreprocessor;
189 # Cleared with clearState():
201 public $mIncludeCount;
203 * @var LinkHolderArray
205 public $mLinkHolders;
208 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
209 public $mDefaultSort;
210 public $mTplRedirCache, $mHeadings, $mDoubleUnderscores;
211 public $mExpensiveFunctionCount; # number of expensive parser function calls
212 public $mShowToc, $mForceTocPosition;
214 public $mTplDomCache;
219 public $mUser; # User object; only used when doing pre-save transform
222 # These are variables reset at least once per parse regardless of $clearState
232 public $mTitle; # Title context, used for self-link rendering and similar things
233 public $mOutputType; # Output type, one of the OT_xxx constants
234 public $ot; # Shortcut alias, see setOutputType()
235 public $mRevisionObject; # The revision object of the specified revision ID
236 public $mRevisionId; # ID to display in {{REVISIONID}} tags
237 public $mRevisionTimestamp; # The timestamp of the specified revision ID
238 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
239 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
240 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
241 public $mInputSize = false; # For {{PAGESIZE}} on current page.
244 * @var array Array with the language name of each language link (i.e. the
245 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
246 * duplicate language links to the ParserOutput.
248 public $mLangLinkLanguages;
251 * @var MapCacheLRU|null
254 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
256 public $currentRevisionCache;
259 * @var bool|string Recursive call protection.
260 * This variable should be treated as if it were private.
262 public $mInParse = false;
264 /** @var SectionProfiler */
265 protected $mProfiler;
270 protected $mLinkRenderer;
272 /** @var MagicWordFactory */
273 private $magicWordFactory;
278 /** @var ParserFactory */
281 /** @var SpecialPageFactory */
282 private $specialPageFactory;
285 * This is called $svcOptions instead of $options like elsewhere to avoid confusion with
286 * $mOptions, which is public and widely used, and also with the local variable $options used
287 * for ParserOptions throughout this file.
289 * @var ServiceOptions
293 /** @var LinkRendererFactory */
294 private $linkRendererFactory;
296 /** @var NamespaceInfo */
299 /** @var LoggerInterface */
303 * TODO Make this a const when HHVM support is dropped (T192166)
308 public static $constructorOptions = [
309 // See $wgParserConf documentation
312 // See documentation for the corresponding config options
314 'EnableScaryTranscluding',
315 'ExtraInterlanguageLinkPrefixes',
327 'TranscludeCacheExpiry',
331 * Constructing parsers directly is deprecated! Use a ParserFactory.
333 * @param ServiceOptions|null $svcOptions
334 * @param MagicWordFactory|null $magicWordFactory
335 * @param Language|null $contLang Content language
336 * @param ParserFactory|null $factory
337 * @param string|null $urlProtocols As returned from wfUrlProtocols()
338 * @param SpecialPageFactory|null $spFactory
339 * @param LinkRendererFactory|null $linkRendererFactory
340 * @param NamespaceInfo|null $nsInfo
341 * @param LoggerInterface|null $logger
343 public function __construct(
345 MagicWordFactory
$magicWordFactory = null,
346 Language
$contLang = null,
347 ParserFactory
$factory = null,
348 $urlProtocols = null,
349 SpecialPageFactory
$spFactory = null,
350 $linkRendererFactory = null,
354 $services = MediaWikiServices
::getInstance();
355 if ( !$svcOptions ||
is_array( $svcOptions ) ) {
356 // Pre-1.34 calling convention is the first parameter is just ParserConf, the seventh is
357 // Config, and the eighth is LinkRendererFactory.
358 $this->mConf
= (array)$svcOptions;
359 if ( empty( $this->mConf
['class'] ) ) {
360 $this->mConf
['class'] = self
::class;
362 if ( empty( $this->mConf
['preprocessorClass'] ) ) {
363 $this->mConf
['preprocessorClass'] = self
::getDefaultPreprocessorClass();
365 $this->svcOptions
= new ServiceOptions( self
::$constructorOptions,
367 func_num_args() > 6 ?
func_get_arg( 6 ) : $services->getMainConfig()
369 $linkRendererFactory = func_num_args() > 7 ?
func_get_arg( 7 ) : null;
370 $nsInfo = func_num_args() > 8 ?
func_get_arg( 8 ) : null;
372 // New calling convention
373 $svcOptions->assertRequiredOptions( self
::$constructorOptions );
374 // $this->mConf is public, so we'll keep those two options there as well for
375 // compatibility until it's removed
377 'class' => $svcOptions->get( 'class' ),
378 'preprocessorClass' => $svcOptions->get( 'preprocessorClass' ),
380 $this->svcOptions
= $svcOptions;
383 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
384 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
385 self
::EXT_LINK_ADDR
.
386 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
388 $this->magicWordFactory
= $magicWordFactory ??
389 $services->getMagicWordFactory();
391 $this->contLang
= $contLang ??
$services->getContentLanguage();
393 $this->factory
= $factory ??
$services->getParserFactory();
394 $this->specialPageFactory
= $spFactory ??
$services->getSpecialPageFactory();
395 $this->linkRendererFactory
= $linkRendererFactory ??
$services->getLinkRendererFactory();
396 $this->nsInfo
= $nsInfo ??
$services->getNamespaceInfo();
397 $this->logger
= $logger ?
: new NullLogger();
401 * Reduce memory usage to reduce the impact of circular references
403 public function __destruct() {
404 if ( isset( $this->mLinkHolders
) ) {
405 unset( $this->mLinkHolders
);
407 foreach ( $this as $name => $value ) {
408 unset( $this->$name );
413 * Allow extensions to clean up when the parser is cloned
415 public function __clone() {
416 $this->mInParse
= false;
418 // T58226: When you create a reference "to" an object field, that
419 // makes the object field itself be a reference too (until the other
420 // reference goes out of scope). When cloning, any field that's a
421 // reference is copied as a reference in the new object. Both of these
422 // are defined PHP5 behaviors, as inconvenient as it is for us when old
423 // hooks from PHP4 days are passing fields by reference.
424 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
425 // Make a non-reference copy of the field, then rebind the field to
426 // reference the new copy.
432 Hooks
::run( 'ParserCloned', [ $this ] );
436 * Which class should we use for the preprocessor if not otherwise specified?
439 * @deprecated since 1.34, removing configurability of preprocessor
442 public static function getDefaultPreprocessorClass() {
443 return Preprocessor_Hash
::class;
447 * Do various kinds of initialisation on the first call of the parser
449 public function firstCallInit() {
450 if ( !$this->mFirstCall
) {
453 $this->mFirstCall
= false;
455 CoreParserFunctions
::register( $this );
456 CoreTagHooks
::register( $this );
457 $this->initialiseVariables();
459 // Avoid PHP 7.1 warning from passing $this by reference
461 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
469 public function clearState() {
470 $this->firstCallInit();
471 $this->mOutput
= new ParserOutput
;
472 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
473 $this->mAutonumber
= 0;
474 $this->mIncludeCount
= [];
475 $this->mLinkHolders
= new LinkHolderArray( $this );
477 $this->mRevisionObject
= $this->mRevisionTimestamp
=
478 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
479 $this->mVarCache
= [];
481 $this->mLangLinkLanguages
= [];
482 $this->currentRevisionCache
= null;
484 $this->mStripState
= new StripState( $this );
486 # Clear these on every parse, T6549
487 $this->mTplRedirCache
= $this->mTplDomCache
= [];
489 $this->mShowToc
= true;
490 $this->mForceTocPosition
= false;
491 $this->mIncludeSizes
= [
495 $this->mPPNodeCount
= 0;
496 $this->mGeneratedPPNodeCount
= 0;
497 $this->mHighestExpansionDepth
= 0;
498 $this->mDefaultSort
= false;
499 $this->mHeadings
= [];
500 $this->mDoubleUnderscores
= [];
501 $this->mExpensiveFunctionCount
= 0;
504 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
505 $this->mPreprocessor
= null;
508 $this->mProfiler
= new SectionProfiler();
510 // Avoid PHP 7.1 warning from passing $this by reference
512 Hooks
::run( 'ParserClearState', [ &$parser ] );
516 * Convert wikitext to HTML
517 * Do not call this function recursively.
519 * @param string $text Text we want to parse
520 * @param-taint $text escapes_htmlnoent
521 * @param Title $title
522 * @param ParserOptions $options
523 * @param bool $linestart
524 * @param bool $clearState
525 * @param int|null $revid Number to pass in {{REVISIONID}}
526 * @return ParserOutput A ParserOutput
527 * @return-taint escaped
529 public function parse(
530 $text, Title
$title, ParserOptions
$options,
531 $linestart = true, $clearState = true, $revid = null
534 // We use U+007F DELETE to construct strip markers, so we have to make
535 // sure that this character does not occur in the input text.
536 $text = strtr( $text, "\x7f", "?" );
537 $magicScopeVariable = $this->lock();
539 // Strip U+0000 NULL (T159174)
540 $text = str_replace( "\000", '', $text );
542 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
544 $this->currentRevisionCache
= null;
545 $this->mInputSize
= strlen( $text );
546 if ( $this->mOptions
->getEnableLimitReport() ) {
547 $this->mOutput
->resetParseStartTime();
550 $oldRevisionId = $this->mRevisionId
;
551 $oldRevisionObject = $this->mRevisionObject
;
552 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
553 $oldRevisionUser = $this->mRevisionUser
;
554 $oldRevisionSize = $this->mRevisionSize
;
555 if ( $revid !== null ) {
556 $this->mRevisionId
= $revid;
557 $this->mRevisionObject
= null;
558 $this->mRevisionTimestamp
= null;
559 $this->mRevisionUser
= null;
560 $this->mRevisionSize
= null;
563 // Avoid PHP 7.1 warning from passing $this by reference
565 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
567 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
568 $text = $this->internalParse( $text );
569 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
571 $text = $this->internalParseHalfParsed( $text, true, $linestart );
574 * A converted title will be provided in the output object if title and
575 * content conversion are enabled, the article text does not contain
576 * a conversion-suppressing double-underscore tag, and no
577 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
578 * automatic link conversion.
580 if ( !( $options->getDisableTitleConversion()
581 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
582 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
583 ||
$this->mOutput
->getDisplayTitle() !== false )
585 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
586 if ( $convruletitle ) {
587 $this->mOutput
->setTitleText( $convruletitle );
589 $titleText = $this->getTargetLanguage()->convertTitle( $title );
590 $this->mOutput
->setTitleText( $titleText );
594 # Compute runtime adaptive expiry if set
595 $this->mOutput
->finalizeAdaptiveCacheExpiry();
597 # Warn if too many heavyweight parser functions were used
598 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
599 $this->limitationWarn( 'expensive-parserfunction',
600 $this->mExpensiveFunctionCount
,
601 $this->mOptions
->getExpensiveParserFunctionLimit()
605 # Information on limits, for the benefit of users who try to skirt them
606 if ( $this->mOptions
->getEnableLimitReport() ) {
607 $text .= $this->makeLimitReport();
610 # Wrap non-interface parser output in a <div> so it can be targeted
612 $class = $this->mOptions
->getWrapOutputClass();
613 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
614 $this->mOutput
->addWrapperDivClass( $class );
617 $this->mOutput
->setText( $text );
619 $this->mRevisionId
= $oldRevisionId;
620 $this->mRevisionObject
= $oldRevisionObject;
621 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
622 $this->mRevisionUser
= $oldRevisionUser;
623 $this->mRevisionSize
= $oldRevisionSize;
624 $this->mInputSize
= false;
625 $this->currentRevisionCache
= null;
627 return $this->mOutput
;
631 * Set the limit report data in the current ParserOutput, and return the
632 * limit report HTML comment.
636 protected function makeLimitReport() {
637 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
639 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
640 if ( $cpuTime !== null ) {
641 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
642 sprintf( "%.3f", $cpuTime )
646 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
647 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
648 sprintf( "%.3f", $wallTime )
651 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
652 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
654 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
655 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
657 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
658 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
660 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
661 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
663 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
664 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
666 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
667 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
670 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
671 $this->mOutput
->setLimitReportData( $key, $value );
674 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
676 $limitReport = "NewPP limit report\n";
677 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
678 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
680 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
681 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
682 $limitReport .= 'Dynamic content: ' .
683 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
685 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput
->getAllFlags() ) . "]\n";
687 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
688 if ( Hooks
::run( 'ParserLimitReportFormat',
689 [ $key, &$value, &$limitReport, false, false ]
691 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
692 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
693 ->inLanguage( 'en' )->useDatabase( false );
694 if ( !$valueMsg->exists() ) {
695 $valueMsg = new RawMessage( '$1' );
697 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
698 $valueMsg->params( $value );
699 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
703 // Since we're not really outputting HTML, decode the entities and
704 // then re-encode the things that need hiding inside HTML comments.
705 $limitReport = htmlspecialchars_decode( $limitReport );
707 // Sanitize for comment. Note '‐' in the replacement is U+2010,
708 // which looks much like the problematic '-'.
709 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
710 $text = "\n<!-- \n$limitReport-->\n";
712 // Add on template profiling data in human/machine readable way
713 $dataByFunc = $this->mProfiler
->getFunctionStats();
714 uasort( $dataByFunc, function ( $a, $b ) {
715 return $b['real'] <=> $a['real']; // descending order
718 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
719 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
720 $item['%real'], $item['real'], $item['calls'],
721 htmlspecialchars( $item['name'] ) );
723 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
724 $text .= implode( "\n", $profileReport ) . "\n-->\n";
726 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
728 // Add other cache related metadata
729 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
730 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
732 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
733 $this->mOutput
->getCacheTime() );
734 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
735 $this->mOutput
->getCacheExpiry() );
736 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
737 $this->mOutput
->hasDynamicContent() );
739 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
740 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
741 $this->mTitle
->getPrefixedDBkey() );
747 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
748 * can be called from an extension tag hook.
750 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
751 * instead, which means that lists and links have not been fully parsed yet,
752 * and strip markers are still present.
754 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
756 * Use this function if you're a parser tag hook and you want to parse
757 * wikitext before or after applying additional transformations, and you
758 * intend to *return the result as hook output*, which will cause it to go
759 * through the rest of parsing process automatically.
761 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
762 * $text are not expanded
764 * @param string $text Text extension wants to have parsed
765 * @param-taint $text escapes_htmlnoent
766 * @param bool|PPFrame $frame The frame to use for expanding any template variables
767 * @return string UNSAFE half-parsed HTML
768 * @return-taint escaped
770 public function recursiveTagParse( $text, $frame = false ) {
771 // Avoid PHP 7.1 warning from passing $this by reference
773 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
774 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
775 $text = $this->internalParse( $text, false, $frame );
780 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
781 * point can be called from an extension tag hook.
783 * The output of this function is fully-parsed HTML that is safe for output.
784 * If you're a parser tag hook, you might want to use recursiveTagParse()
787 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
788 * $text are not expanded
792 * @param string $text Text extension wants to have parsed
793 * @param-taint $text escapes_htmlnoent
794 * @param bool|PPFrame $frame The frame to use for expanding any template variables
795 * @return string Fully parsed HTML
796 * @return-taint escaped
798 public function recursiveTagParseFully( $text, $frame = false ) {
799 $text = $this->recursiveTagParse( $text, $frame );
800 $text = $this->internalParseHalfParsed( $text, false );
805 * Expand templates and variables in the text, producing valid, static wikitext.
806 * Also removes comments.
807 * Do not call this function recursively.
808 * @param string $text
809 * @param Title|null $title
810 * @param ParserOptions $options
811 * @param int|null $revid
812 * @param bool|PPFrame $frame
813 * @return mixed|string
815 public function preprocess( $text, Title
$title = null,
816 ParserOptions
$options, $revid = null, $frame = false
818 $magicScopeVariable = $this->lock();
819 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
820 if ( $revid !== null ) {
821 $this->mRevisionId
= $revid;
823 // Avoid PHP 7.1 warning from passing $this by reference
825 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
826 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
827 $text = $this->replaceVariables( $text, $frame );
828 $text = $this->mStripState
->unstripBoth( $text );
833 * Recursive parser entry point that can be called from an extension tag
836 * @param string $text Text to be expanded
837 * @param bool|PPFrame $frame The frame to use for expanding any template variables
841 public function recursivePreprocess( $text, $frame = false ) {
842 $text = $this->replaceVariables( $text, $frame );
843 $text = $this->mStripState
->unstripBoth( $text );
848 * Process the wikitext for the "?preload=" feature. (T7210)
850 * "<noinclude>", "<includeonly>" etc. are parsed as for template
851 * transclusion, comments, templates, arguments, tags hooks and parser
852 * functions are untouched.
854 * @param string $text
855 * @param Title $title
856 * @param ParserOptions $options
857 * @param array $params
860 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
861 $msg = new RawMessage( $text );
862 $text = $msg->params( $params )->plain();
864 # Parser (re)initialisation
865 $magicScopeVariable = $this->lock();
866 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
868 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
869 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
870 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
871 $text = $this->mStripState
->unstripBoth( $text );
876 * Set the current user.
877 * Should only be used when doing pre-save transform.
879 * @param User|null $user User object or null (to reset)
881 public function setUser( $user ) {
882 $this->mUser
= $user;
886 * Set the context title
890 public function setTitle( $t ) {
892 $t = Title
::newFromText( 'NO TITLE' );
895 if ( $t->hasFragment() ) {
896 # Strip the fragment to avoid various odd effects
897 $this->mTitle
= $t->createFragmentTarget( '' );
904 * Accessor for the Title object
908 public function getTitle() {
909 return $this->mTitle
;
913 * Accessor/mutator for the Title object
915 * @param Title|null $x Title object or null to just get the current one
918 public function Title( $x = null ) {
919 return wfSetVar( $this->mTitle
, $x );
923 * Set the output type
925 * @param int $ot New value
927 public function setOutputType( $ot ) {
928 $this->mOutputType
= $ot;
931 'html' => $ot == self
::OT_HTML
,
932 'wiki' => $ot == self
::OT_WIKI
,
933 'pre' => $ot == self
::OT_PREPROCESS
,
934 'plain' => $ot == self
::OT_PLAIN
,
939 * Accessor/mutator for the output type
941 * @param int|null $x New value or null to just get the current one
944 public function OutputType( $x = null ) {
945 return wfSetVar( $this->mOutputType
, $x );
949 * Get the ParserOutput object
951 * @return ParserOutput
953 public function getOutput() {
954 return $this->mOutput
;
958 * Get the ParserOptions object
960 * @return ParserOptions
962 public function getOptions() {
963 return $this->mOptions
;
967 * Accessor/mutator for the ParserOptions object
969 * @param ParserOptions|null $x New value or null to just get the current one
970 * @return ParserOptions Current ParserOptions object
972 public function Options( $x = null ) {
973 return wfSetVar( $this->mOptions
, $x );
979 public function nextLinkID() {
980 return $this->mLinkID++
;
986 public function setLinkID( $id ) {
987 $this->mLinkID
= $id;
991 * Get a language object for use in parser functions such as {{FORMATNUM:}}
994 public function getFunctionLang() {
995 return $this->getTargetLanguage();
999 * Get the target language for the content being parsed. This is usually the
1000 * language that the content is in.
1004 * @throws MWException
1007 public function getTargetLanguage() {
1008 $target = $this->mOptions
->getTargetLanguage();
1010 if ( $target !== null ) {
1012 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
1013 return $this->mOptions
->getUserLangObj();
1014 } elseif ( is_null( $this->mTitle
) ) {
1015 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
1018 return $this->mTitle
->getPageLanguage();
1022 * Get the language object for language conversion
1023 * @deprecated since 1.32, just use getTargetLanguage()
1024 * @return Language|null
1026 public function getConverterLanguage() {
1027 return $this->getTargetLanguage();
1031 * Get a User object either from $this->mUser, if set, or from the
1032 * ParserOptions object otherwise
1036 public function getUser() {
1037 if ( !is_null( $this->mUser
) ) {
1038 return $this->mUser
;
1040 return $this->mOptions
->getUser();
1044 * Get a preprocessor object
1046 * @return Preprocessor
1048 public function getPreprocessor() {
1049 if ( !isset( $this->mPreprocessor
) ) {
1050 $class = $this->svcOptions
->get( 'preprocessorClass' );
1051 $this->mPreprocessor
= new $class( $this );
1053 return $this->mPreprocessor
;
1057 * Get a LinkRenderer instance to make links with
1060 * @return LinkRenderer
1062 public function getLinkRenderer() {
1063 // XXX We make the LinkRenderer with current options and then cache it forever
1064 if ( !$this->mLinkRenderer
) {
1065 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
1066 $this->mLinkRenderer
->setStubThreshold(
1067 $this->getOptions()->getStubThreshold()
1071 return $this->mLinkRenderer
;
1075 * Get the MagicWordFactory that this Parser is using
1078 * @return MagicWordFactory
1080 public function getMagicWordFactory() {
1081 return $this->magicWordFactory
;
1085 * Get the content language that this Parser is using
1090 public function getContentLanguage() {
1091 return $this->contLang
;
1095 * Replaces all occurrences of HTML-style comments and the given tags
1096 * in the text with a random marker and returns the next text. The output
1097 * parameter $matches will be an associative array filled with data in
1104 * [ 'param' => 'x' ],
1105 * '<element param="x">tag content</element>' ]
1108 * @param array $elements List of element names. Comments are always extracted.
1109 * @param string $text Source text string.
1110 * @param array &$matches Out parameter, Array: extracted tags
1111 * @return string Stripped text
1113 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1118 $taglist = implode( '|', $elements );
1119 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1121 while ( $text != '' ) {
1122 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1124 if ( count( $p ) < 5 ) {
1127 if ( count( $p ) > 5 ) {
1135 list( , $element, $attributes, $close, $inside ) = $p;
1138 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1139 $stripped .= $marker;
1141 if ( $close === '/>' ) {
1142 # Empty element tag, <tag />
1147 if ( $element === '!--' ) {
1150 $end = "/(<\\/$element\\s*>)/i";
1152 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1154 if ( count( $q ) < 3 ) {
1155 # No end tag -- let it run out to the end of the text.
1159 list( , $tail, $text ) = $q;
1163 $matches[$marker] = [ $element,
1165 Sanitizer
::decodeTagAttributes( $attributes ),
1166 "<$element$attributes$close$content$tail" ];
1172 * Get a list of strippable XML-like elements
1176 public function getStripList() {
1177 return $this->mStripList
;
1181 * Add an item to the strip state
1182 * Returns the unique tag which must be inserted into the stripped text
1183 * The tag will be replaced with the original text in unstrip()
1185 * @param string $text
1189 public function insertStripItem( $text ) {
1190 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1191 $this->mMarkerIndex++
;
1192 $this->mStripState
->addGeneral( $marker, $text );
1197 * parse the wiki syntax used to render tables
1200 * @param string $text
1203 public function doTableStuff( $text ) {
1204 $lines = StringUtils
::explode( "\n", $text );
1206 $td_history = []; # Is currently a td tag open?
1207 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1208 $tr_history = []; # Is currently a tr tag open?
1209 $tr_attributes = []; # history of tr attributes
1210 $has_opened_tr = []; # Did this table open a <tr> element?
1211 $indent_level = 0; # indent level of the table
1213 foreach ( $lines as $outLine ) {
1214 $line = trim( $outLine );
1216 if ( $line === '' ) { # empty line, go to next line
1217 $out .= $outLine . "\n";
1221 $first_character = $line[0];
1222 $first_two = substr( $line, 0, 2 );
1225 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1226 # First check if we are starting a new table
1227 $indent_level = strlen( $matches[1] );
1229 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1230 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1232 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1233 array_push( $td_history, false );
1234 array_push( $last_tag_history, '' );
1235 array_push( $tr_history, false );
1236 array_push( $tr_attributes, '' );
1237 array_push( $has_opened_tr, false );
1238 } elseif ( count( $td_history ) == 0 ) {
1239 # Don't do any of the following
1240 $out .= $outLine . "\n";
1242 } elseif ( $first_two === '|}' ) {
1243 # We are ending a table
1244 $line = '</table>' . substr( $line, 2 );
1245 $last_tag = array_pop( $last_tag_history );
1247 if ( !array_pop( $has_opened_tr ) ) {
1248 $line = "<tr><td></td></tr>{$line}";
1251 if ( array_pop( $tr_history ) ) {
1252 $line = "</tr>{$line}";
1255 if ( array_pop( $td_history ) ) {
1256 $line = "</{$last_tag}>{$line}";
1258 array_pop( $tr_attributes );
1259 if ( $indent_level > 0 ) {
1260 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1264 } elseif ( $first_two === '|-' ) {
1265 # Now we have a table row
1266 $line = preg_replace( '#^\|-+#', '', $line );
1268 # Whats after the tag is now only attributes
1269 $attributes = $this->mStripState
->unstripBoth( $line );
1270 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1271 array_pop( $tr_attributes );
1272 array_push( $tr_attributes, $attributes );
1275 $last_tag = array_pop( $last_tag_history );
1276 array_pop( $has_opened_tr );
1277 array_push( $has_opened_tr, true );
1279 if ( array_pop( $tr_history ) ) {
1283 if ( array_pop( $td_history ) ) {
1284 $line = "</{$last_tag}>{$line}";
1288 array_push( $tr_history, false );
1289 array_push( $td_history, false );
1290 array_push( $last_tag_history, '' );
1291 } elseif ( $first_character === '|'
1292 ||
$first_character === '!'
1293 ||
$first_two === '|+'
1295 # This might be cell elements, td, th or captions
1296 if ( $first_two === '|+' ) {
1297 $first_character = '+';
1298 $line = substr( $line, 2 );
1300 $line = substr( $line, 1 );
1303 // Implies both are valid for table headings.
1304 if ( $first_character === '!' ) {
1305 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1308 # Split up multiple cells on the same line.
1309 # FIXME : This can result in improper nesting of tags processed
1310 # by earlier parser steps.
1311 $cells = explode( '||', $line );
1315 # Loop through each table cell
1316 foreach ( $cells as $cell ) {
1318 if ( $first_character !== '+' ) {
1319 $tr_after = array_pop( $tr_attributes );
1320 if ( !array_pop( $tr_history ) ) {
1321 $previous = "<tr{$tr_after}>\n";
1323 array_push( $tr_history, true );
1324 array_push( $tr_attributes, '' );
1325 array_pop( $has_opened_tr );
1326 array_push( $has_opened_tr, true );
1329 $last_tag = array_pop( $last_tag_history );
1331 if ( array_pop( $td_history ) ) {
1332 $previous = "</{$last_tag}>\n{$previous}";
1335 if ( $first_character === '|' ) {
1337 } elseif ( $first_character === '!' ) {
1339 } elseif ( $first_character === '+' ) {
1340 $last_tag = 'caption';
1345 array_push( $last_tag_history, $last_tag );
1347 # A cell could contain both parameters and data
1348 $cell_data = explode( '|', $cell, 2 );
1350 # T2553: Note that a '|' inside an invalid link should not
1351 # be mistaken as delimiting cell parameters
1352 # Bug T153140: Neither should language converter markup.
1353 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1354 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1355 } elseif ( count( $cell_data ) == 1 ) {
1356 // Whitespace in cells is trimmed
1357 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1359 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1360 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1361 // Whitespace in cells is trimmed
1362 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1366 array_push( $td_history, true );
1369 $out .= $outLine . "\n";
1372 # Closing open td, tr && table
1373 while ( count( $td_history ) > 0 ) {
1374 if ( array_pop( $td_history ) ) {
1377 if ( array_pop( $tr_history ) ) {
1380 if ( !array_pop( $has_opened_tr ) ) {
1381 $out .= "<tr><td></td></tr>\n";
1384 $out .= "</table>\n";
1387 # Remove trailing line-ending (b/c)
1388 if ( substr( $out, -1 ) === "\n" ) {
1389 $out = substr( $out, 0, -1 );
1392 # special case: don't return empty table
1393 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1401 * Helper function for parse() that transforms wiki markup into half-parsed
1402 * HTML. Only called for $mOutputType == self::OT_HTML.
1406 * @param string $text The text to parse
1407 * @param-taint $text escapes_html
1408 * @param bool $isMain Whether this is being called from the main parse() function
1409 * @param PPFrame|bool $frame A pre-processor frame
1413 public function internalParse( $text, $isMain = true, $frame = false ) {
1416 // Avoid PHP 7.1 warning from passing $this by reference
1419 # Hook to suspend the parser in this state
1420 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1424 # if $frame is provided, then use $frame for replacing any variables
1426 # use frame depth to infer how include/noinclude tags should be handled
1427 # depth=0 means this is the top-level document; otherwise it's an included document
1428 if ( !$frame->depth
) {
1431 $flag = self
::PTD_FOR_INCLUSION
;
1433 $dom = $this->preprocessToDom( $text, $flag );
1434 $text = $frame->expand( $dom );
1436 # if $frame is not provided, then use old-style replaceVariables
1437 $text = $this->replaceVariables( $text );
1440 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1441 $text = Sanitizer
::removeHTMLtags(
1443 [ $this, 'attributeStripCallback' ],
1445 array_keys( $this->mTransparentTagHooks
),
1447 [ $this, 'addTrackingCategory' ]
1449 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1451 # Tables need to come after variable replacement for things to work
1452 # properly; putting them before other transformations should keep
1453 # exciting things like link expansions from showing up in surprising
1455 $text = $this->doTableStuff( $text );
1457 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1459 $text = $this->doDoubleUnderscore( $text );
1461 $text = $this->doHeadings( $text );
1462 $text = $this->replaceInternalLinks( $text );
1463 $text = $this->doAllQuotes( $text );
1464 $text = $this->replaceExternalLinks( $text );
1466 # replaceInternalLinks may sometimes leave behind
1467 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1468 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1470 $text = $this->doMagicLinks( $text );
1471 $text = $this->formatHeadings( $text, $origText, $isMain );
1477 * Helper function for parse() that transforms half-parsed HTML into fully
1480 * @param string $text
1481 * @param bool $isMain
1482 * @param bool $linestart
1485 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1486 $text = $this->mStripState
->unstripGeneral( $text );
1488 // Avoid PHP 7.1 warning from passing $this by reference
1492 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1495 # Clean up special characters, only run once, next-to-last before doBlockLevels
1496 $text = Sanitizer
::armorFrenchSpaces( $text );
1498 $text = $this->doBlockLevels( $text, $linestart );
1500 $this->replaceLinkHolders( $text );
1503 * The input doesn't get language converted if
1505 * b) Content isn't converted
1506 * c) It's a conversion table
1507 * d) it is an interface message (which is in the user language)
1509 if ( !( $this->mOptions
->getDisableContentConversion()
1510 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1511 && !$this->mOptions
->getInterfaceMessage()
1513 # The position of the convert() call should not be changed. it
1514 # assumes that the links are all replaced and the only thing left
1515 # is the <nowiki> mark.
1516 $text = $this->getTargetLanguage()->convert( $text );
1519 $text = $this->mStripState
->unstripNoWiki( $text );
1522 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1525 $text = $this->replaceTransparentTags( $text );
1526 $text = $this->mStripState
->unstripGeneral( $text );
1528 $text = Sanitizer
::normalizeCharReferences( $text );
1530 if ( MWTidy
::isEnabled() ) {
1531 if ( $this->mOptions
->getTidy() ) {
1532 $text = MWTidy
::tidy( $text );
1535 # attempt to sanitize at least some nesting problems
1536 # (T4702 and quite a few others)
1537 # This code path is buggy and deprecated!
1538 wfDeprecated( 'disabling tidy', '1.33' );
1540 # ''Something [http://www.cool.com cool''] -->
1541 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1542 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1543 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1544 # fix up an anchor inside another anchor, only
1545 # at least for a single single nested link (T5695)
1546 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1547 '\\1\\2</a>\\3</a>\\1\\4</a>',
1548 # fix div inside inline elements- doBlockLevels won't wrap a line which
1549 # contains a div, so fix it up here; replace
1550 # div with escaped text
1551 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1552 '\\1\\3<div\\5>\\6</div>\\8\\9',
1553 # remove empty italic or bold tag pairs, some
1554 # introduced by rules above
1555 '/<([bi])><\/\\1>/' => '',
1558 $text = preg_replace(
1559 array_keys( $tidyregs ),
1560 array_values( $tidyregs ),
1565 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1572 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1573 * magic external links.
1578 * @param string $text
1582 public function doMagicLinks( $text ) {
1583 $prots = wfUrlProtocolsWithoutProtRel();
1584 $urlChar = self
::EXT_LINK_URL_CLASS
;
1585 $addr = self
::EXT_LINK_ADDR
;
1586 $space = self
::SPACE_NOT_NL
; # non-newline space
1587 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1588 $spaces = "$space++"; # possessive match of 1 or more spaces
1589 $text = preg_replace_callback(
1591 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1592 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1593 (\b # m[3]: Free external links
1595 ($addr$urlChar*) # m[4]: Post-protocol path
1597 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1599 \bISBN $spaces ( # m[6]: ISBN, capture number
1600 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1601 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1602 [0-9Xx] # check digit
1604 )!xu", [ $this, 'magicLinkCallback' ], $text );
1609 * @throws MWException
1611 * @return string HTML
1613 public function magicLinkCallback( $m ) {
1614 if ( isset( $m[1] ) && $m[1] !== '' ) {
1617 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1620 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1621 # Free external link
1622 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1623 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1625 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1626 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1631 $cssClass = 'mw-magiclink-rfc';
1632 $trackingCat = 'magiclink-tracking-rfc';
1634 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1635 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1639 $urlmsg = 'pubmedurl';
1640 $cssClass = 'mw-magiclink-pmid';
1641 $trackingCat = 'magiclink-tracking-pmid';
1644 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1645 substr( $m[0], 0, 20 ) . '"' );
1647 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1648 $this->addTrackingCategory( $trackingCat );
1649 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1650 } elseif ( isset( $m[6] ) && $m[6] !== ''
1651 && $this->mOptions
->getMagicISBNLinks()
1655 $space = self
::SPACE_NOT_NL
; # non-newline space
1656 $isbn = preg_replace( "/$space/", ' ', $isbn );
1657 $num = strtr( $isbn, [
1662 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1663 return $this->getLinkRenderer()->makeKnownLink(
1664 SpecialPage
::getTitleFor( 'Booksources', $num ),
1667 'class' => 'internal mw-magiclink-isbn',
1668 'title' => false // suppress title attribute
1677 * Make a free external link, given a user-supplied URL
1679 * @param string $url
1680 * @param int $numPostProto
1681 * The number of characters after the protocol.
1682 * @return string HTML
1685 public function makeFreeExternalLink( $url, $numPostProto ) {
1688 # The characters '<' and '>' (which were escaped by
1689 # removeHTMLtags()) should not be included in
1690 # URLs, per RFC 2396.
1691 # Make terminate a URL as well (bug T84937)
1694 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1699 $trail = substr( $url, $m2[0][1] ) . $trail;
1700 $url = substr( $url, 0, $m2[0][1] );
1703 # Move trailing punctuation to $trail
1705 # If there is no left bracket, then consider right brackets fair game too
1706 if ( strpos( $url, '(' ) === false ) {
1710 $urlRev = strrev( $url );
1711 $numSepChars = strspn( $urlRev, $sep );
1712 # Don't break a trailing HTML entity by moving the ; into $trail
1713 # This is in hot code, so use substr_compare to avoid having to
1714 # create a new string object for the comparison
1715 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1716 # more optimization: instead of running preg_match with a $
1717 # anchor, which can be slow, do the match on the reversed
1718 # string starting at the desired offset.
1719 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1720 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1724 if ( $numSepChars ) {
1725 $trail = substr( $url, -$numSepChars ) . $trail;
1726 $url = substr( $url, 0, -$numSepChars );
1729 # Verify that we still have a real URL after trail removal, and
1730 # not just lone protocol
1731 if ( strlen( $trail ) >= $numPostProto ) {
1732 return $url . $trail;
1735 $url = Sanitizer
::cleanUrl( $url );
1737 # Is this an external image?
1738 $text = $this->maybeMakeExternalImage( $url );
1739 if ( $text === false ) {
1740 # Not an image, make a link
1741 $text = Linker
::makeExternalLink( $url,
1742 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1744 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1745 # Register it in the output object...
1746 $this->mOutput
->addExternalLink( $url );
1748 return $text . $trail;
1752 * Parse headers and return html
1756 * @param string $text
1760 public function doHeadings( $text ) {
1761 for ( $i = 6; $i >= 1; --$i ) {
1762 $h = str_repeat( '=', $i );
1763 // Trim non-newline whitespace from headings
1764 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1765 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1771 * Replace single quotes with HTML markup
1774 * @param string $text
1776 * @return string The altered text
1778 public function doAllQuotes( $text ) {
1780 $lines = StringUtils
::explode( "\n", $text );
1781 foreach ( $lines as $line ) {
1782 $outtext .= $this->doQuotes( $line ) . "\n";
1784 $outtext = substr( $outtext, 0, -1 );
1789 * Helper function for doAllQuotes()
1791 * @param string $text
1795 public function doQuotes( $text ) {
1796 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1797 $countarr = count( $arr );
1798 if ( $countarr == 1 ) {
1802 // First, do some preliminary work. This may shift some apostrophes from
1803 // being mark-up to being text. It also counts the number of occurrences
1804 // of bold and italics mark-ups.
1807 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1808 $thislen = strlen( $arr[$i] );
1809 // If there are ever four apostrophes, assume the first is supposed to
1810 // be text, and the remaining three constitute mark-up for bold text.
1811 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1812 if ( $thislen == 4 ) {
1813 $arr[$i - 1] .= "'";
1816 } elseif ( $thislen > 5 ) {
1817 // If there are more than 5 apostrophes in a row, assume they're all
1818 // text except for the last 5.
1819 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1820 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1824 // Count the number of occurrences of bold and italics mark-ups.
1825 if ( $thislen == 2 ) {
1827 } elseif ( $thislen == 3 ) {
1829 } elseif ( $thislen == 5 ) {
1835 // If there is an odd number of both bold and italics, it is likely
1836 // that one of the bold ones was meant to be an apostrophe followed
1837 // by italics. Which one we cannot know for certain, but it is more
1838 // likely to be one that has a single-letter word before it.
1839 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1840 $firstsingleletterword = -1;
1841 $firstmultiletterword = -1;
1843 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1844 if ( strlen( $arr[$i] ) == 3 ) {
1845 $x1 = substr( $arr[$i - 1], -1 );
1846 $x2 = substr( $arr[$i - 1], -2, 1 );
1847 if ( $x1 === ' ' ) {
1848 if ( $firstspace == -1 ) {
1851 } elseif ( $x2 === ' ' ) {
1852 $firstsingleletterword = $i;
1853 // if $firstsingleletterword is set, we don't
1854 // look at the other options, so we can bail early.
1856 } elseif ( $firstmultiletterword == -1 ) {
1857 $firstmultiletterword = $i;
1862 // If there is a single-letter word, use it!
1863 if ( $firstsingleletterword > -1 ) {
1864 $arr[$firstsingleletterword] = "''";
1865 $arr[$firstsingleletterword - 1] .= "'";
1866 } elseif ( $firstmultiletterword > -1 ) {
1867 // If not, but there's a multi-letter word, use that one.
1868 $arr[$firstmultiletterword] = "''";
1869 $arr[$firstmultiletterword - 1] .= "'";
1870 } elseif ( $firstspace > -1 ) {
1871 // ... otherwise use the first one that has neither.
1872 // (notice that it is possible for all three to be -1 if, for example,
1873 // there is only one pentuple-apostrophe in the line)
1874 $arr[$firstspace] = "''";
1875 $arr[$firstspace - 1] .= "'";
1879 // Now let's actually convert our apostrophic mush to HTML!
1884 foreach ( $arr as $r ) {
1885 if ( ( $i %
2 ) == 0 ) {
1886 if ( $state === 'both' ) {
1892 $thislen = strlen( $r );
1893 if ( $thislen == 2 ) {
1894 if ( $state === 'i' ) {
1897 } elseif ( $state === 'bi' ) {
1900 } elseif ( $state === 'ib' ) {
1901 $output .= '</b></i><b>';
1903 } elseif ( $state === 'both' ) {
1904 $output .= '<b><i>' . $buffer . '</i>';
1906 } else { // $state can be 'b' or ''
1910 } elseif ( $thislen == 3 ) {
1911 if ( $state === 'b' ) {
1914 } elseif ( $state === 'bi' ) {
1915 $output .= '</i></b><i>';
1917 } elseif ( $state === 'ib' ) {
1920 } elseif ( $state === 'both' ) {
1921 $output .= '<i><b>' . $buffer . '</b>';
1923 } else { // $state can be 'i' or ''
1927 } elseif ( $thislen == 5 ) {
1928 if ( $state === 'b' ) {
1929 $output .= '</b><i>';
1931 } elseif ( $state === 'i' ) {
1932 $output .= '</i><b>';
1934 } elseif ( $state === 'bi' ) {
1935 $output .= '</i></b>';
1937 } elseif ( $state === 'ib' ) {
1938 $output .= '</b></i>';
1940 } elseif ( $state === 'both' ) {
1941 $output .= '<i><b>' . $buffer . '</b></i>';
1943 } else { // ($state == '')
1951 // Now close all remaining tags. Notice that the order is important.
1952 if ( $state === 'b' ||
$state === 'ib' ) {
1955 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1958 if ( $state === 'bi' ) {
1961 // There might be lonely ''''', so make sure we have a buffer
1962 if ( $state === 'both' && $buffer ) {
1963 $output .= '<b><i>' . $buffer . '</i></b>';
1969 * Replace external links (REL)
1971 * Note: this is all very hackish and the order of execution matters a lot.
1972 * Make sure to run tests/parser/parserTests.php if you change this code.
1976 * @param string $text
1978 * @throws MWException
1981 public function replaceExternalLinks( $text ) {
1982 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1983 if ( $bits === false ) {
1984 throw new MWException( "PCRE needs to be compiled with "
1985 . "--enable-unicode-properties in order for MediaWiki to function" );
1987 $s = array_shift( $bits );
1990 while ( $i < count( $bits ) ) {
1993 $text = $bits[$i++
];
1994 $trail = $bits[$i++
];
1996 # The characters '<' and '>' (which were escaped by
1997 # removeHTMLtags()) should not be included in
1998 # URLs, per RFC 2396.
2000 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
2001 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
2002 $url = substr( $url, 0, $m2[0][1] );
2005 # If the link text is an image URL, replace it with an <img> tag
2006 # This happened by accident in the original parser, but some people used it extensively
2007 $img = $this->maybeMakeExternalImage( $text );
2008 if ( $img !== false ) {
2014 # Set linktype for CSS
2017 # No link text, e.g. [http://domain.tld/some.link]
2018 if ( $text == '' ) {
2020 $langObj = $this->getTargetLanguage();
2021 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
2022 $linktype = 'autonumber';
2024 # Have link text, e.g. [http://domain.tld/some.link text]s
2026 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
2029 // Excluding protocol-relative URLs may avoid many false positives.
2030 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2031 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2034 $url = Sanitizer
::cleanUrl( $url );
2036 # Use the encoded URL
2037 # This means that users can paste URLs directly into the text
2038 # Funny characters like ö aren't valid in URLs anyway
2039 # This was changed in August 2004
2040 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
2041 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
2043 # Register link in the output object.
2044 $this->mOutput
->addExternalLink( $url );
2051 * Get the rel attribute for a particular external link.
2054 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2055 * nofollow if appropriate
2056 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2057 * @return string|null Rel attribute for $url
2059 public static function getExternalLinkRel( $url = false, $title = null ) {
2060 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2061 $ns = $title ?
$title->getNamespace() : false;
2062 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2063 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2071 * Get an associative array of additional HTML attributes appropriate for a
2072 * particular external link. This currently may include rel => nofollow
2073 * (depending on configuration, namespace, and the URL's domain) and/or a
2074 * target attribute (depending on configuration).
2076 * @param string $url URL to extract the domain from for rel =>
2077 * nofollow if appropriate
2078 * @return array Associative array of HTML attributes
2080 public function getExternalLinkAttribs( $url ) {
2082 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2084 $target = $this->mOptions
->getExternalLinkTarget();
2086 $attribs['target'] = $target;
2087 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2088 // T133507. New windows can navigate parent cross-origin.
2089 // Including noreferrer due to lacking browser
2090 // support of noopener. Eventually noreferrer should be removed.
2091 if ( $rel !== '' ) {
2094 $rel .= 'noreferrer noopener';
2097 $attribs['rel'] = $rel;
2102 * Replace unusual escape codes in a URL with their equivalent characters
2104 * This generally follows the syntax defined in RFC 3986, with special
2105 * consideration for HTTP query strings.
2107 * @param string $url
2110 public static function normalizeLinkUrl( $url ) {
2111 # Test for RFC 3986 IPv6 syntax
2112 $scheme = '[a-z][a-z0-9+.-]*:';
2113 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2114 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2115 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2116 IP
::isValid( rawurldecode( $m[1] ) )
2118 $isIPv6 = rawurldecode( $m[1] );
2123 # Make sure unsafe characters are encoded
2124 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2126 return rawurlencode( $m[0] );
2132 $end = strlen( $url );
2134 # Fragment part - 'fragment'
2135 $start = strpos( $url, '#' );
2136 if ( $start !== false && $start < $end ) {
2137 $ret = self
::normalizeUrlComponent(
2138 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2142 # Query part - 'query' minus &=+;
2143 $start = strpos( $url, '?' );
2144 if ( $start !== false && $start < $end ) {
2145 $ret = self
::normalizeUrlComponent(
2146 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2150 # Scheme and path part - 'pchar'
2151 # (we assume no userinfo or encoded colons in the host)
2152 $ret = self
::normalizeUrlComponent(
2153 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2156 if ( $isIPv6 !== false ) {
2157 $ipv6Host = "%5B({$isIPv6})%5D";
2158 $ret = preg_replace(
2159 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2168 private static function normalizeUrlComponent( $component, $unsafe ) {
2169 $callback = function ( $matches ) use ( $unsafe ) {
2170 $char = urldecode( $matches[0] );
2171 $ord = ord( $char );
2172 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2176 # Leave it escaped, but use uppercase for a-f
2177 return strtoupper( $matches[0] );
2180 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2184 * make an image if it's allowed, either through the global
2185 * option, through the exception, or through the on-wiki whitelist
2187 * @param string $url
2191 private function maybeMakeExternalImage( $url ) {
2192 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2193 $imagesexception = !empty( $imagesfrom );
2195 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2196 if ( $imagesexception && is_array( $imagesfrom ) ) {
2197 $imagematch = false;
2198 foreach ( $imagesfrom as $match ) {
2199 if ( strpos( $url, $match ) === 0 ) {
2204 } elseif ( $imagesexception ) {
2205 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2207 $imagematch = false;
2210 if ( $this->mOptions
->getAllowExternalImages()
2211 ||
( $imagesexception && $imagematch )
2213 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2215 $text = Linker
::makeExternalImage( $url );
2218 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2219 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2221 $whitelist = explode(
2223 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2226 foreach ( $whitelist as $entry ) {
2227 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2228 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2231 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2232 # Image matches a whitelist entry
2233 $text = Linker
::makeExternalImage( $url );
2242 * Process [[ ]] wikilinks
2246 * @return string Processed text
2250 public function replaceInternalLinks( $s ) {
2251 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2256 * Process [[ ]] wikilinks (RIL)
2258 * @throws MWException
2259 * @return LinkHolderArray
2263 public function replaceInternalLinks2( &$s ) {
2264 static $tc = false, $e1, $e1_img;
2265 # the % is needed to support urlencoded titles as well
2267 $tc = Title
::legalChars() . '#%';
2268 # Match a link having the form [[namespace:link|alternate]]trail
2269 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2270 # Match cases where there is no "]]", which might still be images
2271 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2274 $holders = new LinkHolderArray( $this );
2276 # split the entire text string on occurrences of [[
2277 $a = StringUtils
::explode( '[[', ' ' . $s );
2278 # get the first element (all text up to first [[), and remove the space we added
2281 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2282 $s = substr( $s, 1 );
2284 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2286 if ( $useLinkPrefixExtension ) {
2287 # Match the end of a line for a word that's not followed by whitespace,
2288 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2289 $charset = $this->contLang
->linkPrefixCharset();
2290 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2293 if ( is_null( $this->mTitle
) ) {
2294 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2296 $nottalk = !$this->mTitle
->isTalkPage();
2298 if ( $useLinkPrefixExtension ) {
2300 if ( preg_match( $e2, $s, $m ) ) {
2301 $first_prefix = $m[2];
2303 $first_prefix = false;
2309 $useSubpages = $this->areSubpagesAllowed();
2311 # Loop for each link
2312 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2313 # Check for excessive memory usage
2314 if ( $holders->isBig() ) {
2316 # Do the existence check, replace the link holders and clear the array
2317 $holders->replace( $s );
2321 if ( $useLinkPrefixExtension ) {
2322 if ( preg_match( $e2, $s, $m ) ) {
2323 list( , $s, $prefix ) = $m;
2328 if ( $first_prefix ) {
2329 $prefix = $first_prefix;
2330 $first_prefix = false;
2334 $might_be_img = false;
2336 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2338 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2339 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2340 # the real problem is with the $e1 regex
2342 # Still some problems for cases where the ] is meant to be outside punctuation,
2343 # and no image is in sight. See T4095.
2345 && substr( $m[3], 0, 1 ) === ']'
2346 && strpos( $text, '[' ) !== false
2348 $text .= ']'; # so that replaceExternalLinks($text) works later
2349 $m[3] = substr( $m[3], 1 );
2351 # fix up urlencoded title texts
2352 if ( strpos( $m[1], '%' ) !== false ) {
2353 # Should anchors '#' also be rejected?
2354 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2357 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2358 # Invalid, but might be an image with a link in its caption
2359 $might_be_img = true;
2361 if ( strpos( $m[1], '%' ) !== false ) {
2362 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2365 } else { # Invalid form; output directly
2366 $s .= $prefix . '[[' . $line;
2370 $origLink = ltrim( $m[1], ' ' );
2372 # Don't allow internal links to pages containing
2373 # PROTO: where PROTO is a valid URL protocol; these
2374 # should be external links.
2375 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2376 $s .= $prefix . '[[' . $line;
2380 # Make subpage if necessary
2381 if ( $useSubpages ) {
2382 $link = $this->maybeDoSubpageLink( $origLink, $text );
2387 // \x7f isn't a default legal title char, so most likely strip
2388 // markers will force us into the "invalid form" path above. But,
2389 // just in case, let's assert that xmlish tags aren't valid in
2390 // the title position.
2391 $unstrip = $this->mStripState
->killMarkers( $link );
2392 $noMarkers = ( $unstrip === $link );
2394 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2395 if ( $nt === null ) {
2396 $s .= $prefix . '[[' . $line;
2400 $ns = $nt->getNamespace();
2401 $iw = $nt->getInterwiki();
2403 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2405 if ( $might_be_img ) { # if this is actually an invalid link
2406 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2409 # look at the next 'line' to see if we can close it there
2411 $next_line = $a->current();
2412 if ( $next_line === false ||
$next_line === null ) {
2415 $m = explode( ']]', $next_line, 3 );
2416 if ( count( $m ) == 3 ) {
2417 # the first ]] closes the inner link, the second the image
2419 $text .= "[[{$m[0]}]]{$m[1]}";
2422 } elseif ( count( $m ) == 2 ) {
2423 # if there's exactly one ]] that's fine, we'll keep looking
2424 $text .= "[[{$m[0]}]]{$m[1]}";
2426 # if $next_line is invalid too, we need look no further
2427 $text .= '[[' . $next_line;
2432 # we couldn't find the end of this imageLink, so output it raw
2433 # but don't ignore what might be perfectly normal links in the text we've examined
2434 $holders->merge( $this->replaceInternalLinks2( $text ) );
2435 $s .= "{$prefix}[[$link|$text";
2436 # note: no $trail, because without an end, there *is* no trail
2439 } else { # it's not an image, so output it raw
2440 $s .= "{$prefix}[[$link|$text";
2441 # note: no $trail, because without an end, there *is* no trail
2446 $wasblank = ( $text == '' );
2450 # Strip off leading ':'
2451 $text = substr( $text, 1 );
2454 # T6598 madness. Handle the quotes only if they come from the alternate part
2455 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2456 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2457 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2458 $text = $this->doQuotes( $text );
2461 # Link not escaped by : , create the various objects
2462 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2465 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2466 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2467 in_array( $iw, $this->svcOptions
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2470 # T26502: filter duplicates
2471 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2472 $this->mLangLinkLanguages
[$iw] = true;
2473 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2477 * Strip the whitespace interwiki links produce, see T10897
2479 $s = rtrim( $s . $prefix ) . $trail; # T175416
2483 if ( $ns == NS_FILE
) {
2484 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2486 # if no parameters were passed, $text
2487 # becomes something like "File:Foo.png",
2488 # which we don't want to pass on to the
2492 # recursively parse links inside the image caption
2493 # actually, this will parse them in any other parameters, too,
2494 # but it might be hard to fix that, and it doesn't matter ATM
2495 $text = $this->replaceExternalLinks( $text );
2496 $holders->merge( $this->replaceInternalLinks2( $text ) );
2498 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2499 $s .= $prefix . $this->armorLinks(
2500 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2503 } elseif ( $ns == NS_CATEGORY
) {
2505 * Strip the whitespace Category links produce, see T2087
2507 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2510 $sortkey = $this->getDefaultSort();
2514 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2515 $sortkey = str_replace( "\n", '', $sortkey );
2516 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2517 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2523 # Self-link checking. For some languages, variants of the title are checked in
2524 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2525 # for linking to a different variant.
2526 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2527 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2531 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2532 # @todo FIXME: Should do batch file existence checks, see comment below
2533 if ( $ns == NS_MEDIA
) {
2534 # Give extensions a chance to select the file revision for us
2537 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2538 [ $this, $nt, &$options, &$descQuery ] );
2539 # Fetch and register the file (file title may be different via hooks)
2540 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2541 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2542 $s .= $prefix . $this->armorLinks(
2543 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2547 # Some titles, such as valid special pages or files in foreign repos, should
2548 # be shown as bluelinks even though they're not included in the page table
2549 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2550 # batch file existence checks for NS_FILE and NS_MEDIA
2551 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2552 $this->mOutput
->addLink( $nt );
2553 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2555 # Links will be added to the output link list after checking
2556 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2563 * Render a forced-blue link inline; protect against double expansion of
2564 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2565 * Since this little disaster has to split off the trail text to avoid
2566 * breaking URLs in the following text without breaking trails on the
2567 * wiki links, it's been made into a horrible function.
2570 * @param string $text
2571 * @param string $trail
2572 * @param string $prefix
2573 * @return string HTML-wikitext mix oh yuck
2575 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2576 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2578 if ( $text == '' ) {
2579 $text = htmlspecialchars( $nt->getPrefixedText() );
2582 $link = $this->getLinkRenderer()->makeKnownLink(
2583 $nt, new HtmlArmor( "$prefix$text$inside" )
2586 return $this->armorLinks( $link ) . $trail;
2590 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2591 * going to go through further parsing steps before inline URL expansion.
2593 * Not needed quite as much as it used to be since free links are a bit
2594 * more sensible these days. But bracketed links are still an issue.
2596 * @param string $text More-or-less HTML
2597 * @return string Less-or-more HTML with NOPARSE bits
2599 public function armorLinks( $text ) {
2600 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2601 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2605 * Return true if subpage links should be expanded on this page.
2608 public function areSubpagesAllowed() {
2609 # Some namespaces don't allow subpages
2610 return $this->nsInfo
->hasSubpages( $this->mTitle
->getNamespace() );
2614 * Handle link to subpage if necessary
2616 * @param string $target The source of the link
2617 * @param string &$text The link text, modified as necessary
2618 * @return string The full name of the link
2621 public function maybeDoSubpageLink( $target, &$text ) {
2622 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2626 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2628 * @param string $text
2629 * @param bool $linestart Whether or not this is at the start of a line.
2631 * @return string The lists rendered as HTML
2633 public function doBlockLevels( $text, $linestart ) {
2634 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2638 * Return value of a magic variable (like PAGENAME)
2642 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2643 * @param bool|PPFrame $frame
2645 * @throws MWException
2648 public function getVariableValue( $index, $frame = false ) {
2649 if ( is_null( $this->mTitle
) ) {
2650 // If no title set, bad things are going to happen
2651 // later. Title should always be set since this
2652 // should only be called in the middle of a parse
2653 // operation (but the unit-tests do funky stuff)
2654 throw new MWException( __METHOD__
. ' Should only be '
2655 . ' called while parsing (no title set)' );
2658 // Avoid PHP 7.1 warning from passing $this by reference
2662 * Some of these require message or data lookups and can be
2663 * expensive to check many times.
2666 Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) &&
2667 isset( $this->mVarCache
[$index] )
2669 return $this->mVarCache
[$index];
2672 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2673 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2675 $pageLang = $this->getFunctionLang();
2681 case 'currentmonth':
2682 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2684 case 'currentmonth1':
2685 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2687 case 'currentmonthname':
2688 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2690 case 'currentmonthnamegen':
2691 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2693 case 'currentmonthabbrev':
2694 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2697 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2700 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2703 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2706 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2708 case 'localmonthname':
2709 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2711 case 'localmonthnamegen':
2712 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2714 case 'localmonthabbrev':
2715 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2718 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2721 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2724 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2727 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2729 case 'fullpagename':
2730 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2732 case 'fullpagenamee':
2733 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2736 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2738 case 'subpagenamee':
2739 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2741 case 'rootpagename':
2742 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2744 case 'rootpagenamee':
2745 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2748 $this->mTitle
->getRootText()
2751 case 'basepagename':
2752 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2754 case 'basepagenamee':
2755 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2758 $this->mTitle
->getBaseText()
2761 case 'talkpagename':
2762 if ( $this->mTitle
->canHaveTalkPage() ) {
2763 $talkPage = $this->mTitle
->getTalkPage();
2764 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2769 case 'talkpagenamee':
2770 if ( $this->mTitle
->canHaveTalkPage() ) {
2771 $talkPage = $this->mTitle
->getTalkPage();
2772 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2777 case 'subjectpagename':
2778 $subjPage = $this->mTitle
->getSubjectPage();
2779 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2781 case 'subjectpagenamee':
2782 $subjPage = $this->mTitle
->getSubjectPage();
2783 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2785 case 'pageid': // requested in T25427
2786 # Inform the edit saving system that getting the canonical output
2787 # after page insertion requires a parse that used that exact page ID
2788 $this->setOutputFlag( 'vary-page-id', '{{PAGEID}} used' );
2789 $value = $this->mTitle
->getArticleID();
2791 $value = $this->mOptions
->getSpeculativePageId();
2793 $this->mOutput
->setSpeculativePageIdUsed( $value );
2799 $this->svcOptions
->get( 'MiserMode' ) &&
2800 !$this->mOptions
->getInterfaceMessage() &&
2801 // @TODO: disallow this word on all namespaces
2802 $this->nsInfo
->isContent( $this->mTitle
->getNamespace() )
2804 // Use a stub result instead of the actual revision ID in order to avoid
2805 // double parses on page save but still allow preview detection (T137900)
2806 if ( $this->getRevisionId() ||
$this->mOptions
->getSpeculativeRevId() ) {
2809 $this->setOutputFlag( 'vary-revision-exists', '{{REVISIONID}} used' );
2813 # Inform the edit saving system that getting the canonical output after
2814 # revision insertion requires a parse that used that exact revision ID
2815 $this->setOutputFlag( 'vary-revision-id', '{{REVISIONID}} used' );
2816 $value = $this->getRevisionId();
2817 if ( $value === 0 ) {
2818 $rev = $this->getRevisionObject();
2819 $value = $rev ?
$rev->getId() : $value;
2822 $value = $this->mOptions
->getSpeculativeRevId();
2824 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2830 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2832 case 'revisionday2':
2833 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2835 case 'revisionmonth':
2836 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2838 case 'revisionmonth1':
2839 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2841 case 'revisionyear':
2842 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2844 case 'revisiontimestamp':
2845 $value = $this->getRevisionTimestampSubstring( 0, 14, self
::MAX_TTS
, $index );
2847 case 'revisionuser':
2848 # Inform the edit saving system that getting the canonical output after
2849 # revision insertion requires a parse that used the actual user ID
2850 $this->setOutputFlag( 'vary-user', '{{REVISIONUSER}} used' );
2851 $value = $this->getRevisionUser();
2853 case 'revisionsize':
2854 $value = $this->getRevisionSize();
2857 $value = str_replace( '_', ' ',
2858 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2861 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2863 case 'namespacenumber':
2864 $value = $this->mTitle
->getNamespace();
2867 $value = $this->mTitle
->canHaveTalkPage()
2868 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2872 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2874 case 'subjectspace':
2875 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2877 case 'subjectspacee':
2878 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2880 case 'currentdayname':
2881 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2884 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2887 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2890 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2893 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2894 # int to remove the padding
2895 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2898 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2900 case 'localdayname':
2901 $value = $pageLang->getWeekdayName(
2902 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2906 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2909 $value = $pageLang->time(
2910 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2916 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2919 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2920 # int to remove the padding
2921 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2924 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2926 case 'numberofarticles':
2927 $value = $pageLang->formatNum( SiteStats
::articles() );
2929 case 'numberoffiles':
2930 $value = $pageLang->formatNum( SiteStats
::images() );
2932 case 'numberofusers':
2933 $value = $pageLang->formatNum( SiteStats
::users() );
2935 case 'numberofactiveusers':
2936 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2938 case 'numberofpages':
2939 $value = $pageLang->formatNum( SiteStats
::pages() );
2941 case 'numberofadmins':
2942 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2944 case 'numberofedits':
2945 $value = $pageLang->formatNum( SiteStats
::edits() );
2947 case 'currenttimestamp':
2948 $value = wfTimestamp( TS_MW
, $ts );
2950 case 'localtimestamp':
2951 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2953 case 'currentversion':
2954 $value = SpecialVersion
::getVersion();
2957 return $this->svcOptions
->get( 'ArticlePath' );
2959 return $this->svcOptions
->get( 'Sitename' );
2961 return $this->svcOptions
->get( 'Server' );
2963 return $this->svcOptions
->get( 'ServerName' );
2965 return $this->svcOptions
->get( 'ScriptPath' );
2967 return $this->svcOptions
->get( 'StylePath' );
2968 case 'directionmark':
2969 return $pageLang->getDirMark();
2970 case 'contentlanguage':
2971 return $this->svcOptions
->get( 'LanguageCode' );
2972 case 'pagelanguage':
2973 $value = $pageLang->getCode();
2975 case 'cascadingsources':
2976 $value = CoreParserFunctions
::cascadingsources( $this );
2981 'ParserGetVariableValueSwitch',
2982 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
2989 $this->mVarCache
[$index] = $value;
2998 * @param int $mtts Max time-till-save; sets vary-revision-timestamp if result changes by then
2999 * @param string $variable Parser variable name
3002 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
3003 # Get the timezone-adjusted timestamp to be used for this revision
3004 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
3005 # Possibly set vary-revision if there is not yet an associated revision
3006 if ( !$this->getRevisionObject() ) {
3007 # Get the timezone-adjusted timestamp $mtts seconds in the future.
3008 # This future is relative to the current time and not that of the
3009 # parser options. The rendered timestamp can be compared to that
3010 # of the timestamp specified by the parser options.
3012 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
3017 if ( $resNow !== $resThen ) {
3018 # Inform the edit saving system that getting the canonical output after
3019 # revision insertion requires a parse that used an actual revision timestamp
3020 $this->setOutputFlag( 'vary-revision-timestamp', "$variable used" );
3028 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3032 public function initialiseVariables() {
3033 $variableIDs = $this->magicWordFactory
->getVariableIDs();
3034 $substIDs = $this->magicWordFactory
->getSubstIDs();
3036 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
3037 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
3041 * Preprocess some wikitext and return the document tree.
3042 * This is the ghost of replace_variables().
3044 * @param string $text The text to parse
3045 * @param int $flags Bitwise combination of:
3046 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3047 * included. Default is to assume a direct page view.
3049 * The generated DOM tree must depend only on the input text and the flags.
3050 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3052 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3053 * change in the DOM tree for a given text, must be passed through the section identifier
3054 * in the section edit link and thus back to extractSections().
3056 * The output of this function is currently only cached in process memory, but a persistent
3057 * cache may be implemented at a later date which takes further advantage of these strict
3058 * dependency requirements.
3062 public function preprocessToDom( $text, $flags = 0 ) {
3063 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3068 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3074 public static function splitWhitespace( $s ) {
3075 $ltrimmed = ltrim( $s );
3076 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3077 $trimmed = rtrim( $ltrimmed );
3078 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3080 $w2 = substr( $ltrimmed, -$diff );
3084 return [ $w1, $trimmed, $w2 ];
3088 * Replace magic variables, templates, and template arguments
3089 * with the appropriate text. Templates are substituted recursively,
3090 * taking care to avoid infinite loops.
3092 * Note that the substitution depends on value of $mOutputType:
3093 * self::OT_WIKI: only {{subst:}} templates
3094 * self::OT_PREPROCESS: templates but not extension tags
3095 * self::OT_HTML: all templates and extension tags
3097 * @param string $text The text to transform
3098 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3099 * template. Arguments may also be provided as an associative array, as
3100 * was the usual case before MW1.12. Providing arguments this way may be
3101 * useful for extensions wishing to perform variable replacement
3103 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3104 * double-brace expansion.
3107 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3108 # Is there any text? Also, Prevent too big inclusions!
3109 $textSize = strlen( $text );
3110 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3114 if ( $frame === false ) {
3115 $frame = $this->getPreprocessor()->newFrame();
3116 } elseif ( !( $frame instanceof PPFrame
) ) {
3117 $this->logger
->debug(
3118 __METHOD__
. " called using plain parameters instead of " .
3119 "a PPFrame instance. Creating custom frame."
3121 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3124 $dom = $this->preprocessToDom( $text );
3125 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3126 $text = $frame->expand( $dom, $flags );
3132 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3134 * @param array $args
3138 public static function createAssocArgs( $args ) {
3141 foreach ( $args as $arg ) {
3142 $eqpos = strpos( $arg, '=' );
3143 if ( $eqpos === false ) {
3144 $assocArgs[$index++
] = $arg;
3146 $name = trim( substr( $arg, 0, $eqpos ) );
3147 $value = trim( substr( $arg, $eqpos +
1 ) );
3148 if ( $value === false ) {
3151 if ( $name !== false ) {
3152 $assocArgs[$name] = $value;
3161 * Warn the user when a parser limitation is reached
3162 * Will warn at most once the user per limitation type
3164 * The results are shown during preview and run through the Parser (See EditPage.php)
3166 * @param string $limitationType Should be one of:
3167 * 'expensive-parserfunction' (corresponding messages:
3168 * 'expensive-parserfunction-warning',
3169 * 'expensive-parserfunction-category')
3170 * 'post-expand-template-argument' (corresponding messages:
3171 * 'post-expand-template-argument-warning',
3172 * 'post-expand-template-argument-category')
3173 * 'post-expand-template-inclusion' (corresponding messages:
3174 * 'post-expand-template-inclusion-warning',
3175 * 'post-expand-template-inclusion-category')
3176 * 'node-count-exceeded' (corresponding messages:
3177 * 'node-count-exceeded-warning',
3178 * 'node-count-exceeded-category')
3179 * 'expansion-depth-exceeded' (corresponding messages:
3180 * 'expansion-depth-exceeded-warning',
3181 * 'expansion-depth-exceeded-category')
3182 * @param string|int|null $current Current value
3183 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3184 * exceeded, provide the values (optional)
3186 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3187 # does no harm if $current and $max are present but are unnecessary for the message
3188 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3189 # only during preview, and that would split the parser cache unnecessarily.
3190 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3192 $this->mOutput
->addWarning( $warning );
3193 $this->addTrackingCategory( "$limitationType-category" );
3197 * Return the text of a template, after recursively
3198 * replacing any variables or templates within the template.
3200 * @param array $piece The parts of the template
3201 * $piece['title']: the title, i.e. the part before the |
3202 * $piece['parts']: the parameter array
3203 * $piece['lineStart']: whether the brace was at the start of a line
3204 * @param PPFrame $frame The current frame, contains template arguments
3206 * @return string|array The text of the template
3208 public function braceSubstitution( $piece, $frame ) {
3211 // $text has been filled
3213 // wiki markup in $text should be escaped
3215 // $text is HTML, armour it against wikitext transformation
3217 // Force interwiki transclusion to be done in raw mode not rendered
3218 $forceRawInterwiki = false;
3219 // $text is a DOM node needing expansion in a child frame
3220 $isChildObj = false;
3221 // $text is a DOM node needing expansion in the current frame
3222 $isLocalObj = false;
3224 # Title object, where $text came from
3227 # $part1 is the bit before the first |, and must contain only title characters.
3228 # Various prefixes will be stripped from it later.
3229 $titleWithSpaces = $frame->expand( $piece['title'] );
3230 $part1 = trim( $titleWithSpaces );
3233 # Original title text preserved for various purposes
3234 $originalTitle = $part1;
3236 # $args is a list of argument nodes, starting from index 0, not including $part1
3237 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3238 # below won't work b/c this $args isn't an object
3239 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3241 $profileSection = null; // profile templates
3245 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3247 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3248 # Decide whether to expand template or keep wikitext as-is.
3249 if ( $this->ot
['wiki'] ) {
3250 if ( $substMatch === false ) {
3251 $literal = true; # literal when in PST with no prefix
3253 $literal = false; # expand when in PST with subst: or safesubst:
3256 if ( $substMatch == 'subst' ) {
3257 $literal = true; # literal when not in PST with plain subst:
3259 $literal = false; # expand when not in PST with safesubst: or no prefix
3263 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3270 if ( !$found && $args->getLength() == 0 ) {
3271 $id = $this->mVariables
->matchStartToEnd( $part1 );
3272 if ( $id !== false ) {
3273 $text = $this->getVariableValue( $id, $frame );
3274 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3275 $this->mOutput
->updateCacheExpiry(
3276 $this->magicWordFactory
->getCacheTTL( $id ) );
3282 # MSG, MSGNW and RAW
3285 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3286 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3289 # Remove obsolete MSG:
3290 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3291 $mwMsg->matchStartAndRemove( $part1 );
3295 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3296 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3297 $forceRawInterwiki = true;
3303 $colonPos = strpos( $part1, ':' );
3304 if ( $colonPos !== false ) {
3305 $func = substr( $part1, 0, $colonPos );
3306 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3307 $argsLength = $args->getLength();
3308 for ( $i = 0; $i < $argsLength; $i++
) {
3309 $funcArgs[] = $args->item( $i );
3312 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3314 // Extract any forwarded flags
3315 if ( isset( $result['title'] ) ) {
3316 $title = $result['title'];
3318 if ( isset( $result['found'] ) ) {
3319 $found = $result['found'];
3321 if ( array_key_exists( 'text', $result ) ) {
3323 $text = $result['text'];
3325 if ( isset( $result['nowiki'] ) ) {
3326 $nowiki = $result['nowiki'];
3328 if ( isset( $result['isHTML'] ) ) {
3329 $isHTML = $result['isHTML'];
3331 if ( isset( $result['forceRawInterwiki'] ) ) {
3332 $forceRawInterwiki = $result['forceRawInterwiki'];
3334 if ( isset( $result['isChildObj'] ) ) {
3335 $isChildObj = $result['isChildObj'];
3337 if ( isset( $result['isLocalObj'] ) ) {
3338 $isLocalObj = $result['isLocalObj'];
3343 # Finish mangling title and then check for loops.
3344 # Set $title to a Title object and $titleText to the PDBK
3347 # Split the title into page and subpage
3349 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3350 if ( $part1 !== $relative ) {
3352 $ns = $this->mTitle
->getNamespace();
3354 $title = Title
::newFromText( $part1, $ns );
3356 $titleText = $title->getPrefixedText();
3357 # Check for language variants if the template is not found
3358 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3359 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3361 # Do recursion depth check
3362 $limit = $this->mOptions
->getMaxTemplateDepth();
3363 if ( $frame->depth
>= $limit ) {
3365 $text = '<span class="error">'
3366 . wfMessage( 'parser-template-recursion-depth-warning' )
3367 ->numParams( $limit )->inContentLanguage()->text()
3373 # Load from database
3374 if ( !$found && $title ) {
3375 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3376 if ( !$title->isExternal() ) {
3377 if ( $title->isSpecialPage()
3378 && $this->mOptions
->getAllowSpecialInclusion()
3379 && $this->ot
['html']
3381 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3382 // Pass the template arguments as URL parameters.
3383 // "uselang" will have no effect since the Language object
3384 // is forced to the one defined in ParserOptions.
3386 $argsLength = $args->getLength();
3387 for ( $i = 0; $i < $argsLength; $i++
) {
3388 $bits = $args->item( $i )->splitArg();
3389 if ( strval( $bits['index'] ) === '' ) {
3390 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3391 $value = trim( $frame->expand( $bits['value'] ) );
3392 $pageArgs[$name] = $value;
3396 // Create a new context to execute the special page
3397 $context = new RequestContext
;
3398 $context->setTitle( $title );
3399 $context->setRequest( new FauxRequest( $pageArgs ) );
3400 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3401 $context->setUser( $this->getUser() );
3403 // If this page is cached, then we better not be per user.
3404 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3406 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3407 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3409 $text = $context->getOutput()->getHTML();
3410 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3413 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3414 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3415 $specialPage->maxIncludeCacheTime()
3419 } elseif ( $this->nsInfo
->isNonincludable( $title->getNamespace() ) ) {
3420 $found = false; # access denied
3421 $this->logger
->debug(
3423 ": template inclusion denied for " . $title->getPrefixedDBkey()
3426 list( $text, $title ) = $this->getTemplateDom( $title );
3427 if ( $text !== false ) {
3433 # If the title is valid but undisplayable, make a link to it
3434 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3435 $text = "[[:$titleText]]";
3438 } elseif ( $title->isTrans() ) {
3439 # Interwiki transclusion
3440 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3441 $text = $this->interwikiTransclude( $title, 'render' );
3444 $text = $this->interwikiTransclude( $title, 'raw' );
3445 # Preprocess it like a template
3446 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3452 # Do infinite loop check
3453 # This has to be done after redirect resolution to avoid infinite loops via redirects
3454 if ( !$frame->loopCheck( $title ) ) {
3456 $text = '<span class="error">'
3457 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3459 $this->addTrackingCategory( 'template-loop-category' );
3460 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3461 wfEscapeWikiText( $titleText ) )->text() );
3462 $this->logger
->debug( __METHOD__
. ": template loop broken at '$titleText'" );
3466 # If we haven't found text to substitute by now, we're done
3467 # Recover the source wikitext and return it
3469 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3470 if ( $profileSection ) {
3471 $this->mProfiler
->scopedProfileOut( $profileSection );
3473 return [ 'object' => $text ];
3476 # Expand DOM-style return values in a child frame
3477 if ( $isChildObj ) {
3478 # Clean up argument array
3479 $newFrame = $frame->newChild( $args, $title );
3482 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3483 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3484 # Expansion is eligible for the empty-frame cache
3485 $text = $newFrame->cachedExpand( $titleText, $text );
3487 # Uncached expansion
3488 $text = $newFrame->expand( $text );
3491 if ( $isLocalObj && $nowiki ) {
3492 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3493 $isLocalObj = false;
3496 if ( $profileSection ) {
3497 $this->mProfiler
->scopedProfileOut( $profileSection );
3500 # Replace raw HTML by a placeholder
3502 $text = $this->insertStripItem( $text );
3503 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3504 # Escape nowiki-style return values
3505 $text = wfEscapeWikiText( $text );
3506 } elseif ( is_string( $text )
3507 && !$piece['lineStart']
3508 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3510 # T2529: if the template begins with a table or block-level
3511 # element, it should be treated as beginning a new line.
3512 # This behavior is somewhat controversial.
3513 $text = "\n" . $text;
3516 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3517 # Error, oversize inclusion
3518 if ( $titleText !== false ) {
3519 # Make a working, properly escaped link if possible (T25588)
3520 $text = "[[:$titleText]]";
3522 # This will probably not be a working link, but at least it may
3523 # provide some hint of where the problem is
3524 preg_replace( '/^:/', '', $originalTitle );
3525 $text = "[[:$originalTitle]]";
3527 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3528 . 'post-expand include size too large -->' );
3529 $this->limitationWarn( 'post-expand-template-inclusion' );
3532 if ( $isLocalObj ) {
3533 $ret = [ 'object' => $text ];
3535 $ret = [ 'text' => $text ];
3542 * Call a parser function and return an array with text and flags.
3544 * The returned array will always contain a boolean 'found', indicating
3545 * whether the parser function was found or not. It may also contain the
3547 * text: string|object, resulting wikitext or PP DOM object
3548 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3549 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3550 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3551 * nowiki: bool, wiki markup in $text should be escaped
3554 * @param PPFrame $frame The current frame, contains template arguments
3555 * @param string $function Function name
3556 * @param array $args Arguments to the function
3557 * @throws MWException
3560 public function callParserFunction( $frame, $function, array $args = [] ) {
3561 # Case sensitive functions
3562 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3563 $function = $this->mFunctionSynonyms
[1][$function];
3565 # Case insensitive functions
3566 $function = $this->contLang
->lc( $function );
3567 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3568 $function = $this->mFunctionSynonyms
[0][$function];
3570 return [ 'found' => false ];
3574 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3576 // Avoid PHP 7.1 warning from passing $this by reference
3579 $allArgs = [ &$parser ];
3580 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3581 # Convert arguments to PPNodes and collect for appending to $allArgs
3583 foreach ( $args as $k => $v ) {
3584 if ( $v instanceof PPNode ||
$k === 0 ) {
3587 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3591 # Add a frame parameter, and pass the arguments as an array
3592 $allArgs[] = $frame;
3593 $allArgs[] = $funcArgs;
3595 # Convert arguments to plain text and append to $allArgs
3596 foreach ( $args as $k => $v ) {
3597 if ( $v instanceof PPNode
) {
3598 $allArgs[] = trim( $frame->expand( $v ) );
3599 } elseif ( is_int( $k ) && $k >= 0 ) {
3600 $allArgs[] = trim( $v );
3602 $allArgs[] = trim( "$k=$v" );
3607 $result = $callback( ...$allArgs );
3609 # The interface for function hooks allows them to return a wikitext
3610 # string or an array containing the string and any flags. This mungs
3611 # things around to match what this method should return.
3612 if ( !is_array( $result ) ) {
3618 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3619 $result['text'] = $result[0];
3621 unset( $result[0] );
3628 $preprocessFlags = 0;
3629 if ( isset( $result['noparse'] ) ) {
3630 $noparse = $result['noparse'];
3632 if ( isset( $result['preprocessFlags'] ) ) {
3633 $preprocessFlags = $result['preprocessFlags'];
3637 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3638 $result['isChildObj'] = true;
3645 * Get the semi-parsed DOM representation of a template with a given title,
3646 * and its redirect destination title. Cached.
3648 * @param Title $title
3652 public function getTemplateDom( $title ) {
3653 $cacheTitle = $title;
3654 $titleText = $title->getPrefixedDBkey();
3656 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3657 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3658 $title = Title
::makeTitle( $ns, $dbk );
3659 $titleText = $title->getPrefixedDBkey();
3661 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3662 return [ $this->mTplDomCache
[$titleText], $title ];
3665 # Cache miss, go to the database
3666 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3668 if ( $text === false ) {
3669 $this->mTplDomCache
[$titleText] = false;
3670 return [ false, $title ];
3673 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3674 $this->mTplDomCache
[$titleText] = $dom;
3676 if ( !$title->equals( $cacheTitle ) ) {
3677 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3678 [ $title->getNamespace(), $title->getDBkey() ];
3681 return [ $dom, $title ];
3685 * Fetch the current revision of a given title. Note that the revision
3686 * (and even the title) may not exist in the database, so everything
3687 * contributing to the output of the parser should use this method
3688 * where possible, rather than getting the revisions themselves. This
3689 * method also caches its results, so using it benefits performance.
3692 * @param Title $title
3695 public function fetchCurrentRevisionOfTitle( $title ) {
3696 $cacheKey = $title->getPrefixedDBkey();
3697 if ( !$this->currentRevisionCache
) {
3698 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3700 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3701 $this->currentRevisionCache
->set( $cacheKey,
3702 // Defaults to Parser::statelessFetchRevision()
3703 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3706 return $this->currentRevisionCache
->get( $cacheKey );
3710 * @param Title $title
3714 public function isCurrentRevisionOfTitleCached( $title ) {
3716 $this->currentRevisionCache
&&
3717 $this->currentRevisionCache
->has( $title->getPrefixedText() )
3722 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3723 * without passing them on to it.
3726 * @param Title $title
3727 * @param Parser|bool $parser
3728 * @return Revision|bool False if missing
3730 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3731 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3737 * Fetch the unparsed text of a template and register a reference to it.
3738 * @param Title $title
3739 * @return array ( string or false, Title )
3741 public function fetchTemplateAndTitle( $title ) {
3742 // Defaults to Parser::statelessFetchTemplate()
3743 $templateCb = $this->mOptions
->getTemplateCallback();
3744 $stuff = call_user_func( $templateCb, $title, $this );
3745 // We use U+007F DELETE to distinguish strip markers from regular text.
3746 $text = $stuff['text'];
3747 if ( is_string( $stuff['text'] ) ) {
3748 $text = strtr( $text, "\x7f", "?" );
3750 $finalTitle = $stuff['finalTitle'] ??
$title;
3751 if ( isset( $stuff['deps'] ) ) {
3752 foreach ( $stuff['deps'] as $dep ) {
3753 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3754 if ( $dep['title']->equals( $this->getTitle() ) ) {
3755 // Self-transclusion; final result may change based on the new page version
3756 $this->setOutputFlag( 'vary-revision', 'Self transclusion' );
3760 return [ $text, $finalTitle ];
3764 * Fetch the unparsed text of a template and register a reference to it.
3765 * @param Title $title
3766 * @return string|bool
3768 public function fetchTemplate( $title ) {
3769 return $this->fetchTemplateAndTitle( $title )[0];
3773 * Static function to get a template
3774 * Can be overridden via ParserOptions::setTemplateCallback().
3776 * @param Title $title
3777 * @param bool|Parser $parser
3781 public static function statelessFetchTemplate( $title, $parser = false ) {
3782 $text = $skip = false;
3783 $finalTitle = $title;
3786 # Loop to fetch the article, with up to 1 redirect
3787 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3788 # Give extensions a chance to select the revision instead
3789 $id = false; # Assume current
3790 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3791 [ $parser, $title, &$skip, &$id ] );
3797 'page_id' => $title->getArticleID(),
3804 $rev = Revision
::newFromId( $id );
3805 } elseif ( $parser ) {
3806 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3808 $rev = Revision
::newFromTitle( $title );
3810 $rev_id = $rev ?
$rev->getId() : 0;
3811 # If there is no current revision, there is no page
3812 if ( $id === false && !$rev ) {
3813 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3814 $linkCache->addBadLinkObj( $title );
3819 'page_id' => $title->getArticleID(),
3820 'rev_id' => $rev_id ];
3821 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3822 # We fetched a rev from a different title; register it too...
3824 'title' => $rev->getTitle(),
3825 'page_id' => $rev->getPage(),
3826 'rev_id' => $rev_id ];
3830 $content = $rev->getContent();
3831 $text = $content ?
$content->getWikitextForTransclusion() : null;
3833 Hooks
::run( 'ParserFetchTemplate',
3834 [ $parser, $title, $rev, &$text, &$deps ] );
3836 if ( $text === false ||
$text === null ) {
3840 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3841 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3842 lcfirst( $title->getText() ) )->inContentLanguage();
3843 if ( !$message->exists() ) {
3847 $content = $message->content();
3848 $text = $message->plain();
3856 $finalTitle = $title;
3857 $title = $content->getRedirectTarget();
3861 'finalTitle' => $finalTitle,
3866 * Fetch a file and its title and register a reference to it.
3867 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3868 * @param Title $title
3869 * @param array $options Array of options to RepoGroup::findFile
3870 * @return array ( File or false, Title of file )
3872 public function fetchFileAndTitle( $title, $options = [] ) {
3873 $file = $this->fetchFileNoRegister( $title, $options );
3875 $time = $file ?
$file->getTimestamp() : false;
3876 $sha1 = $file ?
$file->getSha1() : false;
3877 # Register the file as a dependency...
3878 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3879 if ( $file && !$title->equals( $file->getTitle() ) ) {
3880 # Update fetched file title
3881 $title = $file->getTitle();
3882 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3884 return [ $file, $title ];
3888 * Helper function for fetchFileAndTitle.
3890 * Also useful if you need to fetch a file but not use it yet,
3891 * for example to get the file's handler.
3893 * @param Title $title
3894 * @param array $options Array of options to RepoGroup::findFile
3897 protected function fetchFileNoRegister( $title, $options = [] ) {
3898 if ( isset( $options['broken'] ) ) {
3899 $file = false; // broken thumbnail forced by hook
3900 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3901 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3902 } else { // get by (name,timestamp)
3903 $file = MediaWikiServices
::getInstance()->getRepoGroup()->findFile( $title, $options );
3909 * Transclude an interwiki link.
3911 * @param Title $title
3912 * @param string $action Usually one of (raw, render)
3916 public function interwikiTransclude( $title, $action ) {
3917 if ( !$this->svcOptions
->get( 'EnableScaryTranscluding' ) ) {
3918 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3921 $url = $title->getFullURL( [ 'action' => $action ] );
3922 if ( strlen( $url ) > 1024 ) {
3923 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3926 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3928 $fname = __METHOD__
;
3929 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3931 $data = $cache->getWithSetCallback(
3932 $cache->makeGlobalKey(
3933 'interwiki-transclude',
3934 ( $wikiId !== false ) ?
$wikiId : 'external',
3937 $this->svcOptions
->get( 'TranscludeCacheExpiry' ),
3938 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3939 $req = MWHttpRequest
::factory( $url, [], $fname );
3941 $status = $req->execute(); // Status object
3942 if ( !$status->isOK() ) {
3943 $ttl = $cache::TTL_UNCACHEABLE
;
3944 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3945 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3949 'text' => $status->isOK() ?
$req->getContent() : null,
3950 'code' => $req->getStatus()
3954 'checkKeys' => ( $wikiId !== false )
3955 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3957 'pcGroup' => 'interwiki-transclude:5',
3958 'pcTTL' => $cache::TTL_PROC_LONG
3962 if ( is_string( $data['text'] ) ) {
3963 $text = $data['text'];
3964 } elseif ( $data['code'] != 200 ) {
3965 // Though we failed to fetch the content, this status is useless.
3966 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3967 ->params( $url, $data['code'] )->inContentLanguage()->text();
3969 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3976 * Triple brace replacement -- used for template arguments
3979 * @param array $piece
3980 * @param PPFrame $frame
3984 public function argSubstitution( $piece, $frame ) {
3986 $parts = $piece['parts'];
3987 $nameWithSpaces = $frame->expand( $piece['title'] );
3988 $argName = trim( $nameWithSpaces );
3990 $text = $frame->getArgument( $argName );
3991 if ( $text === false && $parts->getLength() > 0
3992 && ( $this->ot
['html']
3994 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3997 # No match in frame, use the supplied default
3998 $object = $parts->item( 0 )->getChildren();
4000 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4001 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4002 $this->limitationWarn( 'post-expand-template-argument' );
4005 if ( $text === false && $object === false ) {
4007 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4009 if ( $error !== false ) {
4012 if ( $object !== false ) {
4013 $ret = [ 'object' => $object ];
4015 $ret = [ 'text' => $text ];
4022 * Return the text to be used for a given extension tag.
4023 * This is the ghost of strip().
4025 * @param array $params Associative array of parameters:
4026 * name PPNode for the tag name
4027 * attr PPNode for unparsed text where tag attributes are thought to be
4028 * attributes Optional associative array of parsed attributes
4029 * inner Contents of extension element
4030 * noClose Original text did not have a close tag
4031 * @param PPFrame $frame
4033 * @throws MWException
4036 public function extensionSubstitution( $params, $frame ) {
4037 static $errorStr = '<span class="error">';
4038 static $errorLen = 20;
4040 $name = $frame->expand( $params['name'] );
4041 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4042 // Probably expansion depth or node count exceeded. Just punt the
4047 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
4048 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4053 // We can't safely check if the expansion for $content resulted in an
4054 // error, because the content could happen to be the error string
4056 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
4058 $marker = self
::MARKER_PREFIX
. "-$name-"
4059 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
4061 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
4062 ( $this->ot
['html'] ||
$this->ot
['pre'] );
4063 if ( $isFunctionTag ) {
4064 $markerType = 'none';
4066 $markerType = 'general';
4068 if ( $this->ot
['html'] ||
$isFunctionTag ) {
4069 $name = strtolower( $name );
4070 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
4071 if ( isset( $params['attributes'] ) ) {
4072 $attributes +
= $params['attributes'];
4075 if ( isset( $this->mTagHooks
[$name] ) ) {
4076 $output = call_user_func_array( $this->mTagHooks
[$name],
4077 [ $content, $attributes, $this, $frame ] );
4078 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4079 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4081 // Avoid PHP 7.1 warning from passing $this by reference
4083 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4085 $output = '<span class="error">Invalid tag extension name: ' .
4086 htmlspecialchars( $name ) . '</span>';
4089 if ( is_array( $output ) ) {
4092 $output = $flags[0];
4093 if ( isset( $flags['markerType'] ) ) {
4094 $markerType = $flags['markerType'];
4098 if ( is_null( $attrText ) ) {
4101 if ( isset( $params['attributes'] ) ) {
4102 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4103 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4104 htmlspecialchars( $attrValue ) . '"';
4107 if ( $content === null ) {
4108 $output = "<$name$attrText/>";
4110 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4111 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4115 $output = "<$name$attrText>$content$close";
4119 if ( $markerType === 'none' ) {
4121 } elseif ( $markerType === 'nowiki' ) {
4122 $this->mStripState
->addNoWiki( $marker, $output );
4123 } elseif ( $markerType === 'general' ) {
4124 $this->mStripState
->addGeneral( $marker, $output );
4126 throw new MWException( __METHOD__
. ': invalid marker type' );
4132 * Increment an include size counter
4134 * @param string $type The type of expansion
4135 * @param int $size The size of the text
4136 * @return bool False if this inclusion would take it over the maximum, true otherwise
4138 public function incrementIncludeSize( $type, $size ) {
4139 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4142 $this->mIncludeSizes
[$type] +
= $size;
4148 * Increment the expensive function count
4150 * @return bool False if the limit has been exceeded
4152 public function incrementExpensiveFunctionCount() {
4153 $this->mExpensiveFunctionCount++
;
4154 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4158 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4159 * Fills $this->mDoubleUnderscores, returns the modified text
4161 * @param string $text
4165 public function doDoubleUnderscore( $text ) {
4166 # The position of __TOC__ needs to be recorded
4167 $mw = $this->magicWordFactory
->get( 'toc' );
4168 if ( $mw->match( $text ) ) {
4169 $this->mShowToc
= true;
4170 $this->mForceTocPosition
= true;
4172 # Set a placeholder. At the end we'll fill it in with the TOC.
4173 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4175 # Only keep the first one.
4176 $text = $mw->replace( '', $text );
4179 # Now match and remove the rest of them
4180 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4181 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4183 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4184 $this->mOutput
->mNoGallery
= true;
4186 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4187 $this->mShowToc
= false;
4189 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4190 && $this->mTitle
->getNamespace() == NS_CATEGORY
4192 $this->addTrackingCategory( 'hidden-category-category' );
4194 # (T10068) Allow control over whether robots index a page.
4195 # __INDEX__ always overrides __NOINDEX__, see T16899
4196 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4197 $this->mOutput
->setIndexPolicy( 'noindex' );
4198 $this->addTrackingCategory( 'noindex-category' );
4200 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4201 $this->mOutput
->setIndexPolicy( 'index' );
4202 $this->addTrackingCategory( 'index-category' );
4205 # Cache all double underscores in the database
4206 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4207 $this->mOutput
->setProperty( $key, '' );
4214 * @see ParserOutput::addTrackingCategory()
4215 * @param string $msg Message key
4216 * @return bool Whether the addition was successful
4218 public function addTrackingCategory( $msg ) {
4219 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4223 * This function accomplishes several tasks:
4224 * 1) Auto-number headings if that option is enabled
4225 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4226 * 3) Add a Table of contents on the top for users who have enabled the option
4227 * 4) Auto-anchor headings
4229 * It loops through all headlines, collects the necessary data, then splits up the
4230 * string and re-inserts the newly formatted headlines.
4232 * @param string $text
4233 * @param string $origText Original, untouched wikitext
4234 * @param bool $isMain
4235 * @return mixed|string
4238 public function formatHeadings( $text, $origText, $isMain = true ) {
4239 # Inhibit editsection links if requested in the page
4240 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4241 $maybeShowEditLink = false;
4243 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4246 # Get all headlines for numbering them and adding funky stuff like [edit]
4247 # links - this is for later, but we need the number of headlines right now
4248 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4249 # be trimmed here since whitespace in HTML headings is significant.
4251 $numMatches = preg_match_all(
4252 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4257 # if there are fewer than 4 headlines in the article, do not show TOC
4258 # unless it's been explicitly enabled.
4259 $enoughToc = $this->mShowToc
&&
4260 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4262 # Allow user to stipulate that a page should have a "new section"
4263 # link added via __NEWSECTIONLINK__
4264 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4265 $this->mOutput
->setNewSection( true );
4268 # Allow user to remove the "new section"
4269 # link via __NONEWSECTIONLINK__
4270 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4271 $this->mOutput
->hideNewSection( true );
4274 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4275 # override above conditions and always show TOC above first header
4276 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4277 $this->mShowToc
= true;
4285 # Ugh .. the TOC should have neat indentation levels which can be
4286 # passed to the skin functions. These are determined here
4290 $sublevelCount = [];
4296 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4297 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4298 $oldType = $this->mOutputType
;
4299 $this->setOutputType( self
::OT_WIKI
);
4300 $frame = $this->getPreprocessor()->newFrame();
4301 $root = $this->preprocessToDom( $origText );
4302 $node = $root->getFirstChild();
4307 $headlines = $numMatches !== false ?
$matches[3] : [];
4309 $maxTocLevel = $this->svcOptions
->get( 'MaxTocLevel' );
4310 foreach ( $headlines as $headline ) {
4311 $isTemplate = false;
4313 $sectionIndex = false;
4315 $markerMatches = [];
4316 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4317 $serial = $markerMatches[1];
4318 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4319 $isTemplate = ( $titleText != $baseTitleText );
4320 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4324 $prevlevel = $level;
4326 $level = $matches[1][$headlineCount];
4328 if ( $level > $prevlevel ) {
4329 # Increase TOC level
4331 $sublevelCount[$toclevel] = 0;
4332 if ( $toclevel < $maxTocLevel ) {
4333 $prevtoclevel = $toclevel;
4334 $toc .= Linker
::tocIndent();
4337 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4338 # Decrease TOC level, find level to jump to
4340 for ( $i = $toclevel; $i > 0; $i-- ) {
4341 if ( $levelCount[$i] == $level ) {
4342 # Found last matching level
4345 } elseif ( $levelCount[$i] < $level ) {
4346 # Found first matching level below current level
4354 if ( $toclevel < $maxTocLevel ) {
4355 if ( $prevtoclevel < $maxTocLevel ) {
4356 # Unindent only if the previous toc level was shown :p
4357 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4358 $prevtoclevel = $toclevel;
4360 $toc .= Linker
::tocLineEnd();
4364 # No change in level, end TOC line
4365 if ( $toclevel < $maxTocLevel ) {
4366 $toc .= Linker
::tocLineEnd();
4370 $levelCount[$toclevel] = $level;
4372 # count number of headlines for each level
4373 $sublevelCount[$toclevel]++
;
4375 for ( $i = 1; $i <= $toclevel; $i++
) {
4376 if ( !empty( $sublevelCount[$i] ) ) {
4380 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4385 # The safe header is a version of the header text safe to use for links
4387 # Remove link placeholders by the link text.
4388 # <!--LINK number-->
4390 # link text with suffix
4391 # Do this before unstrip since link text can contain strip markers
4392 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4394 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4395 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4397 # Remove any <style> or <script> tags (T198618)
4398 $safeHeadline = preg_replace(
4399 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4404 # Strip out HTML (first regex removes any tag not allowed)
4406 # * <sup> and <sub> (T10393)
4410 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4411 # * <s> and <strike> (T35715)
4412 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4413 # to allow setting directionality in toc items.
4414 $tocline = preg_replace(
4416 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4417 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4423 # Strip '<span></span>', which is the result from the above if
4424 # <span id="foo"></span> is used to produce an additional anchor
4426 $tocline = str_replace( '<span></span>', '', $tocline );
4428 $tocline = trim( $tocline );
4430 # For the anchor, strip out HTML-y stuff period
4431 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4432 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4434 # Save headline for section edit hint before it's escaped
4435 $headlineHint = $safeHeadline;
4437 # Decode HTML entities
4438 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4440 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4442 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4443 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4444 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4445 if ( $fallbackHeadline === $safeHeadline ) {
4446 # No reason to have both (in fact, we can't)
4447 $fallbackHeadline = false;
4450 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4451 # @todo FIXME: We may be changing them depending on the current locale.
4452 $arrayKey = strtolower( $safeHeadline );
4453 if ( $fallbackHeadline === false ) {
4454 $fallbackArrayKey = false;
4456 $fallbackArrayKey = strtolower( $fallbackHeadline );
4459 # Create the anchor for linking from the TOC to the section
4460 $anchor = $safeHeadline;
4461 $fallbackAnchor = $fallbackHeadline;
4462 if ( isset( $refers[$arrayKey] ) ) {
4463 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4464 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4466 $linkAnchor .= "_$i";
4467 $refers["${arrayKey}_$i"] = true;
4469 $refers[$arrayKey] = true;
4471 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4472 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4473 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4474 $fallbackAnchor .= "_$i";
4475 $refers["${fallbackArrayKey}_$i"] = true;
4477 $refers[$fallbackArrayKey] = true;
4480 # Don't number the heading if it is the only one (looks silly)
4481 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4482 # the two are different if the line contains a link
4483 $headline = Html
::element(
4485 [ 'class' => 'mw-headline-number' ],
4487 ) . ' ' . $headline;
4490 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4491 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4492 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4495 # Add the section to the section tree
4496 # Find the DOM node for this header
4497 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4498 while ( $node && !$noOffset ) {
4499 if ( $node->getName() === 'h' ) {
4500 $bits = $node->splitHeading();
4501 if ( $bits['i'] == $sectionIndex ) {
4505 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4506 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4507 $node = $node->getNextSibling();
4510 'toclevel' => $toclevel,
4513 'number' => $numbering,
4514 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4515 'fromtitle' => $titleText,
4516 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4517 'anchor' => $anchor,
4520 # give headline the correct <h#> tag
4521 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4522 // Output edit section links as markers with styles that can be customized by skins
4523 if ( $isTemplate ) {
4524 # Put a T flag in the section identifier, to indicate to extractSections()
4525 # that sections inside <includeonly> should be counted.
4526 $editsectionPage = $titleText;
4527 $editsectionSection = "T-$sectionIndex";
4528 $editsectionContent = null;
4530 $editsectionPage = $this->mTitle
->getPrefixedText();
4531 $editsectionSection = $sectionIndex;
4532 $editsectionContent = $headlineHint;
4534 // We use a bit of pesudo-xml for editsection markers. The
4535 // language converter is run later on. Using a UNIQ style marker
4536 // leads to the converter screwing up the tokens when it
4537 // converts stuff. And trying to insert strip tags fails too. At
4538 // this point all real inputted tags have already been escaped,
4539 // so we don't have to worry about a user trying to input one of
4540 // these markers directly. We use a page and section attribute
4541 // to stop the language converter from converting these
4542 // important bits of data, but put the headline hint inside a
4543 // content block because the language converter is supposed to
4544 // be able to convert that piece of data.
4545 // Gets replaced with html in ParserOutput::getText
4546 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4547 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4548 if ( $editsectionContent !== null ) {
4549 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4556 $head[$headlineCount] = Linker
::makeHeadline( $level,
4557 $matches['attrib'][$headlineCount], $anchor, $headline,
4558 $editlink, $fallbackAnchor );
4563 $this->setOutputType( $oldType );
4565 # Never ever show TOC if no headers
4566 if ( $numVisible < 1 ) {
4571 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4572 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4574 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4575 $this->mOutput
->setTOCHTML( $toc );
4576 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4580 $this->mOutput
->setSections( $tocraw );
4583 # split up and insert constructed headlines
4584 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4587 // build an array of document sections
4589 foreach ( $blocks as $block ) {
4590 // $head is zero-based, sections aren't.
4591 if ( empty( $head[$i - 1] ) ) {
4592 $sections[$i] = $block;
4594 $sections[$i] = $head[$i - 1] . $block;
4598 * Send a hook, one per section.
4599 * The idea here is to be able to make section-level DIVs, but to do so in a
4600 * lower-impact, more correct way than r50769
4603 * $section : the section number
4604 * &$sectionContent : ref to the content of the section
4605 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4607 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4612 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4613 // append the TOC at the beginning
4614 // Top anchor now in skin
4615 $sections[0] .= $toc . "\n";
4618 $full .= implode( '', $sections );
4620 if ( $this->mForceTocPosition
) {
4621 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4628 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4629 * conversion, substituting signatures, {{subst:}} templates, etc.
4631 * @param string $text The text to transform
4632 * @param Title $title The Title object for the current article
4633 * @param User $user The User object describing the current user
4634 * @param ParserOptions $options Parsing options
4635 * @param bool $clearState Whether to clear the parser state first
4636 * @return string The altered wiki markup
4638 public function preSaveTransform( $text, Title
$title, User
$user,
4639 ParserOptions
$options, $clearState = true
4641 if ( $clearState ) {
4642 $magicScopeVariable = $this->lock();
4644 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4645 $this->setUser( $user );
4647 // Strip U+0000 NULL (T159174)
4648 $text = str_replace( "\000", '', $text );
4650 // We still normalize line endings for backwards-compatibility
4651 // with other code that just calls PST, but this should already
4652 // be handled in TextContent subclasses
4653 $text = TextContent
::normalizeLineEndings( $text );
4655 if ( $options->getPreSaveTransform() ) {
4656 $text = $this->pstPass2( $text, $user );
4658 $text = $this->mStripState
->unstripBoth( $text );
4660 $this->setUser( null ); # Reset
4666 * Pre-save transform helper function
4668 * @param string $text
4673 private function pstPass2( $text, $user ) {
4674 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4675 # $this->contLang here in order to give everyone the same signature and use the default one
4676 # rather than the one selected in each user's preferences. (see also T14815)
4677 $ts = $this->mOptions
->getTimestamp();
4678 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4679 $ts = $timestamp->format( 'YmdHis' );
4680 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4682 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4684 # Variable replacement
4685 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4686 $text = $this->replaceVariables( $text );
4688 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4689 # which may corrupt this parser instance via its wfMessage()->text() call-
4692 if ( strpos( $text, '~~~' ) !== false ) {
4693 $sigText = $this->getUserSig( $user );
4694 $text = strtr( $text, [
4696 '~~~~' => "$sigText $d",
4699 # The main two signature forms used above are time-sensitive
4700 $this->setOutputFlag( 'user-signature', 'User signature detected' );
4703 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4704 $tc = '[' . Title
::legalChars() . ']';
4705 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4707 // [[ns:page (context)|]]
4708 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4709 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4710 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4711 // [[ns:page (context), context|]] (using either single or double-width comma)
4712 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4713 // [[|page]] (reverse pipe trick: add context from page title)
4714 $p2 = "/\[\[\\|($tc+)]]/";
4716 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4717 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4718 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4719 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4721 $t = $this->mTitle
->getText();
4723 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4724 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4725 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4726 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4728 # if there's no context, don't bother duplicating the title
4729 $text = preg_replace( $p2, '[[\\1]]', $text );
4736 * Fetch the user's signature text, if any, and normalize to
4737 * validated, ready-to-insert wikitext.
4738 * If you have pre-fetched the nickname or the fancySig option, you can
4739 * specify them here to save a database query.
4740 * Do not reuse this parser instance after calling getUserSig(),
4741 * as it may have changed.
4743 * @param User &$user
4744 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4745 * @param bool|null $fancySig whether the nicknname is the complete signature
4746 * or null to use default value
4749 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4750 $username = $user->getName();
4752 # If not given, retrieve from the user object.
4753 if ( $nickname === false ) {
4754 $nickname = $user->getOption( 'nickname' );
4757 if ( is_null( $fancySig ) ) {
4758 $fancySig = $user->getBoolOption( 'fancysig' );
4761 $nickname = $nickname == null ?
$username : $nickname;
4763 if ( mb_strlen( $nickname ) > $this->svcOptions
->get( 'MaxSigChars' ) ) {
4764 $nickname = $username;
4765 $this->logger
->debug( __METHOD__
. ": $username has overlong signature." );
4766 } elseif ( $fancySig !== false ) {
4767 # Sig. might contain markup; validate this
4768 if ( $this->validateSig( $nickname ) !== false ) {
4769 # Validated; clean up (if needed) and return it
4770 return $this->cleanSig( $nickname, true );
4772 # Failed to validate; fall back to the default
4773 $nickname = $username;
4774 $this->logger
->debug( __METHOD__
. ": $username has bad XML tags in signature." );
4778 # Make sure nickname doesnt get a sig in a sig
4779 $nickname = self
::cleanSigInSig( $nickname );
4781 # If we're still here, make it a link to the user page
4782 $userText = wfEscapeWikiText( $username );
4783 $nickText = wfEscapeWikiText( $nickname );
4784 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4786 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4787 ->title( $this->getTitle() )->text();
4791 * Check that the user's signature contains no bad XML
4793 * @param string $text
4794 * @return string|bool An expanded string, or false if invalid.
4796 public function validateSig( $text ) {
4797 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4801 * Clean up signature text
4803 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4804 * 2) Substitute all transclusions
4806 * @param string $text
4807 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4808 * @return string Signature text
4810 public function cleanSig( $text, $parsing = false ) {
4813 $magicScopeVariable = $this->lock();
4814 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4817 # Option to disable this feature
4818 if ( !$this->mOptions
->getCleanSignatures() ) {
4822 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4823 # => Move this logic to braceSubstitution()
4824 $substWord = $this->magicWordFactory
->get( 'subst' );
4825 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4826 $substText = '{{' . $substWord->getSynonym( 0 );
4828 $text = preg_replace( $substRegex, $substText, $text );
4829 $text = self
::cleanSigInSig( $text );
4830 $dom = $this->preprocessToDom( $text );
4831 $frame = $this->getPreprocessor()->newFrame();
4832 $text = $frame->expand( $dom );
4835 $text = $this->mStripState
->unstripBoth( $text );
4842 * Strip 3, 4 or 5 tildes out of signatures.
4844 * @param string $text
4845 * @return string Signature text with /~{3,5}/ removed
4847 public static function cleanSigInSig( $text ) {
4848 $text = preg_replace( '/~{3,5}/', '', $text );
4853 * Set up some variables which are usually set up in parse()
4854 * so that an external function can call some class members with confidence
4856 * @param Title|null $title
4857 * @param ParserOptions $options
4858 * @param int $outputType
4859 * @param bool $clearState
4861 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4862 $outputType, $clearState = true
4864 $this->startParse( $title, $options, $outputType, $clearState );
4868 * @param Title|null $title
4869 * @param ParserOptions $options
4870 * @param int $outputType
4871 * @param bool $clearState
4873 private function startParse( Title
$title = null, ParserOptions
$options,
4874 $outputType, $clearState = true
4876 $this->setTitle( $title );
4877 $this->mOptions
= $options;
4878 $this->setOutputType( $outputType );
4879 if ( $clearState ) {
4880 $this->clearState();
4885 * Wrapper for preprocess()
4887 * @param string $text The text to preprocess
4888 * @param ParserOptions $options
4889 * @param Title|null $title Title object or null to use $wgTitle
4892 public function transformMsg( $text, $options, $title = null ) {
4893 static $executing = false;
4895 # Guard against infinite recursion
4906 $text = $this->preprocess( $text, $title, $options );
4913 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4914 * The callback should have the following form:
4915 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4917 * Transform and return $text. Use $parser for any required context, e.g. use
4918 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4920 * Hooks may return extended information by returning an array, of which the
4921 * first numbered element (index 0) must be the return string, and all other
4922 * entries are extracted into local variables within an internal function
4923 * in the Parser class.
4925 * This interface (introduced r61913) appears to be undocumented, but
4926 * 'markerType' is used by some core tag hooks to override which strip
4927 * array their results are placed in. **Use great caution if attempting
4928 * this interface, as it is not documented and injudicious use could smash
4929 * private variables.**
4931 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4932 * @param callable $callback The callback function (and object) to use for the tag
4933 * @throws MWException
4934 * @return callable|null The old value of the mTagHooks array associated with the hook
4936 public function setHook( $tag, callable
$callback ) {
4937 $tag = strtolower( $tag );
4938 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4939 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4941 $oldVal = $this->mTagHooks
[$tag] ??
null;
4942 $this->mTagHooks
[$tag] = $callback;
4943 if ( !in_array( $tag, $this->mStripList
) ) {
4944 $this->mStripList
[] = $tag;
4951 * As setHook(), but letting the contents be parsed.
4953 * Transparent tag hooks are like regular XML-style tag hooks, except they
4954 * operate late in the transformation sequence, on HTML instead of wikitext.
4956 * This is probably obsoleted by things dealing with parser frames?
4957 * The only extension currently using it is geoserver.
4960 * @todo better document or deprecate this
4962 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4963 * @param callable $callback The callback function (and object) to use for the tag
4964 * @throws MWException
4965 * @return callable|null The old value of the mTagHooks array associated with the hook
4967 public function setTransparentTagHook( $tag, callable
$callback ) {
4968 $tag = strtolower( $tag );
4969 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4970 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4972 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4973 $this->mTransparentTagHooks
[$tag] = $callback;
4979 * Remove all tag hooks
4981 public function clearTagHooks() {
4982 $this->mTagHooks
= [];
4983 $this->mFunctionTagHooks
= [];
4984 $this->mStripList
= $this->mDefaultStripList
;
4988 * Create a function, e.g. {{sum:1|2|3}}
4989 * The callback function should have the form:
4990 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4992 * Or with Parser::SFH_OBJECT_ARGS:
4993 * function myParserFunction( $parser, $frame, $args ) { ... }
4995 * The callback may either return the text result of the function, or an array with the text
4996 * in element 0, and a number of flags in the other elements. The names of the flags are
4997 * specified in the keys. Valid flags are:
4998 * found The text returned is valid, stop processing the template. This
5000 * nowiki Wiki markup in the return value should be escaped
5001 * isHTML The returned text is HTML, armour it against wikitext transformation
5003 * @param string $id The magic word ID
5004 * @param callable $callback The callback function (and object) to use
5005 * @param int $flags A combination of the following flags:
5006 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5008 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5009 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5010 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5011 * the arguments, and to control the way they are expanded.
5013 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5014 * arguments, for instance:
5015 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5017 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5018 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5019 * working if/when this is changed.
5021 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5024 * Please read the documentation in includes/parser/Preprocessor.php for more information
5025 * about the methods available in PPFrame and PPNode.
5027 * @throws MWException
5028 * @return string|callable The old callback function for this name, if any
5030 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
5031 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
5032 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
5034 # Add to function cache
5035 $mw = $this->magicWordFactory
->get( $id );
5037 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
5040 $synonyms = $mw->getSynonyms();
5041 $sensitive = intval( $mw->isCaseSensitive() );
5043 foreach ( $synonyms as $syn ) {
5045 if ( !$sensitive ) {
5046 $syn = $this->contLang
->lc( $syn );
5049 if ( !( $flags & self
::SFH_NO_HASH
) ) {
5052 # Remove trailing colon
5053 if ( substr( $syn, -1, 1 ) === ':' ) {
5054 $syn = substr( $syn, 0, -1 );
5056 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
5062 * Get all registered function hook identifiers
5066 public function getFunctionHooks() {
5067 $this->firstCallInit();
5068 return array_keys( $this->mFunctionHooks
);
5072 * Create a tag function, e.g. "<test>some stuff</test>".
5073 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5074 * Unlike parser functions, their content is not preprocessed.
5075 * @param string $tag
5076 * @param callable $callback
5078 * @throws MWException
5081 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5082 $tag = strtolower( $tag );
5083 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5084 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5086 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5087 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5089 if ( !in_array( $tag, $this->mStripList
) ) {
5090 $this->mStripList
[] = $tag;
5097 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5098 * Placeholders created in Linker::link()
5100 * @param string &$text
5101 * @param int $options
5103 public function replaceLinkHolders( &$text, $options = 0 ) {
5104 $this->mLinkHolders
->replace( $text );
5108 * Replace "<!--LINK-->" link placeholders with plain text of links
5109 * (not HTML-formatted).
5111 * @param string $text
5114 public function replaceLinkHoldersText( $text ) {
5115 return $this->mLinkHolders
->replaceText( $text );
5119 * Renders an image gallery from a text with one line per image.
5120 * text labels may be given by using |-style alternative text. E.g.
5121 * Image:one.jpg|The number "1"
5122 * Image:tree.jpg|A tree
5123 * given as text will return the HTML of a gallery with two images,
5124 * labeled 'The number "1"' and
5127 * @param string $text
5128 * @param array $params
5129 * @return string HTML
5131 public function renderImageGallery( $text, $params ) {
5133 if ( isset( $params['mode'] ) ) {
5134 $mode = $params['mode'];
5138 $ig = ImageGalleryBase
::factory( $mode );
5139 } catch ( Exception
$e ) {
5140 // If invalid type set, fallback to default.
5141 $ig = ImageGalleryBase
::factory( false );
5144 $ig->setContextTitle( $this->mTitle
);
5145 $ig->setShowBytes( false );
5146 $ig->setShowDimensions( false );
5147 $ig->setShowFilename( false );
5148 $ig->setParser( $this );
5149 $ig->setHideBadImages();
5150 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5152 if ( isset( $params['showfilename'] ) ) {
5153 $ig->setShowFilename( true );
5155 $ig->setShowFilename( false );
5157 if ( isset( $params['caption'] ) ) {
5158 // NOTE: We aren't passing a frame here or below. Frame info
5159 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5160 // See T107332#4030581
5161 $caption = $this->recursiveTagParse( $params['caption'] );
5162 $ig->setCaptionHtml( $caption );
5164 if ( isset( $params['perrow'] ) ) {
5165 $ig->setPerRow( $params['perrow'] );
5167 if ( isset( $params['widths'] ) ) {
5168 $ig->setWidths( $params['widths'] );
5170 if ( isset( $params['heights'] ) ) {
5171 $ig->setHeights( $params['heights'] );
5173 $ig->setAdditionalOptions( $params );
5175 // Avoid PHP 7.1 warning from passing $this by reference
5177 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5179 $lines = StringUtils
::explode( "\n", $text );
5180 foreach ( $lines as $line ) {
5181 # match lines like these:
5182 # Image:someimage.jpg|This is some image
5184 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5186 if ( count( $matches ) == 0 ) {
5190 if ( strpos( $matches[0], '%' ) !== false ) {
5191 $matches[1] = rawurldecode( $matches[1] );
5193 $title = Title
::newFromText( $matches[1], NS_FILE
);
5194 if ( is_null( $title ) ) {
5195 # Bogus title. Ignore these so we don't bomb out later.
5199 # We need to get what handler the file uses, to figure out parameters.
5200 # Note, a hook can overide the file name, and chose an entirely different
5201 # file (which potentially could be of a different type and have different handler).
5204 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5205 [ $this, $title, &$options, &$descQuery ] );
5206 # Don't register it now, as TraditionalImageGallery does that later.
5207 $file = $this->fetchFileNoRegister( $title, $options );
5208 $handler = $file ?
$file->getHandler() : false;
5211 'img_alt' => 'gallery-internal-alt',
5212 'img_link' => 'gallery-internal-link',
5215 $paramMap +
= $handler->getParamMap();
5216 // We don't want people to specify per-image widths.
5217 // Additionally the width parameter would need special casing anyhow.
5218 unset( $paramMap['img_width'] );
5221 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5226 $handlerOptions = [];
5227 if ( isset( $matches[3] ) ) {
5228 // look for an |alt= definition while trying not to break existing
5229 // captions with multiple pipes (|) in it, until a more sensible grammar
5230 // is defined for images in galleries
5232 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5233 // splitting on '|' is a bit odd, and different from makeImage.
5234 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5235 // Protect LanguageConverter markup
5236 $parameterMatches = StringUtils
::delimiterExplode(
5237 '-{', '}-', '|', $matches[3], true /* nested */
5240 foreach ( $parameterMatches as $parameterMatch ) {
5241 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5243 $paramName = $paramMap[$magicName];
5245 switch ( $paramName ) {
5246 case 'gallery-internal-alt':
5247 $alt = $this->stripAltText( $match, false );
5249 case 'gallery-internal-link':
5250 $linkValue = $this->stripAltText( $match, false );
5251 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5252 // Result of LanguageConverter::markNoConversion
5253 // invoked on an external link.
5254 $linkValue = substr( $linkValue, 4, -2 );
5256 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5257 if ( $type === 'link-url' ) {
5259 $this->mOutput
->addExternalLink( $target );
5260 } elseif ( $type === 'link-title' ) {
5261 $link = $target->getLinkURL();
5262 $this->mOutput
->addLink( $target );
5266 // Must be a handler specific parameter.
5267 if ( $handler->validateParam( $paramName, $match ) ) {
5268 $handlerOptions[$paramName] = $match;
5270 // Guess not, consider it as caption.
5271 $this->logger
->debug(
5272 "$parameterMatch failed parameter validation" );
5273 $label = $parameterMatch;
5279 $label = $parameterMatch;
5284 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5286 $html = $ig->toHTML();
5287 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5292 * @param MediaHandler $handler
5295 public function getImageParams( $handler ) {
5297 $handlerClass = get_class( $handler );
5301 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5302 # Initialise static lists
5303 static $internalParamNames = [
5304 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5305 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5306 'bottom', 'text-bottom' ],
5307 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5308 'upright', 'border', 'link', 'alt', 'class' ],
5310 static $internalParamMap;
5311 if ( !$internalParamMap ) {
5312 $internalParamMap = [];
5313 foreach ( $internalParamNames as $type => $names ) {
5314 foreach ( $names as $name ) {
5315 // For grep: img_left, img_right, img_center, img_none,
5316 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5317 // img_bottom, img_text_bottom,
5318 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5319 // img_border, img_link, img_alt, img_class
5320 $magicName = str_replace( '-', '_', "img_$name" );
5321 $internalParamMap[$magicName] = [ $type, $name ];
5326 # Add handler params
5327 $paramMap = $internalParamMap;
5329 $handlerParamMap = $handler->getParamMap();
5330 foreach ( $handlerParamMap as $magic => $paramName ) {
5331 $paramMap[$magic] = [ 'handler', $paramName ];
5334 $this->mImageParams
[$handlerClass] = $paramMap;
5335 $this->mImageParamsMagicArray
[$handlerClass] =
5336 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5338 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5342 * Parse image options text and use it to make an image
5344 * @param Title $title
5345 * @param string $options
5346 * @param LinkHolderArray|bool $holders
5347 * @return string HTML
5349 public function makeImage( $title, $options, $holders = false ) {
5350 # Check if the options text is of the form "options|alt text"
5352 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5353 # * left no resizing, just left align. label is used for alt= only
5354 # * right same, but right aligned
5355 # * none same, but not aligned
5356 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5357 # * center center the image
5358 # * frame Keep original image size, no magnify-button.
5359 # * framed Same as "frame"
5360 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5361 # * upright reduce width for upright images, rounded to full __0 px
5362 # * border draw a 1px border around the image
5363 # * alt Text for HTML alt attribute (defaults to empty)
5364 # * class Set a class for img node
5365 # * link Set the target of the image link. Can be external, interwiki, or local
5366 # vertical-align values (no % or length right now):
5376 # Protect LanguageConverter markup when splitting into parts
5377 $parts = StringUtils
::delimiterExplode(
5378 '-{', '}-', '|', $options, true /* allow nesting */
5381 # Give extensions a chance to select the file revision for us
5384 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5385 [ $this, $title, &$options, &$descQuery ] );
5386 # Fetch and register the file (file title may be different via hooks)
5387 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5390 $handler = $file ?
$file->getHandler() : false;
5392 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5395 $this->addTrackingCategory( 'broken-file-category' );
5398 # Process the input parameters
5400 $params = [ 'frame' => [], 'handler' => [],
5401 'horizAlign' => [], 'vertAlign' => [] ];
5402 $seenformat = false;
5403 foreach ( $parts as $part ) {
5404 $part = trim( $part );
5405 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5407 if ( isset( $paramMap[$magicName] ) ) {
5408 list( $type, $paramName ) = $paramMap[$magicName];
5410 # Special case; width and height come in one variable together
5411 if ( $type === 'handler' && $paramName === 'width' ) {
5412 $parsedWidthParam = self
::parseWidthParam( $value );
5413 if ( isset( $parsedWidthParam['width'] ) ) {
5414 $width = $parsedWidthParam['width'];
5415 if ( $handler->validateParam( 'width', $width ) ) {
5416 $params[$type]['width'] = $width;
5420 if ( isset( $parsedWidthParam['height'] ) ) {
5421 $height = $parsedWidthParam['height'];
5422 if ( $handler->validateParam( 'height', $height ) ) {
5423 $params[$type]['height'] = $height;
5427 # else no validation -- T15436
5429 if ( $type === 'handler' ) {
5430 # Validate handler parameter
5431 $validated = $handler->validateParam( $paramName, $value );
5433 # Validate internal parameters
5434 switch ( $paramName ) {
5438 # @todo FIXME: Possibly check validity here for
5439 # manualthumb? downstream behavior seems odd with
5440 # missing manual thumbs.
5442 $value = $this->stripAltText( $value, $holders );
5445 list( $paramName, $value ) =
5446 $this->parseLinkParameter(
5447 $this->stripAltText( $value, $holders )
5451 if ( $paramName === 'no-link' ) {
5454 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5455 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5462 // use first appearing option, discard others.
5463 $validated = !$seenformat;
5467 # Most other things appear to be empty or numeric...
5468 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5473 $params[$type][$paramName] = $value;
5477 if ( !$validated ) {
5482 # Process alignment parameters
5483 if ( $params['horizAlign'] ) {
5484 $params['frame']['align'] = key( $params['horizAlign'] );
5486 if ( $params['vertAlign'] ) {
5487 $params['frame']['valign'] = key( $params['vertAlign'] );
5490 $params['frame']['caption'] = $caption;
5492 # Will the image be presented in a frame, with the caption below?
5493 $imageIsFramed = isset( $params['frame']['frame'] )
5494 ||
isset( $params['frame']['framed'] )
5495 ||
isset( $params['frame']['thumbnail'] )
5496 ||
isset( $params['frame']['manualthumb'] );
5498 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5499 # came to also set the caption, ordinary text after the image -- which
5500 # makes no sense, because that just repeats the text multiple times in
5501 # screen readers. It *also* came to set the title attribute.
5502 # Now that we have an alt attribute, we should not set the alt text to
5503 # equal the caption: that's worse than useless, it just repeats the
5504 # text. This is the framed/thumbnail case. If there's no caption, we
5505 # use the unnamed parameter for alt text as well, just for the time be-
5506 # ing, if the unnamed param is set and the alt param is not.
5507 # For the future, we need to figure out if we want to tweak this more,
5508 # e.g., introducing a title= parameter for the title; ignoring the un-
5509 # named parameter entirely for images without a caption; adding an ex-
5510 # plicit caption= parameter and preserving the old magic unnamed para-
5512 if ( $imageIsFramed ) { # Framed image
5513 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5514 # No caption or alt text, add the filename as the alt text so
5515 # that screen readers at least get some description of the image
5516 $params['frame']['alt'] = $title->getText();
5518 # Do not set $params['frame']['title'] because tooltips don't make sense
5520 } else { # Inline image
5521 if ( !isset( $params['frame']['alt'] ) ) {
5522 # No alt text, use the "caption" for the alt text
5523 if ( $caption !== '' ) {
5524 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5526 # No caption, fall back to using the filename for the
5528 $params['frame']['alt'] = $title->getText();
5531 # Use the "caption" for the tooltip text
5532 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5534 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5536 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5538 # Linker does the rest
5539 $time = $options['time'] ??
false;
5540 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5541 $time, $descQuery, $this->mOptions
->getThumbSize() );
5543 # Give the handler a chance to modify the parser object
5545 $handler->parserTransformHook( $this, $file );
5552 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5554 * Adds an entry to appropriate link tables.
5557 * @param string $value
5558 * @return array of `[ type, target ]`, where:
5559 * - `type` is one of:
5560 * - `null`: Given value is not a valid link target, use default
5561 * - `'no-link'`: Given value is empty, do not generate a link
5562 * - `'link-url'`: Given value is a valid external link
5563 * - `'link-title'`: Given value is a valid internal link
5565 * - When `type` is `null` or `'no-link'`: `false`
5566 * - When `type` is `'link-url'`: URL string corresponding to given value
5567 * - When `type` is `'link-title'`: Title object corresponding to given value
5569 public function parseLinkParameter( $value ) {
5570 $chars = self
::EXT_LINK_URL_CLASS
;
5571 $addr = self
::EXT_LINK_ADDR
;
5572 $prots = $this->mUrlProtocols
;
5575 if ( $value === '' ) {
5577 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5578 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5579 $this->mOutput
->addExternalLink( $value );
5584 $linkTitle = Title
::newFromText( $value );
5586 $this->mOutput
->addLink( $linkTitle );
5587 $type = 'link-title';
5588 $target = $linkTitle;
5591 return [ $type, $target ];
5595 * @param string $caption
5596 * @param LinkHolderArray|bool $holders
5597 * @return mixed|string
5599 protected function stripAltText( $caption, $holders ) {
5600 # Strip bad stuff out of the title (tooltip). We can't just use
5601 # replaceLinkHoldersText() here, because if this function is called
5602 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5604 $tooltip = $holders->replaceText( $caption );
5606 $tooltip = $this->replaceLinkHoldersText( $caption );
5609 # make sure there are no placeholders in thumbnail attributes
5610 # that are later expanded to html- so expand them now and
5612 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5613 # Compatibility hack! In HTML certain entity references not terminated
5614 # by a semicolon are decoded (but not if we're in an attribute; that's
5615 # how link URLs get away without properly escaping & in queries).
5616 # But wikitext has always required semicolon-termination of entities,
5617 # so encode & where needed to avoid decode of semicolon-less entities.
5619 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5620 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5621 $tooltip = preg_replace( "/
5622 & # 1. entity prefix
5623 (?= # 2. followed by:
5624 (?: # a. one of the legacy semicolon-less named entities
5625 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5626 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5627 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5628 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5629 U(?:acute|circ|grave|uml)|Yacute|
5630 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5631 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5632 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5633 frac(?:1(?:2|4)|34)|
5634 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5635 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5636 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5637 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5638 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5639 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5640 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5641 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5642 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5644 (?:[^;]|$)) # b. and not followed by a semicolon
5645 # S = study, for efficiency
5646 /Sx", '&', $tooltip );
5647 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5653 * Set a flag in the output object indicating that the content is dynamic and
5654 * shouldn't be cached.
5655 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5657 public function disableCache() {
5658 $this->logger
->debug( "Parser output marked as uncacheable." );
5659 if ( !$this->mOutput
) {
5660 throw new MWException( __METHOD__
.
5661 " can only be called when actually parsing something" );
5663 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5667 * Callback from the Sanitizer for expanding items found in HTML attribute
5668 * values, so they can be safely tested and escaped.
5670 * @param string &$text
5671 * @param bool|PPFrame $frame
5674 public function attributeStripCallback( &$text, $frame = false ) {
5675 $text = $this->replaceVariables( $text, $frame );
5676 $text = $this->mStripState
->unstripBoth( $text );
5685 public function getTags() {
5686 $this->firstCallInit();
5688 array_keys( $this->mTransparentTagHooks
),
5689 array_keys( $this->mTagHooks
),
5690 array_keys( $this->mFunctionTagHooks
)
5698 public function getFunctionSynonyms() {
5699 $this->firstCallInit();
5700 return $this->mFunctionSynonyms
;
5707 public function getUrlProtocols() {
5708 return $this->mUrlProtocols
;
5712 * Replace transparent tags in $text with the values given by the callbacks.
5714 * Transparent tag hooks are like regular XML-style tag hooks, except they
5715 * operate late in the transformation sequence, on HTML instead of wikitext.
5717 * @param string $text
5721 public function replaceTransparentTags( $text ) {
5723 $elements = array_keys( $this->mTransparentTagHooks
);
5724 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5727 foreach ( $matches as $marker => $data ) {
5728 list( $element, $content, $params, $tag ) = $data;
5729 $tagName = strtolower( $element );
5730 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5731 $output = call_user_func_array(
5732 $this->mTransparentTagHooks
[$tagName],
5733 [ $content, $params, $this ]
5738 $replacements[$marker] = $output;
5740 return strtr( $text, $replacements );
5744 * Break wikitext input into sections, and either pull or replace
5745 * some particular section's text.
5747 * External callers should use the getSection and replaceSection methods.
5749 * @param string $text Page wikitext
5750 * @param string|int $sectionId A section identifier string of the form:
5751 * "<flag1> - <flag2> - ... - <section number>"
5753 * Currently the only recognised flag is "T", which means the target section number
5754 * was derived during a template inclusion parse, in other words this is a template
5755 * section edit link. If no flags are given, it was an ordinary section edit link.
5756 * This flag is required to avoid a section numbering mismatch when a section is
5757 * enclosed by "<includeonly>" (T8563).
5759 * The section number 0 pulls the text before the first heading; other numbers will
5760 * pull the given section along with its lower-level subsections. If the section is
5761 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5763 * Section 0 is always considered to exist, even if it only contains the empty
5764 * string. If $text is the empty string and section 0 is replaced, $newText is
5767 * @param string $mode One of "get" or "replace"
5768 * @param string $newText Replacement text for section data.
5769 * @return string For "get", the extracted section text.
5770 * for "replace", the whole page with the section replaced.
5772 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5773 global $wgTitle; # not generally used but removes an ugly failure mode
5775 $magicScopeVariable = $this->lock();
5776 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5778 $frame = $this->getPreprocessor()->newFrame();
5780 # Process section extraction flags
5782 $sectionParts = explode( '-', $sectionId );
5783 $sectionIndex = array_pop( $sectionParts );
5784 foreach ( $sectionParts as $part ) {
5785 if ( $part === 'T' ) {
5786 $flags |
= self
::PTD_FOR_INCLUSION
;
5790 # Check for empty input
5791 if ( strval( $text ) === '' ) {
5792 # Only sections 0 and T-0 exist in an empty document
5793 if ( $sectionIndex == 0 ) {
5794 if ( $mode === 'get' ) {
5800 if ( $mode === 'get' ) {
5808 # Preprocess the text
5809 $root = $this->preprocessToDom( $text, $flags );
5811 # <h> nodes indicate section breaks
5812 # They can only occur at the top level, so we can find them by iterating the root's children
5813 $node = $root->getFirstChild();
5815 # Find the target section
5816 if ( $sectionIndex == 0 ) {
5817 # Section zero doesn't nest, level=big
5818 $targetLevel = 1000;
5821 if ( $node->getName() === 'h' ) {
5822 $bits = $node->splitHeading();
5823 if ( $bits['i'] == $sectionIndex ) {
5824 $targetLevel = $bits['level'];
5828 if ( $mode === 'replace' ) {
5829 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5831 $node = $node->getNextSibling();
5837 if ( $mode === 'get' ) {
5844 # Find the end of the section, including nested sections
5846 if ( $node->getName() === 'h' ) {
5847 $bits = $node->splitHeading();
5848 $curLevel = $bits['level'];
5849 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5853 if ( $mode === 'get' ) {
5854 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5856 $node = $node->getNextSibling();
5859 # Write out the remainder (in replace mode only)
5860 if ( $mode === 'replace' ) {
5861 # Output the replacement text
5862 # Add two newlines on -- trailing whitespace in $newText is conventionally
5863 # stripped by the editor, so we need both newlines to restore the paragraph gap
5864 # Only add trailing whitespace if there is newText
5865 if ( $newText != "" ) {
5866 $outText .= $newText . "\n\n";
5870 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5871 $node = $node->getNextSibling();
5875 if ( is_string( $outText ) ) {
5876 # Re-insert stripped tags
5877 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5884 * This function returns the text of a section, specified by a number ($section).
5885 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5886 * the first section before any such heading (section 0).
5888 * If a section contains subsections, these are also returned.
5890 * @param string $text Text to look in
5891 * @param string|int $sectionId Section identifier as a number or string
5892 * (e.g. 0, 1 or 'T-1').
5893 * @param string $defaultText Default to return if section is not found
5895 * @return string Text of the requested section
5897 public function getSection( $text, $sectionId, $defaultText = '' ) {
5898 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5902 * This function returns $oldtext after the content of the section
5903 * specified by $section has been replaced with $text. If the target
5904 * section does not exist, $oldtext is returned unchanged.
5906 * @param string $oldText Former text of the article
5907 * @param string|int $sectionId Section identifier as a number or string
5908 * (e.g. 0, 1 or 'T-1').
5909 * @param string $newText Replacing text
5911 * @return string Modified text
5913 public function replaceSection( $oldText, $sectionId, $newText ) {
5914 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5918 * Get the ID of the revision we are parsing
5920 * The return value will be either:
5921 * - a) Positive, indicating a specific revision ID (current or old)
5922 * - b) Zero, meaning the revision ID is specified by getCurrentRevisionCallback()
5923 * - c) Null, meaning the parse is for preview mode and there is no revision
5927 public function getRevisionId() {
5928 return $this->mRevisionId
;
5932 * Get the revision object for $this->mRevisionId
5934 * @return Revision|null Either a Revision object or null
5935 * @since 1.23 (public since 1.23)
5937 public function getRevisionObject() {
5938 if ( $this->mRevisionObject
) {
5939 return $this->mRevisionObject
;
5942 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5943 // This is useful when parsing a revision that has not yet been saved.
5944 // However, if we get back a saved revision even though we are in
5945 // preview mode, we'll have to ignore it, see below.
5946 // NOTE: This callback may be used to inject an OLD revision that was
5947 // already loaded, so "current" is a bit of a misnomer. We can't just
5948 // skip it if mRevisionId is set.
5949 $rev = call_user_func(
5950 $this->mOptions
->getCurrentRevisionCallback(),
5955 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5956 // We are in preview mode (mRevisionId is null), and the current revision callback
5957 // returned an existing revision. Ignore it and return null, it's probably the page's
5958 // current revision, which is not what we want here. Note that we do want to call the
5959 // callback to allow the unsaved revision to be injected here, e.g. for
5960 // self-transclusion previews.
5964 // If the parse is for a new revision, then the callback should have
5965 // already been set to force the object and should match mRevisionId.
5966 // If not, try to fetch by mRevisionId for sanity.
5967 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5968 $rev = Revision
::newFromId( $this->mRevisionId
);
5971 $this->mRevisionObject
= $rev;
5973 return $this->mRevisionObject
;
5977 * Get the timestamp associated with the current revision, adjusted for
5978 * the default server-local timestamp
5979 * @return string TS_MW timestamp
5981 public function getRevisionTimestamp() {
5982 if ( $this->mRevisionTimestamp
!== null ) {
5983 return $this->mRevisionTimestamp
;
5986 # Use specified revision timestamp, falling back to the current timestamp
5987 $revObject = $this->getRevisionObject();
5988 $timestamp = $revObject ?
$revObject->getTimestamp() : $this->mOptions
->getTimestamp();
5989 $this->mOutput
->setRevisionTimestampUsed( $timestamp ); // unadjusted time zone
5991 # The cryptic '' timezone parameter tells to use the site-default
5992 # timezone offset instead of the user settings.
5993 # Since this value will be saved into the parser cache, served
5994 # to other users, and potentially even used inside links and such,
5995 # it needs to be consistent for all visitors.
5996 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
5998 return $this->mRevisionTimestamp
;
6002 * Get the name of the user that edited the last revision
6004 * @return string User name
6006 public function getRevisionUser() {
6007 if ( is_null( $this->mRevisionUser
) ) {
6008 $revObject = $this->getRevisionObject();
6010 # if this template is subst: the revision id will be blank,
6011 # so just use the current user's name
6013 $this->mRevisionUser
= $revObject->getUserText();
6014 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
6015 $this->mRevisionUser
= $this->getUser()->getName();
6018 return $this->mRevisionUser
;
6022 * Get the size of the revision
6024 * @return int|null Revision size
6026 public function getRevisionSize() {
6027 if ( is_null( $this->mRevisionSize
) ) {
6028 $revObject = $this->getRevisionObject();
6030 # if this variable is subst: the revision id will be blank,
6031 # so just use the parser input size, because the own substituation
6032 # will change the size.
6034 $this->mRevisionSize
= $revObject->getSize();
6036 $this->mRevisionSize
= $this->mInputSize
;
6039 return $this->mRevisionSize
;
6043 * Mutator for $mDefaultSort
6045 * @param string $sort New value
6047 public function setDefaultSort( $sort ) {
6048 $this->mDefaultSort
= $sort;
6049 $this->mOutput
->setProperty( 'defaultsort', $sort );
6053 * Accessor for $mDefaultSort
6054 * Will use the empty string if none is set.
6056 * This value is treated as a prefix, so the
6057 * empty string is equivalent to sorting by
6062 public function getDefaultSort() {
6063 if ( $this->mDefaultSort
!== false ) {
6064 return $this->mDefaultSort
;
6071 * Accessor for $mDefaultSort
6072 * Unlike getDefaultSort(), will return false if none is set
6074 * @return string|bool
6076 public function getCustomDefaultSort() {
6077 return $this->mDefaultSort
;
6080 private static function getSectionNameFromStrippedText( $text ) {
6081 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
6082 $text = Sanitizer
::decodeCharReferences( $text );
6083 $text = self
::normalizeSectionName( $text );
6087 private static function makeAnchor( $sectionName ) {
6088 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6091 private function makeLegacyAnchor( $sectionName ) {
6092 $fragmentMode = $this->svcOptions
->get( 'FragmentMode' );
6093 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6094 // ForAttribute() and ForLink() are the same for legacy encoding
6095 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6097 $id = Sanitizer
::escapeIdForLink( $sectionName );
6104 * Try to guess the section anchor name based on a wikitext fragment
6105 * presumably extracted from a heading, for example "Header" from
6108 * @param string $text
6109 * @return string Anchor (starting with '#')
6111 public function guessSectionNameFromWikiText( $text ) {
6112 # Strip out wikitext links(they break the anchor)
6113 $text = $this->stripSectionName( $text );
6114 $sectionName = self
::getSectionNameFromStrippedText( $text );
6115 return self
::makeAnchor( $sectionName );
6119 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6120 * instead, if possible. For use in redirects, since various versions
6121 * of Microsoft browsers interpret Location: headers as something other
6122 * than UTF-8, resulting in breakage.
6124 * @param string $text The section name
6125 * @return string Anchor (starting with '#')
6127 public function guessLegacySectionNameFromWikiText( $text ) {
6128 # Strip out wikitext links(they break the anchor)
6129 $text = $this->stripSectionName( $text );
6130 $sectionName = self
::getSectionNameFromStrippedText( $text );
6131 return $this->makeLegacyAnchor( $sectionName );
6135 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6136 * @param string $text Section name (plain text)
6137 * @return string Anchor (starting with '#')
6139 public static function guessSectionNameFromStrippedText( $text ) {
6140 $sectionName = self
::getSectionNameFromStrippedText( $text );
6141 return self
::makeAnchor( $sectionName );
6145 * Apply the same normalization as code making links to this section would
6147 * @param string $text
6150 private static function normalizeSectionName( $text ) {
6151 # T90902: ensure the same normalization is applied for IDs as to links
6152 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6155 $parts = $titleParser->splitTitleString( "#$text" );
6156 } catch ( MalformedTitleException
$ex ) {
6159 return $parts['fragment'];
6163 * Strips a text string of wikitext for use in a section anchor
6165 * Accepts a text string and then removes all wikitext from the
6166 * string and leaves only the resultant text (i.e. the result of
6167 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6168 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6169 * to create valid section anchors by mimicing the output of the
6170 * parser when headings are parsed.
6172 * @param string $text Text string to be stripped of wikitext
6173 * for use in a Section anchor
6174 * @return string Filtered text string
6176 public function stripSectionName( $text ) {
6177 # Strip internal link markup
6178 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6179 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6181 # Strip external link markup
6182 # @todo FIXME: Not tolerant to blank link text
6183 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6184 # on how many empty links there are on the page - need to figure that out.
6185 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6187 # Parse wikitext quotes (italics & bold)
6188 $text = $this->doQuotes( $text );
6191 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6196 * strip/replaceVariables/unstrip for preprocessor regression testing
6198 * @param string $text
6199 * @param Title $title
6200 * @param ParserOptions $options
6201 * @param int $outputType
6205 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6206 $outputType = self
::OT_HTML
6208 $magicScopeVariable = $this->lock();
6209 $this->startParse( $title, $options, $outputType, true );
6211 $text = $this->replaceVariables( $text );
6212 $text = $this->mStripState
->unstripBoth( $text );
6213 $text = Sanitizer
::removeHTMLtags( $text );
6218 * @param string $text
6219 * @param Title $title
6220 * @param ParserOptions $options
6223 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6224 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6228 * @param string $text
6229 * @param Title $title
6230 * @param ParserOptions $options
6233 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6234 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6238 * Call a callback function on all regions of the given text that are not
6239 * inside strip markers, and replace those regions with the return value
6240 * of the callback. For example, with input:
6244 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6245 * two strings will be replaced with the value returned by the callback in
6249 * @param callable $callback
6253 public function markerSkipCallback( $s, $callback ) {
6256 while ( $i < strlen( $s ) ) {
6257 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6258 if ( $markerStart === false ) {
6259 $out .= call_user_func( $callback, substr( $s, $i ) );
6262 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6263 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6264 if ( $markerEnd === false ) {
6265 $out .= substr( $s, $markerStart );
6268 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6269 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6278 * Remove any strip markers found in the given text.
6280 * @param string $text
6283 public function killMarkers( $text ) {
6284 return $this->mStripState
->killMarkers( $text );
6288 * Save the parser state required to convert the given half-parsed text to
6289 * HTML. "Half-parsed" in this context means the output of
6290 * recursiveTagParse() or internalParse(). This output has strip markers
6291 * from replaceVariables (extensionSubstitution() etc.), and link
6292 * placeholders from replaceLinkHolders().
6294 * Returns an array which can be serialized and stored persistently. This
6295 * array can later be loaded into another parser instance with
6296 * unserializeHalfParsedText(). The text can then be safely incorporated into
6297 * the return value of a parser hook.
6299 * @deprecated since 1.31
6300 * @param string $text
6304 public function serializeHalfParsedText( $text ) {
6305 wfDeprecated( __METHOD__
, '1.31' );
6308 'version' => self
::HALF_PARSED_VERSION
,
6309 'stripState' => $this->mStripState
->getSubState( $text ),
6310 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6316 * Load the parser state given in the $data array, which is assumed to
6317 * have been generated by serializeHalfParsedText(). The text contents is
6318 * extracted from the array, and its markers are transformed into markers
6319 * appropriate for the current Parser instance. This transformed text is
6320 * returned, and can be safely included in the return value of a parser
6323 * If the $data array has been stored persistently, the caller should first
6324 * check whether it is still valid, by calling isValidHalfParsedText().
6326 * @deprecated since 1.31
6327 * @param array $data Serialized data
6328 * @throws MWException
6331 public function unserializeHalfParsedText( $data ) {
6332 wfDeprecated( __METHOD__
, '1.31' );
6333 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6334 throw new MWException( __METHOD__
. ': invalid version' );
6337 # First, extract the strip state.
6338 $texts = [ $data['text'] ];
6339 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6341 # Now renumber links
6342 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6344 # Should be good to go.
6349 * Returns true if the given array, presumed to be generated by
6350 * serializeHalfParsedText(), is compatible with the current version of the
6353 * @deprecated since 1.31
6354 * @param array $data
6358 public function isValidHalfParsedText( $data ) {
6359 wfDeprecated( __METHOD__
, '1.31' );
6360 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6364 * Parsed a width param of imagelink like 300px or 200x300px
6366 * @param string $value
6367 * @param bool $parseHeight
6372 public static function parseWidthParam( $value, $parseHeight = true ) {
6373 $parsedWidthParam = [];
6374 if ( $value === '' ) {
6375 return $parsedWidthParam;
6378 # (T15500) In both cases (width/height and width only),
6379 # permit trailing "px" for backward compatibility.
6380 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6381 $width = intval( $m[1] );
6382 $height = intval( $m[2] );
6383 $parsedWidthParam['width'] = $width;
6384 $parsedWidthParam['height'] = $height;
6385 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6386 $width = intval( $value );
6387 $parsedWidthParam['width'] = $width;
6389 return $parsedWidthParam;
6393 * Lock the current instance of the parser.
6395 * This is meant to stop someone from calling the parser
6396 * recursively and messing up all the strip state.
6398 * @throws MWException If parser is in a parse
6399 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6401 protected function lock() {
6402 if ( $this->mInParse
) {
6403 throw new MWException( "Parser state cleared while parsing. "
6404 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6407 // Save the backtrace when locking, so that if some code tries locking again,
6408 // we can print the lock owner's backtrace for easier debugging
6410 $this->mInParse
= $e->getTraceAsString();
6412 $recursiveCheck = new ScopedCallback( function () {
6413 $this->mInParse
= false;
6416 return $recursiveCheck;
6420 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6422 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6423 * or if there is more than one <p/> tag in the input HTML.
6425 * @param string $html
6429 public static function stripOuterParagraph( $html ) {
6431 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6439 * Return this parser if it is not doing anything, otherwise
6440 * get a fresh parser. You can use this method by doing
6441 * $newParser = $oldParser->getFreshParser(), or more simply
6442 * $oldParser->getFreshParser()->parse( ... );
6443 * if you're unsure if $oldParser is safe to use.
6446 * @return Parser A parser object that is not parsing anything
6448 public function getFreshParser() {
6449 if ( $this->mInParse
) {
6450 return $this->factory
->create();
6457 * Set's up the PHP implementation of OOUI for use in this request
6458 * and instructs OutputPage to enable OOUI for itself.
6462 public function enableOOUI() {
6463 OutputPage
::setupOOUI();
6464 $this->mOutput
->setEnableOOUI( true );
6468 * @param string $flag
6469 * @param string $reason
6471 protected function setOutputFlag( $flag, $reason ) {
6472 $this->mOutput
->setFlag( $flag );
6473 $name = $this->mTitle
->getPrefixedText();
6474 $this->logger
->debug( __METHOD__
. ": set $flag flag on '$name'; $reason" );