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->resetOutput();
472 $this->mAutonumber
= 0;
473 $this->mIncludeCount
= [];
474 $this->mLinkHolders
= new LinkHolderArray( $this );
476 $this->mRevisionObject
= $this->mRevisionTimestamp
=
477 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
478 $this->mVarCache
= [];
480 $this->mLangLinkLanguages
= [];
481 $this->currentRevisionCache
= null;
483 $this->mStripState
= new StripState( $this );
485 # Clear these on every parse, T6549
486 $this->mTplRedirCache
= $this->mTplDomCache
= [];
488 $this->mShowToc
= true;
489 $this->mForceTocPosition
= false;
490 $this->mIncludeSizes
= [
494 $this->mPPNodeCount
= 0;
495 $this->mGeneratedPPNodeCount
= 0;
496 $this->mHighestExpansionDepth
= 0;
497 $this->mDefaultSort
= false;
498 $this->mHeadings
= [];
499 $this->mDoubleUnderscores
= [];
500 $this->mExpensiveFunctionCount
= 0;
503 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
504 $this->mPreprocessor
= null;
507 $this->mProfiler
= new SectionProfiler();
509 // Avoid PHP 7.1 warning from passing $this by reference
511 Hooks
::run( 'ParserClearState', [ &$parser ] );
515 * Reset the ParserOutput
517 public function resetOutput() {
518 $this->mOutput
= new ParserOutput
;
519 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
523 * Convert wikitext to HTML
524 * Do not call this function recursively.
526 * @param string $text Text we want to parse
527 * @param-taint $text escapes_htmlnoent
528 * @param Title $title
529 * @param ParserOptions $options
530 * @param bool $linestart
531 * @param bool $clearState
532 * @param int|null $revid Number to pass in {{REVISIONID}}
533 * @return ParserOutput A ParserOutput
534 * @return-taint escaped
536 public function parse(
537 $text, Title
$title, ParserOptions
$options,
538 $linestart = true, $clearState = true, $revid = null
541 // We use U+007F DELETE to construct strip markers, so we have to make
542 // sure that this character does not occur in the input text.
543 $text = strtr( $text, "\x7f", "?" );
544 $magicScopeVariable = $this->lock();
546 // Strip U+0000 NULL (T159174)
547 $text = str_replace( "\000", '', $text );
549 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
551 $this->currentRevisionCache
= null;
552 $this->mInputSize
= strlen( $text );
553 if ( $this->mOptions
->getEnableLimitReport() ) {
554 $this->mOutput
->resetParseStartTime();
557 $oldRevisionId = $this->mRevisionId
;
558 $oldRevisionObject = $this->mRevisionObject
;
559 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
560 $oldRevisionUser = $this->mRevisionUser
;
561 $oldRevisionSize = $this->mRevisionSize
;
562 if ( $revid !== null ) {
563 $this->mRevisionId
= $revid;
564 $this->mRevisionObject
= null;
565 $this->mRevisionTimestamp
= null;
566 $this->mRevisionUser
= null;
567 $this->mRevisionSize
= null;
570 // Avoid PHP 7.1 warning from passing $this by reference
572 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
574 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
575 $text = $this->internalParse( $text );
576 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
578 $text = $this->internalParseHalfParsed( $text, true, $linestart );
581 * A converted title will be provided in the output object if title and
582 * content conversion are enabled, the article text does not contain
583 * a conversion-suppressing double-underscore tag, and no
584 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
585 * automatic link conversion.
587 if ( !( $options->getDisableTitleConversion()
588 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
589 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
590 ||
$this->mOutput
->getDisplayTitle() !== false )
592 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
593 if ( $convruletitle ) {
594 $this->mOutput
->setTitleText( $convruletitle );
596 $titleText = $this->getTargetLanguage()->convertTitle( $title );
597 $this->mOutput
->setTitleText( $titleText );
601 # Compute runtime adaptive expiry if set
602 $this->mOutput
->finalizeAdaptiveCacheExpiry();
604 # Warn if too many heavyweight parser functions were used
605 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
606 $this->limitationWarn( 'expensive-parserfunction',
607 $this->mExpensiveFunctionCount
,
608 $this->mOptions
->getExpensiveParserFunctionLimit()
612 # Information on limits, for the benefit of users who try to skirt them
613 if ( $this->mOptions
->getEnableLimitReport() ) {
614 $text .= $this->makeLimitReport();
617 # Wrap non-interface parser output in a <div> so it can be targeted
619 $class = $this->mOptions
->getWrapOutputClass();
620 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
621 $this->mOutput
->addWrapperDivClass( $class );
624 $this->mOutput
->setText( $text );
626 $this->mRevisionId
= $oldRevisionId;
627 $this->mRevisionObject
= $oldRevisionObject;
628 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
629 $this->mRevisionUser
= $oldRevisionUser;
630 $this->mRevisionSize
= $oldRevisionSize;
631 $this->mInputSize
= false;
632 $this->currentRevisionCache
= null;
634 return $this->mOutput
;
638 * Set the limit report data in the current ParserOutput, and return the
639 * limit report HTML comment.
643 protected function makeLimitReport() {
644 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
646 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
647 if ( $cpuTime !== null ) {
648 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
649 sprintf( "%.3f", $cpuTime )
653 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
654 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
655 sprintf( "%.3f", $wallTime )
658 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
659 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
661 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
662 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
664 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
665 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
667 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
668 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
670 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
671 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
673 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
674 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
677 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
678 $this->mOutput
->setLimitReportData( $key, $value );
681 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
683 $limitReport = "NewPP limit report\n";
684 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
685 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
687 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
688 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
689 $limitReport .= 'Dynamic content: ' .
690 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
692 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput
->getAllFlags() ) . "]\n";
694 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
695 if ( Hooks
::run( 'ParserLimitReportFormat',
696 [ $key, &$value, &$limitReport, false, false ]
698 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
699 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
700 ->inLanguage( 'en' )->useDatabase( false );
701 if ( !$valueMsg->exists() ) {
702 $valueMsg = new RawMessage( '$1' );
704 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
705 $valueMsg->params( $value );
706 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
710 // Since we're not really outputting HTML, decode the entities and
711 // then re-encode the things that need hiding inside HTML comments.
712 $limitReport = htmlspecialchars_decode( $limitReport );
714 // Sanitize for comment. Note '‐' in the replacement is U+2010,
715 // which looks much like the problematic '-'.
716 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
717 $text = "\n<!-- \n$limitReport-->\n";
719 // Add on template profiling data in human/machine readable way
720 $dataByFunc = $this->mProfiler
->getFunctionStats();
721 uasort( $dataByFunc, function ( $a, $b ) {
722 return $b['real'] <=> $a['real']; // descending order
725 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
726 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
727 $item['%real'], $item['real'], $item['calls'],
728 htmlspecialchars( $item['name'] ) );
730 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
731 $text .= implode( "\n", $profileReport ) . "\n-->\n";
733 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
735 // Add other cache related metadata
736 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
737 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
739 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
740 $this->mOutput
->getCacheTime() );
741 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
742 $this->mOutput
->getCacheExpiry() );
743 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
744 $this->mOutput
->hasDynamicContent() );
746 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
747 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
748 $this->mTitle
->getPrefixedDBkey() );
754 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
755 * can be called from an extension tag hook.
757 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
758 * instead, which means that lists and links have not been fully parsed yet,
759 * and strip markers are still present.
761 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
763 * Use this function if you're a parser tag hook and you want to parse
764 * wikitext before or after applying additional transformations, and you
765 * intend to *return the result as hook output*, which will cause it to go
766 * through the rest of parsing process automatically.
768 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
769 * $text are not expanded
771 * @param string $text Text extension wants to have parsed
772 * @param-taint $text escapes_htmlnoent
773 * @param bool|PPFrame $frame The frame to use for expanding any template variables
774 * @return string UNSAFE half-parsed HTML
775 * @return-taint escaped
777 public function recursiveTagParse( $text, $frame = false ) {
778 // Avoid PHP 7.1 warning from passing $this by reference
780 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
781 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
782 $text = $this->internalParse( $text, false, $frame );
787 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
788 * point can be called from an extension tag hook.
790 * The output of this function is fully-parsed HTML that is safe for output.
791 * If you're a parser tag hook, you might want to use recursiveTagParse()
794 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
795 * $text are not expanded
799 * @param string $text Text extension wants to have parsed
800 * @param-taint $text escapes_htmlnoent
801 * @param bool|PPFrame $frame The frame to use for expanding any template variables
802 * @return string Fully parsed HTML
803 * @return-taint escaped
805 public function recursiveTagParseFully( $text, $frame = false ) {
806 $text = $this->recursiveTagParse( $text, $frame );
807 $text = $this->internalParseHalfParsed( $text, false );
812 * Expand templates and variables in the text, producing valid, static wikitext.
813 * Also removes comments.
814 * Do not call this function recursively.
815 * @param string $text
816 * @param Title|null $title
817 * @param ParserOptions $options
818 * @param int|null $revid
819 * @param bool|PPFrame $frame
820 * @return mixed|string
822 public function preprocess( $text, Title
$title = null,
823 ParserOptions
$options, $revid = null, $frame = false
825 $magicScopeVariable = $this->lock();
826 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
827 if ( $revid !== null ) {
828 $this->mRevisionId
= $revid;
830 // Avoid PHP 7.1 warning from passing $this by reference
832 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
833 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
834 $text = $this->replaceVariables( $text, $frame );
835 $text = $this->mStripState
->unstripBoth( $text );
840 * Recursive parser entry point that can be called from an extension tag
843 * @param string $text Text to be expanded
844 * @param bool|PPFrame $frame The frame to use for expanding any template variables
848 public function recursivePreprocess( $text, $frame = false ) {
849 $text = $this->replaceVariables( $text, $frame );
850 $text = $this->mStripState
->unstripBoth( $text );
855 * Process the wikitext for the "?preload=" feature. (T7210)
857 * "<noinclude>", "<includeonly>" etc. are parsed as for template
858 * transclusion, comments, templates, arguments, tags hooks and parser
859 * functions are untouched.
861 * @param string $text
862 * @param Title $title
863 * @param ParserOptions $options
864 * @param array $params
867 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
868 $msg = new RawMessage( $text );
869 $text = $msg->params( $params )->plain();
871 # Parser (re)initialisation
872 $magicScopeVariable = $this->lock();
873 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
875 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
876 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
877 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
878 $text = $this->mStripState
->unstripBoth( $text );
883 * Set the current user.
884 * Should only be used when doing pre-save transform.
886 * @param User|null $user User object or null (to reset)
888 public function setUser( $user ) {
889 $this->mUser
= $user;
893 * Set the context title
897 public function setTitle( $t ) {
899 $t = Title
::newFromText( 'NO TITLE' );
902 if ( $t->hasFragment() ) {
903 # Strip the fragment to avoid various odd effects
904 $this->mTitle
= $t->createFragmentTarget( '' );
911 * Accessor for the Title object
915 public function getTitle() {
916 return $this->mTitle
;
920 * Accessor/mutator for the Title object
922 * @param Title|null $x Title object or null to just get the current one
925 public function Title( $x = null ) {
926 return wfSetVar( $this->mTitle
, $x );
930 * Set the output type
932 * @param int $ot New value
934 public function setOutputType( $ot ) {
935 $this->mOutputType
= $ot;
938 'html' => $ot == self
::OT_HTML
,
939 'wiki' => $ot == self
::OT_WIKI
,
940 'pre' => $ot == self
::OT_PREPROCESS
,
941 'plain' => $ot == self
::OT_PLAIN
,
946 * Accessor/mutator for the output type
948 * @param int|null $x New value or null to just get the current one
951 public function OutputType( $x = null ) {
952 return wfSetVar( $this->mOutputType
, $x );
956 * Get the ParserOutput object
958 * @return ParserOutput
960 public function getOutput() {
961 return $this->mOutput
;
965 * Get the ParserOptions object
967 * @return ParserOptions
969 public function getOptions() {
970 return $this->mOptions
;
974 * Accessor/mutator for the ParserOptions object
976 * @param ParserOptions|null $x New value or null to just get the current one
977 * @return ParserOptions Current ParserOptions object
979 public function Options( $x = null ) {
980 return wfSetVar( $this->mOptions
, $x );
986 public function nextLinkID() {
987 return $this->mLinkID++
;
993 public function setLinkID( $id ) {
994 $this->mLinkID
= $id;
998 * Get a language object for use in parser functions such as {{FORMATNUM:}}
1001 public function getFunctionLang() {
1002 return $this->getTargetLanguage();
1006 * Get the target language for the content being parsed. This is usually the
1007 * language that the content is in.
1011 * @throws MWException
1014 public function getTargetLanguage() {
1015 $target = $this->mOptions
->getTargetLanguage();
1017 if ( $target !== null ) {
1019 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
1020 return $this->mOptions
->getUserLangObj();
1021 } elseif ( is_null( $this->mTitle
) ) {
1022 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
1025 return $this->mTitle
->getPageLanguage();
1029 * Get the language object for language conversion
1030 * @deprecated since 1.32, just use getTargetLanguage()
1031 * @return Language|null
1033 public function getConverterLanguage() {
1034 return $this->getTargetLanguage();
1038 * Get a User object either from $this->mUser, if set, or from the
1039 * ParserOptions object otherwise
1043 public function getUser() {
1044 if ( !is_null( $this->mUser
) ) {
1045 return $this->mUser
;
1047 return $this->mOptions
->getUser();
1051 * Get a preprocessor object
1053 * @return Preprocessor
1055 public function getPreprocessor() {
1056 if ( !isset( $this->mPreprocessor
) ) {
1057 $class = $this->svcOptions
->get( 'preprocessorClass' );
1058 $this->mPreprocessor
= new $class( $this );
1060 return $this->mPreprocessor
;
1064 * Get a LinkRenderer instance to make links with
1067 * @return LinkRenderer
1069 public function getLinkRenderer() {
1070 // XXX We make the LinkRenderer with current options and then cache it forever
1071 if ( !$this->mLinkRenderer
) {
1072 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
1073 $this->mLinkRenderer
->setStubThreshold(
1074 $this->getOptions()->getStubThreshold()
1078 return $this->mLinkRenderer
;
1082 * Get the MagicWordFactory that this Parser is using
1085 * @return MagicWordFactory
1087 public function getMagicWordFactory() {
1088 return $this->magicWordFactory
;
1092 * Get the content language that this Parser is using
1097 public function getContentLanguage() {
1098 return $this->contLang
;
1102 * Replaces all occurrences of HTML-style comments and the given tags
1103 * in the text with a random marker and returns the next text. The output
1104 * parameter $matches will be an associative array filled with data in
1111 * [ 'param' => 'x' ],
1112 * '<element param="x">tag content</element>' ]
1115 * @param array $elements List of element names. Comments are always extracted.
1116 * @param string $text Source text string.
1117 * @param array &$matches Out parameter, Array: extracted tags
1118 * @return string Stripped text
1120 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1125 $taglist = implode( '|', $elements );
1126 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1128 while ( $text != '' ) {
1129 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1131 if ( count( $p ) < 5 ) {
1134 if ( count( $p ) > 5 ) {
1142 list( , $element, $attributes, $close, $inside ) = $p;
1145 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1146 $stripped .= $marker;
1148 if ( $close === '/>' ) {
1149 # Empty element tag, <tag />
1154 if ( $element === '!--' ) {
1157 $end = "/(<\\/$element\\s*>)/i";
1159 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1161 if ( count( $q ) < 3 ) {
1162 # No end tag -- let it run out to the end of the text.
1166 list( , $tail, $text ) = $q;
1170 $matches[$marker] = [ $element,
1172 Sanitizer
::decodeTagAttributes( $attributes ),
1173 "<$element$attributes$close$content$tail" ];
1179 * Get a list of strippable XML-like elements
1183 public function getStripList() {
1184 return $this->mStripList
;
1188 * Get the StripState
1190 * @return StripState
1192 public function getStripState() {
1193 return $this->mStripState
;
1197 * Add an item to the strip state
1198 * Returns the unique tag which must be inserted into the stripped text
1199 * The tag will be replaced with the original text in unstrip()
1201 * @param string $text
1205 public function insertStripItem( $text ) {
1206 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1207 $this->mMarkerIndex++
;
1208 $this->mStripState
->addGeneral( $marker, $text );
1213 * parse the wiki syntax used to render tables
1216 * @param string $text
1219 public function doTableStuff( $text ) {
1220 $lines = StringUtils
::explode( "\n", $text );
1222 $td_history = []; # Is currently a td tag open?
1223 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1224 $tr_history = []; # Is currently a tr tag open?
1225 $tr_attributes = []; # history of tr attributes
1226 $has_opened_tr = []; # Did this table open a <tr> element?
1227 $indent_level = 0; # indent level of the table
1229 foreach ( $lines as $outLine ) {
1230 $line = trim( $outLine );
1232 if ( $line === '' ) { # empty line, go to next line
1233 $out .= $outLine . "\n";
1237 $first_character = $line[0];
1238 $first_two = substr( $line, 0, 2 );
1241 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1242 # First check if we are starting a new table
1243 $indent_level = strlen( $matches[1] );
1245 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1246 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1248 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1249 array_push( $td_history, false );
1250 array_push( $last_tag_history, '' );
1251 array_push( $tr_history, false );
1252 array_push( $tr_attributes, '' );
1253 array_push( $has_opened_tr, false );
1254 } elseif ( count( $td_history ) == 0 ) {
1255 # Don't do any of the following
1256 $out .= $outLine . "\n";
1258 } elseif ( $first_two === '|}' ) {
1259 # We are ending a table
1260 $line = '</table>' . substr( $line, 2 );
1261 $last_tag = array_pop( $last_tag_history );
1263 if ( !array_pop( $has_opened_tr ) ) {
1264 $line = "<tr><td></td></tr>{$line}";
1267 if ( array_pop( $tr_history ) ) {
1268 $line = "</tr>{$line}";
1271 if ( array_pop( $td_history ) ) {
1272 $line = "</{$last_tag}>{$line}";
1274 array_pop( $tr_attributes );
1275 if ( $indent_level > 0 ) {
1276 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1280 } elseif ( $first_two === '|-' ) {
1281 # Now we have a table row
1282 $line = preg_replace( '#^\|-+#', '', $line );
1284 # Whats after the tag is now only attributes
1285 $attributes = $this->mStripState
->unstripBoth( $line );
1286 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1287 array_pop( $tr_attributes );
1288 array_push( $tr_attributes, $attributes );
1291 $last_tag = array_pop( $last_tag_history );
1292 array_pop( $has_opened_tr );
1293 array_push( $has_opened_tr, true );
1295 if ( array_pop( $tr_history ) ) {
1299 if ( array_pop( $td_history ) ) {
1300 $line = "</{$last_tag}>{$line}";
1304 array_push( $tr_history, false );
1305 array_push( $td_history, false );
1306 array_push( $last_tag_history, '' );
1307 } elseif ( $first_character === '|'
1308 ||
$first_character === '!'
1309 ||
$first_two === '|+'
1311 # This might be cell elements, td, th or captions
1312 if ( $first_two === '|+' ) {
1313 $first_character = '+';
1314 $line = substr( $line, 2 );
1316 $line = substr( $line, 1 );
1319 // Implies both are valid for table headings.
1320 if ( $first_character === '!' ) {
1321 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1324 # Split up multiple cells on the same line.
1325 # FIXME : This can result in improper nesting of tags processed
1326 # by earlier parser steps.
1327 $cells = explode( '||', $line );
1331 # Loop through each table cell
1332 foreach ( $cells as $cell ) {
1334 if ( $first_character !== '+' ) {
1335 $tr_after = array_pop( $tr_attributes );
1336 if ( !array_pop( $tr_history ) ) {
1337 $previous = "<tr{$tr_after}>\n";
1339 array_push( $tr_history, true );
1340 array_push( $tr_attributes, '' );
1341 array_pop( $has_opened_tr );
1342 array_push( $has_opened_tr, true );
1345 $last_tag = array_pop( $last_tag_history );
1347 if ( array_pop( $td_history ) ) {
1348 $previous = "</{$last_tag}>\n{$previous}";
1351 if ( $first_character === '|' ) {
1353 } elseif ( $first_character === '!' ) {
1355 } elseif ( $first_character === '+' ) {
1356 $last_tag = 'caption';
1361 array_push( $last_tag_history, $last_tag );
1363 # A cell could contain both parameters and data
1364 $cell_data = explode( '|', $cell, 2 );
1366 # T2553: Note that a '|' inside an invalid link should not
1367 # be mistaken as delimiting cell parameters
1368 # Bug T153140: Neither should language converter markup.
1369 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1370 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1371 } elseif ( count( $cell_data ) == 1 ) {
1372 // Whitespace in cells is trimmed
1373 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1375 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1376 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1377 // Whitespace in cells is trimmed
1378 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1382 array_push( $td_history, true );
1385 $out .= $outLine . "\n";
1388 # Closing open td, tr && table
1389 while ( count( $td_history ) > 0 ) {
1390 if ( array_pop( $td_history ) ) {
1393 if ( array_pop( $tr_history ) ) {
1396 if ( !array_pop( $has_opened_tr ) ) {
1397 $out .= "<tr><td></td></tr>\n";
1400 $out .= "</table>\n";
1403 # Remove trailing line-ending (b/c)
1404 if ( substr( $out, -1 ) === "\n" ) {
1405 $out = substr( $out, 0, -1 );
1408 # special case: don't return empty table
1409 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1417 * Helper function for parse() that transforms wiki markup into half-parsed
1418 * HTML. Only called for $mOutputType == self::OT_HTML.
1422 * @param string $text The text to parse
1423 * @param-taint $text escapes_html
1424 * @param bool $isMain Whether this is being called from the main parse() function
1425 * @param PPFrame|bool $frame A pre-processor frame
1429 public function internalParse( $text, $isMain = true, $frame = false ) {
1432 // Avoid PHP 7.1 warning from passing $this by reference
1435 # Hook to suspend the parser in this state
1436 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1440 # if $frame is provided, then use $frame for replacing any variables
1442 # use frame depth to infer how include/noinclude tags should be handled
1443 # depth=0 means this is the top-level document; otherwise it's an included document
1444 if ( !$frame->depth
) {
1447 $flag = self
::PTD_FOR_INCLUSION
;
1449 $dom = $this->preprocessToDom( $text, $flag );
1450 $text = $frame->expand( $dom );
1452 # if $frame is not provided, then use old-style replaceVariables
1453 $text = $this->replaceVariables( $text );
1456 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1457 $text = Sanitizer
::removeHTMLtags(
1459 [ $this, 'attributeStripCallback' ],
1461 array_keys( $this->mTransparentTagHooks
),
1463 [ $this, 'addTrackingCategory' ]
1465 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1467 # Tables need to come after variable replacement for things to work
1468 # properly; putting them before other transformations should keep
1469 # exciting things like link expansions from showing up in surprising
1471 $text = $this->doTableStuff( $text );
1473 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1475 $text = $this->doDoubleUnderscore( $text );
1477 $text = $this->doHeadings( $text );
1478 $text = $this->replaceInternalLinks( $text );
1479 $text = $this->doAllQuotes( $text );
1480 $text = $this->replaceExternalLinks( $text );
1482 # replaceInternalLinks may sometimes leave behind
1483 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1484 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1486 $text = $this->doMagicLinks( $text );
1487 $text = $this->formatHeadings( $text, $origText, $isMain );
1493 * Helper function for parse() that transforms half-parsed HTML into fully
1496 * @param string $text
1497 * @param bool $isMain
1498 * @param bool $linestart
1501 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1502 $text = $this->mStripState
->unstripGeneral( $text );
1504 // Avoid PHP 7.1 warning from passing $this by reference
1508 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1511 # Clean up special characters, only run once, next-to-last before doBlockLevels
1512 $text = Sanitizer
::armorFrenchSpaces( $text );
1514 $text = $this->doBlockLevels( $text, $linestart );
1516 $this->replaceLinkHolders( $text );
1519 * The input doesn't get language converted if
1521 * b) Content isn't converted
1522 * c) It's a conversion table
1523 * d) it is an interface message (which is in the user language)
1525 if ( !( $this->mOptions
->getDisableContentConversion()
1526 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1527 && !$this->mOptions
->getInterfaceMessage()
1529 # The position of the convert() call should not be changed. it
1530 # assumes that the links are all replaced and the only thing left
1531 # is the <nowiki> mark.
1532 $text = $this->getTargetLanguage()->convert( $text );
1535 $text = $this->mStripState
->unstripNoWiki( $text );
1538 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1541 $text = $this->replaceTransparentTags( $text );
1542 $text = $this->mStripState
->unstripGeneral( $text );
1544 $text = Sanitizer
::normalizeCharReferences( $text );
1546 if ( MWTidy
::isEnabled() ) {
1547 if ( $this->mOptions
->getTidy() ) {
1548 $text = MWTidy
::tidy( $text );
1551 # attempt to sanitize at least some nesting problems
1552 # (T4702 and quite a few others)
1553 # This code path is buggy and deprecated!
1554 wfDeprecated( 'disabling tidy', '1.33' );
1556 # ''Something [http://www.cool.com cool''] -->
1557 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1558 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1559 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1560 # fix up an anchor inside another anchor, only
1561 # at least for a single single nested link (T5695)
1562 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1563 '\\1\\2</a>\\3</a>\\1\\4</a>',
1564 # fix div inside inline elements- doBlockLevels won't wrap a line which
1565 # contains a div, so fix it up here; replace
1566 # div with escaped text
1567 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1568 '\\1\\3<div\\5>\\6</div>\\8\\9',
1569 # remove empty italic or bold tag pairs, some
1570 # introduced by rules above
1571 '/<([bi])><\/\\1>/' => '',
1574 $text = preg_replace(
1575 array_keys( $tidyregs ),
1576 array_values( $tidyregs ),
1581 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1588 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1589 * magic external links.
1594 * @param string $text
1598 public function doMagicLinks( $text ) {
1599 $prots = wfUrlProtocolsWithoutProtRel();
1600 $urlChar = self
::EXT_LINK_URL_CLASS
;
1601 $addr = self
::EXT_LINK_ADDR
;
1602 $space = self
::SPACE_NOT_NL
; # non-newline space
1603 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1604 $spaces = "$space++"; # possessive match of 1 or more spaces
1605 $text = preg_replace_callback(
1607 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1608 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1609 (\b # m[3]: Free external links
1611 ($addr$urlChar*) # m[4]: Post-protocol path
1613 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1615 \bISBN $spaces ( # m[6]: ISBN, capture number
1616 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1617 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1618 [0-9Xx] # check digit
1620 )!xu", [ $this, 'magicLinkCallback' ], $text );
1625 * @throws MWException
1627 * @return string HTML
1629 public function magicLinkCallback( $m ) {
1630 if ( isset( $m[1] ) && $m[1] !== '' ) {
1633 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1636 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1637 # Free external link
1638 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1639 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1641 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1642 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1647 $cssClass = 'mw-magiclink-rfc';
1648 $trackingCat = 'magiclink-tracking-rfc';
1650 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1651 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1655 $urlmsg = 'pubmedurl';
1656 $cssClass = 'mw-magiclink-pmid';
1657 $trackingCat = 'magiclink-tracking-pmid';
1660 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1661 substr( $m[0], 0, 20 ) . '"' );
1663 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1664 $this->addTrackingCategory( $trackingCat );
1665 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1666 } elseif ( isset( $m[6] ) && $m[6] !== ''
1667 && $this->mOptions
->getMagicISBNLinks()
1671 $space = self
::SPACE_NOT_NL
; # non-newline space
1672 $isbn = preg_replace( "/$space/", ' ', $isbn );
1673 $num = strtr( $isbn, [
1678 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1679 return $this->getLinkRenderer()->makeKnownLink(
1680 SpecialPage
::getTitleFor( 'Booksources', $num ),
1683 'class' => 'internal mw-magiclink-isbn',
1684 'title' => false // suppress title attribute
1693 * Make a free external link, given a user-supplied URL
1695 * @param string $url
1696 * @param int $numPostProto
1697 * The number of characters after the protocol.
1698 * @return string HTML
1701 public function makeFreeExternalLink( $url, $numPostProto ) {
1704 # The characters '<' and '>' (which were escaped by
1705 # removeHTMLtags()) should not be included in
1706 # URLs, per RFC 2396.
1707 # Make terminate a URL as well (bug T84937)
1710 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1715 $trail = substr( $url, $m2[0][1] ) . $trail;
1716 $url = substr( $url, 0, $m2[0][1] );
1719 # Move trailing punctuation to $trail
1721 # If there is no left bracket, then consider right brackets fair game too
1722 if ( strpos( $url, '(' ) === false ) {
1726 $urlRev = strrev( $url );
1727 $numSepChars = strspn( $urlRev, $sep );
1728 # Don't break a trailing HTML entity by moving the ; into $trail
1729 # This is in hot code, so use substr_compare to avoid having to
1730 # create a new string object for the comparison
1731 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1732 # more optimization: instead of running preg_match with a $
1733 # anchor, which can be slow, do the match on the reversed
1734 # string starting at the desired offset.
1735 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1736 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1740 if ( $numSepChars ) {
1741 $trail = substr( $url, -$numSepChars ) . $trail;
1742 $url = substr( $url, 0, -$numSepChars );
1745 # Verify that we still have a real URL after trail removal, and
1746 # not just lone protocol
1747 if ( strlen( $trail ) >= $numPostProto ) {
1748 return $url . $trail;
1751 $url = Sanitizer
::cleanUrl( $url );
1753 # Is this an external image?
1754 $text = $this->maybeMakeExternalImage( $url );
1755 if ( $text === false ) {
1756 # Not an image, make a link
1757 $text = Linker
::makeExternalLink( $url,
1758 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1760 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1761 # Register it in the output object...
1762 $this->mOutput
->addExternalLink( $url );
1764 return $text . $trail;
1768 * Parse headers and return html
1772 * @param string $text
1776 public function doHeadings( $text ) {
1777 for ( $i = 6; $i >= 1; --$i ) {
1778 $h = str_repeat( '=', $i );
1779 // Trim non-newline whitespace from headings
1780 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1781 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1787 * Replace single quotes with HTML markup
1790 * @param string $text
1792 * @return string The altered text
1794 public function doAllQuotes( $text ) {
1796 $lines = StringUtils
::explode( "\n", $text );
1797 foreach ( $lines as $line ) {
1798 $outtext .= $this->doQuotes( $line ) . "\n";
1800 $outtext = substr( $outtext, 0, -1 );
1805 * Helper function for doAllQuotes()
1807 * @param string $text
1811 public function doQuotes( $text ) {
1812 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1813 $countarr = count( $arr );
1814 if ( $countarr == 1 ) {
1818 // First, do some preliminary work. This may shift some apostrophes from
1819 // being mark-up to being text. It also counts the number of occurrences
1820 // of bold and italics mark-ups.
1823 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1824 $thislen = strlen( $arr[$i] );
1825 // If there are ever four apostrophes, assume the first is supposed to
1826 // be text, and the remaining three constitute mark-up for bold text.
1827 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1828 if ( $thislen == 4 ) {
1829 $arr[$i - 1] .= "'";
1832 } elseif ( $thislen > 5 ) {
1833 // If there are more than 5 apostrophes in a row, assume they're all
1834 // text except for the last 5.
1835 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1836 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1840 // Count the number of occurrences of bold and italics mark-ups.
1841 if ( $thislen == 2 ) {
1843 } elseif ( $thislen == 3 ) {
1845 } elseif ( $thislen == 5 ) {
1851 // If there is an odd number of both bold and italics, it is likely
1852 // that one of the bold ones was meant to be an apostrophe followed
1853 // by italics. Which one we cannot know for certain, but it is more
1854 // likely to be one that has a single-letter word before it.
1855 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1856 $firstsingleletterword = -1;
1857 $firstmultiletterword = -1;
1859 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1860 if ( strlen( $arr[$i] ) == 3 ) {
1861 $x1 = substr( $arr[$i - 1], -1 );
1862 $x2 = substr( $arr[$i - 1], -2, 1 );
1863 if ( $x1 === ' ' ) {
1864 if ( $firstspace == -1 ) {
1867 } elseif ( $x2 === ' ' ) {
1868 $firstsingleletterword = $i;
1869 // if $firstsingleletterword is set, we don't
1870 // look at the other options, so we can bail early.
1872 } elseif ( $firstmultiletterword == -1 ) {
1873 $firstmultiletterword = $i;
1878 // If there is a single-letter word, use it!
1879 if ( $firstsingleletterword > -1 ) {
1880 $arr[$firstsingleletterword] = "''";
1881 $arr[$firstsingleletterword - 1] .= "'";
1882 } elseif ( $firstmultiletterword > -1 ) {
1883 // If not, but there's a multi-letter word, use that one.
1884 $arr[$firstmultiletterword] = "''";
1885 $arr[$firstmultiletterword - 1] .= "'";
1886 } elseif ( $firstspace > -1 ) {
1887 // ... otherwise use the first one that has neither.
1888 // (notice that it is possible for all three to be -1 if, for example,
1889 // there is only one pentuple-apostrophe in the line)
1890 $arr[$firstspace] = "''";
1891 $arr[$firstspace - 1] .= "'";
1895 // Now let's actually convert our apostrophic mush to HTML!
1900 foreach ( $arr as $r ) {
1901 if ( ( $i %
2 ) == 0 ) {
1902 if ( $state === 'both' ) {
1908 $thislen = strlen( $r );
1909 if ( $thislen == 2 ) {
1910 if ( $state === 'i' ) {
1913 } elseif ( $state === 'bi' ) {
1916 } elseif ( $state === 'ib' ) {
1917 $output .= '</b></i><b>';
1919 } elseif ( $state === 'both' ) {
1920 $output .= '<b><i>' . $buffer . '</i>';
1922 } else { // $state can be 'b' or ''
1926 } elseif ( $thislen == 3 ) {
1927 if ( $state === 'b' ) {
1930 } elseif ( $state === 'bi' ) {
1931 $output .= '</i></b><i>';
1933 } elseif ( $state === 'ib' ) {
1936 } elseif ( $state === 'both' ) {
1937 $output .= '<i><b>' . $buffer . '</b>';
1939 } else { // $state can be 'i' or ''
1943 } elseif ( $thislen == 5 ) {
1944 if ( $state === 'b' ) {
1945 $output .= '</b><i>';
1947 } elseif ( $state === 'i' ) {
1948 $output .= '</i><b>';
1950 } elseif ( $state === 'bi' ) {
1951 $output .= '</i></b>';
1953 } elseif ( $state === 'ib' ) {
1954 $output .= '</b></i>';
1956 } elseif ( $state === 'both' ) {
1957 $output .= '<i><b>' . $buffer . '</b></i>';
1959 } else { // ($state == '')
1967 // Now close all remaining tags. Notice that the order is important.
1968 if ( $state === 'b' ||
$state === 'ib' ) {
1971 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1974 if ( $state === 'bi' ) {
1977 // There might be lonely ''''', so make sure we have a buffer
1978 if ( $state === 'both' && $buffer ) {
1979 $output .= '<b><i>' . $buffer . '</i></b>';
1985 * Replace external links (REL)
1987 * Note: this is all very hackish and the order of execution matters a lot.
1988 * Make sure to run tests/parser/parserTests.php if you change this code.
1992 * @param string $text
1994 * @throws MWException
1997 public function replaceExternalLinks( $text ) {
1998 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1999 if ( $bits === false ) {
2000 throw new MWException( "PCRE needs to be compiled with "
2001 . "--enable-unicode-properties in order for MediaWiki to function" );
2003 $s = array_shift( $bits );
2006 while ( $i < count( $bits ) ) {
2009 $text = $bits[$i++
];
2010 $trail = $bits[$i++
];
2012 # The characters '<' and '>' (which were escaped by
2013 # removeHTMLtags()) should not be included in
2014 # URLs, per RFC 2396.
2016 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
2017 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
2018 $url = substr( $url, 0, $m2[0][1] );
2021 # If the link text is an image URL, replace it with an <img> tag
2022 # This happened by accident in the original parser, but some people used it extensively
2023 $img = $this->maybeMakeExternalImage( $text );
2024 if ( $img !== false ) {
2030 # Set linktype for CSS
2033 # No link text, e.g. [http://domain.tld/some.link]
2034 if ( $text == '' ) {
2036 $langObj = $this->getTargetLanguage();
2037 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
2038 $linktype = 'autonumber';
2040 # Have link text, e.g. [http://domain.tld/some.link text]s
2042 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
2045 // Excluding protocol-relative URLs may avoid many false positives.
2046 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2047 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2050 $url = Sanitizer
::cleanUrl( $url );
2052 # Use the encoded URL
2053 # This means that users can paste URLs directly into the text
2054 # Funny characters like ö aren't valid in URLs anyway
2055 # This was changed in August 2004
2056 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
2057 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
2059 # Register link in the output object.
2060 $this->mOutput
->addExternalLink( $url );
2067 * Get the rel attribute for a particular external link.
2070 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2071 * nofollow if appropriate
2072 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2073 * @return string|null Rel attribute for $url
2075 public static function getExternalLinkRel( $url = false, $title = null ) {
2076 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2077 $ns = $title ?
$title->getNamespace() : false;
2078 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2079 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2087 * Get an associative array of additional HTML attributes appropriate for a
2088 * particular external link. This currently may include rel => nofollow
2089 * (depending on configuration, namespace, and the URL's domain) and/or a
2090 * target attribute (depending on configuration).
2092 * @param string $url URL to extract the domain from for rel =>
2093 * nofollow if appropriate
2094 * @return array Associative array of HTML attributes
2096 public function getExternalLinkAttribs( $url ) {
2098 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2100 $target = $this->mOptions
->getExternalLinkTarget();
2102 $attribs['target'] = $target;
2103 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2104 // T133507. New windows can navigate parent cross-origin.
2105 // Including noreferrer due to lacking browser
2106 // support of noopener. Eventually noreferrer should be removed.
2107 if ( $rel !== '' ) {
2110 $rel .= 'noreferrer noopener';
2113 $attribs['rel'] = $rel;
2118 * Replace unusual escape codes in a URL with their equivalent characters
2120 * This generally follows the syntax defined in RFC 3986, with special
2121 * consideration for HTTP query strings.
2123 * @param string $url
2126 public static function normalizeLinkUrl( $url ) {
2127 # Test for RFC 3986 IPv6 syntax
2128 $scheme = '[a-z][a-z0-9+.-]*:';
2129 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2130 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2131 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2132 IP
::isValid( rawurldecode( $m[1] ) )
2134 $isIPv6 = rawurldecode( $m[1] );
2139 # Make sure unsafe characters are encoded
2140 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2142 return rawurlencode( $m[0] );
2148 $end = strlen( $url );
2150 # Fragment part - 'fragment'
2151 $start = strpos( $url, '#' );
2152 if ( $start !== false && $start < $end ) {
2153 $ret = self
::normalizeUrlComponent(
2154 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2158 # Query part - 'query' minus &=+;
2159 $start = strpos( $url, '?' );
2160 if ( $start !== false && $start < $end ) {
2161 $ret = self
::normalizeUrlComponent(
2162 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2166 # Scheme and path part - 'pchar'
2167 # (we assume no userinfo or encoded colons in the host)
2168 $ret = self
::normalizeUrlComponent(
2169 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2172 if ( $isIPv6 !== false ) {
2173 $ipv6Host = "%5B({$isIPv6})%5D";
2174 $ret = preg_replace(
2175 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2184 private static function normalizeUrlComponent( $component, $unsafe ) {
2185 $callback = function ( $matches ) use ( $unsafe ) {
2186 $char = urldecode( $matches[0] );
2187 $ord = ord( $char );
2188 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2192 # Leave it escaped, but use uppercase for a-f
2193 return strtoupper( $matches[0] );
2196 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2200 * make an image if it's allowed, either through the global
2201 * option, through the exception, or through the on-wiki whitelist
2203 * @param string $url
2207 private function maybeMakeExternalImage( $url ) {
2208 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2209 $imagesexception = !empty( $imagesfrom );
2211 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2212 if ( $imagesexception && is_array( $imagesfrom ) ) {
2213 $imagematch = false;
2214 foreach ( $imagesfrom as $match ) {
2215 if ( strpos( $url, $match ) === 0 ) {
2220 } elseif ( $imagesexception ) {
2221 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2223 $imagematch = false;
2226 if ( $this->mOptions
->getAllowExternalImages()
2227 ||
( $imagesexception && $imagematch )
2229 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2231 $text = Linker
::makeExternalImage( $url );
2234 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2235 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2237 $whitelist = explode(
2239 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2242 foreach ( $whitelist as $entry ) {
2243 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2244 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2247 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2248 # Image matches a whitelist entry
2249 $text = Linker
::makeExternalImage( $url );
2258 * Process [[ ]] wikilinks
2262 * @return string Processed text
2266 public function replaceInternalLinks( $s ) {
2267 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2272 * Process [[ ]] wikilinks (RIL)
2274 * @throws MWException
2275 * @return LinkHolderArray
2279 public function replaceInternalLinks2( &$s ) {
2280 static $tc = false, $e1, $e1_img;
2281 # the % is needed to support urlencoded titles as well
2283 $tc = Title
::legalChars() . '#%';
2284 # Match a link having the form [[namespace:link|alternate]]trail
2285 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2286 # Match cases where there is no "]]", which might still be images
2287 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2290 $holders = new LinkHolderArray( $this );
2292 # split the entire text string on occurrences of [[
2293 $a = StringUtils
::explode( '[[', ' ' . $s );
2294 # get the first element (all text up to first [[), and remove the space we added
2297 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2298 $s = substr( $s, 1 );
2300 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2302 if ( $useLinkPrefixExtension ) {
2303 # Match the end of a line for a word that's not followed by whitespace,
2304 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2305 $charset = $this->contLang
->linkPrefixCharset();
2306 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2309 if ( is_null( $this->mTitle
) ) {
2310 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2312 $nottalk = !$this->mTitle
->isTalkPage();
2314 if ( $useLinkPrefixExtension ) {
2316 if ( preg_match( $e2, $s, $m ) ) {
2317 $first_prefix = $m[2];
2319 $first_prefix = false;
2325 $useSubpages = $this->areSubpagesAllowed();
2327 # Loop for each link
2328 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2329 # Check for excessive memory usage
2330 if ( $holders->isBig() ) {
2332 # Do the existence check, replace the link holders and clear the array
2333 $holders->replace( $s );
2337 if ( $useLinkPrefixExtension ) {
2338 if ( preg_match( $e2, $s, $m ) ) {
2339 list( , $s, $prefix ) = $m;
2344 if ( $first_prefix ) {
2345 $prefix = $first_prefix;
2346 $first_prefix = false;
2350 $might_be_img = false;
2352 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2354 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2355 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2356 # the real problem is with the $e1 regex
2358 # Still some problems for cases where the ] is meant to be outside punctuation,
2359 # and no image is in sight. See T4095.
2361 && substr( $m[3], 0, 1 ) === ']'
2362 && strpos( $text, '[' ) !== false
2364 $text .= ']'; # so that replaceExternalLinks($text) works later
2365 $m[3] = substr( $m[3], 1 );
2367 # fix up urlencoded title texts
2368 if ( strpos( $m[1], '%' ) !== false ) {
2369 # Should anchors '#' also be rejected?
2370 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2373 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2374 # Invalid, but might be an image with a link in its caption
2375 $might_be_img = true;
2377 if ( strpos( $m[1], '%' ) !== false ) {
2378 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2381 } else { # Invalid form; output directly
2382 $s .= $prefix . '[[' . $line;
2386 $origLink = ltrim( $m[1], ' ' );
2388 # Don't allow internal links to pages containing
2389 # PROTO: where PROTO is a valid URL protocol; these
2390 # should be external links.
2391 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2392 $s .= $prefix . '[[' . $line;
2396 # Make subpage if necessary
2397 if ( $useSubpages ) {
2398 $link = $this->maybeDoSubpageLink( $origLink, $text );
2403 // \x7f isn't a default legal title char, so most likely strip
2404 // markers will force us into the "invalid form" path above. But,
2405 // just in case, let's assert that xmlish tags aren't valid in
2406 // the title position.
2407 $unstrip = $this->mStripState
->killMarkers( $link );
2408 $noMarkers = ( $unstrip === $link );
2410 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2411 if ( $nt === null ) {
2412 $s .= $prefix . '[[' . $line;
2416 $ns = $nt->getNamespace();
2417 $iw = $nt->getInterwiki();
2419 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2421 if ( $might_be_img ) { # if this is actually an invalid link
2422 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2425 # look at the next 'line' to see if we can close it there
2427 $next_line = $a->current();
2428 if ( $next_line === false ||
$next_line === null ) {
2431 $m = explode( ']]', $next_line, 3 );
2432 if ( count( $m ) == 3 ) {
2433 # the first ]] closes the inner link, the second the image
2435 $text .= "[[{$m[0]}]]{$m[1]}";
2438 } elseif ( count( $m ) == 2 ) {
2439 # if there's exactly one ]] that's fine, we'll keep looking
2440 $text .= "[[{$m[0]}]]{$m[1]}";
2442 # if $next_line is invalid too, we need look no further
2443 $text .= '[[' . $next_line;
2448 # we couldn't find the end of this imageLink, so output it raw
2449 # but don't ignore what might be perfectly normal links in the text we've examined
2450 $holders->merge( $this->replaceInternalLinks2( $text ) );
2451 $s .= "{$prefix}[[$link|$text";
2452 # note: no $trail, because without an end, there *is* no trail
2455 } else { # it's not an image, so output it raw
2456 $s .= "{$prefix}[[$link|$text";
2457 # note: no $trail, because without an end, there *is* no trail
2462 $wasblank = ( $text == '' );
2466 # Strip off leading ':'
2467 $text = substr( $text, 1 );
2470 # T6598 madness. Handle the quotes only if they come from the alternate part
2471 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2472 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2473 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2474 $text = $this->doQuotes( $text );
2477 # Link not escaped by : , create the various objects
2478 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2481 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2482 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2483 in_array( $iw, $this->svcOptions
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2486 # T26502: filter duplicates
2487 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2488 $this->mLangLinkLanguages
[$iw] = true;
2489 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2493 * Strip the whitespace interwiki links produce, see T10897
2495 $s = rtrim( $s . $prefix ) . $trail; # T175416
2499 if ( $ns == NS_FILE
) {
2500 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2502 # if no parameters were passed, $text
2503 # becomes something like "File:Foo.png",
2504 # which we don't want to pass on to the
2508 # recursively parse links inside the image caption
2509 # actually, this will parse them in any other parameters, too,
2510 # but it might be hard to fix that, and it doesn't matter ATM
2511 $text = $this->replaceExternalLinks( $text );
2512 $holders->merge( $this->replaceInternalLinks2( $text ) );
2514 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2515 $s .= $prefix . $this->armorLinks(
2516 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2519 } elseif ( $ns == NS_CATEGORY
) {
2521 * Strip the whitespace Category links produce, see T2087
2523 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2526 $sortkey = $this->getDefaultSort();
2530 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2531 $sortkey = str_replace( "\n", '', $sortkey );
2532 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2533 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2539 # Self-link checking. For some languages, variants of the title are checked in
2540 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2541 # for linking to a different variant.
2542 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2543 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2547 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2548 # @todo FIXME: Should do batch file existence checks, see comment below
2549 if ( $ns == NS_MEDIA
) {
2550 # Give extensions a chance to select the file revision for us
2553 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2554 [ $this, $nt, &$options, &$descQuery ] );
2555 # Fetch and register the file (file title may be different via hooks)
2556 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2557 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2558 $s .= $prefix . $this->armorLinks(
2559 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2563 # Some titles, such as valid special pages or files in foreign repos, should
2564 # be shown as bluelinks even though they're not included in the page table
2565 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2566 # batch file existence checks for NS_FILE and NS_MEDIA
2567 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2568 $this->mOutput
->addLink( $nt );
2569 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2571 # Links will be added to the output link list after checking
2572 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2579 * Render a forced-blue link inline; protect against double expansion of
2580 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2581 * Since this little disaster has to split off the trail text to avoid
2582 * breaking URLs in the following text without breaking trails on the
2583 * wiki links, it's been made into a horrible function.
2586 * @param string $text
2587 * @param string $trail
2588 * @param string $prefix
2589 * @return string HTML-wikitext mix oh yuck
2591 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2592 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2594 if ( $text == '' ) {
2595 $text = htmlspecialchars( $nt->getPrefixedText() );
2598 $link = $this->getLinkRenderer()->makeKnownLink(
2599 $nt, new HtmlArmor( "$prefix$text$inside" )
2602 return $this->armorLinks( $link ) . $trail;
2606 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2607 * going to go through further parsing steps before inline URL expansion.
2609 * Not needed quite as much as it used to be since free links are a bit
2610 * more sensible these days. But bracketed links are still an issue.
2612 * @param string $text More-or-less HTML
2613 * @return string Less-or-more HTML with NOPARSE bits
2615 public function armorLinks( $text ) {
2616 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2617 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2621 * Return true if subpage links should be expanded on this page.
2624 public function areSubpagesAllowed() {
2625 # Some namespaces don't allow subpages
2626 return $this->nsInfo
->hasSubpages( $this->mTitle
->getNamespace() );
2630 * Handle link to subpage if necessary
2632 * @param string $target The source of the link
2633 * @param string &$text The link text, modified as necessary
2634 * @return string The full name of the link
2637 public function maybeDoSubpageLink( $target, &$text ) {
2638 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2642 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2644 * @param string $text
2645 * @param bool $linestart Whether or not this is at the start of a line.
2647 * @return string The lists rendered as HTML
2649 public function doBlockLevels( $text, $linestart ) {
2650 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2654 * Return value of a magic variable (like PAGENAME)
2658 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2659 * @param bool|PPFrame $frame
2661 * @throws MWException
2664 public function getVariableValue( $index, $frame = false ) {
2665 if ( is_null( $this->mTitle
) ) {
2666 // If no title set, bad things are going to happen
2667 // later. Title should always be set since this
2668 // should only be called in the middle of a parse
2669 // operation (but the unit-tests do funky stuff)
2670 throw new MWException( __METHOD__
. ' Should only be '
2671 . ' called while parsing (no title set)' );
2674 // Avoid PHP 7.1 warning from passing $this by reference
2678 * Some of these require message or data lookups and can be
2679 * expensive to check many times.
2682 Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) &&
2683 isset( $this->mVarCache
[$index] )
2685 return $this->mVarCache
[$index];
2688 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2689 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2691 $pageLang = $this->getFunctionLang();
2697 case 'currentmonth':
2698 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2700 case 'currentmonth1':
2701 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2703 case 'currentmonthname':
2704 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2706 case 'currentmonthnamegen':
2707 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2709 case 'currentmonthabbrev':
2710 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2713 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2716 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2719 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2722 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2724 case 'localmonthname':
2725 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2727 case 'localmonthnamegen':
2728 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2730 case 'localmonthabbrev':
2731 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2734 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2737 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2740 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2743 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2745 case 'fullpagename':
2746 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2748 case 'fullpagenamee':
2749 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2752 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2754 case 'subpagenamee':
2755 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2757 case 'rootpagename':
2758 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2760 case 'rootpagenamee':
2761 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2764 $this->mTitle
->getRootText()
2767 case 'basepagename':
2768 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2770 case 'basepagenamee':
2771 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2774 $this->mTitle
->getBaseText()
2777 case 'talkpagename':
2778 if ( $this->mTitle
->canHaveTalkPage() ) {
2779 $talkPage = $this->mTitle
->getTalkPage();
2780 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2785 case 'talkpagenamee':
2786 if ( $this->mTitle
->canHaveTalkPage() ) {
2787 $talkPage = $this->mTitle
->getTalkPage();
2788 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2793 case 'subjectpagename':
2794 $subjPage = $this->mTitle
->getSubjectPage();
2795 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2797 case 'subjectpagenamee':
2798 $subjPage = $this->mTitle
->getSubjectPage();
2799 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2801 case 'pageid': // requested in T25427
2802 # Inform the edit saving system that getting the canonical output
2803 # after page insertion requires a parse that used that exact page ID
2804 $this->setOutputFlag( 'vary-page-id', '{{PAGEID}} used' );
2805 $value = $this->mTitle
->getArticleID();
2807 $value = $this->mOptions
->getSpeculativePageId();
2809 $this->mOutput
->setSpeculativePageIdUsed( $value );
2815 $this->svcOptions
->get( 'MiserMode' ) &&
2816 !$this->mOptions
->getInterfaceMessage() &&
2817 // @TODO: disallow this word on all namespaces
2818 $this->nsInfo
->isContent( $this->mTitle
->getNamespace() )
2820 // Use a stub result instead of the actual revision ID in order to avoid
2821 // double parses on page save but still allow preview detection (T137900)
2822 if ( $this->getRevisionId() ||
$this->mOptions
->getSpeculativeRevId() ) {
2825 $this->setOutputFlag( 'vary-revision-exists', '{{REVISIONID}} used' );
2829 # Inform the edit saving system that getting the canonical output after
2830 # revision insertion requires a parse that used that exact revision ID
2831 $this->setOutputFlag( 'vary-revision-id', '{{REVISIONID}} used' );
2832 $value = $this->getRevisionId();
2833 if ( $value === 0 ) {
2834 $rev = $this->getRevisionObject();
2835 $value = $rev ?
$rev->getId() : $value;
2838 $value = $this->mOptions
->getSpeculativeRevId();
2840 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2846 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2848 case 'revisionday2':
2849 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2851 case 'revisionmonth':
2852 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2854 case 'revisionmonth1':
2855 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2857 case 'revisionyear':
2858 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2860 case 'revisiontimestamp':
2861 $value = $this->getRevisionTimestampSubstring( 0, 14, self
::MAX_TTS
, $index );
2863 case 'revisionuser':
2864 # Inform the edit saving system that getting the canonical output after
2865 # revision insertion requires a parse that used the actual user ID
2866 $this->setOutputFlag( 'vary-user', '{{REVISIONUSER}} used' );
2867 $value = $this->getRevisionUser();
2869 case 'revisionsize':
2870 $value = $this->getRevisionSize();
2873 $value = str_replace( '_', ' ',
2874 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2877 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2879 case 'namespacenumber':
2880 $value = $this->mTitle
->getNamespace();
2883 $value = $this->mTitle
->canHaveTalkPage()
2884 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2888 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2890 case 'subjectspace':
2891 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2893 case 'subjectspacee':
2894 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2896 case 'currentdayname':
2897 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2900 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2903 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2906 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2909 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2910 # int to remove the padding
2911 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2914 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2916 case 'localdayname':
2917 $value = $pageLang->getWeekdayName(
2918 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2922 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2925 $value = $pageLang->time(
2926 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2932 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2935 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2936 # int to remove the padding
2937 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2940 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2942 case 'numberofarticles':
2943 $value = $pageLang->formatNum( SiteStats
::articles() );
2945 case 'numberoffiles':
2946 $value = $pageLang->formatNum( SiteStats
::images() );
2948 case 'numberofusers':
2949 $value = $pageLang->formatNum( SiteStats
::users() );
2951 case 'numberofactiveusers':
2952 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2954 case 'numberofpages':
2955 $value = $pageLang->formatNum( SiteStats
::pages() );
2957 case 'numberofadmins':
2958 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2960 case 'numberofedits':
2961 $value = $pageLang->formatNum( SiteStats
::edits() );
2963 case 'currenttimestamp':
2964 $value = wfTimestamp( TS_MW
, $ts );
2966 case 'localtimestamp':
2967 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2969 case 'currentversion':
2970 $value = SpecialVersion
::getVersion();
2973 return $this->svcOptions
->get( 'ArticlePath' );
2975 return $this->svcOptions
->get( 'Sitename' );
2977 return $this->svcOptions
->get( 'Server' );
2979 return $this->svcOptions
->get( 'ServerName' );
2981 return $this->svcOptions
->get( 'ScriptPath' );
2983 return $this->svcOptions
->get( 'StylePath' );
2984 case 'directionmark':
2985 return $pageLang->getDirMark();
2986 case 'contentlanguage':
2987 return $this->svcOptions
->get( 'LanguageCode' );
2988 case 'pagelanguage':
2989 $value = $pageLang->getCode();
2991 case 'cascadingsources':
2992 $value = CoreParserFunctions
::cascadingsources( $this );
2997 'ParserGetVariableValueSwitch',
2998 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
3005 $this->mVarCache
[$index] = $value;
3014 * @param int $mtts Max time-till-save; sets vary-revision-timestamp if result changes by then
3015 * @param string $variable Parser variable name
3018 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
3019 # Get the timezone-adjusted timestamp to be used for this revision
3020 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
3021 # Possibly set vary-revision if there is not yet an associated revision
3022 if ( !$this->getRevisionObject() ) {
3023 # Get the timezone-adjusted timestamp $mtts seconds in the future.
3024 # This future is relative to the current time and not that of the
3025 # parser options. The rendered timestamp can be compared to that
3026 # of the timestamp specified by the parser options.
3028 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
3033 if ( $resNow !== $resThen ) {
3034 # Inform the edit saving system that getting the canonical output after
3035 # revision insertion requires a parse that used an actual revision timestamp
3036 $this->setOutputFlag( 'vary-revision-timestamp', "$variable used" );
3044 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3048 public function initialiseVariables() {
3049 $variableIDs = $this->magicWordFactory
->getVariableIDs();
3050 $substIDs = $this->magicWordFactory
->getSubstIDs();
3052 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
3053 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
3057 * Preprocess some wikitext and return the document tree.
3058 * This is the ghost of replace_variables().
3060 * @param string $text The text to parse
3061 * @param int $flags Bitwise combination of:
3062 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3063 * included. Default is to assume a direct page view.
3065 * The generated DOM tree must depend only on the input text and the flags.
3066 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3068 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3069 * change in the DOM tree for a given text, must be passed through the section identifier
3070 * in the section edit link and thus back to extractSections().
3072 * The output of this function is currently only cached in process memory, but a persistent
3073 * cache may be implemented at a later date which takes further advantage of these strict
3074 * dependency requirements.
3078 public function preprocessToDom( $text, $flags = 0 ) {
3079 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3084 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3090 public static function splitWhitespace( $s ) {
3091 $ltrimmed = ltrim( $s );
3092 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3093 $trimmed = rtrim( $ltrimmed );
3094 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3096 $w2 = substr( $ltrimmed, -$diff );
3100 return [ $w1, $trimmed, $w2 ];
3104 * Replace magic variables, templates, and template arguments
3105 * with the appropriate text. Templates are substituted recursively,
3106 * taking care to avoid infinite loops.
3108 * Note that the substitution depends on value of $mOutputType:
3109 * self::OT_WIKI: only {{subst:}} templates
3110 * self::OT_PREPROCESS: templates but not extension tags
3111 * self::OT_HTML: all templates and extension tags
3113 * @param string $text The text to transform
3114 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3115 * template. Arguments may also be provided as an associative array, as
3116 * was the usual case before MW1.12. Providing arguments this way may be
3117 * useful for extensions wishing to perform variable replacement
3119 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3120 * double-brace expansion.
3123 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3124 # Is there any text? Also, Prevent too big inclusions!
3125 $textSize = strlen( $text );
3126 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3130 if ( $frame === false ) {
3131 $frame = $this->getPreprocessor()->newFrame();
3132 } elseif ( !( $frame instanceof PPFrame
) ) {
3133 $this->logger
->debug(
3134 __METHOD__
. " called using plain parameters instead of " .
3135 "a PPFrame instance. Creating custom frame."
3137 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3140 $dom = $this->preprocessToDom( $text );
3141 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3142 $text = $frame->expand( $dom, $flags );
3148 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3150 * @param array $args
3154 public static function createAssocArgs( $args ) {
3157 foreach ( $args as $arg ) {
3158 $eqpos = strpos( $arg, '=' );
3159 if ( $eqpos === false ) {
3160 $assocArgs[$index++
] = $arg;
3162 $name = trim( substr( $arg, 0, $eqpos ) );
3163 $value = trim( substr( $arg, $eqpos +
1 ) );
3164 if ( $value === false ) {
3167 if ( $name !== false ) {
3168 $assocArgs[$name] = $value;
3177 * Warn the user when a parser limitation is reached
3178 * Will warn at most once the user per limitation type
3180 * The results are shown during preview and run through the Parser (See EditPage.php)
3182 * @param string $limitationType Should be one of:
3183 * 'expensive-parserfunction' (corresponding messages:
3184 * 'expensive-parserfunction-warning',
3185 * 'expensive-parserfunction-category')
3186 * 'post-expand-template-argument' (corresponding messages:
3187 * 'post-expand-template-argument-warning',
3188 * 'post-expand-template-argument-category')
3189 * 'post-expand-template-inclusion' (corresponding messages:
3190 * 'post-expand-template-inclusion-warning',
3191 * 'post-expand-template-inclusion-category')
3192 * 'node-count-exceeded' (corresponding messages:
3193 * 'node-count-exceeded-warning',
3194 * 'node-count-exceeded-category')
3195 * 'expansion-depth-exceeded' (corresponding messages:
3196 * 'expansion-depth-exceeded-warning',
3197 * 'expansion-depth-exceeded-category')
3198 * @param string|int|null $current Current value
3199 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3200 * exceeded, provide the values (optional)
3202 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3203 # does no harm if $current and $max are present but are unnecessary for the message
3204 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3205 # only during preview, and that would split the parser cache unnecessarily.
3206 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3208 $this->mOutput
->addWarning( $warning );
3209 $this->addTrackingCategory( "$limitationType-category" );
3213 * Return the text of a template, after recursively
3214 * replacing any variables or templates within the template.
3216 * @param array $piece The parts of the template
3217 * $piece['title']: the title, i.e. the part before the |
3218 * $piece['parts']: the parameter array
3219 * $piece['lineStart']: whether the brace was at the start of a line
3220 * @param PPFrame $frame The current frame, contains template arguments
3222 * @return string|array The text of the template
3224 public function braceSubstitution( $piece, $frame ) {
3227 // $text has been filled
3229 // wiki markup in $text should be escaped
3231 // $text is HTML, armour it against wikitext transformation
3233 // Force interwiki transclusion to be done in raw mode not rendered
3234 $forceRawInterwiki = false;
3235 // $text is a DOM node needing expansion in a child frame
3236 $isChildObj = false;
3237 // $text is a DOM node needing expansion in the current frame
3238 $isLocalObj = false;
3240 # Title object, where $text came from
3243 # $part1 is the bit before the first |, and must contain only title characters.
3244 # Various prefixes will be stripped from it later.
3245 $titleWithSpaces = $frame->expand( $piece['title'] );
3246 $part1 = trim( $titleWithSpaces );
3249 # Original title text preserved for various purposes
3250 $originalTitle = $part1;
3252 # $args is a list of argument nodes, starting from index 0, not including $part1
3253 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3254 # below won't work b/c this $args isn't an object
3255 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3257 $profileSection = null; // profile templates
3261 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3263 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3264 # Decide whether to expand template or keep wikitext as-is.
3265 if ( $this->ot
['wiki'] ) {
3266 if ( $substMatch === false ) {
3267 $literal = true; # literal when in PST with no prefix
3269 $literal = false; # expand when in PST with subst: or safesubst:
3272 if ( $substMatch == 'subst' ) {
3273 $literal = true; # literal when not in PST with plain subst:
3275 $literal = false; # expand when not in PST with safesubst: or no prefix
3279 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3286 if ( !$found && $args->getLength() == 0 ) {
3287 $id = $this->mVariables
->matchStartToEnd( $part1 );
3288 if ( $id !== false ) {
3289 $text = $this->getVariableValue( $id, $frame );
3290 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3291 $this->mOutput
->updateCacheExpiry(
3292 $this->magicWordFactory
->getCacheTTL( $id ) );
3298 # MSG, MSGNW and RAW
3301 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3302 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3305 # Remove obsolete MSG:
3306 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3307 $mwMsg->matchStartAndRemove( $part1 );
3311 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3312 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3313 $forceRawInterwiki = true;
3319 $colonPos = strpos( $part1, ':' );
3320 if ( $colonPos !== false ) {
3321 $func = substr( $part1, 0, $colonPos );
3322 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3323 $argsLength = $args->getLength();
3324 for ( $i = 0; $i < $argsLength; $i++
) {
3325 $funcArgs[] = $args->item( $i );
3328 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3330 // Extract any forwarded flags
3331 if ( isset( $result['title'] ) ) {
3332 $title = $result['title'];
3334 if ( isset( $result['found'] ) ) {
3335 $found = $result['found'];
3337 if ( array_key_exists( 'text', $result ) ) {
3339 $text = $result['text'];
3341 if ( isset( $result['nowiki'] ) ) {
3342 $nowiki = $result['nowiki'];
3344 if ( isset( $result['isHTML'] ) ) {
3345 $isHTML = $result['isHTML'];
3347 if ( isset( $result['forceRawInterwiki'] ) ) {
3348 $forceRawInterwiki = $result['forceRawInterwiki'];
3350 if ( isset( $result['isChildObj'] ) ) {
3351 $isChildObj = $result['isChildObj'];
3353 if ( isset( $result['isLocalObj'] ) ) {
3354 $isLocalObj = $result['isLocalObj'];
3359 # Finish mangling title and then check for loops.
3360 # Set $title to a Title object and $titleText to the PDBK
3363 # Split the title into page and subpage
3365 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3366 if ( $part1 !== $relative ) {
3368 $ns = $this->mTitle
->getNamespace();
3370 $title = Title
::newFromText( $part1, $ns );
3372 $titleText = $title->getPrefixedText();
3373 # Check for language variants if the template is not found
3374 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3375 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3377 # Do recursion depth check
3378 $limit = $this->mOptions
->getMaxTemplateDepth();
3379 if ( $frame->depth
>= $limit ) {
3381 $text = '<span class="error">'
3382 . wfMessage( 'parser-template-recursion-depth-warning' )
3383 ->numParams( $limit )->inContentLanguage()->text()
3389 # Load from database
3390 if ( !$found && $title ) {
3391 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3392 if ( !$title->isExternal() ) {
3393 if ( $title->isSpecialPage()
3394 && $this->mOptions
->getAllowSpecialInclusion()
3395 && $this->ot
['html']
3397 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3398 // Pass the template arguments as URL parameters.
3399 // "uselang" will have no effect since the Language object
3400 // is forced to the one defined in ParserOptions.
3402 $argsLength = $args->getLength();
3403 for ( $i = 0; $i < $argsLength; $i++
) {
3404 $bits = $args->item( $i )->splitArg();
3405 if ( strval( $bits['index'] ) === '' ) {
3406 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3407 $value = trim( $frame->expand( $bits['value'] ) );
3408 $pageArgs[$name] = $value;
3412 // Create a new context to execute the special page
3413 $context = new RequestContext
;
3414 $context->setTitle( $title );
3415 $context->setRequest( new FauxRequest( $pageArgs ) );
3416 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3417 $context->setUser( $this->getUser() );
3419 // If this page is cached, then we better not be per user.
3420 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3422 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3423 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3425 $text = $context->getOutput()->getHTML();
3426 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3429 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3430 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3431 $specialPage->maxIncludeCacheTime()
3435 } elseif ( $this->nsInfo
->isNonincludable( $title->getNamespace() ) ) {
3436 $found = false; # access denied
3437 $this->logger
->debug(
3439 ": template inclusion denied for " . $title->getPrefixedDBkey()
3442 list( $text, $title ) = $this->getTemplateDom( $title );
3443 if ( $text !== false ) {
3449 # If the title is valid but undisplayable, make a link to it
3450 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3451 $text = "[[:$titleText]]";
3454 } elseif ( $title->isTrans() ) {
3455 # Interwiki transclusion
3456 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3457 $text = $this->interwikiTransclude( $title, 'render' );
3460 $text = $this->interwikiTransclude( $title, 'raw' );
3461 # Preprocess it like a template
3462 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3468 # Do infinite loop check
3469 # This has to be done after redirect resolution to avoid infinite loops via redirects
3470 if ( !$frame->loopCheck( $title ) ) {
3472 $text = '<span class="error">'
3473 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3475 $this->addTrackingCategory( 'template-loop-category' );
3476 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3477 wfEscapeWikiText( $titleText ) )->text() );
3478 $this->logger
->debug( __METHOD__
. ": template loop broken at '$titleText'" );
3482 # If we haven't found text to substitute by now, we're done
3483 # Recover the source wikitext and return it
3485 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3486 if ( $profileSection ) {
3487 $this->mProfiler
->scopedProfileOut( $profileSection );
3489 return [ 'object' => $text ];
3492 # Expand DOM-style return values in a child frame
3493 if ( $isChildObj ) {
3494 # Clean up argument array
3495 $newFrame = $frame->newChild( $args, $title );
3498 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3499 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3500 # Expansion is eligible for the empty-frame cache
3501 $text = $newFrame->cachedExpand( $titleText, $text );
3503 # Uncached expansion
3504 $text = $newFrame->expand( $text );
3507 if ( $isLocalObj && $nowiki ) {
3508 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3509 $isLocalObj = false;
3512 if ( $profileSection ) {
3513 $this->mProfiler
->scopedProfileOut( $profileSection );
3516 # Replace raw HTML by a placeholder
3518 $text = $this->insertStripItem( $text );
3519 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3520 # Escape nowiki-style return values
3521 $text = wfEscapeWikiText( $text );
3522 } elseif ( is_string( $text )
3523 && !$piece['lineStart']
3524 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3526 # T2529: if the template begins with a table or block-level
3527 # element, it should be treated as beginning a new line.
3528 # This behavior is somewhat controversial.
3529 $text = "\n" . $text;
3532 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3533 # Error, oversize inclusion
3534 if ( $titleText !== false ) {
3535 # Make a working, properly escaped link if possible (T25588)
3536 $text = "[[:$titleText]]";
3538 # This will probably not be a working link, but at least it may
3539 # provide some hint of where the problem is
3540 preg_replace( '/^:/', '', $originalTitle );
3541 $text = "[[:$originalTitle]]";
3543 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3544 . 'post-expand include size too large -->' );
3545 $this->limitationWarn( 'post-expand-template-inclusion' );
3548 if ( $isLocalObj ) {
3549 $ret = [ 'object' => $text ];
3551 $ret = [ 'text' => $text ];
3558 * Call a parser function and return an array with text and flags.
3560 * The returned array will always contain a boolean 'found', indicating
3561 * whether the parser function was found or not. It may also contain the
3563 * text: string|object, resulting wikitext or PP DOM object
3564 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3565 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3566 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3567 * nowiki: bool, wiki markup in $text should be escaped
3570 * @param PPFrame $frame The current frame, contains template arguments
3571 * @param string $function Function name
3572 * @param array $args Arguments to the function
3573 * @throws MWException
3576 public function callParserFunction( $frame, $function, array $args = [] ) {
3577 # Case sensitive functions
3578 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3579 $function = $this->mFunctionSynonyms
[1][$function];
3581 # Case insensitive functions
3582 $function = $this->contLang
->lc( $function );
3583 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3584 $function = $this->mFunctionSynonyms
[0][$function];
3586 return [ 'found' => false ];
3590 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3592 // Avoid PHP 7.1 warning from passing $this by reference
3595 $allArgs = [ &$parser ];
3596 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3597 # Convert arguments to PPNodes and collect for appending to $allArgs
3599 foreach ( $args as $k => $v ) {
3600 if ( $v instanceof PPNode ||
$k === 0 ) {
3603 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3607 # Add a frame parameter, and pass the arguments as an array
3608 $allArgs[] = $frame;
3609 $allArgs[] = $funcArgs;
3611 # Convert arguments to plain text and append to $allArgs
3612 foreach ( $args as $k => $v ) {
3613 if ( $v instanceof PPNode
) {
3614 $allArgs[] = trim( $frame->expand( $v ) );
3615 } elseif ( is_int( $k ) && $k >= 0 ) {
3616 $allArgs[] = trim( $v );
3618 $allArgs[] = trim( "$k=$v" );
3623 $result = $callback( ...$allArgs );
3625 # The interface for function hooks allows them to return a wikitext
3626 # string or an array containing the string and any flags. This mungs
3627 # things around to match what this method should return.
3628 if ( !is_array( $result ) ) {
3634 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3635 $result['text'] = $result[0];
3637 unset( $result[0] );
3644 $preprocessFlags = 0;
3645 if ( isset( $result['noparse'] ) ) {
3646 $noparse = $result['noparse'];
3648 if ( isset( $result['preprocessFlags'] ) ) {
3649 $preprocessFlags = $result['preprocessFlags'];
3653 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3654 $result['isChildObj'] = true;
3661 * Get the semi-parsed DOM representation of a template with a given title,
3662 * and its redirect destination title. Cached.
3664 * @param Title $title
3668 public function getTemplateDom( $title ) {
3669 $cacheTitle = $title;
3670 $titleText = $title->getPrefixedDBkey();
3672 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3673 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3674 $title = Title
::makeTitle( $ns, $dbk );
3675 $titleText = $title->getPrefixedDBkey();
3677 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3678 return [ $this->mTplDomCache
[$titleText], $title ];
3681 # Cache miss, go to the database
3682 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3684 if ( $text === false ) {
3685 $this->mTplDomCache
[$titleText] = false;
3686 return [ false, $title ];
3689 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3690 $this->mTplDomCache
[$titleText] = $dom;
3692 if ( !$title->equals( $cacheTitle ) ) {
3693 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3694 [ $title->getNamespace(), $title->getDBkey() ];
3697 return [ $dom, $title ];
3701 * Fetch the current revision of a given title. Note that the revision
3702 * (and even the title) may not exist in the database, so everything
3703 * contributing to the output of the parser should use this method
3704 * where possible, rather than getting the revisions themselves. This
3705 * method also caches its results, so using it benefits performance.
3708 * @param Title $title
3711 public function fetchCurrentRevisionOfTitle( $title ) {
3712 $cacheKey = $title->getPrefixedDBkey();
3713 if ( !$this->currentRevisionCache
) {
3714 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3716 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3717 $this->currentRevisionCache
->set( $cacheKey,
3718 // Defaults to Parser::statelessFetchRevision()
3719 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3722 return $this->currentRevisionCache
->get( $cacheKey );
3726 * @param Title $title
3730 public function isCurrentRevisionOfTitleCached( $title ) {
3732 $this->currentRevisionCache
&&
3733 $this->currentRevisionCache
->has( $title->getPrefixedText() )
3738 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3739 * without passing them on to it.
3742 * @param Title $title
3743 * @param Parser|bool $parser
3744 * @return Revision|bool False if missing
3746 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3747 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3753 * Fetch the unparsed text of a template and register a reference to it.
3754 * @param Title $title
3755 * @return array ( string or false, Title )
3757 public function fetchTemplateAndTitle( $title ) {
3758 // Defaults to Parser::statelessFetchTemplate()
3759 $templateCb = $this->mOptions
->getTemplateCallback();
3760 $stuff = call_user_func( $templateCb, $title, $this );
3761 $rev = $stuff['revision'] ??
null;
3762 $text = $stuff['text'];
3763 if ( is_string( $stuff['text'] ) ) {
3764 // We use U+007F DELETE to distinguish strip markers from regular text
3765 $text = strtr( $text, "\x7f", "?" );
3767 $finalTitle = $stuff['finalTitle'] ??
$title;
3768 foreach ( ( $stuff['deps'] ??
[] ) as $dep ) {
3769 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3770 if ( $dep['title']->equals( $this->getTitle() ) && $rev instanceof Revision
) {
3771 // Self-transclusion; final result may change based on the new page version
3772 $this->setOutputFlag( 'vary-revision-sha1', 'Self transclusion' );
3773 $this->getOutput()->setRevisionUsedSha1Base36( $rev->getSha1() );
3777 return [ $text, $finalTitle ];
3781 * Fetch the unparsed text of a template and register a reference to it.
3782 * @param Title $title
3783 * @return string|bool
3785 public function fetchTemplate( $title ) {
3786 return $this->fetchTemplateAndTitle( $title )[0];
3790 * Static function to get a template
3791 * Can be overridden via ParserOptions::setTemplateCallback().
3793 * @param Title $title
3794 * @param bool|Parser $parser
3798 public static function statelessFetchTemplate( $title, $parser = false ) {
3799 $text = $skip = false;
3800 $finalTitle = $title;
3804 # Loop to fetch the article, with up to 1 redirect
3805 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3806 # Give extensions a chance to select the revision instead
3807 $id = false; # Assume current
3808 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3809 [ $parser, $title, &$skip, &$id ] );
3815 'page_id' => $title->getArticleID(),
3822 $rev = Revision
::newFromId( $id );
3823 } elseif ( $parser ) {
3824 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3826 $rev = Revision
::newFromTitle( $title );
3828 $rev_id = $rev ?
$rev->getId() : 0;
3829 # If there is no current revision, there is no page
3830 if ( $id === false && !$rev ) {
3831 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3832 $linkCache->addBadLinkObj( $title );
3837 'page_id' => $title->getArticleID(),
3840 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3841 # We fetched a rev from a different title; register it too...
3843 'title' => $rev->getTitle(),
3844 'page_id' => $rev->getPage(),
3850 $content = $rev->getContent();
3851 $text = $content ?
$content->getWikitextForTransclusion() : null;
3853 Hooks
::run( 'ParserFetchTemplate',
3854 [ $parser, $title, $rev, &$text, &$deps ] );
3856 if ( $text === false ||
$text === null ) {
3860 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3861 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3862 lcfirst( $title->getText() ) )->inContentLanguage();
3863 if ( !$message->exists() ) {
3867 $content = $message->content();
3868 $text = $message->plain();
3876 $finalTitle = $title;
3877 $title = $content->getRedirectTarget();
3882 'finalTitle' => $finalTitle,
3888 * Fetch a file and its title and register a reference to it.
3889 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3890 * @param Title $title
3891 * @param array $options Array of options to RepoGroup::findFile
3892 * @return array ( File or false, Title of file )
3894 public function fetchFileAndTitle( $title, $options = [] ) {
3895 $file = $this->fetchFileNoRegister( $title, $options );
3897 $time = $file ?
$file->getTimestamp() : false;
3898 $sha1 = $file ?
$file->getSha1() : false;
3899 # Register the file as a dependency...
3900 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3901 if ( $file && !$title->equals( $file->getTitle() ) ) {
3902 # Update fetched file title
3903 $title = $file->getTitle();
3904 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3906 return [ $file, $title ];
3910 * Helper function for fetchFileAndTitle.
3912 * Also useful if you need to fetch a file but not use it yet,
3913 * for example to get the file's handler.
3915 * @param Title $title
3916 * @param array $options Array of options to RepoGroup::findFile
3919 protected function fetchFileNoRegister( $title, $options = [] ) {
3920 if ( isset( $options['broken'] ) ) {
3921 $file = false; // broken thumbnail forced by hook
3922 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3923 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3924 } else { // get by (name,timestamp)
3925 $file = MediaWikiServices
::getInstance()->getRepoGroup()->findFile( $title, $options );
3931 * Transclude an interwiki link.
3933 * @param Title $title
3934 * @param string $action Usually one of (raw, render)
3938 public function interwikiTransclude( $title, $action ) {
3939 if ( !$this->svcOptions
->get( 'EnableScaryTranscluding' ) ) {
3940 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3943 $url = $title->getFullURL( [ 'action' => $action ] );
3944 if ( strlen( $url ) > 1024 ) {
3945 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3948 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3950 $fname = __METHOD__
;
3951 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3953 $data = $cache->getWithSetCallback(
3954 $cache->makeGlobalKey(
3955 'interwiki-transclude',
3956 ( $wikiId !== false ) ?
$wikiId : 'external',
3959 $this->svcOptions
->get( 'TranscludeCacheExpiry' ),
3960 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3961 $req = MWHttpRequest
::factory( $url, [], $fname );
3963 $status = $req->execute(); // Status object
3964 if ( !$status->isOK() ) {
3965 $ttl = $cache::TTL_UNCACHEABLE
;
3966 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3967 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3971 'text' => $status->isOK() ?
$req->getContent() : null,
3972 'code' => $req->getStatus()
3976 'checkKeys' => ( $wikiId !== false )
3977 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3979 'pcGroup' => 'interwiki-transclude:5',
3980 'pcTTL' => $cache::TTL_PROC_LONG
3984 if ( is_string( $data['text'] ) ) {
3985 $text = $data['text'];
3986 } elseif ( $data['code'] != 200 ) {
3987 // Though we failed to fetch the content, this status is useless.
3988 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3989 ->params( $url, $data['code'] )->inContentLanguage()->text();
3991 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3998 * Triple brace replacement -- used for template arguments
4001 * @param array $piece
4002 * @param PPFrame $frame
4006 public function argSubstitution( $piece, $frame ) {
4008 $parts = $piece['parts'];
4009 $nameWithSpaces = $frame->expand( $piece['title'] );
4010 $argName = trim( $nameWithSpaces );
4012 $text = $frame->getArgument( $argName );
4013 if ( $text === false && $parts->getLength() > 0
4014 && ( $this->ot
['html']
4016 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
4019 # No match in frame, use the supplied default
4020 $object = $parts->item( 0 )->getChildren();
4022 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4023 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4024 $this->limitationWarn( 'post-expand-template-argument' );
4027 if ( $text === false && $object === false ) {
4029 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4031 if ( $error !== false ) {
4034 if ( $object !== false ) {
4035 $ret = [ 'object' => $object ];
4037 $ret = [ 'text' => $text ];
4044 * Return the text to be used for a given extension tag.
4045 * This is the ghost of strip().
4047 * @param array $params Associative array of parameters:
4048 * name PPNode for the tag name
4049 * attr PPNode for unparsed text where tag attributes are thought to be
4050 * attributes Optional associative array of parsed attributes
4051 * inner Contents of extension element
4052 * noClose Original text did not have a close tag
4053 * @param PPFrame $frame
4055 * @throws MWException
4058 public function extensionSubstitution( $params, $frame ) {
4059 static $errorStr = '<span class="error">';
4060 static $errorLen = 20;
4062 $name = $frame->expand( $params['name'] );
4063 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4064 // Probably expansion depth or node count exceeded. Just punt the
4069 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
4070 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4075 // We can't safely check if the expansion for $content resulted in an
4076 // error, because the content could happen to be the error string
4078 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
4080 $marker = self
::MARKER_PREFIX
. "-$name-"
4081 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
4083 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
4084 ( $this->ot
['html'] ||
$this->ot
['pre'] );
4085 if ( $isFunctionTag ) {
4086 $markerType = 'none';
4088 $markerType = 'general';
4090 if ( $this->ot
['html'] ||
$isFunctionTag ) {
4091 $name = strtolower( $name );
4092 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
4093 if ( isset( $params['attributes'] ) ) {
4094 $attributes +
= $params['attributes'];
4097 if ( isset( $this->mTagHooks
[$name] ) ) {
4098 $output = call_user_func_array( $this->mTagHooks
[$name],
4099 [ $content, $attributes, $this, $frame ] );
4100 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4101 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4103 // Avoid PHP 7.1 warning from passing $this by reference
4105 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4107 $output = '<span class="error">Invalid tag extension name: ' .
4108 htmlspecialchars( $name ) . '</span>';
4111 if ( is_array( $output ) ) {
4114 $output = $flags[0];
4115 if ( isset( $flags['markerType'] ) ) {
4116 $markerType = $flags['markerType'];
4120 if ( is_null( $attrText ) ) {
4123 if ( isset( $params['attributes'] ) ) {
4124 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4125 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4126 htmlspecialchars( $attrValue ) . '"';
4129 if ( $content === null ) {
4130 $output = "<$name$attrText/>";
4132 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4133 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4137 $output = "<$name$attrText>$content$close";
4141 if ( $markerType === 'none' ) {
4143 } elseif ( $markerType === 'nowiki' ) {
4144 $this->mStripState
->addNoWiki( $marker, $output );
4145 } elseif ( $markerType === 'general' ) {
4146 $this->mStripState
->addGeneral( $marker, $output );
4148 throw new MWException( __METHOD__
. ': invalid marker type' );
4154 * Increment an include size counter
4156 * @param string $type The type of expansion
4157 * @param int $size The size of the text
4158 * @return bool False if this inclusion would take it over the maximum, true otherwise
4160 public function incrementIncludeSize( $type, $size ) {
4161 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4164 $this->mIncludeSizes
[$type] +
= $size;
4170 * Increment the expensive function count
4172 * @return bool False if the limit has been exceeded
4174 public function incrementExpensiveFunctionCount() {
4175 $this->mExpensiveFunctionCount++
;
4176 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4180 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4181 * Fills $this->mDoubleUnderscores, returns the modified text
4183 * @param string $text
4187 public function doDoubleUnderscore( $text ) {
4188 # The position of __TOC__ needs to be recorded
4189 $mw = $this->magicWordFactory
->get( 'toc' );
4190 if ( $mw->match( $text ) ) {
4191 $this->mShowToc
= true;
4192 $this->mForceTocPosition
= true;
4194 # Set a placeholder. At the end we'll fill it in with the TOC.
4195 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4197 # Only keep the first one.
4198 $text = $mw->replace( '', $text );
4201 # Now match and remove the rest of them
4202 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4203 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4205 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4206 $this->mOutput
->mNoGallery
= true;
4208 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4209 $this->mShowToc
= false;
4211 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4212 && $this->mTitle
->getNamespace() == NS_CATEGORY
4214 $this->addTrackingCategory( 'hidden-category-category' );
4216 # (T10068) Allow control over whether robots index a page.
4217 # __INDEX__ always overrides __NOINDEX__, see T16899
4218 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4219 $this->mOutput
->setIndexPolicy( 'noindex' );
4220 $this->addTrackingCategory( 'noindex-category' );
4222 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4223 $this->mOutput
->setIndexPolicy( 'index' );
4224 $this->addTrackingCategory( 'index-category' );
4227 # Cache all double underscores in the database
4228 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4229 $this->mOutput
->setProperty( $key, '' );
4236 * @see ParserOutput::addTrackingCategory()
4237 * @param string $msg Message key
4238 * @return bool Whether the addition was successful
4240 public function addTrackingCategory( $msg ) {
4241 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4245 * This function accomplishes several tasks:
4246 * 1) Auto-number headings if that option is enabled
4247 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4248 * 3) Add a Table of contents on the top for users who have enabled the option
4249 * 4) Auto-anchor headings
4251 * It loops through all headlines, collects the necessary data, then splits up the
4252 * string and re-inserts the newly formatted headlines.
4254 * @param string $text
4255 * @param string $origText Original, untouched wikitext
4256 * @param bool $isMain
4257 * @return mixed|string
4260 public function formatHeadings( $text, $origText, $isMain = true ) {
4261 # Inhibit editsection links if requested in the page
4262 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4263 $maybeShowEditLink = false;
4265 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4268 # Get all headlines for numbering them and adding funky stuff like [edit]
4269 # links - this is for later, but we need the number of headlines right now
4270 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4271 # be trimmed here since whitespace in HTML headings is significant.
4273 $numMatches = preg_match_all(
4274 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4279 # if there are fewer than 4 headlines in the article, do not show TOC
4280 # unless it's been explicitly enabled.
4281 $enoughToc = $this->mShowToc
&&
4282 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4284 # Allow user to stipulate that a page should have a "new section"
4285 # link added via __NEWSECTIONLINK__
4286 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4287 $this->mOutput
->setNewSection( true );
4290 # Allow user to remove the "new section"
4291 # link via __NONEWSECTIONLINK__
4292 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4293 $this->mOutput
->hideNewSection( true );
4296 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4297 # override above conditions and always show TOC above first header
4298 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4299 $this->mShowToc
= true;
4307 # Ugh .. the TOC should have neat indentation levels which can be
4308 # passed to the skin functions. These are determined here
4312 $sublevelCount = [];
4318 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4319 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4320 $oldType = $this->mOutputType
;
4321 $this->setOutputType( self
::OT_WIKI
);
4322 $frame = $this->getPreprocessor()->newFrame();
4323 $root = $this->preprocessToDom( $origText );
4324 $node = $root->getFirstChild();
4329 $headlines = $numMatches !== false ?
$matches[3] : [];
4331 $maxTocLevel = $this->svcOptions
->get( 'MaxTocLevel' );
4332 foreach ( $headlines as $headline ) {
4333 $isTemplate = false;
4335 $sectionIndex = false;
4337 $markerMatches = [];
4338 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4339 $serial = $markerMatches[1];
4340 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4341 $isTemplate = ( $titleText != $baseTitleText );
4342 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4346 $prevlevel = $level;
4348 $level = $matches[1][$headlineCount];
4350 if ( $level > $prevlevel ) {
4351 # Increase TOC level
4353 $sublevelCount[$toclevel] = 0;
4354 if ( $toclevel < $maxTocLevel ) {
4355 $prevtoclevel = $toclevel;
4356 $toc .= Linker
::tocIndent();
4359 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4360 # Decrease TOC level, find level to jump to
4362 for ( $i = $toclevel; $i > 0; $i-- ) {
4363 if ( $levelCount[$i] == $level ) {
4364 # Found last matching level
4367 } elseif ( $levelCount[$i] < $level ) {
4368 # Found first matching level below current level
4376 if ( $toclevel < $maxTocLevel ) {
4377 if ( $prevtoclevel < $maxTocLevel ) {
4378 # Unindent only if the previous toc level was shown :p
4379 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4380 $prevtoclevel = $toclevel;
4382 $toc .= Linker
::tocLineEnd();
4386 # No change in level, end TOC line
4387 if ( $toclevel < $maxTocLevel ) {
4388 $toc .= Linker
::tocLineEnd();
4392 $levelCount[$toclevel] = $level;
4394 # count number of headlines for each level
4395 $sublevelCount[$toclevel]++
;
4397 for ( $i = 1; $i <= $toclevel; $i++
) {
4398 if ( !empty( $sublevelCount[$i] ) ) {
4402 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4407 # The safe header is a version of the header text safe to use for links
4409 # Remove link placeholders by the link text.
4410 # <!--LINK number-->
4412 # link text with suffix
4413 # Do this before unstrip since link text can contain strip markers
4414 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4416 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4417 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4419 # Remove any <style> or <script> tags (T198618)
4420 $safeHeadline = preg_replace(
4421 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4426 # Strip out HTML (first regex removes any tag not allowed)
4428 # * <sup> and <sub> (T10393)
4432 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4433 # * <s> and <strike> (T35715)
4434 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4435 # to allow setting directionality in toc items.
4436 $tocline = preg_replace(
4438 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4439 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4445 # Strip '<span></span>', which is the result from the above if
4446 # <span id="foo"></span> is used to produce an additional anchor
4448 $tocline = str_replace( '<span></span>', '', $tocline );
4450 $tocline = trim( $tocline );
4452 # For the anchor, strip out HTML-y stuff period
4453 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4454 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4456 # Save headline for section edit hint before it's escaped
4457 $headlineHint = $safeHeadline;
4459 # Decode HTML entities
4460 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4462 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4464 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4465 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4466 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4467 if ( $fallbackHeadline === $safeHeadline ) {
4468 # No reason to have both (in fact, we can't)
4469 $fallbackHeadline = false;
4472 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4473 # @todo FIXME: We may be changing them depending on the current locale.
4474 $arrayKey = strtolower( $safeHeadline );
4475 if ( $fallbackHeadline === false ) {
4476 $fallbackArrayKey = false;
4478 $fallbackArrayKey = strtolower( $fallbackHeadline );
4481 # Create the anchor for linking from the TOC to the section
4482 $anchor = $safeHeadline;
4483 $fallbackAnchor = $fallbackHeadline;
4484 if ( isset( $refers[$arrayKey] ) ) {
4485 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4486 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4488 $linkAnchor .= "_$i";
4489 $refers["${arrayKey}_$i"] = true;
4491 $refers[$arrayKey] = true;
4493 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4494 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4495 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4496 $fallbackAnchor .= "_$i";
4497 $refers["${fallbackArrayKey}_$i"] = true;
4499 $refers[$fallbackArrayKey] = true;
4502 # Don't number the heading if it is the only one (looks silly)
4503 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4504 # the two are different if the line contains a link
4505 $headline = Html
::element(
4507 [ 'class' => 'mw-headline-number' ],
4509 ) . ' ' . $headline;
4512 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4513 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4514 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4517 # Add the section to the section tree
4518 # Find the DOM node for this header
4519 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4520 while ( $node && !$noOffset ) {
4521 if ( $node->getName() === 'h' ) {
4522 $bits = $node->splitHeading();
4523 if ( $bits['i'] == $sectionIndex ) {
4527 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4528 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4529 $node = $node->getNextSibling();
4532 'toclevel' => $toclevel,
4535 'number' => $numbering,
4536 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4537 'fromtitle' => $titleText,
4538 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4539 'anchor' => $anchor,
4542 # give headline the correct <h#> tag
4543 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4544 // Output edit section links as markers with styles that can be customized by skins
4545 if ( $isTemplate ) {
4546 # Put a T flag in the section identifier, to indicate to extractSections()
4547 # that sections inside <includeonly> should be counted.
4548 $editsectionPage = $titleText;
4549 $editsectionSection = "T-$sectionIndex";
4550 $editsectionContent = null;
4552 $editsectionPage = $this->mTitle
->getPrefixedText();
4553 $editsectionSection = $sectionIndex;
4554 $editsectionContent = $headlineHint;
4556 // We use a bit of pesudo-xml for editsection markers. The
4557 // language converter is run later on. Using a UNIQ style marker
4558 // leads to the converter screwing up the tokens when it
4559 // converts stuff. And trying to insert strip tags fails too. At
4560 // this point all real inputted tags have already been escaped,
4561 // so we don't have to worry about a user trying to input one of
4562 // these markers directly. We use a page and section attribute
4563 // to stop the language converter from converting these
4564 // important bits of data, but put the headline hint inside a
4565 // content block because the language converter is supposed to
4566 // be able to convert that piece of data.
4567 // Gets replaced with html in ParserOutput::getText
4568 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4569 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4570 if ( $editsectionContent !== null ) {
4571 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4578 $head[$headlineCount] = Linker
::makeHeadline( $level,
4579 $matches['attrib'][$headlineCount], $anchor, $headline,
4580 $editlink, $fallbackAnchor );
4585 $this->setOutputType( $oldType );
4587 # Never ever show TOC if no headers
4588 if ( $numVisible < 1 ) {
4593 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4594 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4596 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4597 $this->mOutput
->setTOCHTML( $toc );
4598 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4602 $this->mOutput
->setSections( $tocraw );
4605 # split up and insert constructed headlines
4606 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4609 // build an array of document sections
4611 foreach ( $blocks as $block ) {
4612 // $head is zero-based, sections aren't.
4613 if ( empty( $head[$i - 1] ) ) {
4614 $sections[$i] = $block;
4616 $sections[$i] = $head[$i - 1] . $block;
4620 * Send a hook, one per section.
4621 * The idea here is to be able to make section-level DIVs, but to do so in a
4622 * lower-impact, more correct way than r50769
4625 * $section : the section number
4626 * &$sectionContent : ref to the content of the section
4627 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4629 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4634 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4635 // append the TOC at the beginning
4636 // Top anchor now in skin
4637 $sections[0] .= $toc . "\n";
4640 $full .= implode( '', $sections );
4642 if ( $this->mForceTocPosition
) {
4643 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4650 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4651 * conversion, substituting signatures, {{subst:}} templates, etc.
4653 * @param string $text The text to transform
4654 * @param Title $title The Title object for the current article
4655 * @param User $user The User object describing the current user
4656 * @param ParserOptions $options Parsing options
4657 * @param bool $clearState Whether to clear the parser state first
4658 * @return string The altered wiki markup
4660 public function preSaveTransform( $text, Title
$title, User
$user,
4661 ParserOptions
$options, $clearState = true
4663 if ( $clearState ) {
4664 $magicScopeVariable = $this->lock();
4666 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4667 $this->setUser( $user );
4669 // Strip U+0000 NULL (T159174)
4670 $text = str_replace( "\000", '', $text );
4672 // We still normalize line endings for backwards-compatibility
4673 // with other code that just calls PST, but this should already
4674 // be handled in TextContent subclasses
4675 $text = TextContent
::normalizeLineEndings( $text );
4677 if ( $options->getPreSaveTransform() ) {
4678 $text = $this->pstPass2( $text, $user );
4680 $text = $this->mStripState
->unstripBoth( $text );
4682 $this->setUser( null ); # Reset
4688 * Pre-save transform helper function
4690 * @param string $text
4695 private function pstPass2( $text, $user ) {
4696 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4697 # $this->contLang here in order to give everyone the same signature and use the default one
4698 # rather than the one selected in each user's preferences. (see also T14815)
4699 $ts = $this->mOptions
->getTimestamp();
4700 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4701 $ts = $timestamp->format( 'YmdHis' );
4702 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4704 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4706 # Variable replacement
4707 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4708 $text = $this->replaceVariables( $text );
4710 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4711 # which may corrupt this parser instance via its wfMessage()->text() call-
4714 if ( strpos( $text, '~~~' ) !== false ) {
4715 $sigText = $this->getUserSig( $user );
4716 $text = strtr( $text, [
4718 '~~~~' => "$sigText $d",
4721 # The main two signature forms used above are time-sensitive
4722 $this->setOutputFlag( 'user-signature', 'User signature detected' );
4725 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4726 $tc = '[' . Title
::legalChars() . ']';
4727 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4729 // [[ns:page (context)|]]
4730 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4731 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4732 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4733 // [[ns:page (context), context|]] (using either single or double-width comma)
4734 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4735 // [[|page]] (reverse pipe trick: add context from page title)
4736 $p2 = "/\[\[\\|($tc+)]]/";
4738 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4739 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4740 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4741 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4743 $t = $this->mTitle
->getText();
4745 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4746 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4747 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4748 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4750 # if there's no context, don't bother duplicating the title
4751 $text = preg_replace( $p2, '[[\\1]]', $text );
4758 * Fetch the user's signature text, if any, and normalize to
4759 * validated, ready-to-insert wikitext.
4760 * If you have pre-fetched the nickname or the fancySig option, you can
4761 * specify them here to save a database query.
4762 * Do not reuse this parser instance after calling getUserSig(),
4763 * as it may have changed.
4765 * @param User &$user
4766 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4767 * @param bool|null $fancySig whether the nicknname is the complete signature
4768 * or null to use default value
4771 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4772 $username = $user->getName();
4774 # If not given, retrieve from the user object.
4775 if ( $nickname === false ) {
4776 $nickname = $user->getOption( 'nickname' );
4779 if ( is_null( $fancySig ) ) {
4780 $fancySig = $user->getBoolOption( 'fancysig' );
4783 $nickname = $nickname == null ?
$username : $nickname;
4785 if ( mb_strlen( $nickname ) > $this->svcOptions
->get( 'MaxSigChars' ) ) {
4786 $nickname = $username;
4787 $this->logger
->debug( __METHOD__
. ": $username has overlong signature." );
4788 } elseif ( $fancySig !== false ) {
4789 # Sig. might contain markup; validate this
4790 if ( $this->validateSig( $nickname ) !== false ) {
4791 # Validated; clean up (if needed) and return it
4792 return $this->cleanSig( $nickname, true );
4794 # Failed to validate; fall back to the default
4795 $nickname = $username;
4796 $this->logger
->debug( __METHOD__
. ": $username has bad XML tags in signature." );
4800 # Make sure nickname doesnt get a sig in a sig
4801 $nickname = self
::cleanSigInSig( $nickname );
4803 # If we're still here, make it a link to the user page
4804 $userText = wfEscapeWikiText( $username );
4805 $nickText = wfEscapeWikiText( $nickname );
4806 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4808 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4809 ->title( $this->getTitle() )->text();
4813 * Check that the user's signature contains no bad XML
4815 * @param string $text
4816 * @return string|bool An expanded string, or false if invalid.
4818 public function validateSig( $text ) {
4819 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4823 * Clean up signature text
4825 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4826 * 2) Substitute all transclusions
4828 * @param string $text
4829 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4830 * @return string Signature text
4832 public function cleanSig( $text, $parsing = false ) {
4835 $magicScopeVariable = $this->lock();
4836 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4839 # Option to disable this feature
4840 if ( !$this->mOptions
->getCleanSignatures() ) {
4844 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4845 # => Move this logic to braceSubstitution()
4846 $substWord = $this->magicWordFactory
->get( 'subst' );
4847 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4848 $substText = '{{' . $substWord->getSynonym( 0 );
4850 $text = preg_replace( $substRegex, $substText, $text );
4851 $text = self
::cleanSigInSig( $text );
4852 $dom = $this->preprocessToDom( $text );
4853 $frame = $this->getPreprocessor()->newFrame();
4854 $text = $frame->expand( $dom );
4857 $text = $this->mStripState
->unstripBoth( $text );
4864 * Strip 3, 4 or 5 tildes out of signatures.
4866 * @param string $text
4867 * @return string Signature text with /~{3,5}/ removed
4869 public static function cleanSigInSig( $text ) {
4870 $text = preg_replace( '/~{3,5}/', '', $text );
4875 * Set up some variables which are usually set up in parse()
4876 * so that an external function can call some class members with confidence
4878 * @param Title|null $title
4879 * @param ParserOptions $options
4880 * @param int $outputType
4881 * @param bool $clearState
4882 * @param int|null $revId
4884 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4885 $outputType, $clearState = true, $revId = null
4887 $this->startParse( $title, $options, $outputType, $clearState );
4888 if ( $revId !== null ) {
4889 $this->mRevisionId
= $revId;
4894 * @param Title|null $title
4895 * @param ParserOptions $options
4896 * @param int $outputType
4897 * @param bool $clearState
4899 private function startParse( Title
$title = null, ParserOptions
$options,
4900 $outputType, $clearState = true
4902 $this->setTitle( $title );
4903 $this->mOptions
= $options;
4904 $this->setOutputType( $outputType );
4905 if ( $clearState ) {
4906 $this->clearState();
4911 * Wrapper for preprocess()
4913 * @param string $text The text to preprocess
4914 * @param ParserOptions $options
4915 * @param Title|null $title Title object or null to use $wgTitle
4918 public function transformMsg( $text, $options, $title = null ) {
4919 static $executing = false;
4921 # Guard against infinite recursion
4932 $text = $this->preprocess( $text, $title, $options );
4939 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4940 * The callback should have the following form:
4941 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4943 * Transform and return $text. Use $parser for any required context, e.g. use
4944 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4946 * Hooks may return extended information by returning an array, of which the
4947 * first numbered element (index 0) must be the return string, and all other
4948 * entries are extracted into local variables within an internal function
4949 * in the Parser class.
4951 * This interface (introduced r61913) appears to be undocumented, but
4952 * 'markerType' is used by some core tag hooks to override which strip
4953 * array their results are placed in. **Use great caution if attempting
4954 * this interface, as it is not documented and injudicious use could smash
4955 * private variables.**
4957 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4958 * @param callable $callback The callback function (and object) to use for the tag
4959 * @throws MWException
4960 * @return callable|null The old value of the mTagHooks array associated with the hook
4962 public function setHook( $tag, callable
$callback ) {
4963 $tag = strtolower( $tag );
4964 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4965 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4967 $oldVal = $this->mTagHooks
[$tag] ??
null;
4968 $this->mTagHooks
[$tag] = $callback;
4969 if ( !in_array( $tag, $this->mStripList
) ) {
4970 $this->mStripList
[] = $tag;
4977 * As setHook(), but letting the contents be parsed.
4979 * Transparent tag hooks are like regular XML-style tag hooks, except they
4980 * operate late in the transformation sequence, on HTML instead of wikitext.
4982 * This is probably obsoleted by things dealing with parser frames?
4983 * The only extension currently using it is geoserver.
4986 * @todo better document or deprecate this
4988 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4989 * @param callable $callback The callback function (and object) to use for the tag
4990 * @throws MWException
4991 * @return callable|null The old value of the mTagHooks array associated with the hook
4993 public function setTransparentTagHook( $tag, callable
$callback ) {
4994 $tag = strtolower( $tag );
4995 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4996 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4998 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4999 $this->mTransparentTagHooks
[$tag] = $callback;
5005 * Remove all tag hooks
5007 public function clearTagHooks() {
5008 $this->mTagHooks
= [];
5009 $this->mFunctionTagHooks
= [];
5010 $this->mStripList
= $this->mDefaultStripList
;
5014 * Create a function, e.g. {{sum:1|2|3}}
5015 * The callback function should have the form:
5016 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5018 * Or with Parser::SFH_OBJECT_ARGS:
5019 * function myParserFunction( $parser, $frame, $args ) { ... }
5021 * The callback may either return the text result of the function, or an array with the text
5022 * in element 0, and a number of flags in the other elements. The names of the flags are
5023 * specified in the keys. Valid flags are:
5024 * found The text returned is valid, stop processing the template. This
5026 * nowiki Wiki markup in the return value should be escaped
5027 * isHTML The returned text is HTML, armour it against wikitext transformation
5029 * @param string $id The magic word ID
5030 * @param callable $callback The callback function (and object) to use
5031 * @param int $flags A combination of the following flags:
5032 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5034 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5035 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5036 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5037 * the arguments, and to control the way they are expanded.
5039 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5040 * arguments, for instance:
5041 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5043 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5044 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5045 * working if/when this is changed.
5047 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5050 * Please read the documentation in includes/parser/Preprocessor.php for more information
5051 * about the methods available in PPFrame and PPNode.
5053 * @throws MWException
5054 * @return string|callable The old callback function for this name, if any
5056 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
5057 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
5058 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
5060 # Add to function cache
5061 $mw = $this->magicWordFactory
->get( $id );
5063 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
5066 $synonyms = $mw->getSynonyms();
5067 $sensitive = intval( $mw->isCaseSensitive() );
5069 foreach ( $synonyms as $syn ) {
5071 if ( !$sensitive ) {
5072 $syn = $this->contLang
->lc( $syn );
5075 if ( !( $flags & self
::SFH_NO_HASH
) ) {
5078 # Remove trailing colon
5079 if ( substr( $syn, -1, 1 ) === ':' ) {
5080 $syn = substr( $syn, 0, -1 );
5082 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
5088 * Get all registered function hook identifiers
5092 public function getFunctionHooks() {
5093 $this->firstCallInit();
5094 return array_keys( $this->mFunctionHooks
);
5098 * Create a tag function, e.g. "<test>some stuff</test>".
5099 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5100 * Unlike parser functions, their content is not preprocessed.
5101 * @param string $tag
5102 * @param callable $callback
5104 * @throws MWException
5107 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5108 $tag = strtolower( $tag );
5109 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5110 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5112 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5113 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5115 if ( !in_array( $tag, $this->mStripList
) ) {
5116 $this->mStripList
[] = $tag;
5123 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5124 * Placeholders created in Linker::link()
5126 * @param string &$text
5127 * @param int $options
5129 public function replaceLinkHolders( &$text, $options = 0 ) {
5130 $this->mLinkHolders
->replace( $text );
5134 * Replace "<!--LINK-->" link placeholders with plain text of links
5135 * (not HTML-formatted).
5137 * @param string $text
5140 public function replaceLinkHoldersText( $text ) {
5141 return $this->mLinkHolders
->replaceText( $text );
5145 * Renders an image gallery from a text with one line per image.
5146 * text labels may be given by using |-style alternative text. E.g.
5147 * Image:one.jpg|The number "1"
5148 * Image:tree.jpg|A tree
5149 * given as text will return the HTML of a gallery with two images,
5150 * labeled 'The number "1"' and
5153 * @param string $text
5154 * @param array $params
5155 * @return string HTML
5157 public function renderImageGallery( $text, $params ) {
5159 if ( isset( $params['mode'] ) ) {
5160 $mode = $params['mode'];
5164 $ig = ImageGalleryBase
::factory( $mode );
5165 } catch ( Exception
$e ) {
5166 // If invalid type set, fallback to default.
5167 $ig = ImageGalleryBase
::factory( false );
5170 $ig->setContextTitle( $this->mTitle
);
5171 $ig->setShowBytes( false );
5172 $ig->setShowDimensions( false );
5173 $ig->setShowFilename( false );
5174 $ig->setParser( $this );
5175 $ig->setHideBadImages();
5176 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5178 if ( isset( $params['showfilename'] ) ) {
5179 $ig->setShowFilename( true );
5181 $ig->setShowFilename( false );
5183 if ( isset( $params['caption'] ) ) {
5184 // NOTE: We aren't passing a frame here or below. Frame info
5185 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5186 // See T107332#4030581
5187 $caption = $this->recursiveTagParse( $params['caption'] );
5188 $ig->setCaptionHtml( $caption );
5190 if ( isset( $params['perrow'] ) ) {
5191 $ig->setPerRow( $params['perrow'] );
5193 if ( isset( $params['widths'] ) ) {
5194 $ig->setWidths( $params['widths'] );
5196 if ( isset( $params['heights'] ) ) {
5197 $ig->setHeights( $params['heights'] );
5199 $ig->setAdditionalOptions( $params );
5201 // Avoid PHP 7.1 warning from passing $this by reference
5203 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5205 $lines = StringUtils
::explode( "\n", $text );
5206 foreach ( $lines as $line ) {
5207 # match lines like these:
5208 # Image:someimage.jpg|This is some image
5210 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5212 if ( count( $matches ) == 0 ) {
5216 if ( strpos( $matches[0], '%' ) !== false ) {
5217 $matches[1] = rawurldecode( $matches[1] );
5219 $title = Title
::newFromText( $matches[1], NS_FILE
);
5220 if ( is_null( $title ) ) {
5221 # Bogus title. Ignore these so we don't bomb out later.
5225 # We need to get what handler the file uses, to figure out parameters.
5226 # Note, a hook can overide the file name, and chose an entirely different
5227 # file (which potentially could be of a different type and have different handler).
5230 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5231 [ $this, $title, &$options, &$descQuery ] );
5232 # Don't register it now, as TraditionalImageGallery does that later.
5233 $file = $this->fetchFileNoRegister( $title, $options );
5234 $handler = $file ?
$file->getHandler() : false;
5237 'img_alt' => 'gallery-internal-alt',
5238 'img_link' => 'gallery-internal-link',
5241 $paramMap +
= $handler->getParamMap();
5242 // We don't want people to specify per-image widths.
5243 // Additionally the width parameter would need special casing anyhow.
5244 unset( $paramMap['img_width'] );
5247 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5252 $handlerOptions = [];
5253 if ( isset( $matches[3] ) ) {
5254 // look for an |alt= definition while trying not to break existing
5255 // captions with multiple pipes (|) in it, until a more sensible grammar
5256 // is defined for images in galleries
5258 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5259 // splitting on '|' is a bit odd, and different from makeImage.
5260 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5261 // Protect LanguageConverter markup
5262 $parameterMatches = StringUtils
::delimiterExplode(
5263 '-{', '}-', '|', $matches[3], true /* nested */
5266 foreach ( $parameterMatches as $parameterMatch ) {
5267 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5269 $paramName = $paramMap[$magicName];
5271 switch ( $paramName ) {
5272 case 'gallery-internal-alt':
5273 $alt = $this->stripAltText( $match, false );
5275 case 'gallery-internal-link':
5276 $linkValue = $this->stripAltText( $match, false );
5277 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5278 // Result of LanguageConverter::markNoConversion
5279 // invoked on an external link.
5280 $linkValue = substr( $linkValue, 4, -2 );
5282 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5283 if ( $type === 'link-url' ) {
5285 $this->mOutput
->addExternalLink( $target );
5286 } elseif ( $type === 'link-title' ) {
5287 $link = $target->getLinkURL();
5288 $this->mOutput
->addLink( $target );
5292 // Must be a handler specific parameter.
5293 if ( $handler->validateParam( $paramName, $match ) ) {
5294 $handlerOptions[$paramName] = $match;
5296 // Guess not, consider it as caption.
5297 $this->logger
->debug(
5298 "$parameterMatch failed parameter validation" );
5299 $label = $parameterMatch;
5305 $label = $parameterMatch;
5310 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5312 $html = $ig->toHTML();
5313 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5318 * @param MediaHandler $handler
5321 public function getImageParams( $handler ) {
5323 $handlerClass = get_class( $handler );
5327 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5328 # Initialise static lists
5329 static $internalParamNames = [
5330 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5331 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5332 'bottom', 'text-bottom' ],
5333 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5334 'upright', 'border', 'link', 'alt', 'class' ],
5336 static $internalParamMap;
5337 if ( !$internalParamMap ) {
5338 $internalParamMap = [];
5339 foreach ( $internalParamNames as $type => $names ) {
5340 foreach ( $names as $name ) {
5341 // For grep: img_left, img_right, img_center, img_none,
5342 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5343 // img_bottom, img_text_bottom,
5344 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5345 // img_border, img_link, img_alt, img_class
5346 $magicName = str_replace( '-', '_', "img_$name" );
5347 $internalParamMap[$magicName] = [ $type, $name ];
5352 # Add handler params
5353 $paramMap = $internalParamMap;
5355 $handlerParamMap = $handler->getParamMap();
5356 foreach ( $handlerParamMap as $magic => $paramName ) {
5357 $paramMap[$magic] = [ 'handler', $paramName ];
5360 $this->mImageParams
[$handlerClass] = $paramMap;
5361 $this->mImageParamsMagicArray
[$handlerClass] =
5362 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5364 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5368 * Parse image options text and use it to make an image
5370 * @param Title $title
5371 * @param string $options
5372 * @param LinkHolderArray|bool $holders
5373 * @return string HTML
5375 public function makeImage( $title, $options, $holders = false ) {
5376 # Check if the options text is of the form "options|alt text"
5378 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5379 # * left no resizing, just left align. label is used for alt= only
5380 # * right same, but right aligned
5381 # * none same, but not aligned
5382 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5383 # * center center the image
5384 # * frame Keep original image size, no magnify-button.
5385 # * framed Same as "frame"
5386 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5387 # * upright reduce width for upright images, rounded to full __0 px
5388 # * border draw a 1px border around the image
5389 # * alt Text for HTML alt attribute (defaults to empty)
5390 # * class Set a class for img node
5391 # * link Set the target of the image link. Can be external, interwiki, or local
5392 # vertical-align values (no % or length right now):
5402 # Protect LanguageConverter markup when splitting into parts
5403 $parts = StringUtils
::delimiterExplode(
5404 '-{', '}-', '|', $options, true /* allow nesting */
5407 # Give extensions a chance to select the file revision for us
5410 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5411 [ $this, $title, &$options, &$descQuery ] );
5412 # Fetch and register the file (file title may be different via hooks)
5413 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5416 $handler = $file ?
$file->getHandler() : false;
5418 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5421 $this->addTrackingCategory( 'broken-file-category' );
5424 # Process the input parameters
5426 $params = [ 'frame' => [], 'handler' => [],
5427 'horizAlign' => [], 'vertAlign' => [] ];
5428 $seenformat = false;
5429 foreach ( $parts as $part ) {
5430 $part = trim( $part );
5431 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5433 if ( isset( $paramMap[$magicName] ) ) {
5434 list( $type, $paramName ) = $paramMap[$magicName];
5436 # Special case; width and height come in one variable together
5437 if ( $type === 'handler' && $paramName === 'width' ) {
5438 $parsedWidthParam = self
::parseWidthParam( $value );
5439 if ( isset( $parsedWidthParam['width'] ) ) {
5440 $width = $parsedWidthParam['width'];
5441 if ( $handler->validateParam( 'width', $width ) ) {
5442 $params[$type]['width'] = $width;
5446 if ( isset( $parsedWidthParam['height'] ) ) {
5447 $height = $parsedWidthParam['height'];
5448 if ( $handler->validateParam( 'height', $height ) ) {
5449 $params[$type]['height'] = $height;
5453 # else no validation -- T15436
5455 if ( $type === 'handler' ) {
5456 # Validate handler parameter
5457 $validated = $handler->validateParam( $paramName, $value );
5459 # Validate internal parameters
5460 switch ( $paramName ) {
5464 # @todo FIXME: Possibly check validity here for
5465 # manualthumb? downstream behavior seems odd with
5466 # missing manual thumbs.
5468 $value = $this->stripAltText( $value, $holders );
5471 list( $paramName, $value ) =
5472 $this->parseLinkParameter(
5473 $this->stripAltText( $value, $holders )
5477 if ( $paramName === 'no-link' ) {
5480 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5481 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5488 // use first appearing option, discard others.
5489 $validated = !$seenformat;
5493 # Most other things appear to be empty or numeric...
5494 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5499 $params[$type][$paramName] = $value;
5503 if ( !$validated ) {
5508 # Process alignment parameters
5509 if ( $params['horizAlign'] ) {
5510 $params['frame']['align'] = key( $params['horizAlign'] );
5512 if ( $params['vertAlign'] ) {
5513 $params['frame']['valign'] = key( $params['vertAlign'] );
5516 $params['frame']['caption'] = $caption;
5518 # Will the image be presented in a frame, with the caption below?
5519 $imageIsFramed = isset( $params['frame']['frame'] )
5520 ||
isset( $params['frame']['framed'] )
5521 ||
isset( $params['frame']['thumbnail'] )
5522 ||
isset( $params['frame']['manualthumb'] );
5524 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5525 # came to also set the caption, ordinary text after the image -- which
5526 # makes no sense, because that just repeats the text multiple times in
5527 # screen readers. It *also* came to set the title attribute.
5528 # Now that we have an alt attribute, we should not set the alt text to
5529 # equal the caption: that's worse than useless, it just repeats the
5530 # text. This is the framed/thumbnail case. If there's no caption, we
5531 # use the unnamed parameter for alt text as well, just for the time be-
5532 # ing, if the unnamed param is set and the alt param is not.
5533 # For the future, we need to figure out if we want to tweak this more,
5534 # e.g., introducing a title= parameter for the title; ignoring the un-
5535 # named parameter entirely for images without a caption; adding an ex-
5536 # plicit caption= parameter and preserving the old magic unnamed para-
5538 if ( $imageIsFramed ) { # Framed image
5539 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5540 # No caption or alt text, add the filename as the alt text so
5541 # that screen readers at least get some description of the image
5542 $params['frame']['alt'] = $title->getText();
5544 # Do not set $params['frame']['title'] because tooltips don't make sense
5546 } else { # Inline image
5547 if ( !isset( $params['frame']['alt'] ) ) {
5548 # No alt text, use the "caption" for the alt text
5549 if ( $caption !== '' ) {
5550 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5552 # No caption, fall back to using the filename for the
5554 $params['frame']['alt'] = $title->getText();
5557 # Use the "caption" for the tooltip text
5558 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5560 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5562 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5564 # Linker does the rest
5565 $time = $options['time'] ??
false;
5566 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5567 $time, $descQuery, $this->mOptions
->getThumbSize() );
5569 # Give the handler a chance to modify the parser object
5571 $handler->parserTransformHook( $this, $file );
5578 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5580 * Adds an entry to appropriate link tables.
5583 * @param string $value
5584 * @return array of `[ type, target ]`, where:
5585 * - `type` is one of:
5586 * - `null`: Given value is not a valid link target, use default
5587 * - `'no-link'`: Given value is empty, do not generate a link
5588 * - `'link-url'`: Given value is a valid external link
5589 * - `'link-title'`: Given value is a valid internal link
5591 * - When `type` is `null` or `'no-link'`: `false`
5592 * - When `type` is `'link-url'`: URL string corresponding to given value
5593 * - When `type` is `'link-title'`: Title object corresponding to given value
5595 public function parseLinkParameter( $value ) {
5596 $chars = self
::EXT_LINK_URL_CLASS
;
5597 $addr = self
::EXT_LINK_ADDR
;
5598 $prots = $this->mUrlProtocols
;
5601 if ( $value === '' ) {
5603 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5604 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5605 $this->mOutput
->addExternalLink( $value );
5610 $linkTitle = Title
::newFromText( $value );
5612 $this->mOutput
->addLink( $linkTitle );
5613 $type = 'link-title';
5614 $target = $linkTitle;
5617 return [ $type, $target ];
5621 * @param string $caption
5622 * @param LinkHolderArray|bool $holders
5623 * @return mixed|string
5625 protected function stripAltText( $caption, $holders ) {
5626 # Strip bad stuff out of the title (tooltip). We can't just use
5627 # replaceLinkHoldersText() here, because if this function is called
5628 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5630 $tooltip = $holders->replaceText( $caption );
5632 $tooltip = $this->replaceLinkHoldersText( $caption );
5635 # make sure there are no placeholders in thumbnail attributes
5636 # that are later expanded to html- so expand them now and
5638 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5639 # Compatibility hack! In HTML certain entity references not terminated
5640 # by a semicolon are decoded (but not if we're in an attribute; that's
5641 # how link URLs get away without properly escaping & in queries).
5642 # But wikitext has always required semicolon-termination of entities,
5643 # so encode & where needed to avoid decode of semicolon-less entities.
5645 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5646 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5647 $tooltip = preg_replace( "/
5648 & # 1. entity prefix
5649 (?= # 2. followed by:
5650 (?: # a. one of the legacy semicolon-less named entities
5651 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5652 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5653 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5654 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5655 U(?:acute|circ|grave|uml)|Yacute|
5656 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5657 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5658 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5659 frac(?:1(?:2|4)|34)|
5660 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5661 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5662 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5663 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5664 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5665 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5666 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5667 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5668 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5670 (?:[^;]|$)) # b. and not followed by a semicolon
5671 # S = study, for efficiency
5672 /Sx", '&', $tooltip );
5673 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5679 * Set a flag in the output object indicating that the content is dynamic and
5680 * shouldn't be cached.
5681 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5683 public function disableCache() {
5684 $this->logger
->debug( "Parser output marked as uncacheable." );
5685 if ( !$this->mOutput
) {
5686 throw new MWException( __METHOD__
.
5687 " can only be called when actually parsing something" );
5689 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5693 * Callback from the Sanitizer for expanding items found in HTML attribute
5694 * values, so they can be safely tested and escaped.
5696 * @param string &$text
5697 * @param bool|PPFrame $frame
5700 public function attributeStripCallback( &$text, $frame = false ) {
5701 $text = $this->replaceVariables( $text, $frame );
5702 $text = $this->mStripState
->unstripBoth( $text );
5711 public function getTags() {
5712 $this->firstCallInit();
5714 array_keys( $this->mTransparentTagHooks
),
5715 array_keys( $this->mTagHooks
),
5716 array_keys( $this->mFunctionTagHooks
)
5724 public function getFunctionSynonyms() {
5725 $this->firstCallInit();
5726 return $this->mFunctionSynonyms
;
5733 public function getUrlProtocols() {
5734 return $this->mUrlProtocols
;
5738 * Replace transparent tags in $text with the values given by the callbacks.
5740 * Transparent tag hooks are like regular XML-style tag hooks, except they
5741 * operate late in the transformation sequence, on HTML instead of wikitext.
5743 * @param string $text
5747 public function replaceTransparentTags( $text ) {
5749 $elements = array_keys( $this->mTransparentTagHooks
);
5750 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5753 foreach ( $matches as $marker => $data ) {
5754 list( $element, $content, $params, $tag ) = $data;
5755 $tagName = strtolower( $element );
5756 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5757 $output = call_user_func_array(
5758 $this->mTransparentTagHooks
[$tagName],
5759 [ $content, $params, $this ]
5764 $replacements[$marker] = $output;
5766 return strtr( $text, $replacements );
5770 * Break wikitext input into sections, and either pull or replace
5771 * some particular section's text.
5773 * External callers should use the getSection and replaceSection methods.
5775 * @param string $text Page wikitext
5776 * @param string|int $sectionId A section identifier string of the form:
5777 * "<flag1> - <flag2> - ... - <section number>"
5779 * Currently the only recognised flag is "T", which means the target section number
5780 * was derived during a template inclusion parse, in other words this is a template
5781 * section edit link. If no flags are given, it was an ordinary section edit link.
5782 * This flag is required to avoid a section numbering mismatch when a section is
5783 * enclosed by "<includeonly>" (T8563).
5785 * The section number 0 pulls the text before the first heading; other numbers will
5786 * pull the given section along with its lower-level subsections. If the section is
5787 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5789 * Section 0 is always considered to exist, even if it only contains the empty
5790 * string. If $text is the empty string and section 0 is replaced, $newText is
5793 * @param string $mode One of "get" or "replace"
5794 * @param string $newText Replacement text for section data.
5795 * @return string For "get", the extracted section text.
5796 * for "replace", the whole page with the section replaced.
5798 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5799 global $wgTitle; # not generally used but removes an ugly failure mode
5801 $magicScopeVariable = $this->lock();
5802 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5804 $frame = $this->getPreprocessor()->newFrame();
5806 # Process section extraction flags
5808 $sectionParts = explode( '-', $sectionId );
5809 $sectionIndex = array_pop( $sectionParts );
5810 foreach ( $sectionParts as $part ) {
5811 if ( $part === 'T' ) {
5812 $flags |
= self
::PTD_FOR_INCLUSION
;
5816 # Check for empty input
5817 if ( strval( $text ) === '' ) {
5818 # Only sections 0 and T-0 exist in an empty document
5819 if ( $sectionIndex == 0 ) {
5820 if ( $mode === 'get' ) {
5826 if ( $mode === 'get' ) {
5834 # Preprocess the text
5835 $root = $this->preprocessToDom( $text, $flags );
5837 # <h> nodes indicate section breaks
5838 # They can only occur at the top level, so we can find them by iterating the root's children
5839 $node = $root->getFirstChild();
5841 # Find the target section
5842 if ( $sectionIndex == 0 ) {
5843 # Section zero doesn't nest, level=big
5844 $targetLevel = 1000;
5847 if ( $node->getName() === 'h' ) {
5848 $bits = $node->splitHeading();
5849 if ( $bits['i'] == $sectionIndex ) {
5850 $targetLevel = $bits['level'];
5854 if ( $mode === 'replace' ) {
5855 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5857 $node = $node->getNextSibling();
5863 if ( $mode === 'get' ) {
5870 # Find the end of the section, including nested sections
5872 if ( $node->getName() === 'h' ) {
5873 $bits = $node->splitHeading();
5874 $curLevel = $bits['level'];
5875 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5879 if ( $mode === 'get' ) {
5880 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5882 $node = $node->getNextSibling();
5885 # Write out the remainder (in replace mode only)
5886 if ( $mode === 'replace' ) {
5887 # Output the replacement text
5888 # Add two newlines on -- trailing whitespace in $newText is conventionally
5889 # stripped by the editor, so we need both newlines to restore the paragraph gap
5890 # Only add trailing whitespace if there is newText
5891 if ( $newText != "" ) {
5892 $outText .= $newText . "\n\n";
5896 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5897 $node = $node->getNextSibling();
5901 if ( is_string( $outText ) ) {
5902 # Re-insert stripped tags
5903 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5910 * This function returns the text of a section, specified by a number ($section).
5911 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5912 * the first section before any such heading (section 0).
5914 * If a section contains subsections, these are also returned.
5916 * @param string $text Text to look in
5917 * @param string|int $sectionId Section identifier as a number or string
5918 * (e.g. 0, 1 or 'T-1').
5919 * @param string $defaultText Default to return if section is not found
5921 * @return string Text of the requested section
5923 public function getSection( $text, $sectionId, $defaultText = '' ) {
5924 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5928 * This function returns $oldtext after the content of the section
5929 * specified by $section has been replaced with $text. If the target
5930 * section does not exist, $oldtext is returned unchanged.
5932 * @param string $oldText Former text of the article
5933 * @param string|int $sectionId Section identifier as a number or string
5934 * (e.g. 0, 1 or 'T-1').
5935 * @param string $newText Replacing text
5937 * @return string Modified text
5939 public function replaceSection( $oldText, $sectionId, $newText ) {
5940 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5944 * Get the ID of the revision we are parsing
5946 * The return value will be either:
5947 * - a) Positive, indicating a specific revision ID (current or old)
5948 * - b) Zero, meaning the revision ID is specified by getCurrentRevisionCallback()
5949 * - c) Null, meaning the parse is for preview mode and there is no revision
5953 public function getRevisionId() {
5954 return $this->mRevisionId
;
5958 * Get the revision object for $this->mRevisionId
5960 * @return Revision|null Either a Revision object or null
5961 * @since 1.23 (public since 1.23)
5963 public function getRevisionObject() {
5964 if ( $this->mRevisionObject
) {
5965 return $this->mRevisionObject
;
5968 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5969 // This is useful when parsing a revision that has not yet been saved.
5970 // However, if we get back a saved revision even though we are in
5971 // preview mode, we'll have to ignore it, see below.
5972 // NOTE: This callback may be used to inject an OLD revision that was
5973 // already loaded, so "current" is a bit of a misnomer. We can't just
5974 // skip it if mRevisionId is set.
5975 $rev = call_user_func(
5976 $this->mOptions
->getCurrentRevisionCallback(),
5981 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5982 // We are in preview mode (mRevisionId is null), and the current revision callback
5983 // returned an existing revision. Ignore it and return null, it's probably the page's
5984 // current revision, which is not what we want here. Note that we do want to call the
5985 // callback to allow the unsaved revision to be injected here, e.g. for
5986 // self-transclusion previews.
5990 // If the parse is for a new revision, then the callback should have
5991 // already been set to force the object and should match mRevisionId.
5992 // If not, try to fetch by mRevisionId for sanity.
5993 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5994 $rev = Revision
::newFromId( $this->mRevisionId
);
5997 $this->mRevisionObject
= $rev;
5999 return $this->mRevisionObject
;
6003 * Get the timestamp associated with the current revision, adjusted for
6004 * the default server-local timestamp
6005 * @return string TS_MW timestamp
6007 public function getRevisionTimestamp() {
6008 if ( $this->mRevisionTimestamp
!== null ) {
6009 return $this->mRevisionTimestamp
;
6012 # Use specified revision timestamp, falling back to the current timestamp
6013 $revObject = $this->getRevisionObject();
6014 $timestamp = $revObject ?
$revObject->getTimestamp() : $this->mOptions
->getTimestamp();
6015 $this->mOutput
->setRevisionTimestampUsed( $timestamp ); // unadjusted time zone
6017 # The cryptic '' timezone parameter tells to use the site-default
6018 # timezone offset instead of the user settings.
6019 # Since this value will be saved into the parser cache, served
6020 # to other users, and potentially even used inside links and such,
6021 # it needs to be consistent for all visitors.
6022 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
6024 return $this->mRevisionTimestamp
;
6028 * Get the name of the user that edited the last revision
6030 * @return string User name
6032 public function getRevisionUser() {
6033 if ( is_null( $this->mRevisionUser
) ) {
6034 $revObject = $this->getRevisionObject();
6036 # if this template is subst: the revision id will be blank,
6037 # so just use the current user's name
6039 $this->mRevisionUser
= $revObject->getUserText();
6040 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
6041 $this->mRevisionUser
= $this->getUser()->getName();
6044 return $this->mRevisionUser
;
6048 * Get the size of the revision
6050 * @return int|null Revision size
6052 public function getRevisionSize() {
6053 if ( is_null( $this->mRevisionSize
) ) {
6054 $revObject = $this->getRevisionObject();
6056 # if this variable is subst: the revision id will be blank,
6057 # so just use the parser input size, because the own substituation
6058 # will change the size.
6060 $this->mRevisionSize
= $revObject->getSize();
6062 $this->mRevisionSize
= $this->mInputSize
;
6065 return $this->mRevisionSize
;
6069 * Mutator for $mDefaultSort
6071 * @param string $sort New value
6073 public function setDefaultSort( $sort ) {
6074 $this->mDefaultSort
= $sort;
6075 $this->mOutput
->setProperty( 'defaultsort', $sort );
6079 * Accessor for $mDefaultSort
6080 * Will use the empty string if none is set.
6082 * This value is treated as a prefix, so the
6083 * empty string is equivalent to sorting by
6088 public function getDefaultSort() {
6089 if ( $this->mDefaultSort
!== false ) {
6090 return $this->mDefaultSort
;
6097 * Accessor for $mDefaultSort
6098 * Unlike getDefaultSort(), will return false if none is set
6100 * @return string|bool
6102 public function getCustomDefaultSort() {
6103 return $this->mDefaultSort
;
6106 private static function getSectionNameFromStrippedText( $text ) {
6107 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
6108 $text = Sanitizer
::decodeCharReferences( $text );
6109 $text = self
::normalizeSectionName( $text );
6113 private static function makeAnchor( $sectionName ) {
6114 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6117 private function makeLegacyAnchor( $sectionName ) {
6118 $fragmentMode = $this->svcOptions
->get( 'FragmentMode' );
6119 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6120 // ForAttribute() and ForLink() are the same for legacy encoding
6121 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6123 $id = Sanitizer
::escapeIdForLink( $sectionName );
6130 * Try to guess the section anchor name based on a wikitext fragment
6131 * presumably extracted from a heading, for example "Header" from
6134 * @param string $text
6135 * @return string Anchor (starting with '#')
6137 public function guessSectionNameFromWikiText( $text ) {
6138 # Strip out wikitext links(they break the anchor)
6139 $text = $this->stripSectionName( $text );
6140 $sectionName = self
::getSectionNameFromStrippedText( $text );
6141 return self
::makeAnchor( $sectionName );
6145 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6146 * instead, if possible. For use in redirects, since various versions
6147 * of Microsoft browsers interpret Location: headers as something other
6148 * than UTF-8, resulting in breakage.
6150 * @param string $text The section name
6151 * @return string Anchor (starting with '#')
6153 public function guessLegacySectionNameFromWikiText( $text ) {
6154 # Strip out wikitext links(they break the anchor)
6155 $text = $this->stripSectionName( $text );
6156 $sectionName = self
::getSectionNameFromStrippedText( $text );
6157 return $this->makeLegacyAnchor( $sectionName );
6161 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6162 * @param string $text Section name (plain text)
6163 * @return string Anchor (starting with '#')
6165 public static function guessSectionNameFromStrippedText( $text ) {
6166 $sectionName = self
::getSectionNameFromStrippedText( $text );
6167 return self
::makeAnchor( $sectionName );
6171 * Apply the same normalization as code making links to this section would
6173 * @param string $text
6176 private static function normalizeSectionName( $text ) {
6177 # T90902: ensure the same normalization is applied for IDs as to links
6178 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6181 $parts = $titleParser->splitTitleString( "#$text" );
6182 } catch ( MalformedTitleException
$ex ) {
6185 return $parts['fragment'];
6189 * Strips a text string of wikitext for use in a section anchor
6191 * Accepts a text string and then removes all wikitext from the
6192 * string and leaves only the resultant text (i.e. the result of
6193 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6194 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6195 * to create valid section anchors by mimicing the output of the
6196 * parser when headings are parsed.
6198 * @param string $text Text string to be stripped of wikitext
6199 * for use in a Section anchor
6200 * @return string Filtered text string
6202 public function stripSectionName( $text ) {
6203 # Strip internal link markup
6204 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6205 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6207 # Strip external link markup
6208 # @todo FIXME: Not tolerant to blank link text
6209 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6210 # on how many empty links there are on the page - need to figure that out.
6211 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6213 # Parse wikitext quotes (italics & bold)
6214 $text = $this->doQuotes( $text );
6217 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6222 * strip/replaceVariables/unstrip for preprocessor regression testing
6224 * @param string $text
6225 * @param Title $title
6226 * @param ParserOptions $options
6227 * @param int $outputType
6231 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6232 $outputType = self
::OT_HTML
6234 $magicScopeVariable = $this->lock();
6235 $this->startParse( $title, $options, $outputType, true );
6237 $text = $this->replaceVariables( $text );
6238 $text = $this->mStripState
->unstripBoth( $text );
6239 $text = Sanitizer
::removeHTMLtags( $text );
6244 * @param string $text
6245 * @param Title $title
6246 * @param ParserOptions $options
6249 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6250 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6254 * @param string $text
6255 * @param Title $title
6256 * @param ParserOptions $options
6259 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6260 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6264 * Call a callback function on all regions of the given text that are not
6265 * inside strip markers, and replace those regions with the return value
6266 * of the callback. For example, with input:
6270 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6271 * two strings will be replaced with the value returned by the callback in
6275 * @param callable $callback
6279 public function markerSkipCallback( $s, $callback ) {
6282 while ( $i < strlen( $s ) ) {
6283 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6284 if ( $markerStart === false ) {
6285 $out .= call_user_func( $callback, substr( $s, $i ) );
6288 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6289 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6290 if ( $markerEnd === false ) {
6291 $out .= substr( $s, $markerStart );
6294 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6295 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6304 * Remove any strip markers found in the given text.
6306 * @param string $text
6309 public function killMarkers( $text ) {
6310 return $this->mStripState
->killMarkers( $text );
6314 * Save the parser state required to convert the given half-parsed text to
6315 * HTML. "Half-parsed" in this context means the output of
6316 * recursiveTagParse() or internalParse(). This output has strip markers
6317 * from replaceVariables (extensionSubstitution() etc.), and link
6318 * placeholders from replaceLinkHolders().
6320 * Returns an array which can be serialized and stored persistently. This
6321 * array can later be loaded into another parser instance with
6322 * unserializeHalfParsedText(). The text can then be safely incorporated into
6323 * the return value of a parser hook.
6325 * @deprecated since 1.31
6326 * @param string $text
6330 public function serializeHalfParsedText( $text ) {
6331 wfDeprecated( __METHOD__
, '1.31' );
6334 'version' => self
::HALF_PARSED_VERSION
,
6335 'stripState' => $this->mStripState
->getSubState( $text ),
6336 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6342 * Load the parser state given in the $data array, which is assumed to
6343 * have been generated by serializeHalfParsedText(). The text contents is
6344 * extracted from the array, and its markers are transformed into markers
6345 * appropriate for the current Parser instance. This transformed text is
6346 * returned, and can be safely included in the return value of a parser
6349 * If the $data array has been stored persistently, the caller should first
6350 * check whether it is still valid, by calling isValidHalfParsedText().
6352 * @deprecated since 1.31
6353 * @param array $data Serialized data
6354 * @throws MWException
6357 public function unserializeHalfParsedText( $data ) {
6358 wfDeprecated( __METHOD__
, '1.31' );
6359 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6360 throw new MWException( __METHOD__
. ': invalid version' );
6363 # First, extract the strip state.
6364 $texts = [ $data['text'] ];
6365 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6367 # Now renumber links
6368 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6370 # Should be good to go.
6375 * Returns true if the given array, presumed to be generated by
6376 * serializeHalfParsedText(), is compatible with the current version of the
6379 * @deprecated since 1.31
6380 * @param array $data
6384 public function isValidHalfParsedText( $data ) {
6385 wfDeprecated( __METHOD__
, '1.31' );
6386 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6390 * Parsed a width param of imagelink like 300px or 200x300px
6392 * @param string $value
6393 * @param bool $parseHeight
6398 public static function parseWidthParam( $value, $parseHeight = true ) {
6399 $parsedWidthParam = [];
6400 if ( $value === '' ) {
6401 return $parsedWidthParam;
6404 # (T15500) In both cases (width/height and width only),
6405 # permit trailing "px" for backward compatibility.
6406 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6407 $width = intval( $m[1] );
6408 $height = intval( $m[2] );
6409 $parsedWidthParam['width'] = $width;
6410 $parsedWidthParam['height'] = $height;
6411 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6412 $width = intval( $value );
6413 $parsedWidthParam['width'] = $width;
6415 return $parsedWidthParam;
6419 * Lock the current instance of the parser.
6421 * This is meant to stop someone from calling the parser
6422 * recursively and messing up all the strip state.
6424 * @throws MWException If parser is in a parse
6425 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6427 protected function lock() {
6428 if ( $this->mInParse
) {
6429 throw new MWException( "Parser state cleared while parsing. "
6430 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6433 // Save the backtrace when locking, so that if some code tries locking again,
6434 // we can print the lock owner's backtrace for easier debugging
6436 $this->mInParse
= $e->getTraceAsString();
6438 $recursiveCheck = new ScopedCallback( function () {
6439 $this->mInParse
= false;
6442 return $recursiveCheck;
6446 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6448 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6449 * or if there is more than one <p/> tag in the input HTML.
6451 * @param string $html
6455 public static function stripOuterParagraph( $html ) {
6457 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6465 * Return this parser if it is not doing anything, otherwise
6466 * get a fresh parser. You can use this method by doing
6467 * $newParser = $oldParser->getFreshParser(), or more simply
6468 * $oldParser->getFreshParser()->parse( ... );
6469 * if you're unsure if $oldParser is safe to use.
6472 * @return Parser A parser object that is not parsing anything
6474 public function getFreshParser() {
6475 if ( $this->mInParse
) {
6476 return $this->factory
->create();
6483 * Set's up the PHP implementation of OOUI for use in this request
6484 * and instructs OutputPage to enable OOUI for itself.
6488 public function enableOOUI() {
6489 OutputPage
::setupOOUI();
6490 $this->mOutput
->setEnableOOUI( true );
6494 * @param string $flag
6495 * @param string $reason
6497 protected function setOutputFlag( $flag, $reason ) {
6498 $this->mOutput
->setFlag( $flag );
6499 $name = $this->mTitle
->getPrefixedText();
6500 $this->logger
->debug( __METHOD__
. ": set $flag flag on '$name'; $reason" );