3 * PHP parser that converts wiki markup to HTML.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
23 use MediaWiki\Config\ServiceOptions
;
24 use MediaWiki\Linker\LinkRenderer
;
25 use MediaWiki\Linker\LinkRendererFactory
;
26 use MediaWiki\Linker\LinkTarget
;
27 use MediaWiki\MediaWikiServices
;
28 use MediaWiki\Special\SpecialPageFactory
;
29 use Wikimedia\ScopedCallback
;
32 * @defgroup Parser Parser
36 * PHP Parser - Processes wiki markup (which uses a more user-friendly
37 * syntax, such as "[[link]]" for making links), and provides a one-way
38 * transformation of that wiki markup it into (X)HTML output / markup
39 * (which in turn the browser understands, and can display).
41 * There are seven main entry points into the Parser class:
44 * produces HTML output
45 * - Parser::preSaveTransform()
46 * produces altered wiki markup
47 * - Parser::preprocess()
48 * removes HTML comments and expands templates
49 * - Parser::cleanSig() and Parser::cleanSigInSig()
50 * cleans a signature before saving it to preferences
51 * - Parser::getSection()
52 * return the content of a section from an article for section editing
53 * - Parser::replaceSection()
54 * replaces a section by number inside an article
55 * - Parser::getPreloadText()
56 * removes <noinclude> sections and <includeonly> tags
58 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
61 * $wgNamespacesWithSubpages
63 * @par Settings only within ParserOptions:
64 * $wgAllowExternalImages
65 * $wgAllowSpecialInclusion
73 * Update this version number when the ParserOutput format
74 * changes in an incompatible way, so the parser cache
75 * can automatically discard old data.
77 const VERSION
= '1.6.4';
80 * Update this version number when the output of serialiseHalfParsedText()
81 * changes in an incompatible way
83 const HALF_PARSED_VERSION
= 2;
85 # Flags for Parser::setFunctionHook
86 const SFH_NO_HASH
= 1;
87 const SFH_OBJECT_ARGS
= 2;
89 # Constants needed for external link processing
90 # Everything except bracket, space, or control characters
91 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
92 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
93 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
94 # uses to replace invalid HTML characters.
95 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
96 # Simplified expression to match an IPv4 or IPv6 address, or
97 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
98 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
99 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
100 // phpcs:ignore Generic.Files.LineLength
101 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
102 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION
= 1;
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML
= 1; # like parse()
113 const OT_WIKI
= 2; # like preSaveTransform()
114 const OT_PREPROCESS
= 3; # like preprocess()
116 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
135 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
136 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
138 # Markers used for wrapping the table of contents
139 const TOC_START
= '<mw:toc>';
140 const TOC_END
= '</mw:toc>';
142 /** @var int Assume that no output will later be saved this many seconds after parsing */
146 public $mTagHooks = [];
147 public $mTransparentTagHooks = [];
148 public $mFunctionHooks = [];
149 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
150 public $mFunctionTagHooks = [];
151 public $mStripList = [];
152 public $mDefaultStripList = [];
153 public $mVarCache = [];
154 public $mImageParams = [];
155 public $mImageParamsMagicArray = [];
156 public $mMarkerIndex = 0;
158 * @var bool Whether firstCallInit still needs to be called
160 public $mFirstCall = true;
162 # Initialised by initialiseVariables()
165 * @var MagicWordArray
170 * @var MagicWordArray
175 * @deprecated since 1.34, there should be no need to use this
180 # Initialised in constructor
181 public $mExtLinkBracketedRegex, $mUrlProtocols;
183 # Initialized in getPreprocessor()
184 /** @var Preprocessor */
185 public $mPreprocessor;
187 # Cleared with clearState():
199 public $mIncludeCount;
201 * @var LinkHolderArray
203 public $mLinkHolders;
206 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
207 public $mDefaultSort;
208 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
209 public $mExpensiveFunctionCount; # number of expensive parser function calls
210 public $mShowToc, $mForceTocPosition;
215 public $mUser; # User object; only used when doing pre-save transform
218 # These are variables reset at least once per parse regardless of $clearState
228 public $mTitle; # Title context, used for self-link rendering and similar things
229 public $mOutputType; # Output type, one of the OT_xxx constants
230 public $ot; # Shortcut alias, see setOutputType()
231 public $mRevisionObject; # The revision object of the specified revision ID
232 public $mRevisionId; # ID to display in {{REVISIONID}} tags
233 public $mRevisionTimestamp; # The timestamp of the specified revision ID
234 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
235 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
236 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
237 public $mInputSize = false; # For {{PAGESIZE}} on current page.
240 * @var array Array with the language name of each language link (i.e. the
241 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
242 * duplicate language links to the ParserOutput.
244 public $mLangLinkLanguages;
247 * @var MapCacheLRU|null
250 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
252 public $currentRevisionCache;
255 * @var bool|string Recursive call protection.
256 * This variable should be treated as if it were private.
258 public $mInParse = false;
260 /** @var SectionProfiler */
261 protected $mProfiler;
266 protected $mLinkRenderer;
268 /** @var MagicWordFactory */
269 private $magicWordFactory;
274 /** @var ParserFactory */
277 /** @var SpecialPageFactory */
278 private $specialPageFactory;
281 * This is called $svcOptions instead of $options like elsewhere to avoid confusion with
282 * $mOptions, which is public and widely used, and also with the local variable $options used
283 * for ParserOptions throughout this file.
285 * @var ServiceOptions
289 /** @var LinkRendererFactory */
290 private $linkRendererFactory;
292 /** @var NamespaceInfo */
296 * TODO Make this a const when HHVM support is dropped (T192166)
301 public static $constructorOptions = [
302 // See $wgParserConf documentation
305 // See documentation for the corresponding config options
307 'EnableScaryTranscluding',
308 'ExtraInterlanguageLinkPrefixes',
320 'TranscludeCacheExpiry',
324 * Constructing parsers directly is deprecated! Use a ParserFactory.
326 * @param ServiceOptions|null $svcOptions
327 * @param MagicWordFactory|null $magicWordFactory
328 * @param Language|null $contLang Content language
329 * @param ParserFactory|null $factory
330 * @param string|null $urlProtocols As returned from wfUrlProtocols()
331 * @param SpecialPageFactory|null $spFactory
332 * @param LinkRendererFactory|null $linkRendererFactory
333 * @param NamespaceInfo|null $nsInfo
335 public function __construct(
336 $svcOptions = null, MagicWordFactory
$magicWordFactory = null,
337 Language
$contLang = null, ParserFactory
$factory = null, $urlProtocols = null,
338 SpecialPageFactory
$spFactory = null, $linkRendererFactory = null, $nsInfo = null
340 $services = MediaWikiServices
::getInstance();
341 if ( !$svcOptions ||
is_array( $svcOptions ) ) {
342 // Pre-1.34 calling convention is the first parameter is just ParserConf, the seventh is
343 // Config, and the eighth is LinkRendererFactory.
344 $this->mConf
= (array)$svcOptions;
345 if ( empty( $this->mConf
['class'] ) ) {
346 $this->mConf
['class'] = self
::class;
348 if ( empty( $this->mConf
['preprocessorClass'] ) ) {
349 $this->mConf
['preprocessorClass'] = self
::getDefaultPreprocessorClass();
351 $this->svcOptions
= new ServiceOptions( self
::$constructorOptions,
353 func_num_args() > 6 ?
func_get_arg( 6 ) : $services->getMainConfig()
355 $linkRendererFactory = func_num_args() > 7 ?
func_get_arg( 7 ) : null;
356 $nsInfo = func_num_args() > 8 ?
func_get_arg( 8 ) : null;
358 // New calling convention
359 $svcOptions->assertRequiredOptions( self
::$constructorOptions );
360 // $this->mConf is public, so we'll keep those two options there as well for
361 // compatibility until it's removed
363 'class' => $svcOptions->get( 'class' ),
364 'preprocessorClass' => $svcOptions->get( 'preprocessorClass' ),
366 $this->svcOptions
= $svcOptions;
369 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
370 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
371 self
::EXT_LINK_ADDR
.
372 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
374 $this->magicWordFactory
= $magicWordFactory ??
375 $services->getMagicWordFactory();
377 $this->contLang
= $contLang ??
$services->getContentLanguage();
379 $this->factory
= $factory ??
$services->getParserFactory();
380 $this->specialPageFactory
= $spFactory ??
$services->getSpecialPageFactory();
381 $this->linkRendererFactory
= $linkRendererFactory ??
$services->getLinkRendererFactory();
382 $this->nsInfo
= $nsInfo ??
$services->getNamespaceInfo();
386 * Reduce memory usage to reduce the impact of circular references
388 public function __destruct() {
389 if ( isset( $this->mLinkHolders
) ) {
390 unset( $this->mLinkHolders
);
392 foreach ( $this as $name => $value ) {
393 unset( $this->$name );
398 * Allow extensions to clean up when the parser is cloned
400 public function __clone() {
401 $this->mInParse
= false;
403 // T58226: When you create a reference "to" an object field, that
404 // makes the object field itself be a reference too (until the other
405 // reference goes out of scope). When cloning, any field that's a
406 // reference is copied as a reference in the new object. Both of these
407 // are defined PHP5 behaviors, as inconvenient as it is for us when old
408 // hooks from PHP4 days are passing fields by reference.
409 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
410 // Make a non-reference copy of the field, then rebind the field to
411 // reference the new copy.
417 Hooks
::run( 'ParserCloned', [ $this ] );
421 * Which class should we use for the preprocessor if not otherwise specified?
426 public static function getDefaultPreprocessorClass() {
428 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
429 return Preprocessor_Hash
::class;
431 if ( extension_loaded( 'domxml' ) ) {
432 # PECL extension that conflicts with the core DOM extension (T15770)
433 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
434 return Preprocessor_Hash
::class;
436 if ( extension_loaded( 'dom' ) ) {
437 return Preprocessor_DOM
::class;
439 return Preprocessor_Hash
::class;
443 * Do various kinds of initialisation on the first call of the parser
445 public function firstCallInit() {
446 if ( !$this->mFirstCall
) {
449 $this->mFirstCall
= false;
451 CoreParserFunctions
::register( $this );
452 CoreTagHooks
::register( $this );
453 $this->initialiseVariables();
455 // Avoid PHP 7.1 warning from passing $this by reference
457 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
465 public function clearState() {
466 $this->firstCallInit();
467 $this->mOutput
= new ParserOutput
;
468 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
469 $this->mAutonumber
= 0;
470 $this->mIncludeCount
= [];
471 $this->mLinkHolders
= new LinkHolderArray( $this );
473 $this->mRevisionObject
= $this->mRevisionTimestamp
=
474 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
475 $this->mVarCache
= [];
477 $this->mLangLinkLanguages
= [];
478 $this->currentRevisionCache
= null;
480 $this->mStripState
= new StripState( $this );
482 # Clear these on every parse, T6549
483 $this->mTplRedirCache
= $this->mTplDomCache
= [];
485 $this->mShowToc
= true;
486 $this->mForceTocPosition
= false;
487 $this->mIncludeSizes
= [
491 $this->mPPNodeCount
= 0;
492 $this->mGeneratedPPNodeCount
= 0;
493 $this->mHighestExpansionDepth
= 0;
494 $this->mDefaultSort
= false;
495 $this->mHeadings
= [];
496 $this->mDoubleUnderscores
= [];
497 $this->mExpensiveFunctionCount
= 0;
500 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
501 $this->mPreprocessor
= null;
504 $this->mProfiler
= new SectionProfiler();
506 // Avoid PHP 7.1 warning from passing $this by reference
508 Hooks
::run( 'ParserClearState', [ &$parser ] );
512 * Convert wikitext to HTML
513 * Do not call this function recursively.
515 * @param string $text Text we want to parse
516 * @param-taint $text escapes_htmlnoent
517 * @param Title $title
518 * @param ParserOptions $options
519 * @param bool $linestart
520 * @param bool $clearState
521 * @param int|null $revid Number to pass in {{REVISIONID}}
522 * @return ParserOutput A ParserOutput
523 * @return-taint escaped
525 public function parse(
526 $text, Title
$title, ParserOptions
$options,
527 $linestart = true, $clearState = true, $revid = null
530 // We use U+007F DELETE to construct strip markers, so we have to make
531 // sure that this character does not occur in the input text.
532 $text = strtr( $text, "\x7f", "?" );
533 $magicScopeVariable = $this->lock();
535 // Strip U+0000 NULL (T159174)
536 $text = str_replace( "\000", '', $text );
538 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
540 $this->currentRevisionCache
= null;
541 $this->mInputSize
= strlen( $text );
542 if ( $this->mOptions
->getEnableLimitReport() ) {
543 $this->mOutput
->resetParseStartTime();
546 $oldRevisionId = $this->mRevisionId
;
547 $oldRevisionObject = $this->mRevisionObject
;
548 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
549 $oldRevisionUser = $this->mRevisionUser
;
550 $oldRevisionSize = $this->mRevisionSize
;
551 if ( $revid !== null ) {
552 $this->mRevisionId
= $revid;
553 $this->mRevisionObject
= null;
554 $this->mRevisionTimestamp
= null;
555 $this->mRevisionUser
= null;
556 $this->mRevisionSize
= null;
559 // Avoid PHP 7.1 warning from passing $this by reference
561 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
563 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
564 $text = $this->internalParse( $text );
565 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
567 $text = $this->internalParseHalfParsed( $text, true, $linestart );
570 * A converted title will be provided in the output object if title and
571 * content conversion are enabled, the article text does not contain
572 * a conversion-suppressing double-underscore tag, and no
573 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
574 * automatic link conversion.
576 if ( !( $options->getDisableTitleConversion()
577 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
578 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
579 ||
$this->mOutput
->getDisplayTitle() !== false )
581 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
582 if ( $convruletitle ) {
583 $this->mOutput
->setTitleText( $convruletitle );
585 $titleText = $this->getTargetLanguage()->convertTitle( $title );
586 $this->mOutput
->setTitleText( $titleText );
590 # Compute runtime adaptive expiry if set
591 $this->mOutput
->finalizeAdaptiveCacheExpiry();
593 # Warn if too many heavyweight parser functions were used
594 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
595 $this->limitationWarn( 'expensive-parserfunction',
596 $this->mExpensiveFunctionCount
,
597 $this->mOptions
->getExpensiveParserFunctionLimit()
601 # Information on limits, for the benefit of users who try to skirt them
602 if ( $this->mOptions
->getEnableLimitReport() ) {
603 $text .= $this->makeLimitReport();
606 # Wrap non-interface parser output in a <div> so it can be targeted
608 $class = $this->mOptions
->getWrapOutputClass();
609 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
610 $this->mOutput
->addWrapperDivClass( $class );
613 $this->mOutput
->setText( $text );
615 $this->mRevisionId
= $oldRevisionId;
616 $this->mRevisionObject
= $oldRevisionObject;
617 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
618 $this->mRevisionUser
= $oldRevisionUser;
619 $this->mRevisionSize
= $oldRevisionSize;
620 $this->mInputSize
= false;
621 $this->currentRevisionCache
= null;
623 return $this->mOutput
;
627 * Set the limit report data in the current ParserOutput, and return the
628 * limit report HTML comment.
632 protected function makeLimitReport() {
633 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
635 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
636 if ( $cpuTime !== null ) {
637 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
638 sprintf( "%.3f", $cpuTime )
642 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
643 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
644 sprintf( "%.3f", $wallTime )
647 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
648 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
650 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
651 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
653 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
654 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
656 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
657 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
659 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
660 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
662 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
663 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
666 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
667 $this->mOutput
->setLimitReportData( $key, $value );
670 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
672 $limitReport = "NewPP limit report\n";
673 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
674 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
676 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
677 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
678 $limitReport .= 'Dynamic content: ' .
679 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
681 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput
->getAllFlags() ) . "]\n";
683 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
684 if ( Hooks
::run( 'ParserLimitReportFormat',
685 [ $key, &$value, &$limitReport, false, false ]
687 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
688 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
689 ->inLanguage( 'en' )->useDatabase( false );
690 if ( !$valueMsg->exists() ) {
691 $valueMsg = new RawMessage( '$1' );
693 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
694 $valueMsg->params( $value );
695 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
699 // Since we're not really outputting HTML, decode the entities and
700 // then re-encode the things that need hiding inside HTML comments.
701 $limitReport = htmlspecialchars_decode( $limitReport );
703 // Sanitize for comment. Note '‐' in the replacement is U+2010,
704 // which looks much like the problematic '-'.
705 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
706 $text = "\n<!-- \n$limitReport-->\n";
708 // Add on template profiling data in human/machine readable way
709 $dataByFunc = $this->mProfiler
->getFunctionStats();
710 uasort( $dataByFunc, function ( $a, $b ) {
711 return $b['real'] <=> $a['real']; // descending order
714 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
715 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
716 $item['%real'], $item['real'], $item['calls'],
717 htmlspecialchars( $item['name'] ) );
719 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
720 $text .= implode( "\n", $profileReport ) . "\n-->\n";
722 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
724 // Add other cache related metadata
725 if ( $this->svcOptions
->get( 'ShowHostnames' ) ) {
726 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
728 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
729 $this->mOutput
->getCacheTime() );
730 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
731 $this->mOutput
->getCacheExpiry() );
732 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
733 $this->mOutput
->hasDynamicContent() );
735 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
736 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
737 $this->mTitle
->getPrefixedDBkey() );
743 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
744 * can be called from an extension tag hook.
746 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
747 * instead, which means that lists and links have not been fully parsed yet,
748 * and strip markers are still present.
750 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
752 * Use this function if you're a parser tag hook and you want to parse
753 * wikitext before or after applying additional transformations, and you
754 * intend to *return the result as hook output*, which will cause it to go
755 * through the rest of parsing process automatically.
757 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
758 * $text are not expanded
760 * @param string $text Text extension wants to have parsed
761 * @param-taint $text escapes_htmlnoent
762 * @param bool|PPFrame $frame The frame to use for expanding any template variables
763 * @return string UNSAFE half-parsed HTML
764 * @return-taint escaped
766 public function recursiveTagParse( $text, $frame = false ) {
767 // Avoid PHP 7.1 warning from passing $this by reference
769 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
770 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
771 $text = $this->internalParse( $text, false, $frame );
776 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
777 * point can be called from an extension tag hook.
779 * The output of this function is fully-parsed HTML that is safe for output.
780 * If you're a parser tag hook, you might want to use recursiveTagParse()
783 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
784 * $text are not expanded
788 * @param string $text Text extension wants to have parsed
789 * @param-taint $text escapes_htmlnoent
790 * @param bool|PPFrame $frame The frame to use for expanding any template variables
791 * @return string Fully parsed HTML
792 * @return-taint escaped
794 public function recursiveTagParseFully( $text, $frame = false ) {
795 $text = $this->recursiveTagParse( $text, $frame );
796 $text = $this->internalParseHalfParsed( $text, false );
801 * Expand templates and variables in the text, producing valid, static wikitext.
802 * Also removes comments.
803 * Do not call this function recursively.
804 * @param string $text
805 * @param Title|null $title
806 * @param ParserOptions $options
807 * @param int|null $revid
808 * @param bool|PPFrame $frame
809 * @return mixed|string
811 public function preprocess( $text, Title
$title = null,
812 ParserOptions
$options, $revid = null, $frame = false
814 $magicScopeVariable = $this->lock();
815 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
816 if ( $revid !== null ) {
817 $this->mRevisionId
= $revid;
819 // Avoid PHP 7.1 warning from passing $this by reference
821 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
822 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
823 $text = $this->replaceVariables( $text, $frame );
824 $text = $this->mStripState
->unstripBoth( $text );
829 * Recursive parser entry point that can be called from an extension tag
832 * @param string $text Text to be expanded
833 * @param bool|PPFrame $frame The frame to use for expanding any template variables
837 public function recursivePreprocess( $text, $frame = false ) {
838 $text = $this->replaceVariables( $text, $frame );
839 $text = $this->mStripState
->unstripBoth( $text );
844 * Process the wikitext for the "?preload=" feature. (T7210)
846 * "<noinclude>", "<includeonly>" etc. are parsed as for template
847 * transclusion, comments, templates, arguments, tags hooks and parser
848 * functions are untouched.
850 * @param string $text
851 * @param Title $title
852 * @param ParserOptions $options
853 * @param array $params
856 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
857 $msg = new RawMessage( $text );
858 $text = $msg->params( $params )->plain();
860 # Parser (re)initialisation
861 $magicScopeVariable = $this->lock();
862 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
864 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
865 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
866 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
867 $text = $this->mStripState
->unstripBoth( $text );
872 * Set the current user.
873 * Should only be used when doing pre-save transform.
875 * @param User|null $user User object or null (to reset)
877 public function setUser( $user ) {
878 $this->mUser
= $user;
882 * Set the context title
886 public function setTitle( $t ) {
888 $t = Title
::newFromText( 'NO TITLE' );
891 if ( $t->hasFragment() ) {
892 # Strip the fragment to avoid various odd effects
893 $this->mTitle
= $t->createFragmentTarget( '' );
900 * Accessor for the Title object
904 public function getTitle() {
905 return $this->mTitle
;
909 * Accessor/mutator for the Title object
911 * @param Title|null $x Title object or null to just get the current one
914 public function Title( $x = null ) {
915 return wfSetVar( $this->mTitle
, $x );
919 * Set the output type
921 * @param int $ot New value
923 public function setOutputType( $ot ) {
924 $this->mOutputType
= $ot;
927 'html' => $ot == self
::OT_HTML
,
928 'wiki' => $ot == self
::OT_WIKI
,
929 'pre' => $ot == self
::OT_PREPROCESS
,
930 'plain' => $ot == self
::OT_PLAIN
,
935 * Accessor/mutator for the output type
937 * @param int|null $x New value or null to just get the current one
940 public function OutputType( $x = null ) {
941 return wfSetVar( $this->mOutputType
, $x );
945 * Get the ParserOutput object
947 * @return ParserOutput
949 public function getOutput() {
950 return $this->mOutput
;
954 * Get the ParserOptions object
956 * @return ParserOptions
958 public function getOptions() {
959 return $this->mOptions
;
963 * Accessor/mutator for the ParserOptions object
965 * @param ParserOptions|null $x New value or null to just get the current one
966 * @return ParserOptions Current ParserOptions object
968 public function Options( $x = null ) {
969 return wfSetVar( $this->mOptions
, $x );
975 public function nextLinkID() {
976 return $this->mLinkID++
;
982 public function setLinkID( $id ) {
983 $this->mLinkID
= $id;
987 * Get a language object for use in parser functions such as {{FORMATNUM:}}
990 public function getFunctionLang() {
991 return $this->getTargetLanguage();
995 * Get the target language for the content being parsed. This is usually the
996 * language that the content is in.
1000 * @throws MWException
1003 public function getTargetLanguage() {
1004 $target = $this->mOptions
->getTargetLanguage();
1006 if ( $target !== null ) {
1008 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
1009 return $this->mOptions
->getUserLangObj();
1010 } elseif ( is_null( $this->mTitle
) ) {
1011 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
1014 return $this->mTitle
->getPageLanguage();
1018 * Get the language object for language conversion
1019 * @deprecated since 1.32, just use getTargetLanguage()
1020 * @return Language|null
1022 public function getConverterLanguage() {
1023 return $this->getTargetLanguage();
1027 * Get a User object either from $this->mUser, if set, or from the
1028 * ParserOptions object otherwise
1032 public function getUser() {
1033 if ( !is_null( $this->mUser
) ) {
1034 return $this->mUser
;
1036 return $this->mOptions
->getUser();
1040 * Get a preprocessor object
1042 * @return Preprocessor
1044 public function getPreprocessor() {
1045 if ( !isset( $this->mPreprocessor
) ) {
1046 $class = $this->svcOptions
->get( 'preprocessorClass' );
1047 $this->mPreprocessor
= new $class( $this );
1049 return $this->mPreprocessor
;
1053 * Get a LinkRenderer instance to make links with
1056 * @return LinkRenderer
1058 public function getLinkRenderer() {
1059 // XXX We make the LinkRenderer with current options and then cache it forever
1060 if ( !$this->mLinkRenderer
) {
1061 $this->mLinkRenderer
= $this->linkRendererFactory
->create();
1062 $this->mLinkRenderer
->setStubThreshold(
1063 $this->getOptions()->getStubThreshold()
1067 return $this->mLinkRenderer
;
1071 * Get the MagicWordFactory that this Parser is using
1074 * @return MagicWordFactory
1076 public function getMagicWordFactory() {
1077 return $this->magicWordFactory
;
1081 * Get the content language that this Parser is using
1086 public function getContentLanguage() {
1087 return $this->contLang
;
1091 * Replaces all occurrences of HTML-style comments and the given tags
1092 * in the text with a random marker and returns the next text. The output
1093 * parameter $matches will be an associative array filled with data in
1100 * [ 'param' => 'x' ],
1101 * '<element param="x">tag content</element>' ]
1104 * @param array $elements List of element names. Comments are always extracted.
1105 * @param string $text Source text string.
1106 * @param array &$matches Out parameter, Array: extracted tags
1107 * @return string Stripped text
1109 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1114 $taglist = implode( '|', $elements );
1115 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1117 while ( $text != '' ) {
1118 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1120 if ( count( $p ) < 5 ) {
1123 if ( count( $p ) > 5 ) {
1131 list( , $element, $attributes, $close, $inside ) = $p;
1134 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1135 $stripped .= $marker;
1137 if ( $close === '/>' ) {
1138 # Empty element tag, <tag />
1143 if ( $element === '!--' ) {
1146 $end = "/(<\\/$element\\s*>)/i";
1148 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1150 if ( count( $q ) < 3 ) {
1151 # No end tag -- let it run out to the end of the text.
1155 list( , $tail, $text ) = $q;
1159 $matches[$marker] = [ $element,
1161 Sanitizer
::decodeTagAttributes( $attributes ),
1162 "<$element$attributes$close$content$tail" ];
1168 * Get a list of strippable XML-like elements
1172 public function getStripList() {
1173 return $this->mStripList
;
1177 * Add an item to the strip state
1178 * Returns the unique tag which must be inserted into the stripped text
1179 * The tag will be replaced with the original text in unstrip()
1181 * @param string $text
1185 public function insertStripItem( $text ) {
1186 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1187 $this->mMarkerIndex++
;
1188 $this->mStripState
->addGeneral( $marker, $text );
1193 * parse the wiki syntax used to render tables
1196 * @param string $text
1199 public function doTableStuff( $text ) {
1200 $lines = StringUtils
::explode( "\n", $text );
1202 $td_history = []; # Is currently a td tag open?
1203 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1204 $tr_history = []; # Is currently a tr tag open?
1205 $tr_attributes = []; # history of tr attributes
1206 $has_opened_tr = []; # Did this table open a <tr> element?
1207 $indent_level = 0; # indent level of the table
1209 foreach ( $lines as $outLine ) {
1210 $line = trim( $outLine );
1212 if ( $line === '' ) { # empty line, go to next line
1213 $out .= $outLine . "\n";
1217 $first_character = $line[0];
1218 $first_two = substr( $line, 0, 2 );
1221 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1222 # First check if we are starting a new table
1223 $indent_level = strlen( $matches[1] );
1225 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1226 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1228 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1229 array_push( $td_history, false );
1230 array_push( $last_tag_history, '' );
1231 array_push( $tr_history, false );
1232 array_push( $tr_attributes, '' );
1233 array_push( $has_opened_tr, false );
1234 } elseif ( count( $td_history ) == 0 ) {
1235 # Don't do any of the following
1236 $out .= $outLine . "\n";
1238 } elseif ( $first_two === '|}' ) {
1239 # We are ending a table
1240 $line = '</table>' . substr( $line, 2 );
1241 $last_tag = array_pop( $last_tag_history );
1243 if ( !array_pop( $has_opened_tr ) ) {
1244 $line = "<tr><td></td></tr>{$line}";
1247 if ( array_pop( $tr_history ) ) {
1248 $line = "</tr>{$line}";
1251 if ( array_pop( $td_history ) ) {
1252 $line = "</{$last_tag}>{$line}";
1254 array_pop( $tr_attributes );
1255 if ( $indent_level > 0 ) {
1256 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1260 } elseif ( $first_two === '|-' ) {
1261 # Now we have a table row
1262 $line = preg_replace( '#^\|-+#', '', $line );
1264 # Whats after the tag is now only attributes
1265 $attributes = $this->mStripState
->unstripBoth( $line );
1266 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1267 array_pop( $tr_attributes );
1268 array_push( $tr_attributes, $attributes );
1271 $last_tag = array_pop( $last_tag_history );
1272 array_pop( $has_opened_tr );
1273 array_push( $has_opened_tr, true );
1275 if ( array_pop( $tr_history ) ) {
1279 if ( array_pop( $td_history ) ) {
1280 $line = "</{$last_tag}>{$line}";
1284 array_push( $tr_history, false );
1285 array_push( $td_history, false );
1286 array_push( $last_tag_history, '' );
1287 } elseif ( $first_character === '|'
1288 ||
$first_character === '!'
1289 ||
$first_two === '|+'
1291 # This might be cell elements, td, th or captions
1292 if ( $first_two === '|+' ) {
1293 $first_character = '+';
1294 $line = substr( $line, 2 );
1296 $line = substr( $line, 1 );
1299 // Implies both are valid for table headings.
1300 if ( $first_character === '!' ) {
1301 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1304 # Split up multiple cells on the same line.
1305 # FIXME : This can result in improper nesting of tags processed
1306 # by earlier parser steps.
1307 $cells = explode( '||', $line );
1311 # Loop through each table cell
1312 foreach ( $cells as $cell ) {
1314 if ( $first_character !== '+' ) {
1315 $tr_after = array_pop( $tr_attributes );
1316 if ( !array_pop( $tr_history ) ) {
1317 $previous = "<tr{$tr_after}>\n";
1319 array_push( $tr_history, true );
1320 array_push( $tr_attributes, '' );
1321 array_pop( $has_opened_tr );
1322 array_push( $has_opened_tr, true );
1325 $last_tag = array_pop( $last_tag_history );
1327 if ( array_pop( $td_history ) ) {
1328 $previous = "</{$last_tag}>\n{$previous}";
1331 if ( $first_character === '|' ) {
1333 } elseif ( $first_character === '!' ) {
1335 } elseif ( $first_character === '+' ) {
1336 $last_tag = 'caption';
1341 array_push( $last_tag_history, $last_tag );
1343 # A cell could contain both parameters and data
1344 $cell_data = explode( '|', $cell, 2 );
1346 # T2553: Note that a '|' inside an invalid link should not
1347 # be mistaken as delimiting cell parameters
1348 # Bug T153140: Neither should language converter markup.
1349 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1350 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1351 } elseif ( count( $cell_data ) == 1 ) {
1352 // Whitespace in cells is trimmed
1353 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1355 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1356 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1357 // Whitespace in cells is trimmed
1358 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1362 array_push( $td_history, true );
1365 $out .= $outLine . "\n";
1368 # Closing open td, tr && table
1369 while ( count( $td_history ) > 0 ) {
1370 if ( array_pop( $td_history ) ) {
1373 if ( array_pop( $tr_history ) ) {
1376 if ( !array_pop( $has_opened_tr ) ) {
1377 $out .= "<tr><td></td></tr>\n";
1380 $out .= "</table>\n";
1383 # Remove trailing line-ending (b/c)
1384 if ( substr( $out, -1 ) === "\n" ) {
1385 $out = substr( $out, 0, -1 );
1388 # special case: don't return empty table
1389 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1397 * Helper function for parse() that transforms wiki markup into half-parsed
1398 * HTML. Only called for $mOutputType == self::OT_HTML.
1402 * @param string $text The text to parse
1403 * @param-taint $text escapes_html
1404 * @param bool $isMain Whether this is being called from the main parse() function
1405 * @param PPFrame|bool $frame A pre-processor frame
1409 public function internalParse( $text, $isMain = true, $frame = false ) {
1412 // Avoid PHP 7.1 warning from passing $this by reference
1415 # Hook to suspend the parser in this state
1416 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1420 # if $frame is provided, then use $frame for replacing any variables
1422 # use frame depth to infer how include/noinclude tags should be handled
1423 # depth=0 means this is the top-level document; otherwise it's an included document
1424 if ( !$frame->depth
) {
1427 $flag = self
::PTD_FOR_INCLUSION
;
1429 $dom = $this->preprocessToDom( $text, $flag );
1430 $text = $frame->expand( $dom );
1432 # if $frame is not provided, then use old-style replaceVariables
1433 $text = $this->replaceVariables( $text );
1436 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1437 $text = Sanitizer
::removeHTMLtags(
1439 [ $this, 'attributeStripCallback' ],
1441 array_keys( $this->mTransparentTagHooks
),
1443 [ $this, 'addTrackingCategory' ]
1445 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1447 # Tables need to come after variable replacement for things to work
1448 # properly; putting them before other transformations should keep
1449 # exciting things like link expansions from showing up in surprising
1451 $text = $this->doTableStuff( $text );
1453 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1455 $text = $this->doDoubleUnderscore( $text );
1457 $text = $this->doHeadings( $text );
1458 $text = $this->replaceInternalLinks( $text );
1459 $text = $this->doAllQuotes( $text );
1460 $text = $this->replaceExternalLinks( $text );
1462 # replaceInternalLinks may sometimes leave behind
1463 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1464 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1466 $text = $this->doMagicLinks( $text );
1467 $text = $this->formatHeadings( $text, $origText, $isMain );
1473 * Helper function for parse() that transforms half-parsed HTML into fully
1476 * @param string $text
1477 * @param bool $isMain
1478 * @param bool $linestart
1481 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1482 $text = $this->mStripState
->unstripGeneral( $text );
1484 // Avoid PHP 7.1 warning from passing $this by reference
1488 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1491 # Clean up special characters, only run once, next-to-last before doBlockLevels
1492 $text = Sanitizer
::armorFrenchSpaces( $text );
1494 $text = $this->doBlockLevels( $text, $linestart );
1496 $this->replaceLinkHolders( $text );
1499 * The input doesn't get language converted if
1501 * b) Content isn't converted
1502 * c) It's a conversion table
1503 * d) it is an interface message (which is in the user language)
1505 if ( !( $this->mOptions
->getDisableContentConversion()
1506 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1507 && !$this->mOptions
->getInterfaceMessage()
1509 # The position of the convert() call should not be changed. it
1510 # assumes that the links are all replaced and the only thing left
1511 # is the <nowiki> mark.
1512 $text = $this->getTargetLanguage()->convert( $text );
1515 $text = $this->mStripState
->unstripNoWiki( $text );
1518 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1521 $text = $this->replaceTransparentTags( $text );
1522 $text = $this->mStripState
->unstripGeneral( $text );
1524 $text = Sanitizer
::normalizeCharReferences( $text );
1526 if ( MWTidy
::isEnabled() ) {
1527 if ( $this->mOptions
->getTidy() ) {
1528 $text = MWTidy
::tidy( $text );
1531 # attempt to sanitize at least some nesting problems
1532 # (T4702 and quite a few others)
1533 # This code path is buggy and deprecated!
1534 wfDeprecated( 'disabling tidy', '1.33' );
1536 # ''Something [http://www.cool.com cool''] -->
1537 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1538 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1539 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1540 # fix up an anchor inside another anchor, only
1541 # at least for a single single nested link (T5695)
1542 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1543 '\\1\\2</a>\\3</a>\\1\\4</a>',
1544 # fix div inside inline elements- doBlockLevels won't wrap a line which
1545 # contains a div, so fix it up here; replace
1546 # div with escaped text
1547 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1548 '\\1\\3<div\\5>\\6</div>\\8\\9',
1549 # remove empty italic or bold tag pairs, some
1550 # introduced by rules above
1551 '/<([bi])><\/\\1>/' => '',
1554 $text = preg_replace(
1555 array_keys( $tidyregs ),
1556 array_values( $tidyregs ),
1561 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1568 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1569 * magic external links.
1574 * @param string $text
1578 public function doMagicLinks( $text ) {
1579 $prots = wfUrlProtocolsWithoutProtRel();
1580 $urlChar = self
::EXT_LINK_URL_CLASS
;
1581 $addr = self
::EXT_LINK_ADDR
;
1582 $space = self
::SPACE_NOT_NL
; # non-newline space
1583 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1584 $spaces = "$space++"; # possessive match of 1 or more spaces
1585 $text = preg_replace_callback(
1587 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1588 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1589 (\b # m[3]: Free external links
1591 ($addr$urlChar*) # m[4]: Post-protocol path
1593 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1595 \bISBN $spaces ( # m[6]: ISBN, capture number
1596 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1597 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1598 [0-9Xx] # check digit
1600 )!xu", [ $this, 'magicLinkCallback' ], $text );
1605 * @throws MWException
1607 * @return string HTML
1609 public function magicLinkCallback( $m ) {
1610 if ( isset( $m[1] ) && $m[1] !== '' ) {
1613 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1616 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1617 # Free external link
1618 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1619 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1621 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1622 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1627 $cssClass = 'mw-magiclink-rfc';
1628 $trackingCat = 'magiclink-tracking-rfc';
1630 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1631 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1635 $urlmsg = 'pubmedurl';
1636 $cssClass = 'mw-magiclink-pmid';
1637 $trackingCat = 'magiclink-tracking-pmid';
1640 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1641 substr( $m[0], 0, 20 ) . '"' );
1643 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1644 $this->addTrackingCategory( $trackingCat );
1645 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1646 } elseif ( isset( $m[6] ) && $m[6] !== ''
1647 && $this->mOptions
->getMagicISBNLinks()
1651 $space = self
::SPACE_NOT_NL
; # non-newline space
1652 $isbn = preg_replace( "/$space/", ' ', $isbn );
1653 $num = strtr( $isbn, [
1658 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1659 return $this->getLinkRenderer()->makeKnownLink(
1660 SpecialPage
::getTitleFor( 'Booksources', $num ),
1663 'class' => 'internal mw-magiclink-isbn',
1664 'title' => false // suppress title attribute
1673 * Make a free external link, given a user-supplied URL
1675 * @param string $url
1676 * @param int $numPostProto
1677 * The number of characters after the protocol.
1678 * @return string HTML
1681 public function makeFreeExternalLink( $url, $numPostProto ) {
1684 # The characters '<' and '>' (which were escaped by
1685 # removeHTMLtags()) should not be included in
1686 # URLs, per RFC 2396.
1687 # Make terminate a URL as well (bug T84937)
1690 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1695 $trail = substr( $url, $m2[0][1] ) . $trail;
1696 $url = substr( $url, 0, $m2[0][1] );
1699 # Move trailing punctuation to $trail
1701 # If there is no left bracket, then consider right brackets fair game too
1702 if ( strpos( $url, '(' ) === false ) {
1706 $urlRev = strrev( $url );
1707 $numSepChars = strspn( $urlRev, $sep );
1708 # Don't break a trailing HTML entity by moving the ; into $trail
1709 # This is in hot code, so use substr_compare to avoid having to
1710 # create a new string object for the comparison
1711 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1712 # more optimization: instead of running preg_match with a $
1713 # anchor, which can be slow, do the match on the reversed
1714 # string starting at the desired offset.
1715 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1716 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1720 if ( $numSepChars ) {
1721 $trail = substr( $url, -$numSepChars ) . $trail;
1722 $url = substr( $url, 0, -$numSepChars );
1725 # Verify that we still have a real URL after trail removal, and
1726 # not just lone protocol
1727 if ( strlen( $trail ) >= $numPostProto ) {
1728 return $url . $trail;
1731 $url = Sanitizer
::cleanUrl( $url );
1733 # Is this an external image?
1734 $text = $this->maybeMakeExternalImage( $url );
1735 if ( $text === false ) {
1736 # Not an image, make a link
1737 $text = Linker
::makeExternalLink( $url,
1738 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1740 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1741 # Register it in the output object...
1742 $this->mOutput
->addExternalLink( $url );
1744 return $text . $trail;
1748 * Parse headers and return html
1752 * @param string $text
1756 public function doHeadings( $text ) {
1757 for ( $i = 6; $i >= 1; --$i ) {
1758 $h = str_repeat( '=', $i );
1759 // Trim non-newline whitespace from headings
1760 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1761 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1767 * Replace single quotes with HTML markup
1770 * @param string $text
1772 * @return string The altered text
1774 public function doAllQuotes( $text ) {
1776 $lines = StringUtils
::explode( "\n", $text );
1777 foreach ( $lines as $line ) {
1778 $outtext .= $this->doQuotes( $line ) . "\n";
1780 $outtext = substr( $outtext, 0, -1 );
1785 * Helper function for doAllQuotes()
1787 * @param string $text
1791 public function doQuotes( $text ) {
1792 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1793 $countarr = count( $arr );
1794 if ( $countarr == 1 ) {
1798 // First, do some preliminary work. This may shift some apostrophes from
1799 // being mark-up to being text. It also counts the number of occurrences
1800 // of bold and italics mark-ups.
1803 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1804 $thislen = strlen( $arr[$i] );
1805 // If there are ever four apostrophes, assume the first is supposed to
1806 // be text, and the remaining three constitute mark-up for bold text.
1807 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1808 if ( $thislen == 4 ) {
1809 $arr[$i - 1] .= "'";
1812 } elseif ( $thislen > 5 ) {
1813 // If there are more than 5 apostrophes in a row, assume they're all
1814 // text except for the last 5.
1815 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1816 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1820 // Count the number of occurrences of bold and italics mark-ups.
1821 if ( $thislen == 2 ) {
1823 } elseif ( $thislen == 3 ) {
1825 } elseif ( $thislen == 5 ) {
1831 // If there is an odd number of both bold and italics, it is likely
1832 // that one of the bold ones was meant to be an apostrophe followed
1833 // by italics. Which one we cannot know for certain, but it is more
1834 // likely to be one that has a single-letter word before it.
1835 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1836 $firstsingleletterword = -1;
1837 $firstmultiletterword = -1;
1839 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1840 if ( strlen( $arr[$i] ) == 3 ) {
1841 $x1 = substr( $arr[$i - 1], -1 );
1842 $x2 = substr( $arr[$i - 1], -2, 1 );
1843 if ( $x1 === ' ' ) {
1844 if ( $firstspace == -1 ) {
1847 } elseif ( $x2 === ' ' ) {
1848 $firstsingleletterword = $i;
1849 // if $firstsingleletterword is set, we don't
1850 // look at the other options, so we can bail early.
1852 } elseif ( $firstmultiletterword == -1 ) {
1853 $firstmultiletterword = $i;
1858 // If there is a single-letter word, use it!
1859 if ( $firstsingleletterword > -1 ) {
1860 $arr[$firstsingleletterword] = "''";
1861 $arr[$firstsingleletterword - 1] .= "'";
1862 } elseif ( $firstmultiletterword > -1 ) {
1863 // If not, but there's a multi-letter word, use that one.
1864 $arr[$firstmultiletterword] = "''";
1865 $arr[$firstmultiletterword - 1] .= "'";
1866 } elseif ( $firstspace > -1 ) {
1867 // ... otherwise use the first one that has neither.
1868 // (notice that it is possible for all three to be -1 if, for example,
1869 // there is only one pentuple-apostrophe in the line)
1870 $arr[$firstspace] = "''";
1871 $arr[$firstspace - 1] .= "'";
1875 // Now let's actually convert our apostrophic mush to HTML!
1880 foreach ( $arr as $r ) {
1881 if ( ( $i %
2 ) == 0 ) {
1882 if ( $state === 'both' ) {
1888 $thislen = strlen( $r );
1889 if ( $thislen == 2 ) {
1890 if ( $state === 'i' ) {
1893 } elseif ( $state === 'bi' ) {
1896 } elseif ( $state === 'ib' ) {
1897 $output .= '</b></i><b>';
1899 } elseif ( $state === 'both' ) {
1900 $output .= '<b><i>' . $buffer . '</i>';
1902 } else { // $state can be 'b' or ''
1906 } elseif ( $thislen == 3 ) {
1907 if ( $state === 'b' ) {
1910 } elseif ( $state === 'bi' ) {
1911 $output .= '</i></b><i>';
1913 } elseif ( $state === 'ib' ) {
1916 } elseif ( $state === 'both' ) {
1917 $output .= '<i><b>' . $buffer . '</b>';
1919 } else { // $state can be 'i' or ''
1923 } elseif ( $thislen == 5 ) {
1924 if ( $state === 'b' ) {
1925 $output .= '</b><i>';
1927 } elseif ( $state === 'i' ) {
1928 $output .= '</i><b>';
1930 } elseif ( $state === 'bi' ) {
1931 $output .= '</i></b>';
1933 } elseif ( $state === 'ib' ) {
1934 $output .= '</b></i>';
1936 } elseif ( $state === 'both' ) {
1937 $output .= '<i><b>' . $buffer . '</b></i>';
1939 } else { // ($state == '')
1947 // Now close all remaining tags. Notice that the order is important.
1948 if ( $state === 'b' ||
$state === 'ib' ) {
1951 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1954 if ( $state === 'bi' ) {
1957 // There might be lonely ''''', so make sure we have a buffer
1958 if ( $state === 'both' && $buffer ) {
1959 $output .= '<b><i>' . $buffer . '</i></b>';
1965 * Replace external links (REL)
1967 * Note: this is all very hackish and the order of execution matters a lot.
1968 * Make sure to run tests/parser/parserTests.php if you change this code.
1972 * @param string $text
1974 * @throws MWException
1977 public function replaceExternalLinks( $text ) {
1978 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1979 if ( $bits === false ) {
1980 throw new MWException( "PCRE needs to be compiled with "
1981 . "--enable-unicode-properties in order for MediaWiki to function" );
1983 $s = array_shift( $bits );
1986 while ( $i < count( $bits ) ) {
1989 $text = $bits[$i++
];
1990 $trail = $bits[$i++
];
1992 # The characters '<' and '>' (which were escaped by
1993 # removeHTMLtags()) should not be included in
1994 # URLs, per RFC 2396.
1996 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1997 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1998 $url = substr( $url, 0, $m2[0][1] );
2001 # If the link text is an image URL, replace it with an <img> tag
2002 # This happened by accident in the original parser, but some people used it extensively
2003 $img = $this->maybeMakeExternalImage( $text );
2004 if ( $img !== false ) {
2010 # Set linktype for CSS
2013 # No link text, e.g. [http://domain.tld/some.link]
2014 if ( $text == '' ) {
2016 $langObj = $this->getTargetLanguage();
2017 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
2018 $linktype = 'autonumber';
2020 # Have link text, e.g. [http://domain.tld/some.link text]s
2022 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
2025 // Excluding protocol-relative URLs may avoid many false positives.
2026 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2027 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2030 $url = Sanitizer
::cleanUrl( $url );
2032 # Use the encoded URL
2033 # This means that users can paste URLs directly into the text
2034 # Funny characters like ö aren't valid in URLs anyway
2035 # This was changed in August 2004
2036 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
2037 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
2039 # Register link in the output object.
2040 $this->mOutput
->addExternalLink( $url );
2047 * Get the rel attribute for a particular external link.
2050 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2051 * nofollow if appropriate
2052 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2053 * @return string|null Rel attribute for $url
2055 public static function getExternalLinkRel( $url = false, $title = null ) {
2056 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2057 $ns = $title ?
$title->getNamespace() : false;
2058 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2059 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2067 * Get an associative array of additional HTML attributes appropriate for a
2068 * particular external link. This currently may include rel => nofollow
2069 * (depending on configuration, namespace, and the URL's domain) and/or a
2070 * target attribute (depending on configuration).
2072 * @param string $url URL to extract the domain from for rel =>
2073 * nofollow if appropriate
2074 * @return array Associative array of HTML attributes
2076 public function getExternalLinkAttribs( $url ) {
2078 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2080 $target = $this->mOptions
->getExternalLinkTarget();
2082 $attribs['target'] = $target;
2083 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2084 // T133507. New windows can navigate parent cross-origin.
2085 // Including noreferrer due to lacking browser
2086 // support of noopener. Eventually noreferrer should be removed.
2087 if ( $rel !== '' ) {
2090 $rel .= 'noreferrer noopener';
2093 $attribs['rel'] = $rel;
2098 * Replace unusual escape codes in a URL with their equivalent characters
2100 * This generally follows the syntax defined in RFC 3986, with special
2101 * consideration for HTTP query strings.
2103 * @param string $url
2106 public static function normalizeLinkUrl( $url ) {
2107 # Test for RFC 3986 IPv6 syntax
2108 $scheme = '[a-z][a-z0-9+.-]*:';
2109 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2110 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2111 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2112 IP
::isValid( rawurldecode( $m[1] ) )
2114 $isIPv6 = rawurldecode( $m[1] );
2119 # Make sure unsafe characters are encoded
2120 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2122 return rawurlencode( $m[0] );
2128 $end = strlen( $url );
2130 # Fragment part - 'fragment'
2131 $start = strpos( $url, '#' );
2132 if ( $start !== false && $start < $end ) {
2133 $ret = self
::normalizeUrlComponent(
2134 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2138 # Query part - 'query' minus &=+;
2139 $start = strpos( $url, '?' );
2140 if ( $start !== false && $start < $end ) {
2141 $ret = self
::normalizeUrlComponent(
2142 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2146 # Scheme and path part - 'pchar'
2147 # (we assume no userinfo or encoded colons in the host)
2148 $ret = self
::normalizeUrlComponent(
2149 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2152 if ( $isIPv6 !== false ) {
2153 $ipv6Host = "%5B({$isIPv6})%5D";
2154 $ret = preg_replace(
2155 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2164 private static function normalizeUrlComponent( $component, $unsafe ) {
2165 $callback = function ( $matches ) use ( $unsafe ) {
2166 $char = urldecode( $matches[0] );
2167 $ord = ord( $char );
2168 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2172 # Leave it escaped, but use uppercase for a-f
2173 return strtoupper( $matches[0] );
2176 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2180 * make an image if it's allowed, either through the global
2181 * option, through the exception, or through the on-wiki whitelist
2183 * @param string $url
2187 private function maybeMakeExternalImage( $url ) {
2188 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2189 $imagesexception = !empty( $imagesfrom );
2191 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2192 if ( $imagesexception && is_array( $imagesfrom ) ) {
2193 $imagematch = false;
2194 foreach ( $imagesfrom as $match ) {
2195 if ( strpos( $url, $match ) === 0 ) {
2200 } elseif ( $imagesexception ) {
2201 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2203 $imagematch = false;
2206 if ( $this->mOptions
->getAllowExternalImages()
2207 ||
( $imagesexception && $imagematch )
2209 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2211 $text = Linker
::makeExternalImage( $url );
2214 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2215 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2217 $whitelist = explode(
2219 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2222 foreach ( $whitelist as $entry ) {
2223 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2224 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2227 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2228 # Image matches a whitelist entry
2229 $text = Linker
::makeExternalImage( $url );
2238 * Process [[ ]] wikilinks
2242 * @return string Processed text
2246 public function replaceInternalLinks( $s ) {
2247 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2252 * Process [[ ]] wikilinks (RIL)
2254 * @throws MWException
2255 * @return LinkHolderArray
2259 public function replaceInternalLinks2( &$s ) {
2260 static $tc = false, $e1, $e1_img;
2261 # the % is needed to support urlencoded titles as well
2263 $tc = Title
::legalChars() . '#%';
2264 # Match a link having the form [[namespace:link|alternate]]trail
2265 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2266 # Match cases where there is no "]]", which might still be images
2267 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2270 $holders = new LinkHolderArray( $this );
2272 # split the entire text string on occurrences of [[
2273 $a = StringUtils
::explode( '[[', ' ' . $s );
2274 # get the first element (all text up to first [[), and remove the space we added
2277 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2278 $s = substr( $s, 1 );
2280 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2282 if ( $useLinkPrefixExtension ) {
2283 # Match the end of a line for a word that's not followed by whitespace,
2284 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2285 $charset = $this->contLang
->linkPrefixCharset();
2286 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2289 if ( is_null( $this->mTitle
) ) {
2290 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2292 $nottalk = !$this->mTitle
->isTalkPage();
2294 if ( $useLinkPrefixExtension ) {
2296 if ( preg_match( $e2, $s, $m ) ) {
2297 $first_prefix = $m[2];
2299 $first_prefix = false;
2305 $useSubpages = $this->areSubpagesAllowed();
2307 # Loop for each link
2308 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2309 # Check for excessive memory usage
2310 if ( $holders->isBig() ) {
2312 # Do the existence check, replace the link holders and clear the array
2313 $holders->replace( $s );
2317 if ( $useLinkPrefixExtension ) {
2318 if ( preg_match( $e2, $s, $m ) ) {
2319 list( , $s, $prefix ) = $m;
2324 if ( $first_prefix ) {
2325 $prefix = $first_prefix;
2326 $first_prefix = false;
2330 $might_be_img = false;
2332 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2334 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2335 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2336 # the real problem is with the $e1 regex
2338 # Still some problems for cases where the ] is meant to be outside punctuation,
2339 # and no image is in sight. See T4095.
2341 && substr( $m[3], 0, 1 ) === ']'
2342 && strpos( $text, '[' ) !== false
2344 $text .= ']'; # so that replaceExternalLinks($text) works later
2345 $m[3] = substr( $m[3], 1 );
2347 # fix up urlencoded title texts
2348 if ( strpos( $m[1], '%' ) !== false ) {
2349 # Should anchors '#' also be rejected?
2350 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2353 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2354 # Invalid, but might be an image with a link in its caption
2355 $might_be_img = true;
2357 if ( strpos( $m[1], '%' ) !== false ) {
2358 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2361 } else { # Invalid form; output directly
2362 $s .= $prefix . '[[' . $line;
2366 $origLink = ltrim( $m[1], ' ' );
2368 # Don't allow internal links to pages containing
2369 # PROTO: where PROTO is a valid URL protocol; these
2370 # should be external links.
2371 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2372 $s .= $prefix . '[[' . $line;
2376 # Make subpage if necessary
2377 if ( $useSubpages ) {
2378 $link = $this->maybeDoSubpageLink( $origLink, $text );
2383 // \x7f isn't a default legal title char, so most likely strip
2384 // markers will force us into the "invalid form" path above. But,
2385 // just in case, let's assert that xmlish tags aren't valid in
2386 // the title position.
2387 $unstrip = $this->mStripState
->killMarkers( $link );
2388 $noMarkers = ( $unstrip === $link );
2390 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2391 if ( $nt === null ) {
2392 $s .= $prefix . '[[' . $line;
2396 $ns = $nt->getNamespace();
2397 $iw = $nt->getInterwiki();
2399 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2401 if ( $might_be_img ) { # if this is actually an invalid link
2402 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2405 # look at the next 'line' to see if we can close it there
2407 $next_line = $a->current();
2408 if ( $next_line === false ||
$next_line === null ) {
2411 $m = explode( ']]', $next_line, 3 );
2412 if ( count( $m ) == 3 ) {
2413 # the first ]] closes the inner link, the second the image
2415 $text .= "[[{$m[0]}]]{$m[1]}";
2418 } elseif ( count( $m ) == 2 ) {
2419 # if there's exactly one ]] that's fine, we'll keep looking
2420 $text .= "[[{$m[0]}]]{$m[1]}";
2422 # if $next_line is invalid too, we need look no further
2423 $text .= '[[' . $next_line;
2428 # we couldn't find the end of this imageLink, so output it raw
2429 # but don't ignore what might be perfectly normal links in the text we've examined
2430 $holders->merge( $this->replaceInternalLinks2( $text ) );
2431 $s .= "{$prefix}[[$link|$text";
2432 # note: no $trail, because without an end, there *is* no trail
2435 } else { # it's not an image, so output it raw
2436 $s .= "{$prefix}[[$link|$text";
2437 # note: no $trail, because without an end, there *is* no trail
2442 $wasblank = ( $text == '' );
2446 # Strip off leading ':'
2447 $text = substr( $text, 1 );
2450 # T6598 madness. Handle the quotes only if they come from the alternate part
2451 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2452 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2453 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2454 $text = $this->doQuotes( $text );
2457 # Link not escaped by : , create the various objects
2458 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2461 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2462 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2463 in_array( $iw, $this->svcOptions
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2466 # T26502: filter duplicates
2467 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2468 $this->mLangLinkLanguages
[$iw] = true;
2469 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2473 * Strip the whitespace interwiki links produce, see T10897
2475 $s = rtrim( $s . $prefix ) . $trail; # T175416
2479 if ( $ns == NS_FILE
) {
2480 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2482 # if no parameters were passed, $text
2483 # becomes something like "File:Foo.png",
2484 # which we don't want to pass on to the
2488 # recursively parse links inside the image caption
2489 # actually, this will parse them in any other parameters, too,
2490 # but it might be hard to fix that, and it doesn't matter ATM
2491 $text = $this->replaceExternalLinks( $text );
2492 $holders->merge( $this->replaceInternalLinks2( $text ) );
2494 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2495 $s .= $prefix . $this->armorLinks(
2496 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2499 } elseif ( $ns == NS_CATEGORY
) {
2501 * Strip the whitespace Category links produce, see T2087
2503 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2506 $sortkey = $this->getDefaultSort();
2510 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2511 $sortkey = str_replace( "\n", '', $sortkey );
2512 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2513 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2519 # Self-link checking. For some languages, variants of the title are checked in
2520 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2521 # for linking to a different variant.
2522 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2523 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2527 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2528 # @todo FIXME: Should do batch file existence checks, see comment below
2529 if ( $ns == NS_MEDIA
) {
2530 # Give extensions a chance to select the file revision for us
2533 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2534 [ $this, $nt, &$options, &$descQuery ] );
2535 # Fetch and register the file (file title may be different via hooks)
2536 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2537 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2538 $s .= $prefix . $this->armorLinks(
2539 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2543 # Some titles, such as valid special pages or files in foreign repos, should
2544 # be shown as bluelinks even though they're not included in the page table
2545 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2546 # batch file existence checks for NS_FILE and NS_MEDIA
2547 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2548 $this->mOutput
->addLink( $nt );
2549 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2551 # Links will be added to the output link list after checking
2552 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2559 * Render a forced-blue link inline; protect against double expansion of
2560 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2561 * Since this little disaster has to split off the trail text to avoid
2562 * breaking URLs in the following text without breaking trails on the
2563 * wiki links, it's been made into a horrible function.
2566 * @param string $text
2567 * @param string $trail
2568 * @param string $prefix
2569 * @return string HTML-wikitext mix oh yuck
2571 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2572 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2574 if ( $text == '' ) {
2575 $text = htmlspecialchars( $nt->getPrefixedText() );
2578 $link = $this->getLinkRenderer()->makeKnownLink(
2579 $nt, new HtmlArmor( "$prefix$text$inside" )
2582 return $this->armorLinks( $link ) . $trail;
2586 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2587 * going to go through further parsing steps before inline URL expansion.
2589 * Not needed quite as much as it used to be since free links are a bit
2590 * more sensible these days. But bracketed links are still an issue.
2592 * @param string $text More-or-less HTML
2593 * @return string Less-or-more HTML with NOPARSE bits
2595 public function armorLinks( $text ) {
2596 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2597 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2601 * Return true if subpage links should be expanded on this page.
2604 public function areSubpagesAllowed() {
2605 # Some namespaces don't allow subpages
2606 return $this->nsInfo
->hasSubpages( $this->mTitle
->getNamespace() );
2610 * Handle link to subpage if necessary
2612 * @param string $target The source of the link
2613 * @param string &$text The link text, modified as necessary
2614 * @return string The full name of the link
2617 public function maybeDoSubpageLink( $target, &$text ) {
2618 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2622 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2624 * @param string $text
2625 * @param bool $linestart Whether or not this is at the start of a line.
2627 * @return string The lists rendered as HTML
2629 public function doBlockLevels( $text, $linestart ) {
2630 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2634 * Return value of a magic variable (like PAGENAME)
2638 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2639 * @param bool|PPFrame $frame
2641 * @throws MWException
2644 public function getVariableValue( $index, $frame = false ) {
2645 if ( is_null( $this->mTitle
) ) {
2646 // If no title set, bad things are going to happen
2647 // later. Title should always be set since this
2648 // should only be called in the middle of a parse
2649 // operation (but the unit-tests do funky stuff)
2650 throw new MWException( __METHOD__
. ' Should only be '
2651 . ' called while parsing (no title set)' );
2654 // Avoid PHP 7.1 warning from passing $this by reference
2658 * Some of these require message or data lookups and can be
2659 * expensive to check many times.
2662 Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) &&
2663 isset( $this->mVarCache
[$index] )
2665 return $this->mVarCache
[$index];
2668 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2669 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2671 $pageLang = $this->getFunctionLang();
2677 case 'currentmonth':
2678 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2680 case 'currentmonth1':
2681 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2683 case 'currentmonthname':
2684 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2686 case 'currentmonthnamegen':
2687 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2689 case 'currentmonthabbrev':
2690 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2693 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2696 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2699 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2702 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2704 case 'localmonthname':
2705 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2707 case 'localmonthnamegen':
2708 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2710 case 'localmonthabbrev':
2711 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2714 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2717 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2720 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2723 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2725 case 'fullpagename':
2726 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2728 case 'fullpagenamee':
2729 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2732 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2734 case 'subpagenamee':
2735 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2737 case 'rootpagename':
2738 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2740 case 'rootpagenamee':
2741 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2744 $this->mTitle
->getRootText()
2747 case 'basepagename':
2748 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2750 case 'basepagenamee':
2751 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2754 $this->mTitle
->getBaseText()
2757 case 'talkpagename':
2758 if ( $this->mTitle
->canHaveTalkPage() ) {
2759 $talkPage = $this->mTitle
->getTalkPage();
2760 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2765 case 'talkpagenamee':
2766 if ( $this->mTitle
->canHaveTalkPage() ) {
2767 $talkPage = $this->mTitle
->getTalkPage();
2768 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2773 case 'subjectpagename':
2774 $subjPage = $this->mTitle
->getSubjectPage();
2775 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2777 case 'subjectpagenamee':
2778 $subjPage = $this->mTitle
->getSubjectPage();
2779 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2781 case 'pageid': // requested in T25427
2782 $pageid = $this->getTitle()->getArticleID();
2783 if ( $pageid == 0 ) {
2784 # 0 means the page doesn't exist in the database,
2785 # which means the user is previewing a new page.
2786 # The vary-revision flag must be set, because the magic word
2787 # will have a different value once the page is saved.
2788 $this->mOutput
->setFlag( 'vary-revision' );
2789 wfDebug( __METHOD__
. ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2791 $value = $pageid ?
: null;
2795 $this->svcOptions
->get( 'MiserMode' ) &&
2796 !$this->mOptions
->getInterfaceMessage() &&
2797 // @TODO: disallow this word on all namespaces
2798 $this->nsInfo
->isContent( $this->mTitle
->getNamespace() )
2800 // Use a stub result instead of the actual revision ID in order to avoid
2801 // double parses on page save but still allow preview detection (T137900)
2802 if ( $this->getRevisionId() ||
$this->mOptions
->getSpeculativeRevId() ) {
2805 $this->mOutput
->setFlag( 'vary-revision-exists' );
2809 # Inform the edit saving system that getting the canonical output after
2810 # revision insertion requires another parse using the actual revision ID
2811 $this->mOutput
->setFlag( 'vary-revision-id' );
2812 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2813 $value = $this->getRevisionId();
2814 if ( $value === 0 ) {
2815 $rev = $this->getRevisionObject();
2816 $value = $rev ?
$rev->getId() : $value;
2819 $value = $this->mOptions
->getSpeculativeRevId();
2821 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2827 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2829 case 'revisionday2':
2830 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2832 case 'revisionmonth':
2833 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2835 case 'revisionmonth1':
2836 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2838 case 'revisionyear':
2839 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2841 case 'revisiontimestamp':
2842 # Let the edit saving system know we should parse the page
2843 # *after* a revision ID has been assigned. This is for null edits.
2844 $this->mOutput
->setFlag( 'vary-revision' );
2845 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2846 $value = $this->getRevisionTimestamp();
2848 case 'revisionuser':
2849 # Let the edit saving system know we should parse the page
2850 # *after* a revision ID has been assigned for null edits.
2851 $this->mOutput
->setFlag( 'vary-user' );
2852 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-user...\n" );
2853 $value = $this->getRevisionUser();
2855 case 'revisionsize':
2856 $value = $this->getRevisionSize();
2859 $value = str_replace( '_', ' ',
2860 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2863 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2865 case 'namespacenumber':
2866 $value = $this->mTitle
->getNamespace();
2869 $value = $this->mTitle
->canHaveTalkPage()
2870 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2874 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2876 case 'subjectspace':
2877 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2879 case 'subjectspacee':
2880 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2882 case 'currentdayname':
2883 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2886 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2889 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2892 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2895 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2896 # int to remove the padding
2897 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2900 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2902 case 'localdayname':
2903 $value = $pageLang->getWeekdayName(
2904 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2908 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2911 $value = $pageLang->time(
2912 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2918 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2921 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2922 # int to remove the padding
2923 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2926 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2928 case 'numberofarticles':
2929 $value = $pageLang->formatNum( SiteStats
::articles() );
2931 case 'numberoffiles':
2932 $value = $pageLang->formatNum( SiteStats
::images() );
2934 case 'numberofusers':
2935 $value = $pageLang->formatNum( SiteStats
::users() );
2937 case 'numberofactiveusers':
2938 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2940 case 'numberofpages':
2941 $value = $pageLang->formatNum( SiteStats
::pages() );
2943 case 'numberofadmins':
2944 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2946 case 'numberofedits':
2947 $value = $pageLang->formatNum( SiteStats
::edits() );
2949 case 'currenttimestamp':
2950 $value = wfTimestamp( TS_MW
, $ts );
2952 case 'localtimestamp':
2953 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2955 case 'currentversion':
2956 $value = SpecialVersion
::getVersion();
2959 return $this->svcOptions
->get( 'ArticlePath' );
2961 return $this->svcOptions
->get( 'Sitename' );
2963 return $this->svcOptions
->get( 'Server' );
2965 return $this->svcOptions
->get( 'ServerName' );
2967 return $this->svcOptions
->get( 'ScriptPath' );
2969 return $this->svcOptions
->get( 'StylePath' );
2970 case 'directionmark':
2971 return $pageLang->getDirMark();
2972 case 'contentlanguage':
2973 return $this->svcOptions
->get( 'LanguageCode' );
2974 case 'pagelanguage':
2975 $value = $pageLang->getCode();
2977 case 'cascadingsources':
2978 $value = CoreParserFunctions
::cascadingsources( $this );
2983 'ParserGetVariableValueSwitch',
2984 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
2991 $this->mVarCache
[$index] = $value;
3000 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
3001 * @param string $variable Parser variable name
3004 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
3005 # Get the timezone-adjusted timestamp to be used for this revision
3006 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
3007 # Possibly set vary-revision if there is not yet an associated revision
3008 if ( !$this->getRevisionObject() ) {
3009 # Get the timezone-adjusted timestamp $mtts seconds in the future
3011 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
3016 if ( $resNow !== $resThen ) {
3017 # Let the edit saving system know we should parse the page
3018 # *after* a revision ID has been assigned. This is for null edits.
3019 $this->mOutput
->setFlag( 'vary-revision' );
3020 wfDebug( __METHOD__
. ": $variable used, setting vary-revision...\n" );
3028 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3032 public function initialiseVariables() {
3033 $variableIDs = $this->magicWordFactory
->getVariableIDs();
3034 $substIDs = $this->magicWordFactory
->getSubstIDs();
3036 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
3037 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
3041 * Preprocess some wikitext and return the document tree.
3042 * This is the ghost of replace_variables().
3044 * @param string $text The text to parse
3045 * @param int $flags Bitwise combination of:
3046 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3047 * included. Default is to assume a direct page view.
3049 * The generated DOM tree must depend only on the input text and the flags.
3050 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3052 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3053 * change in the DOM tree for a given text, must be passed through the section identifier
3054 * in the section edit link and thus back to extractSections().
3056 * The output of this function is currently only cached in process memory, but a persistent
3057 * cache may be implemented at a later date which takes further advantage of these strict
3058 * dependency requirements.
3062 public function preprocessToDom( $text, $flags = 0 ) {
3063 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3068 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3074 public static function splitWhitespace( $s ) {
3075 $ltrimmed = ltrim( $s );
3076 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3077 $trimmed = rtrim( $ltrimmed );
3078 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3080 $w2 = substr( $ltrimmed, -$diff );
3084 return [ $w1, $trimmed, $w2 ];
3088 * Replace magic variables, templates, and template arguments
3089 * with the appropriate text. Templates are substituted recursively,
3090 * taking care to avoid infinite loops.
3092 * Note that the substitution depends on value of $mOutputType:
3093 * self::OT_WIKI: only {{subst:}} templates
3094 * self::OT_PREPROCESS: templates but not extension tags
3095 * self::OT_HTML: all templates and extension tags
3097 * @param string $text The text to transform
3098 * @param bool|PPFrame $frame Object describing the arguments passed to the
3099 * template. Arguments may also be provided as an associative array, as
3100 * was the usual case before MW1.12. Providing arguments this way may be
3101 * useful for extensions wishing to perform variable replacement
3103 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3104 * double-brace expansion.
3107 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3108 # Is there any text? Also, Prevent too big inclusions!
3109 $textSize = strlen( $text );
3110 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3114 if ( $frame === false ) {
3115 $frame = $this->getPreprocessor()->newFrame();
3116 } elseif ( !( $frame instanceof PPFrame
) ) {
3117 wfDebug( __METHOD__
. " called using plain parameters instead of "
3118 . "a PPFrame instance. Creating custom frame.\n" );
3119 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3122 $dom = $this->preprocessToDom( $text );
3123 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3124 $text = $frame->expand( $dom, $flags );
3130 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3132 * @param array $args
3136 public static function createAssocArgs( $args ) {
3139 foreach ( $args as $arg ) {
3140 $eqpos = strpos( $arg, '=' );
3141 if ( $eqpos === false ) {
3142 $assocArgs[$index++
] = $arg;
3144 $name = trim( substr( $arg, 0, $eqpos ) );
3145 $value = trim( substr( $arg, $eqpos +
1 ) );
3146 if ( $value === false ) {
3149 if ( $name !== false ) {
3150 $assocArgs[$name] = $value;
3159 * Warn the user when a parser limitation is reached
3160 * Will warn at most once the user per limitation type
3162 * The results are shown during preview and run through the Parser (See EditPage.php)
3164 * @param string $limitationType Should be one of:
3165 * 'expensive-parserfunction' (corresponding messages:
3166 * 'expensive-parserfunction-warning',
3167 * 'expensive-parserfunction-category')
3168 * 'post-expand-template-argument' (corresponding messages:
3169 * 'post-expand-template-argument-warning',
3170 * 'post-expand-template-argument-category')
3171 * 'post-expand-template-inclusion' (corresponding messages:
3172 * 'post-expand-template-inclusion-warning',
3173 * 'post-expand-template-inclusion-category')
3174 * 'node-count-exceeded' (corresponding messages:
3175 * 'node-count-exceeded-warning',
3176 * 'node-count-exceeded-category')
3177 * 'expansion-depth-exceeded' (corresponding messages:
3178 * 'expansion-depth-exceeded-warning',
3179 * 'expansion-depth-exceeded-category')
3180 * @param string|int|null $current Current value
3181 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3182 * exceeded, provide the values (optional)
3184 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3185 # does no harm if $current and $max are present but are unnecessary for the message
3186 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3187 # only during preview, and that would split the parser cache unnecessarily.
3188 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3190 $this->mOutput
->addWarning( $warning );
3191 $this->addTrackingCategory( "$limitationType-category" );
3195 * Return the text of a template, after recursively
3196 * replacing any variables or templates within the template.
3198 * @param array $piece The parts of the template
3199 * $piece['title']: the title, i.e. the part before the |
3200 * $piece['parts']: the parameter array
3201 * $piece['lineStart']: whether the brace was at the start of a line
3202 * @param PPFrame $frame The current frame, contains template arguments
3204 * @return string The text of the template
3206 public function braceSubstitution( $piece, $frame ) {
3209 // $text has been filled
3211 // wiki markup in $text should be escaped
3213 // $text is HTML, armour it against wikitext transformation
3215 // Force interwiki transclusion to be done in raw mode not rendered
3216 $forceRawInterwiki = false;
3217 // $text is a DOM node needing expansion in a child frame
3218 $isChildObj = false;
3219 // $text is a DOM node needing expansion in the current frame
3220 $isLocalObj = false;
3222 # Title object, where $text came from
3225 # $part1 is the bit before the first |, and must contain only title characters.
3226 # Various prefixes will be stripped from it later.
3227 $titleWithSpaces = $frame->expand( $piece['title'] );
3228 $part1 = trim( $titleWithSpaces );
3231 # Original title text preserved for various purposes
3232 $originalTitle = $part1;
3234 # $args is a list of argument nodes, starting from index 0, not including $part1
3235 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3236 # below won't work b/c this $args isn't an object
3237 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3239 $profileSection = null; // profile templates
3243 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3245 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3246 # Decide whether to expand template or keep wikitext as-is.
3247 if ( $this->ot
['wiki'] ) {
3248 if ( $substMatch === false ) {
3249 $literal = true; # literal when in PST with no prefix
3251 $literal = false; # expand when in PST with subst: or safesubst:
3254 if ( $substMatch == 'subst' ) {
3255 $literal = true; # literal when not in PST with plain subst:
3257 $literal = false; # expand when not in PST with safesubst: or no prefix
3261 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3268 if ( !$found && $args->getLength() == 0 ) {
3269 $id = $this->mVariables
->matchStartToEnd( $part1 );
3270 if ( $id !== false ) {
3271 $text = $this->getVariableValue( $id, $frame );
3272 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3273 $this->mOutput
->updateCacheExpiry(
3274 $this->magicWordFactory
->getCacheTTL( $id ) );
3280 # MSG, MSGNW and RAW
3283 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3284 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3287 # Remove obsolete MSG:
3288 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3289 $mwMsg->matchStartAndRemove( $part1 );
3293 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3294 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3295 $forceRawInterwiki = true;
3301 $colonPos = strpos( $part1, ':' );
3302 if ( $colonPos !== false ) {
3303 $func = substr( $part1, 0, $colonPos );
3304 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3305 $argsLength = $args->getLength();
3306 for ( $i = 0; $i < $argsLength; $i++
) {
3307 $funcArgs[] = $args->item( $i );
3310 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3312 // Extract any forwarded flags
3313 if ( isset( $result['title'] ) ) {
3314 $title = $result['title'];
3316 if ( isset( $result['found'] ) ) {
3317 $found = $result['found'];
3319 if ( array_key_exists( 'text', $result ) ) {
3321 $text = $result['text'];
3323 if ( isset( $result['nowiki'] ) ) {
3324 $nowiki = $result['nowiki'];
3326 if ( isset( $result['isHTML'] ) ) {
3327 $isHTML = $result['isHTML'];
3329 if ( isset( $result['forceRawInterwiki'] ) ) {
3330 $forceRawInterwiki = $result['forceRawInterwiki'];
3332 if ( isset( $result['isChildObj'] ) ) {
3333 $isChildObj = $result['isChildObj'];
3335 if ( isset( $result['isLocalObj'] ) ) {
3336 $isLocalObj = $result['isLocalObj'];
3341 # Finish mangling title and then check for loops.
3342 # Set $title to a Title object and $titleText to the PDBK
3345 # Split the title into page and subpage
3347 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3348 if ( $part1 !== $relative ) {
3350 $ns = $this->mTitle
->getNamespace();
3352 $title = Title
::newFromText( $part1, $ns );
3354 $titleText = $title->getPrefixedText();
3355 # Check for language variants if the template is not found
3356 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3357 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3359 # Do recursion depth check
3360 $limit = $this->mOptions
->getMaxTemplateDepth();
3361 if ( $frame->depth
>= $limit ) {
3363 $text = '<span class="error">'
3364 . wfMessage( 'parser-template-recursion-depth-warning' )
3365 ->numParams( $limit )->inContentLanguage()->text()
3371 # Load from database
3372 if ( !$found && $title ) {
3373 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3374 if ( !$title->isExternal() ) {
3375 if ( $title->isSpecialPage()
3376 && $this->mOptions
->getAllowSpecialInclusion()
3377 && $this->ot
['html']
3379 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3380 // Pass the template arguments as URL parameters.
3381 // "uselang" will have no effect since the Language object
3382 // is forced to the one defined in ParserOptions.
3384 $argsLength = $args->getLength();
3385 for ( $i = 0; $i < $argsLength; $i++
) {
3386 $bits = $args->item( $i )->splitArg();
3387 if ( strval( $bits['index'] ) === '' ) {
3388 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3389 $value = trim( $frame->expand( $bits['value'] ) );
3390 $pageArgs[$name] = $value;
3394 // Create a new context to execute the special page
3395 $context = new RequestContext
;
3396 $context->setTitle( $title );
3397 $context->setRequest( new FauxRequest( $pageArgs ) );
3398 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3399 $context->setUser( $this->getUser() );
3401 // If this page is cached, then we better not be per user.
3402 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3404 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3405 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3407 $text = $context->getOutput()->getHTML();
3408 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3411 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3412 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3413 $specialPage->maxIncludeCacheTime()
3417 } elseif ( $this->nsInfo
->isNonincludable( $title->getNamespace() ) ) {
3418 $found = false; # access denied
3419 wfDebug( __METHOD__
. ": template inclusion denied for " .
3420 $title->getPrefixedDBkey() . "\n" );
3422 list( $text, $title ) = $this->getTemplateDom( $title );
3423 if ( $text !== false ) {
3429 # If the title is valid but undisplayable, make a link to it
3430 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3431 $text = "[[:$titleText]]";
3434 } elseif ( $title->isTrans() ) {
3435 # Interwiki transclusion
3436 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3437 $text = $this->interwikiTransclude( $title, 'render' );
3440 $text = $this->interwikiTransclude( $title, 'raw' );
3441 # Preprocess it like a template
3442 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3448 # Do infinite loop check
3449 # This has to be done after redirect resolution to avoid infinite loops via redirects
3450 if ( !$frame->loopCheck( $title ) ) {
3452 $text = '<span class="error">'
3453 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3455 $this->addTrackingCategory( 'template-loop-category' );
3456 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3457 wfEscapeWikiText( $titleText ) )->text() );
3458 wfDebug( __METHOD__
. ": template loop broken at '$titleText'\n" );
3462 # If we haven't found text to substitute by now, we're done
3463 # Recover the source wikitext and return it
3465 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3466 if ( $profileSection ) {
3467 $this->mProfiler
->scopedProfileOut( $profileSection );
3469 return [ 'object' => $text ];
3472 # Expand DOM-style return values in a child frame
3473 if ( $isChildObj ) {
3474 # Clean up argument array
3475 $newFrame = $frame->newChild( $args, $title );
3478 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3479 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3480 # Expansion is eligible for the empty-frame cache
3481 $text = $newFrame->cachedExpand( $titleText, $text );
3483 # Uncached expansion
3484 $text = $newFrame->expand( $text );
3487 if ( $isLocalObj && $nowiki ) {
3488 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3489 $isLocalObj = false;
3492 if ( $profileSection ) {
3493 $this->mProfiler
->scopedProfileOut( $profileSection );
3496 # Replace raw HTML by a placeholder
3498 $text = $this->insertStripItem( $text );
3499 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3500 # Escape nowiki-style return values
3501 $text = wfEscapeWikiText( $text );
3502 } elseif ( is_string( $text )
3503 && !$piece['lineStart']
3504 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3506 # T2529: if the template begins with a table or block-level
3507 # element, it should be treated as beginning a new line.
3508 # This behavior is somewhat controversial.
3509 $text = "\n" . $text;
3512 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3513 # Error, oversize inclusion
3514 if ( $titleText !== false ) {
3515 # Make a working, properly escaped link if possible (T25588)
3516 $text = "[[:$titleText]]";
3518 # This will probably not be a working link, but at least it may
3519 # provide some hint of where the problem is
3520 preg_replace( '/^:/', '', $originalTitle );
3521 $text = "[[:$originalTitle]]";
3523 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3524 . 'post-expand include size too large -->' );
3525 $this->limitationWarn( 'post-expand-template-inclusion' );
3528 if ( $isLocalObj ) {
3529 $ret = [ 'object' => $text ];
3531 $ret = [ 'text' => $text ];
3538 * Call a parser function and return an array with text and flags.
3540 * The returned array will always contain a boolean 'found', indicating
3541 * whether the parser function was found or not. It may also contain the
3543 * text: string|object, resulting wikitext or PP DOM object
3544 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3545 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3546 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3547 * nowiki: bool, wiki markup in $text should be escaped
3550 * @param PPFrame $frame The current frame, contains template arguments
3551 * @param string $function Function name
3552 * @param array $args Arguments to the function
3553 * @throws MWException
3556 public function callParserFunction( $frame, $function, array $args = [] ) {
3557 # Case sensitive functions
3558 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3559 $function = $this->mFunctionSynonyms
[1][$function];
3561 # Case insensitive functions
3562 $function = $this->contLang
->lc( $function );
3563 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3564 $function = $this->mFunctionSynonyms
[0][$function];
3566 return [ 'found' => false ];
3570 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3572 // Avoid PHP 7.1 warning from passing $this by reference
3575 $allArgs = [ &$parser ];
3576 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3577 # Convert arguments to PPNodes and collect for appending to $allArgs
3579 foreach ( $args as $k => $v ) {
3580 if ( $v instanceof PPNode ||
$k === 0 ) {
3583 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3587 # Add a frame parameter, and pass the arguments as an array
3588 $allArgs[] = $frame;
3589 $allArgs[] = $funcArgs;
3591 # Convert arguments to plain text and append to $allArgs
3592 foreach ( $args as $k => $v ) {
3593 if ( $v instanceof PPNode
) {
3594 $allArgs[] = trim( $frame->expand( $v ) );
3595 } elseif ( is_int( $k ) && $k >= 0 ) {
3596 $allArgs[] = trim( $v );
3598 $allArgs[] = trim( "$k=$v" );
3603 $result = $callback( ...$allArgs );
3605 # The interface for function hooks allows them to return a wikitext
3606 # string or an array containing the string and any flags. This mungs
3607 # things around to match what this method should return.
3608 if ( !is_array( $result ) ) {
3614 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3615 $result['text'] = $result[0];
3617 unset( $result[0] );
3624 $preprocessFlags = 0;
3625 if ( isset( $result['noparse'] ) ) {
3626 $noparse = $result['noparse'];
3628 if ( isset( $result['preprocessFlags'] ) ) {
3629 $preprocessFlags = $result['preprocessFlags'];
3633 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3634 $result['isChildObj'] = true;
3641 * Get the semi-parsed DOM representation of a template with a given title,
3642 * and its redirect destination title. Cached.
3644 * @param Title $title
3648 public function getTemplateDom( $title ) {
3649 $cacheTitle = $title;
3650 $titleText = $title->getPrefixedDBkey();
3652 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3653 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3654 $title = Title
::makeTitle( $ns, $dbk );
3655 $titleText = $title->getPrefixedDBkey();
3657 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3658 return [ $this->mTplDomCache
[$titleText], $title ];
3661 # Cache miss, go to the database
3662 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3664 if ( $text === false ) {
3665 $this->mTplDomCache
[$titleText] = false;
3666 return [ false, $title ];
3669 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3670 $this->mTplDomCache
[$titleText] = $dom;
3672 if ( !$title->equals( $cacheTitle ) ) {
3673 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3674 [ $title->getNamespace(), $title->getDBkey() ];
3677 return [ $dom, $title ];
3681 * Fetch the current revision of a given title. Note that the revision
3682 * (and even the title) may not exist in the database, so everything
3683 * contributing to the output of the parser should use this method
3684 * where possible, rather than getting the revisions themselves. This
3685 * method also caches its results, so using it benefits performance.
3688 * @param Title $title
3691 public function fetchCurrentRevisionOfTitle( $title ) {
3692 $cacheKey = $title->getPrefixedDBkey();
3693 if ( !$this->currentRevisionCache
) {
3694 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3696 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3697 $this->currentRevisionCache
->set( $cacheKey,
3698 // Defaults to Parser::statelessFetchRevision()
3699 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3702 return $this->currentRevisionCache
->get( $cacheKey );
3706 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3707 * without passing them on to it.
3710 * @param Title $title
3711 * @param Parser|bool $parser
3712 * @return Revision|bool False if missing
3714 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3715 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3721 * Fetch the unparsed text of a template and register a reference to it.
3722 * @param Title $title
3723 * @return array ( string or false, Title )
3725 public function fetchTemplateAndTitle( $title ) {
3726 // Defaults to Parser::statelessFetchTemplate()
3727 $templateCb = $this->mOptions
->getTemplateCallback();
3728 $stuff = call_user_func( $templateCb, $title, $this );
3729 // We use U+007F DELETE to distinguish strip markers from regular text.
3730 $text = $stuff['text'];
3731 if ( is_string( $stuff['text'] ) ) {
3732 $text = strtr( $text, "\x7f", "?" );
3734 $finalTitle = $stuff['finalTitle'] ??
$title;
3735 if ( isset( $stuff['deps'] ) ) {
3736 foreach ( $stuff['deps'] as $dep ) {
3737 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3738 if ( $dep['title']->equals( $this->getTitle() ) ) {
3739 // If we transclude ourselves, the final result
3740 // will change based on the new version of the page
3741 $this->mOutput
->setFlag( 'vary-revision' );
3745 return [ $text, $finalTitle ];
3749 * Fetch the unparsed text of a template and register a reference to it.
3750 * @param Title $title
3751 * @return string|bool
3753 public function fetchTemplate( $title ) {
3754 return $this->fetchTemplateAndTitle( $title )[0];
3758 * Static function to get a template
3759 * Can be overridden via ParserOptions::setTemplateCallback().
3761 * @param Title $title
3762 * @param bool|Parser $parser
3766 public static function statelessFetchTemplate( $title, $parser = false ) {
3767 $text = $skip = false;
3768 $finalTitle = $title;
3771 # Loop to fetch the article, with up to 1 redirect
3772 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3773 # Give extensions a chance to select the revision instead
3774 $id = false; # Assume current
3775 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3776 [ $parser, $title, &$skip, &$id ] );
3782 'page_id' => $title->getArticleID(),
3789 $rev = Revision
::newFromId( $id );
3790 } elseif ( $parser ) {
3791 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3793 $rev = Revision
::newFromTitle( $title );
3795 $rev_id = $rev ?
$rev->getId() : 0;
3796 # If there is no current revision, there is no page
3797 if ( $id === false && !$rev ) {
3798 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3799 $linkCache->addBadLinkObj( $title );
3804 'page_id' => $title->getArticleID(),
3805 'rev_id' => $rev_id ];
3806 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3807 # We fetched a rev from a different title; register it too...
3809 'title' => $rev->getTitle(),
3810 'page_id' => $rev->getPage(),
3811 'rev_id' => $rev_id ];
3815 $content = $rev->getContent();
3816 $text = $content ?
$content->getWikitextForTransclusion() : null;
3818 Hooks
::run( 'ParserFetchTemplate',
3819 [ $parser, $title, $rev, &$text, &$deps ] );
3821 if ( $text === false ||
$text === null ) {
3825 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3826 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3827 lcfirst( $title->getText() ) )->inContentLanguage();
3828 if ( !$message->exists() ) {
3832 $content = $message->content();
3833 $text = $message->plain();
3841 $finalTitle = $title;
3842 $title = $content->getRedirectTarget();
3846 'finalTitle' => $finalTitle,
3851 * Fetch a file and its title and register a reference to it.
3852 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3853 * @param Title $title
3854 * @param array $options Array of options to RepoGroup::findFile
3856 * @deprecated since 1.32, use fetchFileAndTitle instead
3858 public function fetchFile( $title, $options = [] ) {
3859 wfDeprecated( __METHOD__
, '1.32' );
3860 return $this->fetchFileAndTitle( $title, $options )[0];
3864 * Fetch a file and its title and register a reference to it.
3865 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3866 * @param Title $title
3867 * @param array $options Array of options to RepoGroup::findFile
3868 * @return array ( File or false, Title of file )
3870 public function fetchFileAndTitle( $title, $options = [] ) {
3871 $file = $this->fetchFileNoRegister( $title, $options );
3873 $time = $file ?
$file->getTimestamp() : false;
3874 $sha1 = $file ?
$file->getSha1() : false;
3875 # Register the file as a dependency...
3876 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3877 if ( $file && !$title->equals( $file->getTitle() ) ) {
3878 # Update fetched file title
3879 $title = $file->getTitle();
3880 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3882 return [ $file, $title ];
3886 * Helper function for fetchFileAndTitle.
3888 * Also useful if you need to fetch a file but not use it yet,
3889 * for example to get the file's handler.
3891 * @param Title $title
3892 * @param array $options Array of options to RepoGroup::findFile
3895 protected function fetchFileNoRegister( $title, $options = [] ) {
3896 if ( isset( $options['broken'] ) ) {
3897 $file = false; // broken thumbnail forced by hook
3898 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3899 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3900 } else { // get by (name,timestamp)
3901 $file = wfFindFile( $title, $options );
3907 * Transclude an interwiki link.
3909 * @param Title $title
3910 * @param string $action Usually one of (raw, render)
3914 public function interwikiTransclude( $title, $action ) {
3915 if ( !$this->svcOptions
->get( 'EnableScaryTranscluding' ) ) {
3916 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3919 $url = $title->getFullURL( [ 'action' => $action ] );
3920 if ( strlen( $url ) > 1024 ) {
3921 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3924 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3926 $fname = __METHOD__
;
3927 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3929 $data = $cache->getWithSetCallback(
3930 $cache->makeGlobalKey(
3931 'interwiki-transclude',
3932 ( $wikiId !== false ) ?
$wikiId : 'external',
3935 $this->svcOptions
->get( 'TranscludeCacheExpiry' ),
3936 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3937 $req = MWHttpRequest
::factory( $url, [], $fname );
3939 $status = $req->execute(); // Status object
3940 if ( !$status->isOK() ) {
3941 $ttl = $cache::TTL_UNCACHEABLE
;
3942 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3943 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3947 'text' => $status->isOK() ?
$req->getContent() : null,
3948 'code' => $req->getStatus()
3952 'checkKeys' => ( $wikiId !== false )
3953 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3955 'pcGroup' => 'interwiki-transclude:5',
3956 'pcTTL' => $cache::TTL_PROC_LONG
3960 if ( is_string( $data['text'] ) ) {
3961 $text = $data['text'];
3962 } elseif ( $data['code'] != 200 ) {
3963 // Though we failed to fetch the content, this status is useless.
3964 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3965 ->params( $url, $data['code'] )->inContentLanguage()->text();
3967 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3974 * Triple brace replacement -- used for template arguments
3977 * @param array $piece
3978 * @param PPFrame $frame
3982 public function argSubstitution( $piece, $frame ) {
3984 $parts = $piece['parts'];
3985 $nameWithSpaces = $frame->expand( $piece['title'] );
3986 $argName = trim( $nameWithSpaces );
3988 $text = $frame->getArgument( $argName );
3989 if ( $text === false && $parts->getLength() > 0
3990 && ( $this->ot
['html']
3992 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3995 # No match in frame, use the supplied default
3996 $object = $parts->item( 0 )->getChildren();
3998 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3999 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4000 $this->limitationWarn( 'post-expand-template-argument' );
4003 if ( $text === false && $object === false ) {
4005 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4007 if ( $error !== false ) {
4010 if ( $object !== false ) {
4011 $ret = [ 'object' => $object ];
4013 $ret = [ 'text' => $text ];
4020 * Return the text to be used for a given extension tag.
4021 * This is the ghost of strip().
4023 * @param array $params Associative array of parameters:
4024 * name PPNode for the tag name
4025 * attr PPNode for unparsed text where tag attributes are thought to be
4026 * attributes Optional associative array of parsed attributes
4027 * inner Contents of extension element
4028 * noClose Original text did not have a close tag
4029 * @param PPFrame $frame
4031 * @throws MWException
4034 public function extensionSubstitution( $params, $frame ) {
4035 static $errorStr = '<span class="error">';
4036 static $errorLen = 20;
4038 $name = $frame->expand( $params['name'] );
4039 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4040 // Probably expansion depth or node count exceeded. Just punt the
4045 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
4046 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4051 // We can't safely check if the expansion for $content resulted in an
4052 // error, because the content could happen to be the error string
4054 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
4056 $marker = self
::MARKER_PREFIX
. "-$name-"
4057 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
4059 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
4060 ( $this->ot
['html'] ||
$this->ot
['pre'] );
4061 if ( $isFunctionTag ) {
4062 $markerType = 'none';
4064 $markerType = 'general';
4066 if ( $this->ot
['html'] ||
$isFunctionTag ) {
4067 $name = strtolower( $name );
4068 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
4069 if ( isset( $params['attributes'] ) ) {
4070 $attributes +
= $params['attributes'];
4073 if ( isset( $this->mTagHooks
[$name] ) ) {
4074 $output = call_user_func_array( $this->mTagHooks
[$name],
4075 [ $content, $attributes, $this, $frame ] );
4076 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
4077 list( $callback, ) = $this->mFunctionTagHooks
[$name];
4079 // Avoid PHP 7.1 warning from passing $this by reference
4081 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4083 $output = '<span class="error">Invalid tag extension name: ' .
4084 htmlspecialchars( $name ) . '</span>';
4087 if ( is_array( $output ) ) {
4090 $output = $flags[0];
4091 if ( isset( $flags['markerType'] ) ) {
4092 $markerType = $flags['markerType'];
4096 if ( is_null( $attrText ) ) {
4099 if ( isset( $params['attributes'] ) ) {
4100 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4101 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4102 htmlspecialchars( $attrValue ) . '"';
4105 if ( $content === null ) {
4106 $output = "<$name$attrText/>";
4108 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4109 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4113 $output = "<$name$attrText>$content$close";
4117 if ( $markerType === 'none' ) {
4119 } elseif ( $markerType === 'nowiki' ) {
4120 $this->mStripState
->addNoWiki( $marker, $output );
4121 } elseif ( $markerType === 'general' ) {
4122 $this->mStripState
->addGeneral( $marker, $output );
4124 throw new MWException( __METHOD__
. ': invalid marker type' );
4130 * Increment an include size counter
4132 * @param string $type The type of expansion
4133 * @param int $size The size of the text
4134 * @return bool False if this inclusion would take it over the maximum, true otherwise
4136 public function incrementIncludeSize( $type, $size ) {
4137 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4140 $this->mIncludeSizes
[$type] +
= $size;
4146 * Increment the expensive function count
4148 * @return bool False if the limit has been exceeded
4150 public function incrementExpensiveFunctionCount() {
4151 $this->mExpensiveFunctionCount++
;
4152 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4156 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4157 * Fills $this->mDoubleUnderscores, returns the modified text
4159 * @param string $text
4163 public function doDoubleUnderscore( $text ) {
4164 # The position of __TOC__ needs to be recorded
4165 $mw = $this->magicWordFactory
->get( 'toc' );
4166 if ( $mw->match( $text ) ) {
4167 $this->mShowToc
= true;
4168 $this->mForceTocPosition
= true;
4170 # Set a placeholder. At the end we'll fill it in with the TOC.
4171 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4173 # Only keep the first one.
4174 $text = $mw->replace( '', $text );
4177 # Now match and remove the rest of them
4178 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4179 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4181 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4182 $this->mOutput
->mNoGallery
= true;
4184 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4185 $this->mShowToc
= false;
4187 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4188 && $this->mTitle
->getNamespace() == NS_CATEGORY
4190 $this->addTrackingCategory( 'hidden-category-category' );
4192 # (T10068) Allow control over whether robots index a page.
4193 # __INDEX__ always overrides __NOINDEX__, see T16899
4194 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4195 $this->mOutput
->setIndexPolicy( 'noindex' );
4196 $this->addTrackingCategory( 'noindex-category' );
4198 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4199 $this->mOutput
->setIndexPolicy( 'index' );
4200 $this->addTrackingCategory( 'index-category' );
4203 # Cache all double underscores in the database
4204 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4205 $this->mOutput
->setProperty( $key, '' );
4212 * @see ParserOutput::addTrackingCategory()
4213 * @param string $msg Message key
4214 * @return bool Whether the addition was successful
4216 public function addTrackingCategory( $msg ) {
4217 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4221 * This function accomplishes several tasks:
4222 * 1) Auto-number headings if that option is enabled
4223 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4224 * 3) Add a Table of contents on the top for users who have enabled the option
4225 * 4) Auto-anchor headings
4227 * It loops through all headlines, collects the necessary data, then splits up the
4228 * string and re-inserts the newly formatted headlines.
4230 * @param string $text
4231 * @param string $origText Original, untouched wikitext
4232 * @param bool $isMain
4233 * @return mixed|string
4236 public function formatHeadings( $text, $origText, $isMain = true ) {
4237 # Inhibit editsection links if requested in the page
4238 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4239 $maybeShowEditLink = false;
4241 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4244 # Get all headlines for numbering them and adding funky stuff like [edit]
4245 # links - this is for later, but we need the number of headlines right now
4246 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4247 # be trimmed here since whitespace in HTML headings is significant.
4249 $numMatches = preg_match_all(
4250 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4255 # if there are fewer than 4 headlines in the article, do not show TOC
4256 # unless it's been explicitly enabled.
4257 $enoughToc = $this->mShowToc
&&
4258 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4260 # Allow user to stipulate that a page should have a "new section"
4261 # link added via __NEWSECTIONLINK__
4262 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4263 $this->mOutput
->setNewSection( true );
4266 # Allow user to remove the "new section"
4267 # link via __NONEWSECTIONLINK__
4268 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4269 $this->mOutput
->hideNewSection( true );
4272 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4273 # override above conditions and always show TOC above first header
4274 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4275 $this->mShowToc
= true;
4283 # Ugh .. the TOC should have neat indentation levels which can be
4284 # passed to the skin functions. These are determined here
4288 $sublevelCount = [];
4294 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4295 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4296 $oldType = $this->mOutputType
;
4297 $this->setOutputType( self
::OT_WIKI
);
4298 $frame = $this->getPreprocessor()->newFrame();
4299 $root = $this->preprocessToDom( $origText );
4300 $node = $root->getFirstChild();
4305 $headlines = $numMatches !== false ?
$matches[3] : [];
4307 $maxTocLevel = $this->svcOptions
->get( 'MaxTocLevel' );
4308 foreach ( $headlines as $headline ) {
4309 $isTemplate = false;
4311 $sectionIndex = false;
4313 $markerMatches = [];
4314 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4315 $serial = $markerMatches[1];
4316 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4317 $isTemplate = ( $titleText != $baseTitleText );
4318 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4322 $prevlevel = $level;
4324 $level = $matches[1][$headlineCount];
4326 if ( $level > $prevlevel ) {
4327 # Increase TOC level
4329 $sublevelCount[$toclevel] = 0;
4330 if ( $toclevel < $maxTocLevel ) {
4331 $prevtoclevel = $toclevel;
4332 $toc .= Linker
::tocIndent();
4335 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4336 # Decrease TOC level, find level to jump to
4338 for ( $i = $toclevel; $i > 0; $i-- ) {
4339 if ( $levelCount[$i] == $level ) {
4340 # Found last matching level
4343 } elseif ( $levelCount[$i] < $level ) {
4344 # Found first matching level below current level
4352 if ( $toclevel < $maxTocLevel ) {
4353 if ( $prevtoclevel < $maxTocLevel ) {
4354 # Unindent only if the previous toc level was shown :p
4355 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4356 $prevtoclevel = $toclevel;
4358 $toc .= Linker
::tocLineEnd();
4362 # No change in level, end TOC line
4363 if ( $toclevel < $maxTocLevel ) {
4364 $toc .= Linker
::tocLineEnd();
4368 $levelCount[$toclevel] = $level;
4370 # count number of headlines for each level
4371 $sublevelCount[$toclevel]++
;
4373 for ( $i = 1; $i <= $toclevel; $i++
) {
4374 if ( !empty( $sublevelCount[$i] ) ) {
4378 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4383 # The safe header is a version of the header text safe to use for links
4385 # Remove link placeholders by the link text.
4386 # <!--LINK number-->
4388 # link text with suffix
4389 # Do this before unstrip since link text can contain strip markers
4390 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4392 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4393 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4395 # Remove any <style> or <script> tags (T198618)
4396 $safeHeadline = preg_replace(
4397 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4402 # Strip out HTML (first regex removes any tag not allowed)
4404 # * <sup> and <sub> (T10393)
4408 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4409 # * <s> and <strike> (T35715)
4410 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4411 # to allow setting directionality in toc items.
4412 $tocline = preg_replace(
4414 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4415 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4421 # Strip '<span></span>', which is the result from the above if
4422 # <span id="foo"></span> is used to produce an additional anchor
4424 $tocline = str_replace( '<span></span>', '', $tocline );
4426 $tocline = trim( $tocline );
4428 # For the anchor, strip out HTML-y stuff period
4429 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4430 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4432 # Save headline for section edit hint before it's escaped
4433 $headlineHint = $safeHeadline;
4435 # Decode HTML entities
4436 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4438 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4440 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4441 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4442 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4443 if ( $fallbackHeadline === $safeHeadline ) {
4444 # No reason to have both (in fact, we can't)
4445 $fallbackHeadline = false;
4448 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4449 # @todo FIXME: We may be changing them depending on the current locale.
4450 $arrayKey = strtolower( $safeHeadline );
4451 if ( $fallbackHeadline === false ) {
4452 $fallbackArrayKey = false;
4454 $fallbackArrayKey = strtolower( $fallbackHeadline );
4457 # Create the anchor for linking from the TOC to the section
4458 $anchor = $safeHeadline;
4459 $fallbackAnchor = $fallbackHeadline;
4460 if ( isset( $refers[$arrayKey] ) ) {
4461 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4462 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4464 $linkAnchor .= "_$i";
4465 $refers["${arrayKey}_$i"] = true;
4467 $refers[$arrayKey] = true;
4469 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4470 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4471 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4472 $fallbackAnchor .= "_$i";
4473 $refers["${fallbackArrayKey}_$i"] = true;
4475 $refers[$fallbackArrayKey] = true;
4478 # Don't number the heading if it is the only one (looks silly)
4479 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4480 # the two are different if the line contains a link
4481 $headline = Html
::element(
4483 [ 'class' => 'mw-headline-number' ],
4485 ) . ' ' . $headline;
4488 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4489 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4490 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4493 # Add the section to the section tree
4494 # Find the DOM node for this header
4495 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4496 while ( $node && !$noOffset ) {
4497 if ( $node->getName() === 'h' ) {
4498 $bits = $node->splitHeading();
4499 if ( $bits['i'] == $sectionIndex ) {
4503 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4504 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4505 $node = $node->getNextSibling();
4508 'toclevel' => $toclevel,
4511 'number' => $numbering,
4512 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4513 'fromtitle' => $titleText,
4514 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4515 'anchor' => $anchor,
4518 # give headline the correct <h#> tag
4519 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4520 // Output edit section links as markers with styles that can be customized by skins
4521 if ( $isTemplate ) {
4522 # Put a T flag in the section identifier, to indicate to extractSections()
4523 # that sections inside <includeonly> should be counted.
4524 $editsectionPage = $titleText;
4525 $editsectionSection = "T-$sectionIndex";
4526 $editsectionContent = null;
4528 $editsectionPage = $this->mTitle
->getPrefixedText();
4529 $editsectionSection = $sectionIndex;
4530 $editsectionContent = $headlineHint;
4532 // We use a bit of pesudo-xml for editsection markers. The
4533 // language converter is run later on. Using a UNIQ style marker
4534 // leads to the converter screwing up the tokens when it
4535 // converts stuff. And trying to insert strip tags fails too. At
4536 // this point all real inputted tags have already been escaped,
4537 // so we don't have to worry about a user trying to input one of
4538 // these markers directly. We use a page and section attribute
4539 // to stop the language converter from converting these
4540 // important bits of data, but put the headline hint inside a
4541 // content block because the language converter is supposed to
4542 // be able to convert that piece of data.
4543 // Gets replaced with html in ParserOutput::getText
4544 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4545 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4546 if ( $editsectionContent !== null ) {
4547 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4554 $head[$headlineCount] = Linker
::makeHeadline( $level,
4555 $matches['attrib'][$headlineCount], $anchor, $headline,
4556 $editlink, $fallbackAnchor );
4561 $this->setOutputType( $oldType );
4563 # Never ever show TOC if no headers
4564 if ( $numVisible < 1 ) {
4569 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4570 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4572 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4573 $this->mOutput
->setTOCHTML( $toc );
4574 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4578 $this->mOutput
->setSections( $tocraw );
4581 # split up and insert constructed headlines
4582 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4585 // build an array of document sections
4587 foreach ( $blocks as $block ) {
4588 // $head is zero-based, sections aren't.
4589 if ( empty( $head[$i - 1] ) ) {
4590 $sections[$i] = $block;
4592 $sections[$i] = $head[$i - 1] . $block;
4596 * Send a hook, one per section.
4597 * The idea here is to be able to make section-level DIVs, but to do so in a
4598 * lower-impact, more correct way than r50769
4601 * $section : the section number
4602 * &$sectionContent : ref to the content of the section
4603 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4605 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4610 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4611 // append the TOC at the beginning
4612 // Top anchor now in skin
4613 $sections[0] .= $toc . "\n";
4616 $full .= implode( '', $sections );
4618 if ( $this->mForceTocPosition
) {
4619 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4626 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4627 * conversion, substituting signatures, {{subst:}} templates, etc.
4629 * @param string $text The text to transform
4630 * @param Title $title The Title object for the current article
4631 * @param User $user The User object describing the current user
4632 * @param ParserOptions $options Parsing options
4633 * @param bool $clearState Whether to clear the parser state first
4634 * @return string The altered wiki markup
4636 public function preSaveTransform( $text, Title
$title, User
$user,
4637 ParserOptions
$options, $clearState = true
4639 if ( $clearState ) {
4640 $magicScopeVariable = $this->lock();
4642 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4643 $this->setUser( $user );
4645 // Strip U+0000 NULL (T159174)
4646 $text = str_replace( "\000", '', $text );
4648 // We still normalize line endings for backwards-compatibility
4649 // with other code that just calls PST, but this should already
4650 // be handled in TextContent subclasses
4651 $text = TextContent
::normalizeLineEndings( $text );
4653 if ( $options->getPreSaveTransform() ) {
4654 $text = $this->pstPass2( $text, $user );
4656 $text = $this->mStripState
->unstripBoth( $text );
4658 $this->setUser( null ); # Reset
4664 * Pre-save transform helper function
4666 * @param string $text
4671 private function pstPass2( $text, $user ) {
4672 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4673 # $this->contLang here in order to give everyone the same signature and use the default one
4674 # rather than the one selected in each user's preferences. (see also T14815)
4675 $ts = $this->mOptions
->getTimestamp();
4676 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4677 $ts = $timestamp->format( 'YmdHis' );
4678 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4680 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4682 # Variable replacement
4683 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4684 $text = $this->replaceVariables( $text );
4686 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4687 # which may corrupt this parser instance via its wfMessage()->text() call-
4690 if ( strpos( $text, '~~~' ) !== false ) {
4691 $sigText = $this->getUserSig( $user );
4692 $text = strtr( $text, [
4694 '~~~~' => "$sigText $d",
4697 # The main two signature forms used above are time-sensitive
4698 $this->mOutput
->setFlag( 'user-signature' );
4701 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4702 $tc = '[' . Title
::legalChars() . ']';
4703 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4705 // [[ns:page (context)|]]
4706 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4707 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4708 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4709 // [[ns:page (context), context|]] (using either single or double-width comma)
4710 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4711 // [[|page]] (reverse pipe trick: add context from page title)
4712 $p2 = "/\[\[\\|($tc+)]]/";
4714 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4715 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4716 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4717 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4719 $t = $this->mTitle
->getText();
4721 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4722 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4723 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4724 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4726 # if there's no context, don't bother duplicating the title
4727 $text = preg_replace( $p2, '[[\\1]]', $text );
4734 * Fetch the user's signature text, if any, and normalize to
4735 * validated, ready-to-insert wikitext.
4736 * If you have pre-fetched the nickname or the fancySig option, you can
4737 * specify them here to save a database query.
4738 * Do not reuse this parser instance after calling getUserSig(),
4739 * as it may have changed.
4741 * @param User &$user
4742 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4743 * @param bool|null $fancySig whether the nicknname is the complete signature
4744 * or null to use default value
4747 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4748 $username = $user->getName();
4750 # If not given, retrieve from the user object.
4751 if ( $nickname === false ) {
4752 $nickname = $user->getOption( 'nickname' );
4755 if ( is_null( $fancySig ) ) {
4756 $fancySig = $user->getBoolOption( 'fancysig' );
4759 $nickname = $nickname == null ?
$username : $nickname;
4761 if ( mb_strlen( $nickname ) > $this->svcOptions
->get( 'MaxSigChars' ) ) {
4762 $nickname = $username;
4763 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4764 } elseif ( $fancySig !== false ) {
4765 # Sig. might contain markup; validate this
4766 if ( $this->validateSig( $nickname ) !== false ) {
4767 # Validated; clean up (if needed) and return it
4768 return $this->cleanSig( $nickname, true );
4770 # Failed to validate; fall back to the default
4771 $nickname = $username;
4772 wfDebug( __METHOD__
. ": $username has bad XML tags in signature.\n" );
4776 # Make sure nickname doesnt get a sig in a sig
4777 $nickname = self
::cleanSigInSig( $nickname );
4779 # If we're still here, make it a link to the user page
4780 $userText = wfEscapeWikiText( $username );
4781 $nickText = wfEscapeWikiText( $nickname );
4782 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4784 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4785 ->title( $this->getTitle() )->text();
4789 * Check that the user's signature contains no bad XML
4791 * @param string $text
4792 * @return string|bool An expanded string, or false if invalid.
4794 public function validateSig( $text ) {
4795 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4799 * Clean up signature text
4801 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4802 * 2) Substitute all transclusions
4804 * @param string $text
4805 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4806 * @return string Signature text
4808 public function cleanSig( $text, $parsing = false ) {
4811 $magicScopeVariable = $this->lock();
4812 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4815 # Option to disable this feature
4816 if ( !$this->mOptions
->getCleanSignatures() ) {
4820 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4821 # => Move this logic to braceSubstitution()
4822 $substWord = $this->magicWordFactory
->get( 'subst' );
4823 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4824 $substText = '{{' . $substWord->getSynonym( 0 );
4826 $text = preg_replace( $substRegex, $substText, $text );
4827 $text = self
::cleanSigInSig( $text );
4828 $dom = $this->preprocessToDom( $text );
4829 $frame = $this->getPreprocessor()->newFrame();
4830 $text = $frame->expand( $dom );
4833 $text = $this->mStripState
->unstripBoth( $text );
4840 * Strip 3, 4 or 5 tildes out of signatures.
4842 * @param string $text
4843 * @return string Signature text with /~{3,5}/ removed
4845 public static function cleanSigInSig( $text ) {
4846 $text = preg_replace( '/~{3,5}/', '', $text );
4851 * Set up some variables which are usually set up in parse()
4852 * so that an external function can call some class members with confidence
4854 * @param Title|null $title
4855 * @param ParserOptions $options
4856 * @param int $outputType
4857 * @param bool $clearState
4859 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4860 $outputType, $clearState = true
4862 $this->startParse( $title, $options, $outputType, $clearState );
4866 * @param Title|null $title
4867 * @param ParserOptions $options
4868 * @param int $outputType
4869 * @param bool $clearState
4871 private function startParse( Title
$title = null, ParserOptions
$options,
4872 $outputType, $clearState = true
4874 $this->setTitle( $title );
4875 $this->mOptions
= $options;
4876 $this->setOutputType( $outputType );
4877 if ( $clearState ) {
4878 $this->clearState();
4883 * Wrapper for preprocess()
4885 * @param string $text The text to preprocess
4886 * @param ParserOptions $options
4887 * @param Title|null $title Title object or null to use $wgTitle
4890 public function transformMsg( $text, $options, $title = null ) {
4891 static $executing = false;
4893 # Guard against infinite recursion
4904 $text = $this->preprocess( $text, $title, $options );
4911 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4912 * The callback should have the following form:
4913 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4915 * Transform and return $text. Use $parser for any required context, e.g. use
4916 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4918 * Hooks may return extended information by returning an array, of which the
4919 * first numbered element (index 0) must be the return string, and all other
4920 * entries are extracted into local variables within an internal function
4921 * in the Parser class.
4923 * This interface (introduced r61913) appears to be undocumented, but
4924 * 'markerType' is used by some core tag hooks to override which strip
4925 * array their results are placed in. **Use great caution if attempting
4926 * this interface, as it is not documented and injudicious use could smash
4927 * private variables.**
4929 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4930 * @param callable $callback The callback function (and object) to use for the tag
4931 * @throws MWException
4932 * @return callable|null The old value of the mTagHooks array associated with the hook
4934 public function setHook( $tag, callable
$callback ) {
4935 $tag = strtolower( $tag );
4936 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4937 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4939 $oldVal = $this->mTagHooks
[$tag] ??
null;
4940 $this->mTagHooks
[$tag] = $callback;
4941 if ( !in_array( $tag, $this->mStripList
) ) {
4942 $this->mStripList
[] = $tag;
4949 * As setHook(), but letting the contents be parsed.
4951 * Transparent tag hooks are like regular XML-style tag hooks, except they
4952 * operate late in the transformation sequence, on HTML instead of wikitext.
4954 * This is probably obsoleted by things dealing with parser frames?
4955 * The only extension currently using it is geoserver.
4958 * @todo better document or deprecate this
4960 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4961 * @param callable $callback The callback function (and object) to use for the tag
4962 * @throws MWException
4963 * @return callable|null The old value of the mTagHooks array associated with the hook
4965 public function setTransparentTagHook( $tag, callable
$callback ) {
4966 $tag = strtolower( $tag );
4967 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4968 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4970 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4971 $this->mTransparentTagHooks
[$tag] = $callback;
4977 * Remove all tag hooks
4979 public function clearTagHooks() {
4980 $this->mTagHooks
= [];
4981 $this->mFunctionTagHooks
= [];
4982 $this->mStripList
= $this->mDefaultStripList
;
4986 * Create a function, e.g. {{sum:1|2|3}}
4987 * The callback function should have the form:
4988 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4990 * Or with Parser::SFH_OBJECT_ARGS:
4991 * function myParserFunction( $parser, $frame, $args ) { ... }
4993 * The callback may either return the text result of the function, or an array with the text
4994 * in element 0, and a number of flags in the other elements. The names of the flags are
4995 * specified in the keys. Valid flags are:
4996 * found The text returned is valid, stop processing the template. This
4998 * nowiki Wiki markup in the return value should be escaped
4999 * isHTML The returned text is HTML, armour it against wikitext transformation
5001 * @param string $id The magic word ID
5002 * @param callable $callback The callback function (and object) to use
5003 * @param int $flags A combination of the following flags:
5004 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5006 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5007 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5008 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5009 * the arguments, and to control the way they are expanded.
5011 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5012 * arguments, for instance:
5013 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5015 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5016 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5017 * working if/when this is changed.
5019 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5022 * Please read the documentation in includes/parser/Preprocessor.php for more information
5023 * about the methods available in PPFrame and PPNode.
5025 * @throws MWException
5026 * @return string|callable The old callback function for this name, if any
5028 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
5029 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
5030 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
5032 # Add to function cache
5033 $mw = $this->magicWordFactory
->get( $id );
5035 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
5038 $synonyms = $mw->getSynonyms();
5039 $sensitive = intval( $mw->isCaseSensitive() );
5041 foreach ( $synonyms as $syn ) {
5043 if ( !$sensitive ) {
5044 $syn = $this->contLang
->lc( $syn );
5047 if ( !( $flags & self
::SFH_NO_HASH
) ) {
5050 # Remove trailing colon
5051 if ( substr( $syn, -1, 1 ) === ':' ) {
5052 $syn = substr( $syn, 0, -1 );
5054 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
5060 * Get all registered function hook identifiers
5064 public function getFunctionHooks() {
5065 $this->firstCallInit();
5066 return array_keys( $this->mFunctionHooks
);
5070 * Create a tag function, e.g. "<test>some stuff</test>".
5071 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5072 * Unlike parser functions, their content is not preprocessed.
5073 * @param string $tag
5074 * @param callable $callback
5076 * @throws MWException
5079 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
5080 $tag = strtolower( $tag );
5081 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5082 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5084 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5085 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5087 if ( !in_array( $tag, $this->mStripList
) ) {
5088 $this->mStripList
[] = $tag;
5095 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5096 * Placeholders created in Linker::link()
5098 * @param string &$text
5099 * @param int $options
5101 public function replaceLinkHolders( &$text, $options = 0 ) {
5102 $this->mLinkHolders
->replace( $text );
5106 * Replace "<!--LINK-->" link placeholders with plain text of links
5107 * (not HTML-formatted).
5109 * @param string $text
5112 public function replaceLinkHoldersText( $text ) {
5113 return $this->mLinkHolders
->replaceText( $text );
5117 * Renders an image gallery from a text with one line per image.
5118 * text labels may be given by using |-style alternative text. E.g.
5119 * Image:one.jpg|The number "1"
5120 * Image:tree.jpg|A tree
5121 * given as text will return the HTML of a gallery with two images,
5122 * labeled 'The number "1"' and
5125 * @param string $text
5126 * @param array $params
5127 * @return string HTML
5129 public function renderImageGallery( $text, $params ) {
5131 if ( isset( $params['mode'] ) ) {
5132 $mode = $params['mode'];
5136 $ig = ImageGalleryBase
::factory( $mode );
5137 } catch ( Exception
$e ) {
5138 // If invalid type set, fallback to default.
5139 $ig = ImageGalleryBase
::factory( false );
5142 $ig->setContextTitle( $this->mTitle
);
5143 $ig->setShowBytes( false );
5144 $ig->setShowDimensions( false );
5145 $ig->setShowFilename( false );
5146 $ig->setParser( $this );
5147 $ig->setHideBadImages();
5148 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5150 if ( isset( $params['showfilename'] ) ) {
5151 $ig->setShowFilename( true );
5153 $ig->setShowFilename( false );
5155 if ( isset( $params['caption'] ) ) {
5156 // NOTE: We aren't passing a frame here or below. Frame info
5157 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5158 // See T107332#4030581
5159 $caption = $this->recursiveTagParse( $params['caption'] );
5160 $ig->setCaptionHtml( $caption );
5162 if ( isset( $params['perrow'] ) ) {
5163 $ig->setPerRow( $params['perrow'] );
5165 if ( isset( $params['widths'] ) ) {
5166 $ig->setWidths( $params['widths'] );
5168 if ( isset( $params['heights'] ) ) {
5169 $ig->setHeights( $params['heights'] );
5171 $ig->setAdditionalOptions( $params );
5173 // Avoid PHP 7.1 warning from passing $this by reference
5175 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5177 $lines = StringUtils
::explode( "\n", $text );
5178 foreach ( $lines as $line ) {
5179 # match lines like these:
5180 # Image:someimage.jpg|This is some image
5182 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5184 if ( count( $matches ) == 0 ) {
5188 if ( strpos( $matches[0], '%' ) !== false ) {
5189 $matches[1] = rawurldecode( $matches[1] );
5191 $title = Title
::newFromText( $matches[1], NS_FILE
);
5192 if ( is_null( $title ) ) {
5193 # Bogus title. Ignore these so we don't bomb out later.
5197 # We need to get what handler the file uses, to figure out parameters.
5198 # Note, a hook can overide the file name, and chose an entirely different
5199 # file (which potentially could be of a different type and have different handler).
5202 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5203 [ $this, $title, &$options, &$descQuery ] );
5204 # Don't register it now, as TraditionalImageGallery does that later.
5205 $file = $this->fetchFileNoRegister( $title, $options );
5206 $handler = $file ?
$file->getHandler() : false;
5209 'img_alt' => 'gallery-internal-alt',
5210 'img_link' => 'gallery-internal-link',
5213 $paramMap +
= $handler->getParamMap();
5214 // We don't want people to specify per-image widths.
5215 // Additionally the width parameter would need special casing anyhow.
5216 unset( $paramMap['img_width'] );
5219 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5224 $handlerOptions = [];
5225 if ( isset( $matches[3] ) ) {
5226 // look for an |alt= definition while trying not to break existing
5227 // captions with multiple pipes (|) in it, until a more sensible grammar
5228 // is defined for images in galleries
5230 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5231 // splitting on '|' is a bit odd, and different from makeImage.
5232 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5233 // Protect LanguageConverter markup
5234 $parameterMatches = StringUtils
::delimiterExplode(
5235 '-{', '}-', '|', $matches[3], true /* nested */
5238 foreach ( $parameterMatches as $parameterMatch ) {
5239 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5241 $paramName = $paramMap[$magicName];
5243 switch ( $paramName ) {
5244 case 'gallery-internal-alt':
5245 $alt = $this->stripAltText( $match, false );
5247 case 'gallery-internal-link':
5248 $linkValue = $this->stripAltText( $match, false );
5249 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5250 // Result of LanguageConverter::markNoConversion
5251 // invoked on an external link.
5252 $linkValue = substr( $linkValue, 4, -2 );
5254 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5255 if ( $type === 'link-url' ) {
5257 $this->mOutput
->addExternalLink( $target );
5258 } elseif ( $type === 'link-title' ) {
5259 $link = $target->getLinkURL();
5260 $this->mOutput
->addLink( $target );
5264 // Must be a handler specific parameter.
5265 if ( $handler->validateParam( $paramName, $match ) ) {
5266 $handlerOptions[$paramName] = $match;
5268 // Guess not, consider it as caption.
5269 wfDebug( "$parameterMatch failed parameter validation\n" );
5270 $label = $parameterMatch;
5276 $label = $parameterMatch;
5281 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5283 $html = $ig->toHTML();
5284 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5289 * @param MediaHandler $handler
5292 public function getImageParams( $handler ) {
5294 $handlerClass = get_class( $handler );
5298 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5299 # Initialise static lists
5300 static $internalParamNames = [
5301 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5302 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5303 'bottom', 'text-bottom' ],
5304 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5305 'upright', 'border', 'link', 'alt', 'class' ],
5307 static $internalParamMap;
5308 if ( !$internalParamMap ) {
5309 $internalParamMap = [];
5310 foreach ( $internalParamNames as $type => $names ) {
5311 foreach ( $names as $name ) {
5312 // For grep: img_left, img_right, img_center, img_none,
5313 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5314 // img_bottom, img_text_bottom,
5315 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5316 // img_border, img_link, img_alt, img_class
5317 $magicName = str_replace( '-', '_', "img_$name" );
5318 $internalParamMap[$magicName] = [ $type, $name ];
5323 # Add handler params
5324 $paramMap = $internalParamMap;
5326 $handlerParamMap = $handler->getParamMap();
5327 foreach ( $handlerParamMap as $magic => $paramName ) {
5328 $paramMap[$magic] = [ 'handler', $paramName ];
5331 $this->mImageParams
[$handlerClass] = $paramMap;
5332 $this->mImageParamsMagicArray
[$handlerClass] =
5333 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5335 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5339 * Parse image options text and use it to make an image
5341 * @param Title $title
5342 * @param string $options
5343 * @param LinkHolderArray|bool $holders
5344 * @return string HTML
5346 public function makeImage( $title, $options, $holders = false ) {
5347 # Check if the options text is of the form "options|alt text"
5349 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5350 # * left no resizing, just left align. label is used for alt= only
5351 # * right same, but right aligned
5352 # * none same, but not aligned
5353 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5354 # * center center the image
5355 # * frame Keep original image size, no magnify-button.
5356 # * framed Same as "frame"
5357 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5358 # * upright reduce width for upright images, rounded to full __0 px
5359 # * border draw a 1px border around the image
5360 # * alt Text for HTML alt attribute (defaults to empty)
5361 # * class Set a class for img node
5362 # * link Set the target of the image link. Can be external, interwiki, or local
5363 # vertical-align values (no % or length right now):
5373 # Protect LanguageConverter markup when splitting into parts
5374 $parts = StringUtils
::delimiterExplode(
5375 '-{', '}-', '|', $options, true /* allow nesting */
5378 # Give extensions a chance to select the file revision for us
5381 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5382 [ $this, $title, &$options, &$descQuery ] );
5383 # Fetch and register the file (file title may be different via hooks)
5384 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5387 $handler = $file ?
$file->getHandler() : false;
5389 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5392 $this->addTrackingCategory( 'broken-file-category' );
5395 # Process the input parameters
5397 $params = [ 'frame' => [], 'handler' => [],
5398 'horizAlign' => [], 'vertAlign' => [] ];
5399 $seenformat = false;
5400 foreach ( $parts as $part ) {
5401 $part = trim( $part );
5402 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5404 if ( isset( $paramMap[$magicName] ) ) {
5405 list( $type, $paramName ) = $paramMap[$magicName];
5407 # Special case; width and height come in one variable together
5408 if ( $type === 'handler' && $paramName === 'width' ) {
5409 $parsedWidthParam = self
::parseWidthParam( $value );
5410 if ( isset( $parsedWidthParam['width'] ) ) {
5411 $width = $parsedWidthParam['width'];
5412 if ( $handler->validateParam( 'width', $width ) ) {
5413 $params[$type]['width'] = $width;
5417 if ( isset( $parsedWidthParam['height'] ) ) {
5418 $height = $parsedWidthParam['height'];
5419 if ( $handler->validateParam( 'height', $height ) ) {
5420 $params[$type]['height'] = $height;
5424 # else no validation -- T15436
5426 if ( $type === 'handler' ) {
5427 # Validate handler parameter
5428 $validated = $handler->validateParam( $paramName, $value );
5430 # Validate internal parameters
5431 switch ( $paramName ) {
5435 # @todo FIXME: Possibly check validity here for
5436 # manualthumb? downstream behavior seems odd with
5437 # missing manual thumbs.
5439 $value = $this->stripAltText( $value, $holders );
5442 list( $paramName, $value ) =
5443 $this->parseLinkParameter(
5444 $this->stripAltText( $value, $holders )
5448 if ( $paramName === 'no-link' ) {
5451 if ( ( $paramName === 'link-url' ) && $this->mOptions
->getExternalLinkTarget() ) {
5452 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5459 // use first appearing option, discard others.
5460 $validated = !$seenformat;
5464 # Most other things appear to be empty or numeric...
5465 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5470 $params[$type][$paramName] = $value;
5474 if ( !$validated ) {
5479 # Process alignment parameters
5480 if ( $params['horizAlign'] ) {
5481 $params['frame']['align'] = key( $params['horizAlign'] );
5483 if ( $params['vertAlign'] ) {
5484 $params['frame']['valign'] = key( $params['vertAlign'] );
5487 $params['frame']['caption'] = $caption;
5489 # Will the image be presented in a frame, with the caption below?
5490 $imageIsFramed = isset( $params['frame']['frame'] )
5491 ||
isset( $params['frame']['framed'] )
5492 ||
isset( $params['frame']['thumbnail'] )
5493 ||
isset( $params['frame']['manualthumb'] );
5495 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5496 # came to also set the caption, ordinary text after the image -- which
5497 # makes no sense, because that just repeats the text multiple times in
5498 # screen readers. It *also* came to set the title attribute.
5499 # Now that we have an alt attribute, we should not set the alt text to
5500 # equal the caption: that's worse than useless, it just repeats the
5501 # text. This is the framed/thumbnail case. If there's no caption, we
5502 # use the unnamed parameter for alt text as well, just for the time be-
5503 # ing, if the unnamed param is set and the alt param is not.
5504 # For the future, we need to figure out if we want to tweak this more,
5505 # e.g., introducing a title= parameter for the title; ignoring the un-
5506 # named parameter entirely for images without a caption; adding an ex-
5507 # plicit caption= parameter and preserving the old magic unnamed para-
5509 if ( $imageIsFramed ) { # Framed image
5510 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5511 # No caption or alt text, add the filename as the alt text so
5512 # that screen readers at least get some description of the image
5513 $params['frame']['alt'] = $title->getText();
5515 # Do not set $params['frame']['title'] because tooltips don't make sense
5517 } else { # Inline image
5518 if ( !isset( $params['frame']['alt'] ) ) {
5519 # No alt text, use the "caption" for the alt text
5520 if ( $caption !== '' ) {
5521 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5523 # No caption, fall back to using the filename for the
5525 $params['frame']['alt'] = $title->getText();
5528 # Use the "caption" for the tooltip text
5529 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5531 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5533 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5535 # Linker does the rest
5536 $time = $options['time'] ??
false;
5537 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5538 $time, $descQuery, $this->mOptions
->getThumbSize() );
5540 # Give the handler a chance to modify the parser object
5542 $handler->parserTransformHook( $this, $file );
5549 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5551 * Adds an entry to appropriate link tables.
5554 * @param string $value
5555 * @return array of `[ type, target ]`, where:
5556 * - `type` is one of:
5557 * - `null`: Given value is not a valid link target, use default
5558 * - `'no-link'`: Given value is empty, do not generate a link
5559 * - `'link-url'`: Given value is a valid external link
5560 * - `'link-title'`: Given value is a valid internal link
5562 * - When `type` is `null` or `'no-link'`: `false`
5563 * - When `type` is `'link-url'`: URL string corresponding to given value
5564 * - When `type` is `'link-title'`: Title object corresponding to given value
5566 public function parseLinkParameter( $value ) {
5567 $chars = self
::EXT_LINK_URL_CLASS
;
5568 $addr = self
::EXT_LINK_ADDR
;
5569 $prots = $this->mUrlProtocols
;
5572 if ( $value === '' ) {
5574 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5575 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5576 $this->mOutput
->addExternalLink( $value );
5581 $linkTitle = Title
::newFromText( $value );
5583 $this->mOutput
->addLink( $linkTitle );
5584 $type = 'link-title';
5585 $target = $linkTitle;
5588 return [ $type, $target ];
5592 * @param string $caption
5593 * @param LinkHolderArray|bool $holders
5594 * @return mixed|string
5596 protected function stripAltText( $caption, $holders ) {
5597 # Strip bad stuff out of the title (tooltip). We can't just use
5598 # replaceLinkHoldersText() here, because if this function is called
5599 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5601 $tooltip = $holders->replaceText( $caption );
5603 $tooltip = $this->replaceLinkHoldersText( $caption );
5606 # make sure there are no placeholders in thumbnail attributes
5607 # that are later expanded to html- so expand them now and
5609 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5610 # Compatibility hack! In HTML certain entity references not terminated
5611 # by a semicolon are decoded (but not if we're in an attribute; that's
5612 # how link URLs get away without properly escaping & in queries).
5613 # But wikitext has always required semicolon-termination of entities,
5614 # so encode & where needed to avoid decode of semicolon-less entities.
5616 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5617 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5618 $tooltip = preg_replace( "/
5619 & # 1. entity prefix
5620 (?= # 2. followed by:
5621 (?: # a. one of the legacy semicolon-less named entities
5622 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5623 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5624 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5625 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5626 U(?:acute|circ|grave|uml)|Yacute|
5627 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5628 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5629 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5630 frac(?:1(?:2|4)|34)|
5631 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5632 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5633 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5634 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5635 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5636 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5637 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5638 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5639 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5641 (?:[^;]|$)) # b. and not followed by a semicolon
5642 # S = study, for efficiency
5643 /Sx", '&', $tooltip );
5644 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5650 * Set a flag in the output object indicating that the content is dynamic and
5651 * shouldn't be cached.
5652 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5654 public function disableCache() {
5655 wfDebug( "Parser output marked as uncacheable.\n" );
5656 if ( !$this->mOutput
) {
5657 throw new MWException( __METHOD__
.
5658 " can only be called when actually parsing something" );
5660 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5664 * Callback from the Sanitizer for expanding items found in HTML attribute
5665 * values, so they can be safely tested and escaped.
5667 * @param string &$text
5668 * @param bool|PPFrame $frame
5671 public function attributeStripCallback( &$text, $frame = false ) {
5672 $text = $this->replaceVariables( $text, $frame );
5673 $text = $this->mStripState
->unstripBoth( $text );
5682 public function getTags() {
5683 $this->firstCallInit();
5685 array_keys( $this->mTransparentTagHooks
),
5686 array_keys( $this->mTagHooks
),
5687 array_keys( $this->mFunctionTagHooks
)
5695 public function getFunctionSynonyms() {
5696 $this->firstCallInit();
5697 return $this->mFunctionSynonyms
;
5704 public function getUrlProtocols() {
5705 return $this->mUrlProtocols
;
5709 * Replace transparent tags in $text with the values given by the callbacks.
5711 * Transparent tag hooks are like regular XML-style tag hooks, except they
5712 * operate late in the transformation sequence, on HTML instead of wikitext.
5714 * @param string $text
5718 public function replaceTransparentTags( $text ) {
5720 $elements = array_keys( $this->mTransparentTagHooks
);
5721 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5724 foreach ( $matches as $marker => $data ) {
5725 list( $element, $content, $params, $tag ) = $data;
5726 $tagName = strtolower( $element );
5727 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5728 $output = call_user_func_array(
5729 $this->mTransparentTagHooks
[$tagName],
5730 [ $content, $params, $this ]
5735 $replacements[$marker] = $output;
5737 return strtr( $text, $replacements );
5741 * Break wikitext input into sections, and either pull or replace
5742 * some particular section's text.
5744 * External callers should use the getSection and replaceSection methods.
5746 * @param string $text Page wikitext
5747 * @param string|int $sectionId A section identifier string of the form:
5748 * "<flag1> - <flag2> - ... - <section number>"
5750 * Currently the only recognised flag is "T", which means the target section number
5751 * was derived during a template inclusion parse, in other words this is a template
5752 * section edit link. If no flags are given, it was an ordinary section edit link.
5753 * This flag is required to avoid a section numbering mismatch when a section is
5754 * enclosed by "<includeonly>" (T8563).
5756 * The section number 0 pulls the text before the first heading; other numbers will
5757 * pull the given section along with its lower-level subsections. If the section is
5758 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5760 * Section 0 is always considered to exist, even if it only contains the empty
5761 * string. If $text is the empty string and section 0 is replaced, $newText is
5764 * @param string $mode One of "get" or "replace"
5765 * @param string $newText Replacement text for section data.
5766 * @return string For "get", the extracted section text.
5767 * for "replace", the whole page with the section replaced.
5769 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5770 global $wgTitle; # not generally used but removes an ugly failure mode
5772 $magicScopeVariable = $this->lock();
5773 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5775 $frame = $this->getPreprocessor()->newFrame();
5777 # Process section extraction flags
5779 $sectionParts = explode( '-', $sectionId );
5780 $sectionIndex = array_pop( $sectionParts );
5781 foreach ( $sectionParts as $part ) {
5782 if ( $part === 'T' ) {
5783 $flags |
= self
::PTD_FOR_INCLUSION
;
5787 # Check for empty input
5788 if ( strval( $text ) === '' ) {
5789 # Only sections 0 and T-0 exist in an empty document
5790 if ( $sectionIndex == 0 ) {
5791 if ( $mode === 'get' ) {
5797 if ( $mode === 'get' ) {
5805 # Preprocess the text
5806 $root = $this->preprocessToDom( $text, $flags );
5808 # <h> nodes indicate section breaks
5809 # They can only occur at the top level, so we can find them by iterating the root's children
5810 $node = $root->getFirstChild();
5812 # Find the target section
5813 if ( $sectionIndex == 0 ) {
5814 # Section zero doesn't nest, level=big
5815 $targetLevel = 1000;
5818 if ( $node->getName() === 'h' ) {
5819 $bits = $node->splitHeading();
5820 if ( $bits['i'] == $sectionIndex ) {
5821 $targetLevel = $bits['level'];
5825 if ( $mode === 'replace' ) {
5826 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5828 $node = $node->getNextSibling();
5834 if ( $mode === 'get' ) {
5841 # Find the end of the section, including nested sections
5843 if ( $node->getName() === 'h' ) {
5844 $bits = $node->splitHeading();
5845 $curLevel = $bits['level'];
5846 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5850 if ( $mode === 'get' ) {
5851 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5853 $node = $node->getNextSibling();
5856 # Write out the remainder (in replace mode only)
5857 if ( $mode === 'replace' ) {
5858 # Output the replacement text
5859 # Add two newlines on -- trailing whitespace in $newText is conventionally
5860 # stripped by the editor, so we need both newlines to restore the paragraph gap
5861 # Only add trailing whitespace if there is newText
5862 if ( $newText != "" ) {
5863 $outText .= $newText . "\n\n";
5867 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5868 $node = $node->getNextSibling();
5872 if ( is_string( $outText ) ) {
5873 # Re-insert stripped tags
5874 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5881 * This function returns the text of a section, specified by a number ($section).
5882 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5883 * the first section before any such heading (section 0).
5885 * If a section contains subsections, these are also returned.
5887 * @param string $text Text to look in
5888 * @param string|int $sectionId Section identifier as a number or string
5889 * (e.g. 0, 1 or 'T-1').
5890 * @param string $defaultText Default to return if section is not found
5892 * @return string Text of the requested section
5894 public function getSection( $text, $sectionId, $defaultText = '' ) {
5895 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5899 * This function returns $oldtext after the content of the section
5900 * specified by $section has been replaced with $text. If the target
5901 * section does not exist, $oldtext is returned unchanged.
5903 * @param string $oldText Former text of the article
5904 * @param string|int $sectionId Section identifier as a number or string
5905 * (e.g. 0, 1 or 'T-1').
5906 * @param string $newText Replacing text
5908 * @return string Modified text
5910 public function replaceSection( $oldText, $sectionId, $newText ) {
5911 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5915 * Get the ID of the revision we are parsing
5917 * The return value will be either:
5918 * - a) Positive, indicating a specific revision ID (current or old)
5919 * - b) Zero, meaning the revision ID specified by getCurrentRevisionCallback()
5920 * - c) Null, meaning the parse is for preview mode and there is no revision
5924 public function getRevisionId() {
5925 return $this->mRevisionId
;
5929 * Get the revision object for $this->mRevisionId
5931 * @return Revision|null Either a Revision object or null
5932 * @since 1.23 (public since 1.23)
5934 public function getRevisionObject() {
5935 if ( !is_null( $this->mRevisionObject
) ) {
5936 return $this->mRevisionObject
;
5939 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5940 // This is useful when parsing revision that has not yet been saved.
5941 // However, if we get back a saved revision even though we are in
5942 // preview mode, we'll have to ignore it, see below.
5943 // NOTE: This callback may be used to inject an OLD revision that was
5944 // already loaded, so "current" is a bit of a misnomer. We can't just
5945 // skip it if mRevisionId is set.
5946 $rev = call_user_func(
5947 $this->mOptions
->getCurrentRevisionCallback(), $this->getTitle(), $this
5950 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5951 // We are in preview mode (mRevisionId is null), and the current revision callback
5952 // returned an existing revision. Ignore it and return null, it's probably the page's
5953 // current revision, which is not what we want here. Note that we do want to call the
5954 // callback to allow the unsaved revision to be injected here, e.g. for
5955 // self-transclusion previews.
5959 // If the parse is for a new revision, then the callback should have
5960 // already been set to force the object and should match mRevisionId.
5961 // If not, try to fetch by mRevisionId for sanity.
5962 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5963 $rev = Revision
::newFromId( $this->mRevisionId
);
5966 $this->mRevisionObject
= $rev;
5968 return $this->mRevisionObject
;
5972 * Get the timestamp associated with the current revision, adjusted for
5973 * the default server-local timestamp
5976 public function getRevisionTimestamp() {
5977 if ( is_null( $this->mRevisionTimestamp
) ) {
5978 $revObject = $this->getRevisionObject();
5979 $timestamp = $revObject ?
$revObject->getTimestamp() : wfTimestampNow();
5981 # The cryptic '' timezone parameter tells to use the site-default
5982 # timezone offset instead of the user settings.
5983 # Since this value will be saved into the parser cache, served
5984 # to other users, and potentially even used inside links and such,
5985 # it needs to be consistent for all visitors.
5986 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
5988 return $this->mRevisionTimestamp
;
5992 * Get the name of the user that edited the last revision
5994 * @return string User name
5996 public function getRevisionUser() {
5997 if ( is_null( $this->mRevisionUser
) ) {
5998 $revObject = $this->getRevisionObject();
6000 # if this template is subst: the revision id will be blank,
6001 # so just use the current user's name
6003 $this->mRevisionUser
= $revObject->getUserText();
6004 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
6005 $this->mRevisionUser
= $this->getUser()->getName();
6008 return $this->mRevisionUser
;
6012 * Get the size of the revision
6014 * @return int|null Revision size
6016 public function getRevisionSize() {
6017 if ( is_null( $this->mRevisionSize
) ) {
6018 $revObject = $this->getRevisionObject();
6020 # if this variable is subst: the revision id will be blank,
6021 # so just use the parser input size, because the own substituation
6022 # will change the size.
6024 $this->mRevisionSize
= $revObject->getSize();
6026 $this->mRevisionSize
= $this->mInputSize
;
6029 return $this->mRevisionSize
;
6033 * Mutator for $mDefaultSort
6035 * @param string $sort New value
6037 public function setDefaultSort( $sort ) {
6038 $this->mDefaultSort
= $sort;
6039 $this->mOutput
->setProperty( 'defaultsort', $sort );
6043 * Accessor for $mDefaultSort
6044 * Will use the empty string if none is set.
6046 * This value is treated as a prefix, so the
6047 * empty string is equivalent to sorting by
6052 public function getDefaultSort() {
6053 if ( $this->mDefaultSort
!== false ) {
6054 return $this->mDefaultSort
;
6061 * Accessor for $mDefaultSort
6062 * Unlike getDefaultSort(), will return false if none is set
6064 * @return string|bool
6066 public function getCustomDefaultSort() {
6067 return $this->mDefaultSort
;
6070 private static function getSectionNameFromStrippedText( $text ) {
6071 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
6072 $text = Sanitizer
::decodeCharReferences( $text );
6073 $text = self
::normalizeSectionName( $text );
6077 private static function makeAnchor( $sectionName ) {
6078 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
6081 private function makeLegacyAnchor( $sectionName ) {
6082 $fragmentMode = $this->svcOptions
->get( 'FragmentMode' );
6083 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6084 // ForAttribute() and ForLink() are the same for legacy encoding
6085 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6087 $id = Sanitizer
::escapeIdForLink( $sectionName );
6094 * Try to guess the section anchor name based on a wikitext fragment
6095 * presumably extracted from a heading, for example "Header" from
6098 * @param string $text
6099 * @return string Anchor (starting with '#')
6101 public function guessSectionNameFromWikiText( $text ) {
6102 # Strip out wikitext links(they break the anchor)
6103 $text = $this->stripSectionName( $text );
6104 $sectionName = self
::getSectionNameFromStrippedText( $text );
6105 return self
::makeAnchor( $sectionName );
6109 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6110 * instead, if possible. For use in redirects, since various versions
6111 * of Microsoft browsers interpret Location: headers as something other
6112 * than UTF-8, resulting in breakage.
6114 * @param string $text The section name
6115 * @return string Anchor (starting with '#')
6117 public function guessLegacySectionNameFromWikiText( $text ) {
6118 # Strip out wikitext links(they break the anchor)
6119 $text = $this->stripSectionName( $text );
6120 $sectionName = self
::getSectionNameFromStrippedText( $text );
6121 return $this->makeLegacyAnchor( $sectionName );
6125 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6126 * @param string $text Section name (plain text)
6127 * @return string Anchor (starting with '#')
6129 public static function guessSectionNameFromStrippedText( $text ) {
6130 $sectionName = self
::getSectionNameFromStrippedText( $text );
6131 return self
::makeAnchor( $sectionName );
6135 * Apply the same normalization as code making links to this section would
6137 * @param string $text
6140 private static function normalizeSectionName( $text ) {
6141 # T90902: ensure the same normalization is applied for IDs as to links
6142 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6145 $parts = $titleParser->splitTitleString( "#$text" );
6146 } catch ( MalformedTitleException
$ex ) {
6149 return $parts['fragment'];
6153 * Strips a text string of wikitext for use in a section anchor
6155 * Accepts a text string and then removes all wikitext from the
6156 * string and leaves only the resultant text (i.e. the result of
6157 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6158 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6159 * to create valid section anchors by mimicing the output of the
6160 * parser when headings are parsed.
6162 * @param string $text Text string to be stripped of wikitext
6163 * for use in a Section anchor
6164 * @return string Filtered text string
6166 public function stripSectionName( $text ) {
6167 # Strip internal link markup
6168 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6169 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6171 # Strip external link markup
6172 # @todo FIXME: Not tolerant to blank link text
6173 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6174 # on how many empty links there are on the page - need to figure that out.
6175 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6177 # Parse wikitext quotes (italics & bold)
6178 $text = $this->doQuotes( $text );
6181 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6186 * strip/replaceVariables/unstrip for preprocessor regression testing
6188 * @param string $text
6189 * @param Title $title
6190 * @param ParserOptions $options
6191 * @param int $outputType
6195 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6196 $outputType = self
::OT_HTML
6198 $magicScopeVariable = $this->lock();
6199 $this->startParse( $title, $options, $outputType, true );
6201 $text = $this->replaceVariables( $text );
6202 $text = $this->mStripState
->unstripBoth( $text );
6203 $text = Sanitizer
::removeHTMLtags( $text );
6208 * @param string $text
6209 * @param Title $title
6210 * @param ParserOptions $options
6213 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6214 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6218 * @param string $text
6219 * @param Title $title
6220 * @param ParserOptions $options
6223 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6224 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6228 * Call a callback function on all regions of the given text that are not
6229 * inside strip markers, and replace those regions with the return value
6230 * of the callback. For example, with input:
6234 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6235 * two strings will be replaced with the value returned by the callback in
6239 * @param callable $callback
6243 public function markerSkipCallback( $s, $callback ) {
6246 while ( $i < strlen( $s ) ) {
6247 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6248 if ( $markerStart === false ) {
6249 $out .= call_user_func( $callback, substr( $s, $i ) );
6252 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6253 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6254 if ( $markerEnd === false ) {
6255 $out .= substr( $s, $markerStart );
6258 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6259 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6268 * Remove any strip markers found in the given text.
6270 * @param string $text
6273 public function killMarkers( $text ) {
6274 return $this->mStripState
->killMarkers( $text );
6278 * Save the parser state required to convert the given half-parsed text to
6279 * HTML. "Half-parsed" in this context means the output of
6280 * recursiveTagParse() or internalParse(). This output has strip markers
6281 * from replaceVariables (extensionSubstitution() etc.), and link
6282 * placeholders from replaceLinkHolders().
6284 * Returns an array which can be serialized and stored persistently. This
6285 * array can later be loaded into another parser instance with
6286 * unserializeHalfParsedText(). The text can then be safely incorporated into
6287 * the return value of a parser hook.
6289 * @deprecated since 1.31
6290 * @param string $text
6294 public function serializeHalfParsedText( $text ) {
6295 wfDeprecated( __METHOD__
, '1.31' );
6298 'version' => self
::HALF_PARSED_VERSION
,
6299 'stripState' => $this->mStripState
->getSubState( $text ),
6300 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6306 * Load the parser state given in the $data array, which is assumed to
6307 * have been generated by serializeHalfParsedText(). The text contents is
6308 * extracted from the array, and its markers are transformed into markers
6309 * appropriate for the current Parser instance. This transformed text is
6310 * returned, and can be safely included in the return value of a parser
6313 * If the $data array has been stored persistently, the caller should first
6314 * check whether it is still valid, by calling isValidHalfParsedText().
6316 * @deprecated since 1.31
6317 * @param array $data Serialized data
6318 * @throws MWException
6321 public function unserializeHalfParsedText( $data ) {
6322 wfDeprecated( __METHOD__
, '1.31' );
6323 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6324 throw new MWException( __METHOD__
. ': invalid version' );
6327 # First, extract the strip state.
6328 $texts = [ $data['text'] ];
6329 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6331 # Now renumber links
6332 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6334 # Should be good to go.
6339 * Returns true if the given array, presumed to be generated by
6340 * serializeHalfParsedText(), is compatible with the current version of the
6343 * @deprecated since 1.31
6344 * @param array $data
6348 public function isValidHalfParsedText( $data ) {
6349 wfDeprecated( __METHOD__
, '1.31' );
6350 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6354 * Parsed a width param of imagelink like 300px or 200x300px
6356 * @param string $value
6357 * @param bool $parseHeight
6362 public static function parseWidthParam( $value, $parseHeight = true ) {
6363 $parsedWidthParam = [];
6364 if ( $value === '' ) {
6365 return $parsedWidthParam;
6368 # (T15500) In both cases (width/height and width only),
6369 # permit trailing "px" for backward compatibility.
6370 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6371 $width = intval( $m[1] );
6372 $height = intval( $m[2] );
6373 $parsedWidthParam['width'] = $width;
6374 $parsedWidthParam['height'] = $height;
6375 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6376 $width = intval( $value );
6377 $parsedWidthParam['width'] = $width;
6379 return $parsedWidthParam;
6383 * Lock the current instance of the parser.
6385 * This is meant to stop someone from calling the parser
6386 * recursively and messing up all the strip state.
6388 * @throws MWException If parser is in a parse
6389 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6391 protected function lock() {
6392 if ( $this->mInParse
) {
6393 throw new MWException( "Parser state cleared while parsing. "
6394 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6397 // Save the backtrace when locking, so that if some code tries locking again,
6398 // we can print the lock owner's backtrace for easier debugging
6400 $this->mInParse
= $e->getTraceAsString();
6402 $recursiveCheck = new ScopedCallback( function () {
6403 $this->mInParse
= false;
6406 return $recursiveCheck;
6410 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6412 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6413 * or if there is more than one <p/> tag in the input HTML.
6415 * @param string $html
6419 public static function stripOuterParagraph( $html ) {
6421 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6429 * Return this parser if it is not doing anything, otherwise
6430 * get a fresh parser. You can use this method by doing
6431 * $newParser = $oldParser->getFreshParser(), or more simply
6432 * $oldParser->getFreshParser()->parse( ... );
6433 * if you're unsure if $oldParser is safe to use.
6436 * @return Parser A parser object that is not parsing anything
6438 public function getFreshParser() {
6439 if ( $this->mInParse
) {
6440 return $this->factory
->create();
6447 * Set's up the PHP implementation of OOUI for use in this request
6448 * and instructs OutputPage to enable OOUI for itself.
6452 public function enableOOUI() {
6453 OutputPage
::setupOOUI();
6454 $this->mOutput
->setEnableOOUI( true );