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\Linker\LinkRenderer
;
24 use MediaWiki\MediaWikiServices
;
25 use MediaWiki\Special\SpecialPageFactory
;
26 use Wikimedia\ScopedCallback
;
29 * @defgroup Parser Parser
33 * PHP Parser - Processes wiki markup (which uses a more user-friendly
34 * syntax, such as "[[link]]" for making links), and provides a one-way
35 * transformation of that wiki markup it into (X)HTML output / markup
36 * (which in turn the browser understands, and can display).
38 * There are seven main entry points into the Parser class:
41 * produces HTML output
42 * - Parser::preSaveTransform()
43 * produces altered wiki markup
44 * - Parser::preprocess()
45 * removes HTML comments and expands templates
46 * - Parser::cleanSig() and Parser::cleanSigInSig()
47 * cleans a signature before saving it to preferences
48 * - Parser::getSection()
49 * return the content of a section from an article for section editing
50 * - Parser::replaceSection()
51 * replaces a section by number inside an article
52 * - Parser::getPreloadText()
53 * removes <noinclude> sections and <includeonly> tags
55 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 * $wgNamespacesWithSubpages
60 * @par Settings only within ParserOptions:
61 * $wgAllowExternalImages
62 * $wgAllowSpecialInclusion
70 * Update this version number when the ParserOutput format
71 * changes in an incompatible way, so the parser cache
72 * can automatically discard old data.
74 const VERSION
= '1.6.4';
77 * Update this version number when the output of serialiseHalfParsedText()
78 * changes in an incompatible way
80 const HALF_PARSED_VERSION
= 2;
82 # Flags for Parser::setFunctionHook
83 const SFH_NO_HASH
= 1;
84 const SFH_OBJECT_ARGS
= 2;
86 # Constants needed for external link processing
87 # Everything except bracket, space, or control characters
88 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
89 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
90 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
91 # uses to replace invalid HTML characters.
92 const EXT_LINK_URL_CLASS
= '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
93 # Simplified expression to match an IPv4 or IPv6 address, or
94 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
95 const EXT_LINK_ADDR
= '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
96 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
97 // phpcs:ignore Generic.Files.LineLength
98 const EXT_IMAGE_REGEX
= '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
99 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL
= '(?:\t| |&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION
= 1;
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML
= 1; # like parse()
110 const OT_WIKI
= 2; # like preSaveTransform()
111 const OT_PREPROCESS
= 3; # like preprocess()
113 const OT_PLAIN
= 4; # like extractSections() - portions of the original are returned unchanged.
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
132 const MARKER_SUFFIX
= "-QINU`\"'\x7f";
133 const MARKER_PREFIX
= "\x7f'\"`UNIQ-";
135 # Markers used for wrapping the table of contents
136 const TOC_START
= '<mw:toc>';
137 const TOC_END
= '</mw:toc>';
139 /** @var int Assume that no output will later be saved this many seconds after parsing */
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
155 * @var bool Whether firstCallInit still needs to be called
157 public $mFirstCall = true;
159 # Initialised by initialiseVariables()
162 * @var MagicWordArray
167 * @var MagicWordArray
170 # Initialised in constructor
171 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
173 # Initialized in getPreprocessor()
174 /** @var Preprocessor */
175 public $mPreprocessor;
177 # Cleared with clearState():
189 public $mIncludeCount;
191 * @var LinkHolderArray
193 public $mLinkHolders;
196 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
197 public $mDefaultSort;
198 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
199 public $mExpensiveFunctionCount; # number of expensive parser function calls
200 public $mShowToc, $mForceTocPosition;
205 public $mUser; # User object; only used when doing pre-save transform
208 # These are variables reset at least once per parse regardless of $clearState
218 public $mTitle; # Title context, used for self-link rendering and similar things
219 public $mOutputType; # Output type, one of the OT_xxx constants
220 public $ot; # Shortcut alias, see setOutputType()
221 public $mRevisionObject; # The revision object of the specified revision ID
222 public $mRevisionId; # ID to display in {{REVISIONID}} tags
223 public $mRevisionTimestamp; # The timestamp of the specified revision ID
224 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
225 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
226 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
227 public $mInputSize = false; # For {{PAGESIZE}} on current page.
230 * @var string Deprecated accessor for the strip marker prefix.
231 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
233 public $mUniqPrefix = self
::MARKER_PREFIX
;
236 * @var array Array with the language name of each language link (i.e. the
237 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
238 * duplicate language links to the ParserOutput.
240 public $mLangLinkLanguages;
243 * @var MapCacheLRU|null
246 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
248 public $currentRevisionCache;
251 * @var bool|string Recursive call protection.
252 * This variable should be treated as if it were private.
254 public $mInParse = false;
256 /** @var SectionProfiler */
257 protected $mProfiler;
262 protected $mLinkRenderer;
264 /** @var MagicWordFactory */
265 private $magicWordFactory;
270 /** @var ParserFactory */
273 /** @var SpecialPageFactory */
274 private $specialPageFactory;
280 * @param array $parserConf See $wgParserConf documentation
281 * @param MagicWordFactory|null $magicWordFactory
282 * @param Language|null $contLang Content language
283 * @param ParserFactory|null $factory
284 * @param string|null $urlProtocols As returned from wfUrlProtocols()
285 * @param SpecialPageFactory|null $spFactory
286 * @param Config|null $siteConfig
288 public function __construct(
289 array $parserConf = [], MagicWordFactory
$magicWordFactory = null,
290 Language
$contLang = null, ParserFactory
$factory = null, $urlProtocols = null,
291 SpecialPageFactory
$spFactory = null, Config
$siteConfig = null
293 $this->mConf
= $parserConf;
294 $this->mUrlProtocols
= $urlProtocols ??
wfUrlProtocols();
295 $this->mExtLinkBracketedRegex
= '/\[(((?i)' . $this->mUrlProtocols
. ')' .
296 self
::EXT_LINK_ADDR
.
297 self
::EXT_LINK_URL_CLASS
. '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
298 if ( isset( $parserConf['preprocessorClass'] ) ) {
299 $this->mPreprocessorClass
= $parserConf['preprocessorClass'];
300 } elseif ( wfIsHHVM() ) {
301 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
302 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
303 } elseif ( extension_loaded( 'domxml' ) ) {
304 # PECL extension that conflicts with the core DOM extension (T15770)
305 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
306 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
307 } elseif ( extension_loaded( 'dom' ) ) {
308 $this->mPreprocessorClass
= Preprocessor_DOM
::class;
310 $this->mPreprocessorClass
= Preprocessor_Hash
::class;
312 wfDebug( __CLASS__
. ": using preprocessor: {$this->mPreprocessorClass}\n" );
314 $services = MediaWikiServices
::getInstance();
315 $this->magicWordFactory
= $magicWordFactory ??
316 $services->getMagicWordFactory();
318 $this->contLang
= $contLang ??
$services->getContentLanguage();
320 $this->factory
= $factory ??
$services->getParserFactory();
321 $this->specialPageFactory
= $spFactory ??
$services->getSpecialPageFactory();
322 $this->siteConfig
= $siteConfig ?? MediaWikiServices
::getInstance()->getMainConfig();
326 * Reduce memory usage to reduce the impact of circular references
328 public function __destruct() {
329 if ( isset( $this->mLinkHolders
) ) {
330 unset( $this->mLinkHolders
);
332 foreach ( $this as $name => $value ) {
333 unset( $this->$name );
338 * Allow extensions to clean up when the parser is cloned
340 public function __clone() {
341 $this->mInParse
= false;
343 // T58226: When you create a reference "to" an object field, that
344 // makes the object field itself be a reference too (until the other
345 // reference goes out of scope). When cloning, any field that's a
346 // reference is copied as a reference in the new object. Both of these
347 // are defined PHP5 behaviors, as inconvenient as it is for us when old
348 // hooks from PHP4 days are passing fields by reference.
349 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
350 // Make a non-reference copy of the field, then rebind the field to
351 // reference the new copy.
357 Hooks
::run( 'ParserCloned', [ $this ] );
361 * Do various kinds of initialisation on the first call of the parser
363 public function firstCallInit() {
364 if ( !$this->mFirstCall
) {
367 $this->mFirstCall
= false;
369 CoreParserFunctions
::register( $this );
370 CoreTagHooks
::register( $this );
371 $this->initialiseVariables();
373 // Avoid PHP 7.1 warning from passing $this by reference
375 Hooks
::run( 'ParserFirstCallInit', [ &$parser ] );
383 public function clearState() {
384 $this->firstCallInit();
385 $this->mOutput
= new ParserOutput
;
386 $this->mOptions
->registerWatcher( [ $this->mOutput
, 'recordOption' ] );
387 $this->mAutonumber
= 0;
388 $this->mIncludeCount
= [];
389 $this->mLinkHolders
= new LinkHolderArray( $this );
391 $this->mRevisionObject
= $this->mRevisionTimestamp
=
392 $this->mRevisionId
= $this->mRevisionUser
= $this->mRevisionSize
= null;
393 $this->mVarCache
= [];
395 $this->mLangLinkLanguages
= [];
396 $this->currentRevisionCache
= null;
398 $this->mStripState
= new StripState( $this );
400 # Clear these on every parse, T6549
401 $this->mTplRedirCache
= $this->mTplDomCache
= [];
403 $this->mShowToc
= true;
404 $this->mForceTocPosition
= false;
405 $this->mIncludeSizes
= [
409 $this->mPPNodeCount
= 0;
410 $this->mGeneratedPPNodeCount
= 0;
411 $this->mHighestExpansionDepth
= 0;
412 $this->mDefaultSort
= false;
413 $this->mHeadings
= [];
414 $this->mDoubleUnderscores
= [];
415 $this->mExpensiveFunctionCount
= 0;
418 if ( isset( $this->mPreprocessor
) && $this->mPreprocessor
->parser
!== $this ) {
419 $this->mPreprocessor
= null;
422 $this->mProfiler
= new SectionProfiler();
424 // Avoid PHP 7.1 warning from passing $this by reference
426 Hooks
::run( 'ParserClearState', [ &$parser ] );
430 * Convert wikitext to HTML
431 * Do not call this function recursively.
433 * @param string $text Text we want to parse
434 * @param-taint $text escapes_htmlnoent
435 * @param Title $title
436 * @param ParserOptions $options
437 * @param bool $linestart
438 * @param bool $clearState
439 * @param int|null $revid Number to pass in {{REVISIONID}}
440 * @return ParserOutput A ParserOutput
441 * @return-taint escaped
443 public function parse(
444 $text, Title
$title, ParserOptions
$options,
445 $linestart = true, $clearState = true, $revid = null
448 // We use U+007F DELETE to construct strip markers, so we have to make
449 // sure that this character does not occur in the input text.
450 $text = strtr( $text, "\x7f", "?" );
451 $magicScopeVariable = $this->lock();
453 // Strip U+0000 NULL (T159174)
454 $text = str_replace( "\000", '', $text );
456 $this->startParse( $title, $options, self
::OT_HTML
, $clearState );
458 $this->currentRevisionCache
= null;
459 $this->mInputSize
= strlen( $text );
460 if ( $this->mOptions
->getEnableLimitReport() ) {
461 $this->mOutput
->resetParseStartTime();
464 $oldRevisionId = $this->mRevisionId
;
465 $oldRevisionObject = $this->mRevisionObject
;
466 $oldRevisionTimestamp = $this->mRevisionTimestamp
;
467 $oldRevisionUser = $this->mRevisionUser
;
468 $oldRevisionSize = $this->mRevisionSize
;
469 if ( $revid !== null ) {
470 $this->mRevisionId
= $revid;
471 $this->mRevisionObject
= null;
472 $this->mRevisionTimestamp
= null;
473 $this->mRevisionUser
= null;
474 $this->mRevisionSize
= null;
477 // Avoid PHP 7.1 warning from passing $this by reference
479 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
481 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
482 $text = $this->internalParse( $text );
483 Hooks
::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState
] );
485 $text = $this->internalParseHalfParsed( $text, true, $linestart );
488 * A converted title will be provided in the output object if title and
489 * content conversion are enabled, the article text does not contain
490 * a conversion-suppressing double-underscore tag, and no
491 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
492 * automatic link conversion.
494 if ( !( $options->getDisableTitleConversion()
495 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] )
496 ||
isset( $this->mDoubleUnderscores
['notitleconvert'] )
497 ||
$this->mOutput
->getDisplayTitle() !== false )
499 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
500 if ( $convruletitle ) {
501 $this->mOutput
->setTitleText( $convruletitle );
503 $titleText = $this->getTargetLanguage()->convertTitle( $title );
504 $this->mOutput
->setTitleText( $titleText );
508 # Compute runtime adaptive expiry if set
509 $this->mOutput
->finalizeAdaptiveCacheExpiry();
511 # Warn if too many heavyweight parser functions were used
512 if ( $this->mExpensiveFunctionCount
> $this->mOptions
->getExpensiveParserFunctionLimit() ) {
513 $this->limitationWarn( 'expensive-parserfunction',
514 $this->mExpensiveFunctionCount
,
515 $this->mOptions
->getExpensiveParserFunctionLimit()
519 # Information on limits, for the benefit of users who try to skirt them
520 if ( $this->mOptions
->getEnableLimitReport() ) {
521 $text .= $this->makeLimitReport();
524 # Wrap non-interface parser output in a <div> so it can be targeted
526 $class = $this->mOptions
->getWrapOutputClass();
527 if ( $class !== false && !$this->mOptions
->getInterfaceMessage() ) {
528 $this->mOutput
->addWrapperDivClass( $class );
531 $this->mOutput
->setText( $text );
533 $this->mRevisionId
= $oldRevisionId;
534 $this->mRevisionObject
= $oldRevisionObject;
535 $this->mRevisionTimestamp
= $oldRevisionTimestamp;
536 $this->mRevisionUser
= $oldRevisionUser;
537 $this->mRevisionSize
= $oldRevisionSize;
538 $this->mInputSize
= false;
539 $this->currentRevisionCache
= null;
541 return $this->mOutput
;
545 * Set the limit report data in the current ParserOutput, and return the
546 * limit report HTML comment.
550 protected function makeLimitReport() {
551 $maxIncludeSize = $this->mOptions
->getMaxIncludeSize();
553 $cpuTime = $this->mOutput
->getTimeSinceStart( 'cpu' );
554 if ( $cpuTime !== null ) {
555 $this->mOutput
->setLimitReportData( 'limitreport-cputime',
556 sprintf( "%.3f", $cpuTime )
560 $wallTime = $this->mOutput
->getTimeSinceStart( 'wall' );
561 $this->mOutput
->setLimitReportData( 'limitreport-walltime',
562 sprintf( "%.3f", $wallTime )
565 $this->mOutput
->setLimitReportData( 'limitreport-ppvisitednodes',
566 [ $this->mPPNodeCount
, $this->mOptions
->getMaxPPNodeCount() ]
568 $this->mOutput
->setLimitReportData( 'limitreport-ppgeneratednodes',
569 [ $this->mGeneratedPPNodeCount
, $this->mOptions
->getMaxGeneratedPPNodeCount() ]
571 $this->mOutput
->setLimitReportData( 'limitreport-postexpandincludesize',
572 [ $this->mIncludeSizes
['post-expand'], $maxIncludeSize ]
574 $this->mOutput
->setLimitReportData( 'limitreport-templateargumentsize',
575 [ $this->mIncludeSizes
['arg'], $maxIncludeSize ]
577 $this->mOutput
->setLimitReportData( 'limitreport-expansiondepth',
578 [ $this->mHighestExpansionDepth
, $this->mOptions
->getMaxPPExpandDepth() ]
580 $this->mOutput
->setLimitReportData( 'limitreport-expensivefunctioncount',
581 [ $this->mExpensiveFunctionCount
, $this->mOptions
->getExpensiveParserFunctionLimit() ]
584 foreach ( $this->mStripState
->getLimitReport() as list( $key, $value ) ) {
585 $this->mOutput
->setLimitReportData( $key, $value );
588 Hooks
::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput
] );
590 $limitReport = "NewPP limit report\n";
591 if ( $this->siteConfig
->get( 'ShowHostnames' ) ) {
592 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
594 $limitReport .= 'Cached time: ' . $this->mOutput
->getCacheTime() . "\n";
595 $limitReport .= 'Cache expiry: ' . $this->mOutput
->getCacheExpiry() . "\n";
596 $limitReport .= 'Dynamic content: ' .
597 ( $this->mOutput
->hasDynamicContent() ?
'true' : 'false' ) .
600 foreach ( $this->mOutput
->getLimitReportData() as $key => $value ) {
601 if ( Hooks
::run( 'ParserLimitReportFormat',
602 [ $key, &$value, &$limitReport, false, false ]
604 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
605 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
606 ->inLanguage( 'en' )->useDatabase( false );
607 if ( !$valueMsg->exists() ) {
608 $valueMsg = new RawMessage( '$1' );
610 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
611 $valueMsg->params( $value );
612 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
616 // Since we're not really outputting HTML, decode the entities and
617 // then re-encode the things that need hiding inside HTML comments.
618 $limitReport = htmlspecialchars_decode( $limitReport );
620 // Sanitize for comment. Note '‐' in the replacement is U+2010,
621 // which looks much like the problematic '-'.
622 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&' ], $limitReport );
623 $text = "\n<!-- \n$limitReport-->\n";
625 // Add on template profiling data in human/machine readable way
626 $dataByFunc = $this->mProfiler
->getFunctionStats();
627 uasort( $dataByFunc, function ( $a, $b ) {
628 return $b['real'] <=> $a['real']; // descending order
631 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
632 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
633 $item['%real'], $item['real'], $item['calls'],
634 htmlspecialchars( $item['name'] ) );
636 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
637 $text .= implode( "\n", $profileReport ) . "\n-->\n";
639 $this->mOutput
->setLimitReportData( 'limitreport-timingprofile', $profileReport );
641 // Add other cache related metadata
642 if ( $this->siteConfig
->get( 'ShowHostnames' ) ) {
643 $this->mOutput
->setLimitReportData( 'cachereport-origin', wfHostname() );
645 $this->mOutput
->setLimitReportData( 'cachereport-timestamp',
646 $this->mOutput
->getCacheTime() );
647 $this->mOutput
->setLimitReportData( 'cachereport-ttl',
648 $this->mOutput
->getCacheExpiry() );
649 $this->mOutput
->setLimitReportData( 'cachereport-transientcontent',
650 $this->mOutput
->hasDynamicContent() );
652 if ( $this->mGeneratedPPNodeCount
> $this->mOptions
->getMaxGeneratedPPNodeCount() / 10 ) {
653 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount
. ' ' .
654 $this->mTitle
->getPrefixedDBkey() );
660 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
661 * can be called from an extension tag hook.
663 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
664 * instead, which means that lists and links have not been fully parsed yet,
665 * and strip markers are still present.
667 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
669 * Use this function if you're a parser tag hook and you want to parse
670 * wikitext before or after applying additional transformations, and you
671 * intend to *return the result as hook output*, which will cause it to go
672 * through the rest of parsing process automatically.
674 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
675 * $text are not expanded
677 * @param string $text Text extension wants to have parsed
678 * @param-taint $text escapes_htmlnoent
679 * @param bool|PPFrame $frame The frame to use for expanding any template variables
680 * @return string UNSAFE half-parsed HTML
681 * @return-taint escaped
683 public function recursiveTagParse( $text, $frame = false ) {
684 // Avoid PHP 7.1 warning from passing $this by reference
686 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
687 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
688 $text = $this->internalParse( $text, false, $frame );
693 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
694 * point can be called from an extension tag hook.
696 * The output of this function is fully-parsed HTML that is safe for output.
697 * If you're a parser tag hook, you might want to use recursiveTagParse()
700 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
701 * $text are not expanded
705 * @param string $text Text extension wants to have parsed
706 * @param-taint $text escapes_htmlnoent
707 * @param bool|PPFrame $frame The frame to use for expanding any template variables
708 * @return string Fully parsed HTML
709 * @return-taint escaped
711 public function recursiveTagParseFully( $text, $frame = false ) {
712 $text = $this->recursiveTagParse( $text, $frame );
713 $text = $this->internalParseHalfParsed( $text, false );
718 * Expand templates and variables in the text, producing valid, static wikitext.
719 * Also removes comments.
720 * Do not call this function recursively.
721 * @param string $text
722 * @param Title|null $title
723 * @param ParserOptions $options
724 * @param int|null $revid
725 * @param bool|PPFrame $frame
726 * @return mixed|string
728 public function preprocess( $text, Title
$title = null,
729 ParserOptions
$options, $revid = null, $frame = false
731 $magicScopeVariable = $this->lock();
732 $this->startParse( $title, $options, self
::OT_PREPROCESS
, true );
733 if ( $revid !== null ) {
734 $this->mRevisionId
= $revid;
736 // Avoid PHP 7.1 warning from passing $this by reference
738 Hooks
::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState
] );
739 Hooks
::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState
] );
740 $text = $this->replaceVariables( $text, $frame );
741 $text = $this->mStripState
->unstripBoth( $text );
746 * Recursive parser entry point that can be called from an extension tag
749 * @param string $text Text to be expanded
750 * @param bool|PPFrame $frame The frame to use for expanding any template variables
754 public function recursivePreprocess( $text, $frame = false ) {
755 $text = $this->replaceVariables( $text, $frame );
756 $text = $this->mStripState
->unstripBoth( $text );
761 * Process the wikitext for the "?preload=" feature. (T7210)
763 * "<noinclude>", "<includeonly>" etc. are parsed as for template
764 * transclusion, comments, templates, arguments, tags hooks and parser
765 * functions are untouched.
767 * @param string $text
768 * @param Title $title
769 * @param ParserOptions $options
770 * @param array $params
773 public function getPreloadText( $text, Title
$title, ParserOptions
$options, $params = [] ) {
774 $msg = new RawMessage( $text );
775 $text = $msg->params( $params )->plain();
777 # Parser (re)initialisation
778 $magicScopeVariable = $this->lock();
779 $this->startParse( $title, $options, self
::OT_PLAIN
, true );
781 $flags = PPFrame
::NO_ARGS | PPFrame
::NO_TEMPLATES
;
782 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
783 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
784 $text = $this->mStripState
->unstripBoth( $text );
789 * Set the current user.
790 * Should only be used when doing pre-save transform.
792 * @param User|null $user User object or null (to reset)
794 public function setUser( $user ) {
795 $this->mUser
= $user;
799 * Set the context title
803 public function setTitle( $t ) {
805 $t = Title
::newFromText( 'NO TITLE' );
808 if ( $t->hasFragment() ) {
809 # Strip the fragment to avoid various odd effects
810 $this->mTitle
= $t->createFragmentTarget( '' );
817 * Accessor for the Title object
821 public function getTitle() {
822 return $this->mTitle
;
826 * Accessor/mutator for the Title object
828 * @param Title|null $x Title object or null to just get the current one
831 public function Title( $x = null ) {
832 return wfSetVar( $this->mTitle
, $x );
836 * Set the output type
838 * @param int $ot New value
840 public function setOutputType( $ot ) {
841 $this->mOutputType
= $ot;
844 'html' => $ot == self
::OT_HTML
,
845 'wiki' => $ot == self
::OT_WIKI
,
846 'pre' => $ot == self
::OT_PREPROCESS
,
847 'plain' => $ot == self
::OT_PLAIN
,
852 * Accessor/mutator for the output type
854 * @param int|null $x New value or null to just get the current one
857 public function OutputType( $x = null ) {
858 return wfSetVar( $this->mOutputType
, $x );
862 * Get the ParserOutput object
864 * @return ParserOutput
866 public function getOutput() {
867 return $this->mOutput
;
871 * Get the ParserOptions object
873 * @return ParserOptions
875 public function getOptions() {
876 return $this->mOptions
;
880 * Accessor/mutator for the ParserOptions object
882 * @param ParserOptions|null $x New value or null to just get the current one
883 * @return ParserOptions Current ParserOptions object
885 public function Options( $x = null ) {
886 return wfSetVar( $this->mOptions
, $x );
892 public function nextLinkID() {
893 return $this->mLinkID++
;
899 public function setLinkID( $id ) {
900 $this->mLinkID
= $id;
904 * Get a language object for use in parser functions such as {{FORMATNUM:}}
907 public function getFunctionLang() {
908 return $this->getTargetLanguage();
912 * Get the target language for the content being parsed. This is usually the
913 * language that the content is in.
917 * @throws MWException
920 public function getTargetLanguage() {
921 $target = $this->mOptions
->getTargetLanguage();
923 if ( $target !== null ) {
925 } elseif ( $this->mOptions
->getInterfaceMessage() ) {
926 return $this->mOptions
->getUserLangObj();
927 } elseif ( is_null( $this->mTitle
) ) {
928 throw new MWException( __METHOD__
. ': $this->mTitle is null' );
931 return $this->mTitle
->getPageLanguage();
935 * Get the language object for language conversion
936 * @deprecated since 1.32, just use getTargetLanguage()
937 * @return Language|null
939 public function getConverterLanguage() {
940 return $this->getTargetLanguage();
944 * Get a User object either from $this->mUser, if set, or from the
945 * ParserOptions object otherwise
949 public function getUser() {
950 if ( !is_null( $this->mUser
) ) {
953 return $this->mOptions
->getUser();
957 * Get a preprocessor object
959 * @return Preprocessor
961 public function getPreprocessor() {
962 if ( !isset( $this->mPreprocessor
) ) {
963 $class = $this->mPreprocessorClass
;
964 $this->mPreprocessor
= new $class( $this );
966 return $this->mPreprocessor
;
970 * Get a LinkRenderer instance to make links with
973 * @return LinkRenderer
975 public function getLinkRenderer() {
976 if ( !$this->mLinkRenderer
) {
977 $this->mLinkRenderer
= MediaWikiServices
::getInstance()
978 ->getLinkRendererFactory()->create();
979 $this->mLinkRenderer
->setStubThreshold(
980 $this->getOptions()->getStubThreshold()
984 return $this->mLinkRenderer
;
988 * Get the MagicWordFactory that this Parser is using
991 * @return MagicWordFactory
993 public function getMagicWordFactory() {
994 return $this->magicWordFactory
;
998 * Get the content language that this Parser is using
1003 public function getContentLanguage() {
1004 return $this->contLang
;
1008 * Replaces all occurrences of HTML-style comments and the given tags
1009 * in the text with a random marker and returns the next text. The output
1010 * parameter $matches will be an associative array filled with data in
1017 * [ 'param' => 'x' ],
1018 * '<element param="x">tag content</element>' ]
1021 * @param array $elements List of element names. Comments are always extracted.
1022 * @param string $text Source text string.
1023 * @param array &$matches Out parameter, Array: extracted tags
1024 * @return string Stripped text
1026 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1031 $taglist = implode( '|', $elements );
1032 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1034 while ( $text != '' ) {
1035 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE
);
1037 if ( count( $p ) < 5 ) {
1040 if ( count( $p ) > 5 ) {
1048 list( , $element, $attributes, $close, $inside ) = $p;
1051 $marker = self
::MARKER_PREFIX
. "-$element-" . sprintf( '%08X', $n++
) . self
::MARKER_SUFFIX
;
1052 $stripped .= $marker;
1054 if ( $close === '/>' ) {
1055 # Empty element tag, <tag />
1060 if ( $element === '!--' ) {
1063 $end = "/(<\\/$element\\s*>)/i";
1065 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE
);
1067 if ( count( $q ) < 3 ) {
1068 # No end tag -- let it run out to the end of the text.
1072 list( , $tail, $text ) = $q;
1076 $matches[$marker] = [ $element,
1078 Sanitizer
::decodeTagAttributes( $attributes ),
1079 "<$element$attributes$close$content$tail" ];
1085 * Get a list of strippable XML-like elements
1089 public function getStripList() {
1090 return $this->mStripList
;
1094 * Add an item to the strip state
1095 * Returns the unique tag which must be inserted into the stripped text
1096 * The tag will be replaced with the original text in unstrip()
1098 * @param string $text
1102 public function insertStripItem( $text ) {
1103 $marker = self
::MARKER_PREFIX
. "-item-{$this->mMarkerIndex}-" . self
::MARKER_SUFFIX
;
1104 $this->mMarkerIndex++
;
1105 $this->mStripState
->addGeneral( $marker, $text );
1110 * parse the wiki syntax used to render tables
1113 * @param string $text
1116 public function doTableStuff( $text ) {
1117 $lines = StringUtils
::explode( "\n", $text );
1119 $td_history = []; # Is currently a td tag open?
1120 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1121 $tr_history = []; # Is currently a tr tag open?
1122 $tr_attributes = []; # history of tr attributes
1123 $has_opened_tr = []; # Did this table open a <tr> element?
1124 $indent_level = 0; # indent level of the table
1126 foreach ( $lines as $outLine ) {
1127 $line = trim( $outLine );
1129 if ( $line === '' ) { # empty line, go to next line
1130 $out .= $outLine . "\n";
1134 $first_character = $line[0];
1135 $first_two = substr( $line, 0, 2 );
1138 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1139 # First check if we are starting a new table
1140 $indent_level = strlen( $matches[1] );
1142 $attributes = $this->mStripState
->unstripBoth( $matches[2] );
1143 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'table' );
1145 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1146 array_push( $td_history, false );
1147 array_push( $last_tag_history, '' );
1148 array_push( $tr_history, false );
1149 array_push( $tr_attributes, '' );
1150 array_push( $has_opened_tr, false );
1151 } elseif ( count( $td_history ) == 0 ) {
1152 # Don't do any of the following
1153 $out .= $outLine . "\n";
1155 } elseif ( $first_two === '|}' ) {
1156 # We are ending a table
1157 $line = '</table>' . substr( $line, 2 );
1158 $last_tag = array_pop( $last_tag_history );
1160 if ( !array_pop( $has_opened_tr ) ) {
1161 $line = "<tr><td></td></tr>{$line}";
1164 if ( array_pop( $tr_history ) ) {
1165 $line = "</tr>{$line}";
1168 if ( array_pop( $td_history ) ) {
1169 $line = "</{$last_tag}>{$line}";
1171 array_pop( $tr_attributes );
1172 if ( $indent_level > 0 ) {
1173 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1177 } elseif ( $first_two === '|-' ) {
1178 # Now we have a table row
1179 $line = preg_replace( '#^\|-+#', '', $line );
1181 # Whats after the tag is now only attributes
1182 $attributes = $this->mStripState
->unstripBoth( $line );
1183 $attributes = Sanitizer
::fixTagAttributes( $attributes, 'tr' );
1184 array_pop( $tr_attributes );
1185 array_push( $tr_attributes, $attributes );
1188 $last_tag = array_pop( $last_tag_history );
1189 array_pop( $has_opened_tr );
1190 array_push( $has_opened_tr, true );
1192 if ( array_pop( $tr_history ) ) {
1196 if ( array_pop( $td_history ) ) {
1197 $line = "</{$last_tag}>{$line}";
1201 array_push( $tr_history, false );
1202 array_push( $td_history, false );
1203 array_push( $last_tag_history, '' );
1204 } elseif ( $first_character === '|'
1205 ||
$first_character === '!'
1206 ||
$first_two === '|+'
1208 # This might be cell elements, td, th or captions
1209 if ( $first_two === '|+' ) {
1210 $first_character = '+';
1211 $line = substr( $line, 2 );
1213 $line = substr( $line, 1 );
1216 // Implies both are valid for table headings.
1217 if ( $first_character === '!' ) {
1218 $line = StringUtils
::replaceMarkup( '!!', '||', $line );
1221 # Split up multiple cells on the same line.
1222 # FIXME : This can result in improper nesting of tags processed
1223 # by earlier parser steps.
1224 $cells = explode( '||', $line );
1228 # Loop through each table cell
1229 foreach ( $cells as $cell ) {
1231 if ( $first_character !== '+' ) {
1232 $tr_after = array_pop( $tr_attributes );
1233 if ( !array_pop( $tr_history ) ) {
1234 $previous = "<tr{$tr_after}>\n";
1236 array_push( $tr_history, true );
1237 array_push( $tr_attributes, '' );
1238 array_pop( $has_opened_tr );
1239 array_push( $has_opened_tr, true );
1242 $last_tag = array_pop( $last_tag_history );
1244 if ( array_pop( $td_history ) ) {
1245 $previous = "</{$last_tag}>\n{$previous}";
1248 if ( $first_character === '|' ) {
1250 } elseif ( $first_character === '!' ) {
1252 } elseif ( $first_character === '+' ) {
1253 $last_tag = 'caption';
1258 array_push( $last_tag_history, $last_tag );
1260 # A cell could contain both parameters and data
1261 $cell_data = explode( '|', $cell, 2 );
1263 # T2553: Note that a '|' inside an invalid link should not
1264 # be mistaken as delimiting cell parameters
1265 # Bug T153140: Neither should language converter markup.
1266 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1267 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1268 } elseif ( count( $cell_data ) == 1 ) {
1269 // Whitespace in cells is trimmed
1270 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1272 $attributes = $this->mStripState
->unstripBoth( $cell_data[0] );
1273 $attributes = Sanitizer
::fixTagAttributes( $attributes, $last_tag );
1274 // Whitespace in cells is trimmed
1275 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1279 array_push( $td_history, true );
1282 $out .= $outLine . "\n";
1285 # Closing open td, tr && table
1286 while ( count( $td_history ) > 0 ) {
1287 if ( array_pop( $td_history ) ) {
1290 if ( array_pop( $tr_history ) ) {
1293 if ( !array_pop( $has_opened_tr ) ) {
1294 $out .= "<tr><td></td></tr>\n";
1297 $out .= "</table>\n";
1300 # Remove trailing line-ending (b/c)
1301 if ( substr( $out, -1 ) === "\n" ) {
1302 $out = substr( $out, 0, -1 );
1305 # special case: don't return empty table
1306 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1314 * Helper function for parse() that transforms wiki markup into half-parsed
1315 * HTML. Only called for $mOutputType == self::OT_HTML.
1319 * @param string $text The text to parse
1320 * @param-taint $text escapes_html
1321 * @param bool $isMain Whether this is being called from the main parse() function
1322 * @param PPFrame|bool $frame A pre-processor frame
1326 public function internalParse( $text, $isMain = true, $frame = false ) {
1329 // Avoid PHP 7.1 warning from passing $this by reference
1332 # Hook to suspend the parser in this state
1333 if ( !Hooks
::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState
] ) ) {
1337 # if $frame is provided, then use $frame for replacing any variables
1339 # use frame depth to infer how include/noinclude tags should be handled
1340 # depth=0 means this is the top-level document; otherwise it's an included document
1341 if ( !$frame->depth
) {
1344 $flag = self
::PTD_FOR_INCLUSION
;
1346 $dom = $this->preprocessToDom( $text, $flag );
1347 $text = $frame->expand( $dom );
1349 # if $frame is not provided, then use old-style replaceVariables
1350 $text = $this->replaceVariables( $text );
1353 Hooks
::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState
] );
1354 $text = Sanitizer
::removeHTMLtags(
1356 [ $this, 'attributeStripCallback' ],
1358 array_keys( $this->mTransparentTagHooks
),
1360 [ $this, 'addTrackingCategory' ]
1362 Hooks
::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState
] );
1364 # Tables need to come after variable replacement for things to work
1365 # properly; putting them before other transformations should keep
1366 # exciting things like link expansions from showing up in surprising
1368 $text = $this->doTableStuff( $text );
1370 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1372 $text = $this->doDoubleUnderscore( $text );
1374 $text = $this->doHeadings( $text );
1375 $text = $this->replaceInternalLinks( $text );
1376 $text = $this->doAllQuotes( $text );
1377 $text = $this->replaceExternalLinks( $text );
1379 # replaceInternalLinks may sometimes leave behind
1380 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1381 $text = str_replace( self
::MARKER_PREFIX
. 'NOPARSE', '', $text );
1383 $text = $this->doMagicLinks( $text );
1384 $text = $this->formatHeadings( $text, $origText, $isMain );
1390 * Helper function for parse() that transforms half-parsed HTML into fully
1393 * @param string $text
1394 * @param bool $isMain
1395 * @param bool $linestart
1398 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1399 $text = $this->mStripState
->unstripGeneral( $text );
1401 // Avoid PHP 7.1 warning from passing $this by reference
1405 Hooks
::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1408 # Clean up special characters, only run once, next-to-last before doBlockLevels
1409 $text = Sanitizer
::armorFrenchSpaces( $text );
1411 $text = $this->doBlockLevels( $text, $linestart );
1413 $this->replaceLinkHolders( $text );
1416 * The input doesn't get language converted if
1418 * b) Content isn't converted
1419 * c) It's a conversion table
1420 * d) it is an interface message (which is in the user language)
1422 if ( !( $this->mOptions
->getDisableContentConversion()
1423 ||
isset( $this->mDoubleUnderscores
['nocontentconvert'] ) )
1425 if ( !$this->mOptions
->getInterfaceMessage() ) {
1426 # The position of the convert() call should not be changed. it
1427 # assumes that the links are all replaced and the only thing left
1428 # is the <nowiki> mark.
1429 $text = $this->getTargetLanguage()->convert( $text );
1433 $text = $this->mStripState
->unstripNoWiki( $text );
1436 Hooks
::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1439 $text = $this->replaceTransparentTags( $text );
1440 $text = $this->mStripState
->unstripGeneral( $text );
1442 $text = Sanitizer
::normalizeCharReferences( $text );
1444 if ( MWTidy
::isEnabled() ) {
1445 if ( $this->mOptions
->getTidy() ) {
1446 $text = MWTidy
::tidy( $text );
1449 # attempt to sanitize at least some nesting problems
1450 # (T4702 and quite a few others)
1451 # This code path is buggy and deprecated!
1452 wfDeprecated( 'disabling tidy', '1.33' );
1454 # ''Something [http://www.cool.com cool''] -->
1455 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1456 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1457 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1458 # fix up an anchor inside another anchor, only
1459 # at least for a single single nested link (T5695)
1460 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1461 '\\1\\2</a>\\3</a>\\1\\4</a>',
1462 # fix div inside inline elements- doBlockLevels won't wrap a line which
1463 # contains a div, so fix it up here; replace
1464 # div with escaped text
1465 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1466 '\\1\\3<div\\5>\\6</div>\\8\\9',
1467 # remove empty italic or bold tag pairs, some
1468 # introduced by rules above
1469 '/<([bi])><\/\\1>/' => '',
1472 $text = preg_replace(
1473 array_keys( $tidyregs ),
1474 array_values( $tidyregs ),
1479 Hooks
::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1486 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1487 * magic external links.
1492 * @param string $text
1496 public function doMagicLinks( $text ) {
1497 $prots = wfUrlProtocolsWithoutProtRel();
1498 $urlChar = self
::EXT_LINK_URL_CLASS
;
1499 $addr = self
::EXT_LINK_ADDR
;
1500 $space = self
::SPACE_NOT_NL
; # non-newline space
1501 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1502 $spaces = "$space++"; # possessive match of 1 or more spaces
1503 $text = preg_replace_callback(
1505 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1506 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1507 (\b # m[3]: Free external links
1509 ($addr$urlChar*) # m[4]: Post-protocol path
1511 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1513 \bISBN $spaces ( # m[6]: ISBN, capture number
1514 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1515 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1516 [0-9Xx] # check digit
1518 )!xu", [ $this, 'magicLinkCallback' ], $text );
1523 * @throws MWException
1525 * @return string HTML
1527 public function magicLinkCallback( $m ) {
1528 if ( isset( $m[1] ) && $m[1] !== '' ) {
1531 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1534 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1535 # Free external link
1536 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1537 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1539 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1540 if ( !$this->mOptions
->getMagicRFCLinks() ) {
1545 $cssClass = 'mw-magiclink-rfc';
1546 $trackingCat = 'magiclink-tracking-rfc';
1548 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1549 if ( !$this->mOptions
->getMagicPMIDLinks() ) {
1553 $urlmsg = 'pubmedurl';
1554 $cssClass = 'mw-magiclink-pmid';
1555 $trackingCat = 'magiclink-tracking-pmid';
1558 throw new MWException( __METHOD__
. ': unrecognised match type "' .
1559 substr( $m[0], 0, 20 ) . '"' );
1561 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1562 $this->addTrackingCategory( $trackingCat );
1563 return Linker
::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle
);
1564 } elseif ( isset( $m[6] ) && $m[6] !== ''
1565 && $this->mOptions
->getMagicISBNLinks()
1569 $space = self
::SPACE_NOT_NL
; # non-newline space
1570 $isbn = preg_replace( "/$space/", ' ', $isbn );
1571 $num = strtr( $isbn, [
1576 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1577 return $this->getLinkRenderer()->makeKnownLink(
1578 SpecialPage
::getTitleFor( 'Booksources', $num ),
1581 'class' => 'internal mw-magiclink-isbn',
1582 'title' => false // suppress title attribute
1591 * Make a free external link, given a user-supplied URL
1593 * @param string $url
1594 * @param int $numPostProto
1595 * The number of characters after the protocol.
1596 * @return string HTML
1599 public function makeFreeExternalLink( $url, $numPostProto ) {
1602 # The characters '<' and '>' (which were escaped by
1603 # removeHTMLtags()) should not be included in
1604 # URLs, per RFC 2396.
1605 # Make terminate a URL as well (bug T84937)
1608 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1613 $trail = substr( $url, $m2[0][1] ) . $trail;
1614 $url = substr( $url, 0, $m2[0][1] );
1617 # Move trailing punctuation to $trail
1619 # If there is no left bracket, then consider right brackets fair game too
1620 if ( strpos( $url, '(' ) === false ) {
1624 $urlRev = strrev( $url );
1625 $numSepChars = strspn( $urlRev, $sep );
1626 # Don't break a trailing HTML entity by moving the ; into $trail
1627 # This is in hot code, so use substr_compare to avoid having to
1628 # create a new string object for the comparison
1629 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1630 # more optimization: instead of running preg_match with a $
1631 # anchor, which can be slow, do the match on the reversed
1632 # string starting at the desired offset.
1633 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1634 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1638 if ( $numSepChars ) {
1639 $trail = substr( $url, -$numSepChars ) . $trail;
1640 $url = substr( $url, 0, -$numSepChars );
1643 # Verify that we still have a real URL after trail removal, and
1644 # not just lone protocol
1645 if ( strlen( $trail ) >= $numPostProto ) {
1646 return $url . $trail;
1649 $url = Sanitizer
::cleanUrl( $url );
1651 # Is this an external image?
1652 $text = $this->maybeMakeExternalImage( $url );
1653 if ( $text === false ) {
1654 # Not an image, make a link
1655 $text = Linker
::makeExternalLink( $url,
1656 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1658 $this->getExternalLinkAttribs( $url ), $this->mTitle
);
1659 # Register it in the output object...
1660 $this->mOutput
->addExternalLink( $url );
1662 return $text . $trail;
1666 * Parse headers and return html
1670 * @param string $text
1674 public function doHeadings( $text ) {
1675 for ( $i = 6; $i >= 1; --$i ) {
1676 $h = str_repeat( '=', $i );
1677 // Trim non-newline whitespace from headings
1678 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1679 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1685 * Replace single quotes with HTML markup
1688 * @param string $text
1690 * @return string The altered text
1692 public function doAllQuotes( $text ) {
1694 $lines = StringUtils
::explode( "\n", $text );
1695 foreach ( $lines as $line ) {
1696 $outtext .= $this->doQuotes( $line ) . "\n";
1698 $outtext = substr( $outtext, 0, -1 );
1703 * Helper function for doAllQuotes()
1705 * @param string $text
1709 public function doQuotes( $text ) {
1710 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1711 $countarr = count( $arr );
1712 if ( $countarr == 1 ) {
1716 // First, do some preliminary work. This may shift some apostrophes from
1717 // being mark-up to being text. It also counts the number of occurrences
1718 // of bold and italics mark-ups.
1721 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1722 $thislen = strlen( $arr[$i] );
1723 // If there are ever four apostrophes, assume the first is supposed to
1724 // be text, and the remaining three constitute mark-up for bold text.
1725 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1726 if ( $thislen == 4 ) {
1727 $arr[$i - 1] .= "'";
1730 } elseif ( $thislen > 5 ) {
1731 // If there are more than 5 apostrophes in a row, assume they're all
1732 // text except for the last 5.
1733 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1734 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1738 // Count the number of occurrences of bold and italics mark-ups.
1739 if ( $thislen == 2 ) {
1741 } elseif ( $thislen == 3 ) {
1743 } elseif ( $thislen == 5 ) {
1749 // If there is an odd number of both bold and italics, it is likely
1750 // that one of the bold ones was meant to be an apostrophe followed
1751 // by italics. Which one we cannot know for certain, but it is more
1752 // likely to be one that has a single-letter word before it.
1753 if ( ( $numbold %
2 == 1 ) && ( $numitalics %
2 == 1 ) ) {
1754 $firstsingleletterword = -1;
1755 $firstmultiletterword = -1;
1757 for ( $i = 1; $i < $countarr; $i +
= 2 ) {
1758 if ( strlen( $arr[$i] ) == 3 ) {
1759 $x1 = substr( $arr[$i - 1], -1 );
1760 $x2 = substr( $arr[$i - 1], -2, 1 );
1761 if ( $x1 === ' ' ) {
1762 if ( $firstspace == -1 ) {
1765 } elseif ( $x2 === ' ' ) {
1766 $firstsingleletterword = $i;
1767 // if $firstsingleletterword is set, we don't
1768 // look at the other options, so we can bail early.
1771 if ( $firstmultiletterword == -1 ) {
1772 $firstmultiletterword = $i;
1778 // If there is a single-letter word, use it!
1779 if ( $firstsingleletterword > -1 ) {
1780 $arr[$firstsingleletterword] = "''";
1781 $arr[$firstsingleletterword - 1] .= "'";
1782 } elseif ( $firstmultiletterword > -1 ) {
1783 // If not, but there's a multi-letter word, use that one.
1784 $arr[$firstmultiletterword] = "''";
1785 $arr[$firstmultiletterword - 1] .= "'";
1786 } elseif ( $firstspace > -1 ) {
1787 // ... otherwise use the first one that has neither.
1788 // (notice that it is possible for all three to be -1 if, for example,
1789 // there is only one pentuple-apostrophe in the line)
1790 $arr[$firstspace] = "''";
1791 $arr[$firstspace - 1] .= "'";
1795 // Now let's actually convert our apostrophic mush to HTML!
1800 foreach ( $arr as $r ) {
1801 if ( ( $i %
2 ) == 0 ) {
1802 if ( $state === 'both' ) {
1808 $thislen = strlen( $r );
1809 if ( $thislen == 2 ) {
1810 if ( $state === 'i' ) {
1813 } elseif ( $state === 'bi' ) {
1816 } elseif ( $state === 'ib' ) {
1817 $output .= '</b></i><b>';
1819 } elseif ( $state === 'both' ) {
1820 $output .= '<b><i>' . $buffer . '</i>';
1822 } else { // $state can be 'b' or ''
1826 } elseif ( $thislen == 3 ) {
1827 if ( $state === 'b' ) {
1830 } elseif ( $state === 'bi' ) {
1831 $output .= '</i></b><i>';
1833 } elseif ( $state === 'ib' ) {
1836 } elseif ( $state === 'both' ) {
1837 $output .= '<i><b>' . $buffer . '</b>';
1839 } else { // $state can be 'i' or ''
1843 } elseif ( $thislen == 5 ) {
1844 if ( $state === 'b' ) {
1845 $output .= '</b><i>';
1847 } elseif ( $state === 'i' ) {
1848 $output .= '</i><b>';
1850 } elseif ( $state === 'bi' ) {
1851 $output .= '</i></b>';
1853 } elseif ( $state === 'ib' ) {
1854 $output .= '</b></i>';
1856 } elseif ( $state === 'both' ) {
1857 $output .= '<i><b>' . $buffer . '</b></i>';
1859 } else { // ($state == '')
1867 // Now close all remaining tags. Notice that the order is important.
1868 if ( $state === 'b' ||
$state === 'ib' ) {
1871 if ( $state === 'i' ||
$state === 'bi' ||
$state === 'ib' ) {
1874 if ( $state === 'bi' ) {
1877 // There might be lonely ''''', so make sure we have a buffer
1878 if ( $state === 'both' && $buffer ) {
1879 $output .= '<b><i>' . $buffer . '</i></b>';
1885 * Replace external links (REL)
1887 * Note: this is all very hackish and the order of execution matters a lot.
1888 * Make sure to run tests/parser/parserTests.php if you change this code.
1892 * @param string $text
1894 * @throws MWException
1897 public function replaceExternalLinks( $text ) {
1898 $bits = preg_split( $this->mExtLinkBracketedRegex
, $text, -1, PREG_SPLIT_DELIM_CAPTURE
);
1899 if ( $bits === false ) {
1900 throw new MWException( "PCRE needs to be compiled with "
1901 . "--enable-unicode-properties in order for MediaWiki to function" );
1903 $s = array_shift( $bits );
1906 while ( $i < count( $bits ) ) {
1909 $text = $bits[$i++
];
1910 $trail = $bits[$i++
];
1912 # The characters '<' and '>' (which were escaped by
1913 # removeHTMLtags()) should not be included in
1914 # URLs, per RFC 2396.
1916 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE
) ) {
1917 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1918 $url = substr( $url, 0, $m2[0][1] );
1921 # If the link text is an image URL, replace it with an <img> tag
1922 # This happened by accident in the original parser, but some people used it extensively
1923 $img = $this->maybeMakeExternalImage( $text );
1924 if ( $img !== false ) {
1930 # Set linktype for CSS
1933 # No link text, e.g. [http://domain.tld/some.link]
1934 if ( $text == '' ) {
1936 $langObj = $this->getTargetLanguage();
1937 $text = '[' . $langObj->formatNum( ++
$this->mAutonumber
) . ']';
1938 $linktype = 'autonumber';
1940 # Have link text, e.g. [http://domain.tld/some.link text]s
1942 list( $dtrail, $trail ) = Linker
::splitTrail( $trail );
1945 // Excluding protocol-relative URLs may avoid many false positives.
1946 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1947 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1950 $url = Sanitizer
::cleanUrl( $url );
1952 # Use the encoded URL
1953 # This means that users can paste URLs directly into the text
1954 # Funny characters like ö aren't valid in URLs anyway
1955 # This was changed in August 2004
1956 $s .= Linker
::makeExternalLink( $url, $text, false, $linktype,
1957 $this->getExternalLinkAttribs( $url ), $this->mTitle
) . $dtrail . $trail;
1959 # Register link in the output object.
1960 $this->mOutput
->addExternalLink( $url );
1967 * Get the rel attribute for a particular external link.
1970 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1971 * nofollow if appropriate
1972 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1973 * @return string|null Rel attribute for $url
1975 public static function getExternalLinkRel( $url = false, $title = null ) {
1976 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1977 $ns = $title ?
$title->getNamespace() : false;
1978 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1979 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1987 * Get an associative array of additional HTML attributes appropriate for a
1988 * particular external link. This currently may include rel => nofollow
1989 * (depending on configuration, namespace, and the URL's domain) and/or a
1990 * target attribute (depending on configuration).
1992 * @param string $url URL to extract the domain from for rel =>
1993 * nofollow if appropriate
1994 * @return array Associative array of HTML attributes
1996 public function getExternalLinkAttribs( $url ) {
1998 $rel = self
::getExternalLinkRel( $url, $this->mTitle
);
2000 $target = $this->mOptions
->getExternalLinkTarget();
2002 $attribs['target'] = $target;
2003 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2004 // T133507. New windows can navigate parent cross-origin.
2005 // Including noreferrer due to lacking browser
2006 // support of noopener. Eventually noreferrer should be removed.
2007 if ( $rel !== '' ) {
2010 $rel .= 'noreferrer noopener';
2013 $attribs['rel'] = $rel;
2018 * Replace unusual escape codes in a URL with their equivalent characters
2020 * This generally follows the syntax defined in RFC 3986, with special
2021 * consideration for HTTP query strings.
2023 * @param string $url
2026 public static function normalizeLinkUrl( $url ) {
2027 # Test for RFC 3986 IPv6 syntax
2028 $scheme = '[a-z][a-z0-9+.-]*:';
2029 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2030 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2031 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2032 IP
::isValid( rawurldecode( $m[1] ) )
2034 $isIPv6 = rawurldecode( $m[1] );
2039 # Make sure unsafe characters are encoded
2040 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2042 return rawurlencode( $m[0] );
2048 $end = strlen( $url );
2050 # Fragment part - 'fragment'
2051 $start = strpos( $url, '#' );
2052 if ( $start !== false && $start < $end ) {
2053 $ret = self
::normalizeUrlComponent(
2054 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2058 # Query part - 'query' minus &=+;
2059 $start = strpos( $url, '?' );
2060 if ( $start !== false && $start < $end ) {
2061 $ret = self
::normalizeUrlComponent(
2062 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2066 # Scheme and path part - 'pchar'
2067 # (we assume no userinfo or encoded colons in the host)
2068 $ret = self
::normalizeUrlComponent(
2069 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2072 if ( $isIPv6 !== false ) {
2073 $ipv6Host = "%5B({$isIPv6})%5D";
2074 $ret = preg_replace(
2075 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2084 private static function normalizeUrlComponent( $component, $unsafe ) {
2085 $callback = function ( $matches ) use ( $unsafe ) {
2086 $char = urldecode( $matches[0] );
2087 $ord = ord( $char );
2088 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2092 # Leave it escaped, but use uppercase for a-f
2093 return strtoupper( $matches[0] );
2096 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2100 * make an image if it's allowed, either through the global
2101 * option, through the exception, or through the on-wiki whitelist
2103 * @param string $url
2107 private function maybeMakeExternalImage( $url ) {
2108 $imagesfrom = $this->mOptions
->getAllowExternalImagesFrom();
2109 $imagesexception = !empty( $imagesfrom );
2111 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2112 if ( $imagesexception && is_array( $imagesfrom ) ) {
2113 $imagematch = false;
2114 foreach ( $imagesfrom as $match ) {
2115 if ( strpos( $url, $match ) === 0 ) {
2120 } elseif ( $imagesexception ) {
2121 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2123 $imagematch = false;
2126 if ( $this->mOptions
->getAllowExternalImages()
2127 ||
( $imagesexception && $imagematch )
2129 if ( preg_match( self
::EXT_IMAGE_REGEX
, $url ) ) {
2131 $text = Linker
::makeExternalImage( $url );
2134 if ( !$text && $this->mOptions
->getEnableImageWhitelist()
2135 && preg_match( self
::EXT_IMAGE_REGEX
, $url )
2137 $whitelist = explode(
2139 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2142 foreach ( $whitelist as $entry ) {
2143 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2144 if ( strpos( $entry, '#' ) === 0 ||
$entry === '' ) {
2147 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2148 # Image matches a whitelist entry
2149 $text = Linker
::makeExternalImage( $url );
2158 * Process [[ ]] wikilinks
2162 * @return string Processed text
2166 public function replaceInternalLinks( $s ) {
2167 $this->mLinkHolders
->merge( $this->replaceInternalLinks2( $s ) );
2172 * Process [[ ]] wikilinks (RIL)
2174 * @throws MWException
2175 * @return LinkHolderArray
2179 public function replaceInternalLinks2( &$s ) {
2180 static $tc = false, $e1, $e1_img;
2181 # the % is needed to support urlencoded titles as well
2183 $tc = Title
::legalChars() . '#%';
2184 # Match a link having the form [[namespace:link|alternate]]trail
2185 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2186 # Match cases where there is no "]]", which might still be images
2187 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2190 $holders = new LinkHolderArray( $this );
2192 # split the entire text string on occurrences of [[
2193 $a = StringUtils
::explode( '[[', ' ' . $s );
2194 # get the first element (all text up to first [[), and remove the space we added
2197 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2198 $s = substr( $s, 1 );
2200 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2202 if ( $useLinkPrefixExtension ) {
2203 # Match the end of a line for a word that's not followed by whitespace,
2204 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2205 $charset = $this->contLang
->linkPrefixCharset();
2206 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2209 if ( is_null( $this->mTitle
) ) {
2210 throw new MWException( __METHOD__
. ": \$this->mTitle is null\n" );
2212 $nottalk = !$this->mTitle
->isTalkPage();
2214 if ( $useLinkPrefixExtension ) {
2216 if ( preg_match( $e2, $s, $m ) ) {
2217 $first_prefix = $m[2];
2219 $first_prefix = false;
2225 $useSubpages = $this->areSubpagesAllowed();
2227 # Loop for each link
2228 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2229 # Check for excessive memory usage
2230 if ( $holders->isBig() ) {
2232 # Do the existence check, replace the link holders and clear the array
2233 $holders->replace( $s );
2237 if ( $useLinkPrefixExtension ) {
2238 if ( preg_match( $e2, $s, $m ) ) {
2239 list( , $s, $prefix ) = $m;
2244 if ( $first_prefix ) {
2245 $prefix = $first_prefix;
2246 $first_prefix = false;
2250 $might_be_img = false;
2252 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2254 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2255 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2256 # the real problem is with the $e1 regex
2258 # Still some problems for cases where the ] is meant to be outside punctuation,
2259 # and no image is in sight. See T4095.
2261 && substr( $m[3], 0, 1 ) === ']'
2262 && strpos( $text, '[' ) !== false
2264 $text .= ']'; # so that replaceExternalLinks($text) works later
2265 $m[3] = substr( $m[3], 1 );
2267 # fix up urlencoded title texts
2268 if ( strpos( $m[1], '%' ) !== false ) {
2269 # Should anchors '#' also be rejected?
2270 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2273 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2274 # Invalid, but might be an image with a link in its caption
2275 $might_be_img = true;
2277 if ( strpos( $m[1], '%' ) !== false ) {
2278 $m[1] = str_replace( [ '<', '>' ], [ '<', '>' ], rawurldecode( $m[1] ) );
2281 } else { # Invalid form; output directly
2282 $s .= $prefix . '[[' . $line;
2286 $origLink = ltrim( $m[1], ' ' );
2288 # Don't allow internal links to pages containing
2289 # PROTO: where PROTO is a valid URL protocol; these
2290 # should be external links.
2291 if ( preg_match( '/^(?i:' . $this->mUrlProtocols
. ')/', $origLink ) ) {
2292 $s .= $prefix . '[[' . $line;
2296 # Make subpage if necessary
2297 if ( $useSubpages ) {
2298 $link = $this->maybeDoSubpageLink( $origLink, $text );
2303 // \x7f isn't a default legal title char, so most likely strip
2304 // markers will force us into the "invalid form" path above. But,
2305 // just in case, let's assert that xmlish tags aren't valid in
2306 // the title position.
2307 $unstrip = $this->mStripState
->killMarkers( $link );
2308 $noMarkers = ( $unstrip === $link );
2310 $nt = $noMarkers ? Title
::newFromText( $link ) : null;
2311 if ( $nt === null ) {
2312 $s .= $prefix . '[[' . $line;
2316 $ns = $nt->getNamespace();
2317 $iw = $nt->getInterwiki();
2319 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2321 if ( $might_be_img ) { # if this is actually an invalid link
2322 if ( $ns == NS_FILE
&& $noforce ) { # but might be an image
2325 # look at the next 'line' to see if we can close it there
2327 $next_line = $a->current();
2328 if ( $next_line === false ||
$next_line === null ) {
2331 $m = explode( ']]', $next_line, 3 );
2332 if ( count( $m ) == 3 ) {
2333 # the first ]] closes the inner link, the second the image
2335 $text .= "[[{$m[0]}]]{$m[1]}";
2338 } elseif ( count( $m ) == 2 ) {
2339 # if there's exactly one ]] that's fine, we'll keep looking
2340 $text .= "[[{$m[0]}]]{$m[1]}";
2342 # if $next_line is invalid too, we need look no further
2343 $text .= '[[' . $next_line;
2348 # we couldn't find the end of this imageLink, so output it raw
2349 # but don't ignore what might be perfectly normal links in the text we've examined
2350 $holders->merge( $this->replaceInternalLinks2( $text ) );
2351 $s .= "{$prefix}[[$link|$text";
2352 # note: no $trail, because without an end, there *is* no trail
2355 } else { # it's not an image, so output it raw
2356 $s .= "{$prefix}[[$link|$text";
2357 # note: no $trail, because without an end, there *is* no trail
2362 $wasblank = ( $text == '' );
2366 # Strip off leading ':'
2367 $text = substr( $text, 1 );
2370 # T6598 madness. Handle the quotes only if they come from the alternate part
2371 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2372 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2373 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2374 $text = $this->doQuotes( $text );
2377 # Link not escaped by : , create the various objects
2378 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2381 $iw && $this->mOptions
->getInterwikiMagic() && $nottalk && (
2382 Language
::fetchLanguageName( $iw, null, 'mw' ) ||
2383 in_array( $iw, $this->siteConfig
->get( 'ExtraInterlanguageLinkPrefixes' ) )
2386 # T26502: filter duplicates
2387 if ( !isset( $this->mLangLinkLanguages
[$iw] ) ) {
2388 $this->mLangLinkLanguages
[$iw] = true;
2389 $this->mOutput
->addLanguageLink( $nt->getFullText() );
2393 * Strip the whitespace interwiki links produce, see T10897
2395 $s = rtrim( $s . $prefix ) . $trail; # T175416
2399 if ( $ns == NS_FILE
) {
2400 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle
) ) {
2402 # if no parameters were passed, $text
2403 # becomes something like "File:Foo.png",
2404 # which we don't want to pass on to the
2408 # recursively parse links inside the image caption
2409 # actually, this will parse them in any other parameters, too,
2410 # but it might be hard to fix that, and it doesn't matter ATM
2411 $text = $this->replaceExternalLinks( $text );
2412 $holders->merge( $this->replaceInternalLinks2( $text ) );
2414 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2415 $s .= $prefix . $this->armorLinks(
2416 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2419 } elseif ( $ns == NS_CATEGORY
) {
2421 * Strip the whitespace Category links produce, see T2087
2423 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2426 $sortkey = $this->getDefaultSort();
2430 $sortkey = Sanitizer
::decodeCharReferences( $sortkey );
2431 $sortkey = str_replace( "\n", '', $sortkey );
2432 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2433 $this->mOutput
->addCategory( $nt->getDBkey(), $sortkey );
2439 # Self-link checking. For some languages, variants of the title are checked in
2440 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2441 # for linking to a different variant.
2442 if ( $ns != NS_SPECIAL
&& $nt->equals( $this->mTitle
) && !$nt->hasFragment() ) {
2443 $s .= $prefix . Linker
::makeSelfLinkObj( $nt, $text, '', $trail );
2447 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2448 # @todo FIXME: Should do batch file existence checks, see comment below
2449 if ( $ns == NS_MEDIA
) {
2450 # Give extensions a chance to select the file revision for us
2453 Hooks
::run( 'BeforeParserFetchFileAndTitle',
2454 [ $this, $nt, &$options, &$descQuery ] );
2455 # Fetch and register the file (file title may be different via hooks)
2456 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2457 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2458 $s .= $prefix . $this->armorLinks(
2459 Linker
::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2463 # Some titles, such as valid special pages or files in foreign repos, should
2464 # be shown as bluelinks even though they're not included in the page table
2465 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2466 # batch file existence checks for NS_FILE and NS_MEDIA
2467 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2468 $this->mOutput
->addLink( $nt );
2469 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2471 # Links will be added to the output link list after checking
2472 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2479 * Render a forced-blue link inline; protect against double expansion of
2480 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2481 * Since this little disaster has to split off the trail text to avoid
2482 * breaking URLs in the following text without breaking trails on the
2483 * wiki links, it's been made into a horrible function.
2486 * @param string $text
2487 * @param string $trail
2488 * @param string $prefix
2489 * @return string HTML-wikitext mix oh yuck
2491 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2492 list( $inside, $trail ) = Linker
::splitTrail( $trail );
2494 if ( $text == '' ) {
2495 $text = htmlspecialchars( $nt->getPrefixedText() );
2498 $link = $this->getLinkRenderer()->makeKnownLink(
2499 $nt, new HtmlArmor( "$prefix$text$inside" )
2502 return $this->armorLinks( $link ) . $trail;
2506 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2507 * going to go through further parsing steps before inline URL expansion.
2509 * Not needed quite as much as it used to be since free links are a bit
2510 * more sensible these days. But bracketed links are still an issue.
2512 * @param string $text More-or-less HTML
2513 * @return string Less-or-more HTML with NOPARSE bits
2515 public function armorLinks( $text ) {
2516 return preg_replace( '/\b((?i)' . $this->mUrlProtocols
. ')/',
2517 self
::MARKER_PREFIX
. "NOPARSE$1", $text );
2521 * Return true if subpage links should be expanded on this page.
2524 public function areSubpagesAllowed() {
2525 # Some namespaces don't allow subpages
2526 return MWNamespace
::hasSubpages( $this->mTitle
->getNamespace() );
2530 * Handle link to subpage if necessary
2532 * @param string $target The source of the link
2533 * @param string &$text The link text, modified as necessary
2534 * @return string The full name of the link
2537 public function maybeDoSubpageLink( $target, &$text ) {
2538 return Linker
::normalizeSubpageLink( $this->mTitle
, $target, $text );
2542 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2544 * @param string $text
2545 * @param bool $linestart Whether or not this is at the start of a line.
2547 * @return string The lists rendered as HTML
2549 public function doBlockLevels( $text, $linestart ) {
2550 return BlockLevelPass
::doBlockLevels( $text, $linestart );
2554 * Return value of a magic variable (like PAGENAME)
2558 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2559 * @param bool|PPFrame $frame
2561 * @throws MWException
2564 public function getVariableValue( $index, $frame = false ) {
2565 if ( is_null( $this->mTitle
) ) {
2566 // If no title set, bad things are going to happen
2567 // later. Title should always be set since this
2568 // should only be called in the middle of a parse
2569 // operation (but the unit-tests do funky stuff)
2570 throw new MWException( __METHOD__
. ' Should only be '
2571 . ' called while parsing (no title set)' );
2574 // Avoid PHP 7.1 warning from passing $this by reference
2578 * Some of these require message or data lookups and can be
2579 * expensive to check many times.
2581 if ( Hooks
::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache
] ) ) {
2582 if ( isset( $this->mVarCache
[$index] ) ) {
2583 return $this->mVarCache
[$index];
2587 $ts = wfTimestamp( TS_UNIX
, $this->mOptions
->getTimestamp() );
2588 Hooks
::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2590 // In miser mode, disable words that always cause double-parses on page save (T137900)
2591 static $slowRevWords = [ 'revisionid' => true ]; // @TODO: 'revisiontimestamp'
2593 isset( $slowRevWords[$index] ) &&
2594 $this->siteConfig
->get( 'MiserMode' ) &&
2595 !$this->mOptions
->getInterfaceMessage() &&
2596 // @TODO: disallow this word on all namespaces
2597 MWNamespace
::isContent( $this->mTitle
->getNamespace() )
2599 return $this->mRevisionId ?
'-' : '';
2602 $pageLang = $this->getFunctionLang();
2608 case 'currentmonth':
2609 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'm' ), true );
2611 case 'currentmonth1':
2612 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'n' ), true );
2614 case 'currentmonthname':
2615 $value = $pageLang->getMonthName( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2617 case 'currentmonthnamegen':
2618 $value = $pageLang->getMonthNameGen( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2620 case 'currentmonthabbrev':
2621 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getInstance( $ts )->format( 'n' ) );
2624 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'j' ), true );
2627 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'd' ), true );
2630 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'm' ), true );
2633 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ), true );
2635 case 'localmonthname':
2636 $value = $pageLang->getMonthName( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2638 case 'localmonthnamegen':
2639 $value = $pageLang->getMonthNameGen( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2641 case 'localmonthabbrev':
2642 $value = $pageLang->getMonthAbbreviation( MWTimestamp
::getLocalInstance( $ts )->format( 'n' ) );
2645 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'j' ), true );
2648 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'd' ), true );
2651 $value = wfEscapeWikiText( $this->mTitle
->getText() );
2654 $value = wfEscapeWikiText( $this->mTitle
->getPartialURL() );
2656 case 'fullpagename':
2657 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedText() );
2659 case 'fullpagenamee':
2660 $value = wfEscapeWikiText( $this->mTitle
->getPrefixedURL() );
2663 $value = wfEscapeWikiText( $this->mTitle
->getSubpageText() );
2665 case 'subpagenamee':
2666 $value = wfEscapeWikiText( $this->mTitle
->getSubpageUrlForm() );
2668 case 'rootpagename':
2669 $value = wfEscapeWikiText( $this->mTitle
->getRootText() );
2671 case 'rootpagenamee':
2672 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2675 $this->mTitle
->getRootText()
2678 case 'basepagename':
2679 $value = wfEscapeWikiText( $this->mTitle
->getBaseText() );
2681 case 'basepagenamee':
2682 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2685 $this->mTitle
->getBaseText()
2688 case 'talkpagename':
2689 if ( $this->mTitle
->canHaveTalkPage() ) {
2690 $talkPage = $this->mTitle
->getTalkPage();
2691 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2696 case 'talkpagenamee':
2697 if ( $this->mTitle
->canHaveTalkPage() ) {
2698 $talkPage = $this->mTitle
->getTalkPage();
2699 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2704 case 'subjectpagename':
2705 $subjPage = $this->mTitle
->getSubjectPage();
2706 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2708 case 'subjectpagenamee':
2709 $subjPage = $this->mTitle
->getSubjectPage();
2710 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2712 case 'pageid': // requested in T25427
2713 $pageid = $this->getTitle()->getArticleID();
2714 if ( $pageid == 0 ) {
2715 # 0 means the page doesn't exist in the database,
2716 # which means the user is previewing a new page.
2717 # The vary-revision flag must be set, because the magic word
2718 # will have a different value once the page is saved.
2719 $this->mOutput
->setFlag( 'vary-revision' );
2720 wfDebug( __METHOD__
. ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2722 $value = $pageid ?
: null;
2725 # Let the edit saving system know we should parse the page
2726 # *after* a revision ID has been assigned.
2727 $this->mOutput
->setFlag( 'vary-revision-id' );
2728 wfDebug( __METHOD__
. ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2729 $value = $this->mRevisionId
;
2732 $rev = $this->getRevisionObject();
2734 $value = $rev->getId();
2739 $value = $this->mOptions
->getSpeculativeRevId();
2741 $this->mOutput
->setSpeculativeRevIdUsed( $value );
2746 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2748 case 'revisionday2':
2749 $value = $this->getRevisionTimestampSubstring( 6, 2, self
::MAX_TTS
, $index );
2751 case 'revisionmonth':
2752 $value = $this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2754 case 'revisionmonth1':
2755 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self
::MAX_TTS
, $index );
2757 case 'revisionyear':
2758 $value = $this->getRevisionTimestampSubstring( 0, 4, self
::MAX_TTS
, $index );
2760 case 'revisiontimestamp':
2761 # Let the edit saving system know we should parse the page
2762 # *after* a revision ID has been assigned. This is for null edits.
2763 $this->mOutput
->setFlag( 'vary-revision' );
2764 wfDebug( __METHOD__
. ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2765 $value = $this->getRevisionTimestamp();
2767 case 'revisionuser':
2768 # Let the edit saving system know we should parse the page
2769 # *after* a revision ID has been assigned for null edits.
2770 $this->mOutput
->setFlag( 'vary-user' );
2771 wfDebug( __METHOD__
. ": {{REVISIONUSER}} used, setting vary-user...\n" );
2772 $value = $this->getRevisionUser();
2774 case 'revisionsize':
2775 $value = $this->getRevisionSize();
2778 $value = str_replace( '_', ' ',
2779 $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2782 $value = wfUrlencode( $this->contLang
->getNsText( $this->mTitle
->getNamespace() ) );
2784 case 'namespacenumber':
2785 $value = $this->mTitle
->getNamespace();
2788 $value = $this->mTitle
->canHaveTalkPage()
2789 ?
str_replace( '_', ' ', $this->mTitle
->getTalkNsText() )
2793 $value = $this->mTitle
->canHaveTalkPage() ?
wfUrlencode( $this->mTitle
->getTalkNsText() ) : '';
2795 case 'subjectspace':
2796 $value = str_replace( '_', ' ', $this->mTitle
->getSubjectNsText() );
2798 case 'subjectspacee':
2799 $value = ( wfUrlencode( $this->mTitle
->getSubjectNsText() ) );
2801 case 'currentdayname':
2802 $value = $pageLang->getWeekdayName( (int)MWTimestamp
::getInstance( $ts )->format( 'w' ) +
1 );
2805 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'Y' ), true );
2808 $value = $pageLang->time( wfTimestamp( TS_MW
, $ts ), false, false );
2811 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'H' ), true );
2814 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2815 # int to remove the padding
2816 $value = $pageLang->formatNum( (int)MWTimestamp
::getInstance( $ts )->format( 'W' ) );
2819 $value = $pageLang->formatNum( MWTimestamp
::getInstance( $ts )->format( 'w' ) );
2821 case 'localdayname':
2822 $value = $pageLang->getWeekdayName(
2823 (int)MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) +
1
2827 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'Y' ), true );
2830 $value = $pageLang->time(
2831 MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' ),
2837 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'H' ), true );
2840 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2841 # int to remove the padding
2842 $value = $pageLang->formatNum( (int)MWTimestamp
::getLocalInstance( $ts )->format( 'W' ) );
2845 $value = $pageLang->formatNum( MWTimestamp
::getLocalInstance( $ts )->format( 'w' ) );
2847 case 'numberofarticles':
2848 $value = $pageLang->formatNum( SiteStats
::articles() );
2850 case 'numberoffiles':
2851 $value = $pageLang->formatNum( SiteStats
::images() );
2853 case 'numberofusers':
2854 $value = $pageLang->formatNum( SiteStats
::users() );
2856 case 'numberofactiveusers':
2857 $value = $pageLang->formatNum( SiteStats
::activeUsers() );
2859 case 'numberofpages':
2860 $value = $pageLang->formatNum( SiteStats
::pages() );
2862 case 'numberofadmins':
2863 $value = $pageLang->formatNum( SiteStats
::numberingroup( 'sysop' ) );
2865 case 'numberofedits':
2866 $value = $pageLang->formatNum( SiteStats
::edits() );
2868 case 'currenttimestamp':
2869 $value = wfTimestamp( TS_MW
, $ts );
2871 case 'localtimestamp':
2872 $value = MWTimestamp
::getLocalInstance( $ts )->format( 'YmdHis' );
2874 case 'currentversion':
2875 $value = SpecialVersion
::getVersion();
2878 return $this->siteConfig
->get( 'ArticlePath' );
2880 return $this->siteConfig
->get( 'Sitename' );
2882 return $this->siteConfig
->get( 'Server' );
2884 return $this->siteConfig
->get( 'ServerName' );
2886 return $this->siteConfig
->get( 'ScriptPath' );
2888 return $this->siteConfig
->get( 'StylePath' );
2889 case 'directionmark':
2890 return $pageLang->getDirMark();
2891 case 'contentlanguage':
2892 return $this->siteConfig
->get( 'LanguageCode' );
2893 case 'pagelanguage':
2894 $value = $pageLang->getCode();
2896 case 'cascadingsources':
2897 $value = CoreParserFunctions
::cascadingsources( $this );
2902 'ParserGetVariableValueSwitch',
2903 [ &$parser, &$this->mVarCache
, &$index, &$ret, &$frame ]
2910 $this->mVarCache
[$index] = $value;
2919 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2920 * @param string $variable Parser variable name
2923 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2924 # Get the timezone-adjusted timestamp to be used for this revision
2925 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2926 # Possibly set vary-revision if there is not yet an associated revision
2927 if ( !$this->getRevisionObject() ) {
2928 # Get the timezone-adjusted timestamp $mtts seconds in the future
2930 $this->contLang
->userAdjust( wfTimestamp( TS_MW
, time() +
$mtts ), '' ),
2935 if ( $resNow !== $resThen ) {
2936 # Let the edit saving system know we should parse the page
2937 # *after* a revision ID has been assigned. This is for null edits.
2938 $this->mOutput
->setFlag( 'vary-revision' );
2939 wfDebug( __METHOD__
. ": $variable used, setting vary-revision...\n" );
2947 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2951 public function initialiseVariables() {
2952 $variableIDs = $this->magicWordFactory
->getVariableIDs();
2953 $substIDs = $this->magicWordFactory
->getSubstIDs();
2955 $this->mVariables
= $this->magicWordFactory
->newArray( $variableIDs );
2956 $this->mSubstWords
= $this->magicWordFactory
->newArray( $substIDs );
2960 * Preprocess some wikitext and return the document tree.
2961 * This is the ghost of replace_variables().
2963 * @param string $text The text to parse
2964 * @param int $flags Bitwise combination of:
2965 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2966 * included. Default is to assume a direct page view.
2968 * The generated DOM tree must depend only on the input text and the flags.
2969 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2971 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2972 * change in the DOM tree for a given text, must be passed through the section identifier
2973 * in the section edit link and thus back to extractSections().
2975 * The output of this function is currently only cached in process memory, but a persistent
2976 * cache may be implemented at a later date which takes further advantage of these strict
2977 * dependency requirements.
2981 public function preprocessToDom( $text, $flags = 0 ) {
2982 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2987 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2993 public static function splitWhitespace( $s ) {
2994 $ltrimmed = ltrim( $s );
2995 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2996 $trimmed = rtrim( $ltrimmed );
2997 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2999 $w2 = substr( $ltrimmed, -$diff );
3003 return [ $w1, $trimmed, $w2 ];
3007 * Replace magic variables, templates, and template arguments
3008 * with the appropriate text. Templates are substituted recursively,
3009 * taking care to avoid infinite loops.
3011 * Note that the substitution depends on value of $mOutputType:
3012 * self::OT_WIKI: only {{subst:}} templates
3013 * self::OT_PREPROCESS: templates but not extension tags
3014 * self::OT_HTML: all templates and extension tags
3016 * @param string $text The text to transform
3017 * @param bool|PPFrame $frame Object describing the arguments passed to the
3018 * template. Arguments may also be provided as an associative array, as
3019 * was the usual case before MW1.12. Providing arguments this way may be
3020 * useful for extensions wishing to perform variable replacement
3022 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3023 * double-brace expansion.
3026 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3027 # Is there any text? Also, Prevent too big inclusions!
3028 $textSize = strlen( $text );
3029 if ( $textSize < 1 ||
$textSize > $this->mOptions
->getMaxIncludeSize() ) {
3033 if ( $frame === false ) {
3034 $frame = $this->getPreprocessor()->newFrame();
3035 } elseif ( !( $frame instanceof PPFrame
) ) {
3036 wfDebug( __METHOD__
. " called using plain parameters instead of "
3037 . "a PPFrame instance. Creating custom frame.\n" );
3038 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3041 $dom = $this->preprocessToDom( $text );
3042 $flags = $argsOnly ? PPFrame
::NO_TEMPLATES
: 0;
3043 $text = $frame->expand( $dom, $flags );
3049 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3051 * @param array $args
3055 public static function createAssocArgs( $args ) {
3058 foreach ( $args as $arg ) {
3059 $eqpos = strpos( $arg, '=' );
3060 if ( $eqpos === false ) {
3061 $assocArgs[$index++
] = $arg;
3063 $name = trim( substr( $arg, 0, $eqpos ) );
3064 $value = trim( substr( $arg, $eqpos +
1 ) );
3065 if ( $value === false ) {
3068 if ( $name !== false ) {
3069 $assocArgs[$name] = $value;
3078 * Warn the user when a parser limitation is reached
3079 * Will warn at most once the user per limitation type
3081 * The results are shown during preview and run through the Parser (See EditPage.php)
3083 * @param string $limitationType Should be one of:
3084 * 'expensive-parserfunction' (corresponding messages:
3085 * 'expensive-parserfunction-warning',
3086 * 'expensive-parserfunction-category')
3087 * 'post-expand-template-argument' (corresponding messages:
3088 * 'post-expand-template-argument-warning',
3089 * 'post-expand-template-argument-category')
3090 * 'post-expand-template-inclusion' (corresponding messages:
3091 * 'post-expand-template-inclusion-warning',
3092 * 'post-expand-template-inclusion-category')
3093 * 'node-count-exceeded' (corresponding messages:
3094 * 'node-count-exceeded-warning',
3095 * 'node-count-exceeded-category')
3096 * 'expansion-depth-exceeded' (corresponding messages:
3097 * 'expansion-depth-exceeded-warning',
3098 * 'expansion-depth-exceeded-category')
3099 * @param string|int|null $current Current value
3100 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3101 * exceeded, provide the values (optional)
3103 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3104 # does no harm if $current and $max are present but are unnecessary for the message
3105 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3106 # only during preview, and that would split the parser cache unnecessarily.
3107 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3109 $this->mOutput
->addWarning( $warning );
3110 $this->addTrackingCategory( "$limitationType-category" );
3114 * Return the text of a template, after recursively
3115 * replacing any variables or templates within the template.
3117 * @param array $piece The parts of the template
3118 * $piece['title']: the title, i.e. the part before the |
3119 * $piece['parts']: the parameter array
3120 * $piece['lineStart']: whether the brace was at the start of a line
3121 * @param PPFrame $frame The current frame, contains template arguments
3123 * @return string The text of the template
3125 public function braceSubstitution( $piece, $frame ) {
3128 // $text has been filled
3130 // wiki markup in $text should be escaped
3132 // $text is HTML, armour it against wikitext transformation
3134 // Force interwiki transclusion to be done in raw mode not rendered
3135 $forceRawInterwiki = false;
3136 // $text is a DOM node needing expansion in a child frame
3137 $isChildObj = false;
3138 // $text is a DOM node needing expansion in the current frame
3139 $isLocalObj = false;
3141 # Title object, where $text came from
3144 # $part1 is the bit before the first |, and must contain only title characters.
3145 # Various prefixes will be stripped from it later.
3146 $titleWithSpaces = $frame->expand( $piece['title'] );
3147 $part1 = trim( $titleWithSpaces );
3150 # Original title text preserved for various purposes
3151 $originalTitle = $part1;
3153 # $args is a list of argument nodes, starting from index 0, not including $part1
3154 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3155 # below won't work b/c this $args isn't an object
3156 $args = ( $piece['parts'] == null ) ?
[] : $piece['parts'];
3158 $profileSection = null; // profile templates
3162 $substMatch = $this->mSubstWords
->matchStartAndRemove( $part1 );
3164 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3165 # Decide whether to expand template or keep wikitext as-is.
3166 if ( $this->ot
['wiki'] ) {
3167 if ( $substMatch === false ) {
3168 $literal = true; # literal when in PST with no prefix
3170 $literal = false; # expand when in PST with subst: or safesubst:
3173 if ( $substMatch == 'subst' ) {
3174 $literal = true; # literal when not in PST with plain subst:
3176 $literal = false; # expand when not in PST with safesubst: or no prefix
3180 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3187 if ( !$found && $args->getLength() == 0 ) {
3188 $id = $this->mVariables
->matchStartToEnd( $part1 );
3189 if ( $id !== false ) {
3190 $text = $this->getVariableValue( $id, $frame );
3191 if ( $this->magicWordFactory
->getCacheTTL( $id ) > -1 ) {
3192 $this->mOutput
->updateCacheExpiry(
3193 $this->magicWordFactory
->getCacheTTL( $id ) );
3199 # MSG, MSGNW and RAW
3202 $mwMsgnw = $this->magicWordFactory
->get( 'msgnw' );
3203 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3206 # Remove obsolete MSG:
3207 $mwMsg = $this->magicWordFactory
->get( 'msg' );
3208 $mwMsg->matchStartAndRemove( $part1 );
3212 $mwRaw = $this->magicWordFactory
->get( 'raw' );
3213 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3214 $forceRawInterwiki = true;
3220 $colonPos = strpos( $part1, ':' );
3221 if ( $colonPos !== false ) {
3222 $func = substr( $part1, 0, $colonPos );
3223 $funcArgs = [ trim( substr( $part1, $colonPos +
1 ) ) ];
3224 $argsLength = $args->getLength();
3225 for ( $i = 0; $i < $argsLength; $i++
) {
3226 $funcArgs[] = $args->item( $i );
3229 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3231 // Extract any forwarded flags
3232 if ( isset( $result['title'] ) ) {
3233 $title = $result['title'];
3235 if ( isset( $result['found'] ) ) {
3236 $found = $result['found'];
3238 if ( array_key_exists( 'text', $result ) ) {
3240 $text = $result['text'];
3242 if ( isset( $result['nowiki'] ) ) {
3243 $nowiki = $result['nowiki'];
3245 if ( isset( $result['isHTML'] ) ) {
3246 $isHTML = $result['isHTML'];
3248 if ( isset( $result['forceRawInterwiki'] ) ) {
3249 $forceRawInterwiki = $result['forceRawInterwiki'];
3251 if ( isset( $result['isChildObj'] ) ) {
3252 $isChildObj = $result['isChildObj'];
3254 if ( isset( $result['isLocalObj'] ) ) {
3255 $isLocalObj = $result['isLocalObj'];
3260 # Finish mangling title and then check for loops.
3261 # Set $title to a Title object and $titleText to the PDBK
3264 # Split the title into page and subpage
3266 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3267 if ( $part1 !== $relative ) {
3269 $ns = $this->mTitle
->getNamespace();
3271 $title = Title
::newFromText( $part1, $ns );
3273 $titleText = $title->getPrefixedText();
3274 # Check for language variants if the template is not found
3275 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3276 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3278 # Do recursion depth check
3279 $limit = $this->mOptions
->getMaxTemplateDepth();
3280 if ( $frame->depth
>= $limit ) {
3282 $text = '<span class="error">'
3283 . wfMessage( 'parser-template-recursion-depth-warning' )
3284 ->numParams( $limit )->inContentLanguage()->text()
3290 # Load from database
3291 if ( !$found && $title ) {
3292 $profileSection = $this->mProfiler
->scopedProfileIn( $title->getPrefixedDBkey() );
3293 if ( !$title->isExternal() ) {
3294 if ( $title->isSpecialPage()
3295 && $this->mOptions
->getAllowSpecialInclusion()
3296 && $this->ot
['html']
3298 $specialPage = $this->specialPageFactory
->getPage( $title->getDBkey() );
3299 // Pass the template arguments as URL parameters.
3300 // "uselang" will have no effect since the Language object
3301 // is forced to the one defined in ParserOptions.
3303 $argsLength = $args->getLength();
3304 for ( $i = 0; $i < $argsLength; $i++
) {
3305 $bits = $args->item( $i )->splitArg();
3306 if ( strval( $bits['index'] ) === '' ) {
3307 $name = trim( $frame->expand( $bits['name'], PPFrame
::STRIP_COMMENTS
) );
3308 $value = trim( $frame->expand( $bits['value'] ) );
3309 $pageArgs[$name] = $value;
3313 // Create a new context to execute the special page
3314 $context = new RequestContext
;
3315 $context->setTitle( $title );
3316 $context->setRequest( new FauxRequest( $pageArgs ) );
3317 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3318 $context->setUser( $this->getUser() );
3320 // If this page is cached, then we better not be per user.
3321 $context->setUser( User
::newFromName( '127.0.0.1', false ) );
3323 $context->setLanguage( $this->mOptions
->getUserLangObj() );
3324 $ret = $this->specialPageFactory
->capturePath( $title, $context, $this->getLinkRenderer() );
3326 $text = $context->getOutput()->getHTML();
3327 $this->mOutput
->addOutputPageMetadata( $context->getOutput() );
3330 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3331 $this->mOutput
->updateRuntimeAdaptiveExpiry(
3332 $specialPage->maxIncludeCacheTime()
3336 } elseif ( MWNamespace
::isNonincludable( $title->getNamespace() ) ) {
3337 $found = false; # access denied
3338 wfDebug( __METHOD__
. ": template inclusion denied for " .
3339 $title->getPrefixedDBkey() . "\n" );
3341 list( $text, $title ) = $this->getTemplateDom( $title );
3342 if ( $text !== false ) {
3348 # If the title is valid but undisplayable, make a link to it
3349 if ( !$found && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3350 $text = "[[:$titleText]]";
3353 } elseif ( $title->isTrans() ) {
3354 # Interwiki transclusion
3355 if ( $this->ot
['html'] && !$forceRawInterwiki ) {
3356 $text = $this->interwikiTransclude( $title, 'render' );
3359 $text = $this->interwikiTransclude( $title, 'raw' );
3360 # Preprocess it like a template
3361 $text = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3367 # Do infinite loop check
3368 # This has to be done after redirect resolution to avoid infinite loops via redirects
3369 if ( !$frame->loopCheck( $title ) ) {
3371 $text = '<span class="error">'
3372 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3374 $this->addTrackingCategory( 'template-loop-category' );
3375 $this->mOutput
->addWarning( wfMessage( 'template-loop-warning',
3376 wfEscapeWikiText( $titleText ) )->text() );
3377 wfDebug( __METHOD__
. ": template loop broken at '$titleText'\n" );
3381 # If we haven't found text to substitute by now, we're done
3382 # Recover the source wikitext and return it
3384 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3385 if ( $profileSection ) {
3386 $this->mProfiler
->scopedProfileOut( $profileSection );
3388 return [ 'object' => $text ];
3391 # Expand DOM-style return values in a child frame
3392 if ( $isChildObj ) {
3393 # Clean up argument array
3394 $newFrame = $frame->newChild( $args, $title );
3397 $text = $newFrame->expand( $text, PPFrame
::RECOVER_ORIG
);
3398 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3399 # Expansion is eligible for the empty-frame cache
3400 $text = $newFrame->cachedExpand( $titleText, $text );
3402 # Uncached expansion
3403 $text = $newFrame->expand( $text );
3406 if ( $isLocalObj && $nowiki ) {
3407 $text = $frame->expand( $text, PPFrame
::RECOVER_ORIG
);
3408 $isLocalObj = false;
3411 if ( $profileSection ) {
3412 $this->mProfiler
->scopedProfileOut( $profileSection );
3415 # Replace raw HTML by a placeholder
3417 $text = $this->insertStripItem( $text );
3418 } elseif ( $nowiki && ( $this->ot
['html'] ||
$this->ot
['pre'] ) ) {
3419 # Escape nowiki-style return values
3420 $text = wfEscapeWikiText( $text );
3421 } elseif ( is_string( $text )
3422 && !$piece['lineStart']
3423 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3425 # T2529: if the template begins with a table or block-level
3426 # element, it should be treated as beginning a new line.
3427 # This behavior is somewhat controversial.
3428 $text = "\n" . $text;
3431 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3432 # Error, oversize inclusion
3433 if ( $titleText !== false ) {
3434 # Make a working, properly escaped link if possible (T25588)
3435 $text = "[[:$titleText]]";
3437 # This will probably not be a working link, but at least it may
3438 # provide some hint of where the problem is
3439 preg_replace( '/^:/', '', $originalTitle );
3440 $text = "[[:$originalTitle]]";
3442 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3443 . 'post-expand include size too large -->' );
3444 $this->limitationWarn( 'post-expand-template-inclusion' );
3447 if ( $isLocalObj ) {
3448 $ret = [ 'object' => $text ];
3450 $ret = [ 'text' => $text ];
3457 * Call a parser function and return an array with text and flags.
3459 * The returned array will always contain a boolean 'found', indicating
3460 * whether the parser function was found or not. It may also contain the
3462 * text: string|object, resulting wikitext or PP DOM object
3463 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3464 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3465 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3466 * nowiki: bool, wiki markup in $text should be escaped
3469 * @param PPFrame $frame The current frame, contains template arguments
3470 * @param string $function Function name
3471 * @param array $args Arguments to the function
3472 * @throws MWException
3475 public function callParserFunction( $frame, $function, array $args = [] ) {
3476 # Case sensitive functions
3477 if ( isset( $this->mFunctionSynonyms
[1][$function] ) ) {
3478 $function = $this->mFunctionSynonyms
[1][$function];
3480 # Case insensitive functions
3481 $function = $this->contLang
->lc( $function );
3482 if ( isset( $this->mFunctionSynonyms
[0][$function] ) ) {
3483 $function = $this->mFunctionSynonyms
[0][$function];
3485 return [ 'found' => false ];
3489 list( $callback, $flags ) = $this->mFunctionHooks
[$function];
3491 // Avoid PHP 7.1 warning from passing $this by reference
3494 $allArgs = [ &$parser ];
3495 if ( $flags & self
::SFH_OBJECT_ARGS
) {
3496 # Convert arguments to PPNodes and collect for appending to $allArgs
3498 foreach ( $args as $k => $v ) {
3499 if ( $v instanceof PPNode ||
$k === 0 ) {
3502 $funcArgs[] = $this->mPreprocessor
->newPartNodeArray( [ $k => $v ] )->item( 0 );
3506 # Add a frame parameter, and pass the arguments as an array
3507 $allArgs[] = $frame;
3508 $allArgs[] = $funcArgs;
3510 # Convert arguments to plain text and append to $allArgs
3511 foreach ( $args as $k => $v ) {
3512 if ( $v instanceof PPNode
) {
3513 $allArgs[] = trim( $frame->expand( $v ) );
3514 } elseif ( is_int( $k ) && $k >= 0 ) {
3515 $allArgs[] = trim( $v );
3517 $allArgs[] = trim( "$k=$v" );
3522 $result = $callback( ...$allArgs );
3524 # The interface for function hooks allows them to return a wikitext
3525 # string or an array containing the string and any flags. This mungs
3526 # things around to match what this method should return.
3527 if ( !is_array( $result ) ) {
3533 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3534 $result['text'] = $result[0];
3536 unset( $result[0] );
3543 $preprocessFlags = 0;
3544 if ( isset( $result['noparse'] ) ) {
3545 $noparse = $result['noparse'];
3547 if ( isset( $result['preprocessFlags'] ) ) {
3548 $preprocessFlags = $result['preprocessFlags'];
3552 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3553 $result['isChildObj'] = true;
3560 * Get the semi-parsed DOM representation of a template with a given title,
3561 * and its redirect destination title. Cached.
3563 * @param Title $title
3567 public function getTemplateDom( $title ) {
3568 $cacheTitle = $title;
3569 $titleText = $title->getPrefixedDBkey();
3571 if ( isset( $this->mTplRedirCache
[$titleText] ) ) {
3572 list( $ns, $dbk ) = $this->mTplRedirCache
[$titleText];
3573 $title = Title
::makeTitle( $ns, $dbk );
3574 $titleText = $title->getPrefixedDBkey();
3576 if ( isset( $this->mTplDomCache
[$titleText] ) ) {
3577 return [ $this->mTplDomCache
[$titleText], $title ];
3580 # Cache miss, go to the database
3581 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3583 if ( $text === false ) {
3584 $this->mTplDomCache
[$titleText] = false;
3585 return [ false, $title ];
3588 $dom = $this->preprocessToDom( $text, self
::PTD_FOR_INCLUSION
);
3589 $this->mTplDomCache
[$titleText] = $dom;
3591 if ( !$title->equals( $cacheTitle ) ) {
3592 $this->mTplRedirCache
[$cacheTitle->getPrefixedDBkey()] =
3593 [ $title->getNamespace(), $title->getDBkey() ];
3596 return [ $dom, $title ];
3600 * Fetch the current revision of a given title. Note that the revision
3601 * (and even the title) may not exist in the database, so everything
3602 * contributing to the output of the parser should use this method
3603 * where possible, rather than getting the revisions themselves. This
3604 * method also caches its results, so using it benefits performance.
3607 * @param Title $title
3610 public function fetchCurrentRevisionOfTitle( $title ) {
3611 $cacheKey = $title->getPrefixedDBkey();
3612 if ( !$this->currentRevisionCache
) {
3613 $this->currentRevisionCache
= new MapCacheLRU( 100 );
3615 if ( !$this->currentRevisionCache
->has( $cacheKey ) ) {
3616 $this->currentRevisionCache
->set( $cacheKey,
3617 // Defaults to Parser::statelessFetchRevision()
3618 call_user_func( $this->mOptions
->getCurrentRevisionCallback(), $title, $this )
3621 return $this->currentRevisionCache
->get( $cacheKey );
3625 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3626 * without passing them on to it.
3629 * @param Title $title
3630 * @param Parser|bool $parser
3631 * @return Revision|bool False if missing
3633 public static function statelessFetchRevision( Title
$title, $parser = false ) {
3634 $rev = Revision
::newKnownCurrent( wfGetDB( DB_REPLICA
), $title );
3640 * Fetch the unparsed text of a template and register a reference to it.
3641 * @param Title $title
3642 * @return array ( string or false, Title )
3644 public function fetchTemplateAndTitle( $title ) {
3645 // Defaults to Parser::statelessFetchTemplate()
3646 $templateCb = $this->mOptions
->getTemplateCallback();
3647 $stuff = call_user_func( $templateCb, $title, $this );
3648 // We use U+007F DELETE to distinguish strip markers from regular text.
3649 $text = $stuff['text'];
3650 if ( is_string( $stuff['text'] ) ) {
3651 $text = strtr( $text, "\x7f", "?" );
3653 $finalTitle = $stuff['finalTitle'] ??
$title;
3654 if ( isset( $stuff['deps'] ) ) {
3655 foreach ( $stuff['deps'] as $dep ) {
3656 $this->mOutput
->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3657 if ( $dep['title']->equals( $this->getTitle() ) ) {
3658 // If we transclude ourselves, the final result
3659 // will change based on the new version of the page
3660 $this->mOutput
->setFlag( 'vary-revision' );
3664 return [ $text, $finalTitle ];
3668 * Fetch the unparsed text of a template and register a reference to it.
3669 * @param Title $title
3670 * @return string|bool
3672 public function fetchTemplate( $title ) {
3673 return $this->fetchTemplateAndTitle( $title )[0];
3677 * Static function to get a template
3678 * Can be overridden via ParserOptions::setTemplateCallback().
3680 * @param Title $title
3681 * @param bool|Parser $parser
3685 public static function statelessFetchTemplate( $title, $parser = false ) {
3686 $text = $skip = false;
3687 $finalTitle = $title;
3690 # Loop to fetch the article, with up to 1 redirect
3691 for ( $i = 0; $i < 2 && is_object( $title ); $i++
) {
3692 # Give extensions a chance to select the revision instead
3693 $id = false; # Assume current
3694 Hooks
::run( 'BeforeParserFetchTemplateAndtitle',
3695 [ $parser, $title, &$skip, &$id ] );
3701 'page_id' => $title->getArticleID(),
3708 $rev = Revision
::newFromId( $id );
3709 } elseif ( $parser ) {
3710 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3712 $rev = Revision
::newFromTitle( $title );
3714 $rev_id = $rev ?
$rev->getId() : 0;
3715 # If there is no current revision, there is no page
3716 if ( $id === false && !$rev ) {
3717 $linkCache = MediaWikiServices
::getInstance()->getLinkCache();
3718 $linkCache->addBadLinkObj( $title );
3723 'page_id' => $title->getArticleID(),
3724 'rev_id' => $rev_id ];
3725 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3726 # We fetched a rev from a different title; register it too...
3728 'title' => $rev->getTitle(),
3729 'page_id' => $rev->getPage(),
3730 'rev_id' => $rev_id ];
3734 $content = $rev->getContent();
3735 $text = $content ?
$content->getWikitextForTransclusion() : null;
3737 Hooks
::run( 'ParserFetchTemplate',
3738 [ $parser, $title, $rev, &$text, &$deps ] );
3740 if ( $text === false ||
$text === null ) {
3744 } elseif ( $title->getNamespace() == NS_MEDIAWIKI
) {
3745 $message = wfMessage( MediaWikiServices
::getInstance()->getContentLanguage()->
3746 lcfirst( $title->getText() ) )->inContentLanguage();
3747 if ( !$message->exists() ) {
3751 $content = $message->content();
3752 $text = $message->plain();
3760 $finalTitle = $title;
3761 $title = $content->getRedirectTarget();
3765 'finalTitle' => $finalTitle,
3770 * Fetch a file and its title and register a reference to it.
3771 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3772 * @param Title $title
3773 * @param array $options Array of options to RepoGroup::findFile
3775 * @deprecated since 1.32, use fetchFileAndTitle instead
3777 public function fetchFile( $title, $options = [] ) {
3778 wfDeprecated( __METHOD__
, '1.32' );
3779 return $this->fetchFileAndTitle( $title, $options )[0];
3783 * Fetch a file and its title and register a reference to it.
3784 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3785 * @param Title $title
3786 * @param array $options Array of options to RepoGroup::findFile
3787 * @return array ( File or false, Title of file )
3789 public function fetchFileAndTitle( $title, $options = [] ) {
3790 $file = $this->fetchFileNoRegister( $title, $options );
3792 $time = $file ?
$file->getTimestamp() : false;
3793 $sha1 = $file ?
$file->getSha1() : false;
3794 # Register the file as a dependency...
3795 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3796 if ( $file && !$title->equals( $file->getTitle() ) ) {
3797 # Update fetched file title
3798 $title = $file->getTitle();
3799 $this->mOutput
->addImage( $title->getDBkey(), $time, $sha1 );
3801 return [ $file, $title ];
3805 * Helper function for fetchFileAndTitle.
3807 * Also useful if you need to fetch a file but not use it yet,
3808 * for example to get the file's handler.
3810 * @param Title $title
3811 * @param array $options Array of options to RepoGroup::findFile
3814 protected function fetchFileNoRegister( $title, $options = [] ) {
3815 if ( isset( $options['broken'] ) ) {
3816 $file = false; // broken thumbnail forced by hook
3817 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3818 $file = RepoGroup
::singleton()->findFileFromKey( $options['sha1'], $options );
3819 } else { // get by (name,timestamp)
3820 $file = wfFindFile( $title, $options );
3826 * Transclude an interwiki link.
3828 * @param Title $title
3829 * @param string $action Usually one of (raw, render)
3833 public function interwikiTransclude( $title, $action ) {
3834 if ( !$this->siteConfig
->get( 'EnableScaryTranscluding' ) ) {
3835 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3838 $url = $title->getFullURL( [ 'action' => $action ] );
3839 if ( strlen( $url ) > 1024 ) {
3840 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3843 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3845 $fname = __METHOD__
;
3846 $cache = MediaWikiServices
::getInstance()->getMainWANObjectCache();
3848 $data = $cache->getWithSetCallback(
3849 $cache->makeGlobalKey(
3850 'interwiki-transclude',
3851 ( $wikiId !== false ) ?
$wikiId : 'external',
3854 $this->siteConfig
->get( 'TranscludeCacheExpiry' ),
3855 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3856 $req = MWHttpRequest
::factory( $url, [], $fname );
3858 $status = $req->execute(); // Status object
3859 if ( !$status->isOK() ) {
3860 $ttl = $cache::TTL_UNCACHEABLE
;
3861 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3862 $ttl = min( $cache::TTL_LAGGED
, $ttl );
3866 'text' => $status->isOK() ?
$req->getContent() : null,
3867 'code' => $req->getStatus()
3871 'checkKeys' => ( $wikiId !== false )
3872 ?
[ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3874 'pcGroup' => 'interwiki-transclude:5',
3875 'pcTTL' => $cache::TTL_PROC_LONG
3879 if ( is_string( $data['text'] ) ) {
3880 $text = $data['text'];
3881 } elseif ( $data['code'] != 200 ) {
3882 // Though we failed to fetch the content, this status is useless.
3883 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3884 ->params( $url, $data['code'] )->inContentLanguage()->text();
3886 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3893 * Triple brace replacement -- used for template arguments
3896 * @param array $piece
3897 * @param PPFrame $frame
3901 public function argSubstitution( $piece, $frame ) {
3903 $parts = $piece['parts'];
3904 $nameWithSpaces = $frame->expand( $piece['title'] );
3905 $argName = trim( $nameWithSpaces );
3907 $text = $frame->getArgument( $argName );
3908 if ( $text === false && $parts->getLength() > 0
3909 && ( $this->ot
['html']
3911 ||
( $this->ot
['wiki'] && $frame->isTemplate() )
3914 # No match in frame, use the supplied default
3915 $object = $parts->item( 0 )->getChildren();
3917 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3918 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3919 $this->limitationWarn( 'post-expand-template-argument' );
3922 if ( $text === false && $object === false ) {
3924 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3926 if ( $error !== false ) {
3929 if ( $object !== false ) {
3930 $ret = [ 'object' => $object ];
3932 $ret = [ 'text' => $text ];
3939 * Return the text to be used for a given extension tag.
3940 * This is the ghost of strip().
3942 * @param array $params Associative array of parameters:
3943 * name PPNode for the tag name
3944 * attr PPNode for unparsed text where tag attributes are thought to be
3945 * attributes Optional associative array of parsed attributes
3946 * inner Contents of extension element
3947 * noClose Original text did not have a close tag
3948 * @param PPFrame $frame
3950 * @throws MWException
3953 public function extensionSubstitution( $params, $frame ) {
3954 static $errorStr = '<span class="error">';
3955 static $errorLen = 20;
3957 $name = $frame->expand( $params['name'] );
3958 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3959 // Probably expansion depth or node count exceeded. Just punt the
3964 $attrText = !isset( $params['attr'] ) ?
null : $frame->expand( $params['attr'] );
3965 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3970 // We can't safely check if the expansion for $content resulted in an
3971 // error, because the content could happen to be the error string
3973 $content = !isset( $params['inner'] ) ?
null : $frame->expand( $params['inner'] );
3975 $marker = self
::MARKER_PREFIX
. "-$name-"
3976 . sprintf( '%08X', $this->mMarkerIndex++
) . self
::MARKER_SUFFIX
;
3978 $isFunctionTag = isset( $this->mFunctionTagHooks
[strtolower( $name )] ) &&
3979 ( $this->ot
['html'] ||
$this->ot
['pre'] );
3980 if ( $isFunctionTag ) {
3981 $markerType = 'none';
3983 $markerType = 'general';
3985 if ( $this->ot
['html'] ||
$isFunctionTag ) {
3986 $name = strtolower( $name );
3987 $attributes = Sanitizer
::decodeTagAttributes( $attrText );
3988 if ( isset( $params['attributes'] ) ) {
3989 $attributes +
= $params['attributes'];
3992 if ( isset( $this->mTagHooks
[$name] ) ) {
3993 $output = call_user_func_array( $this->mTagHooks
[$name],
3994 [ $content, $attributes, $this, $frame ] );
3995 } elseif ( isset( $this->mFunctionTagHooks
[$name] ) ) {
3996 list( $callback, ) = $this->mFunctionTagHooks
[$name];
3998 // Avoid PHP 7.1 warning from passing $this by reference
4000 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4002 $output = '<span class="error">Invalid tag extension name: ' .
4003 htmlspecialchars( $name ) . '</span>';
4006 if ( is_array( $output ) ) {
4009 $output = $flags[0];
4010 if ( isset( $flags['markerType'] ) ) {
4011 $markerType = $flags['markerType'];
4015 if ( is_null( $attrText ) ) {
4018 if ( isset( $params['attributes'] ) ) {
4019 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4020 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4021 htmlspecialchars( $attrValue ) . '"';
4024 if ( $content === null ) {
4025 $output = "<$name$attrText/>";
4027 $close = is_null( $params['close'] ) ?
'' : $frame->expand( $params['close'] );
4028 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4032 $output = "<$name$attrText>$content$close";
4036 if ( $markerType === 'none' ) {
4038 } elseif ( $markerType === 'nowiki' ) {
4039 $this->mStripState
->addNoWiki( $marker, $output );
4040 } elseif ( $markerType === 'general' ) {
4041 $this->mStripState
->addGeneral( $marker, $output );
4043 throw new MWException( __METHOD__
. ': invalid marker type' );
4049 * Increment an include size counter
4051 * @param string $type The type of expansion
4052 * @param int $size The size of the text
4053 * @return bool False if this inclusion would take it over the maximum, true otherwise
4055 public function incrementIncludeSize( $type, $size ) {
4056 if ( $this->mIncludeSizes
[$type] +
$size > $this->mOptions
->getMaxIncludeSize() ) {
4059 $this->mIncludeSizes
[$type] +
= $size;
4065 * Increment the expensive function count
4067 * @return bool False if the limit has been exceeded
4069 public function incrementExpensiveFunctionCount() {
4070 $this->mExpensiveFunctionCount++
;
4071 return $this->mExpensiveFunctionCount
<= $this->mOptions
->getExpensiveParserFunctionLimit();
4075 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4076 * Fills $this->mDoubleUnderscores, returns the modified text
4078 * @param string $text
4082 public function doDoubleUnderscore( $text ) {
4083 # The position of __TOC__ needs to be recorded
4084 $mw = $this->magicWordFactory
->get( 'toc' );
4085 if ( $mw->match( $text ) ) {
4086 $this->mShowToc
= true;
4087 $this->mForceTocPosition
= true;
4089 # Set a placeholder. At the end we'll fill it in with the TOC.
4090 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4092 # Only keep the first one.
4093 $text = $mw->replace( '', $text );
4096 # Now match and remove the rest of them
4097 $mwa = $this->magicWordFactory
->getDoubleUnderscoreArray();
4098 $this->mDoubleUnderscores
= $mwa->matchAndRemove( $text );
4100 if ( isset( $this->mDoubleUnderscores
['nogallery'] ) ) {
4101 $this->mOutput
->mNoGallery
= true;
4103 if ( isset( $this->mDoubleUnderscores
['notoc'] ) && !$this->mForceTocPosition
) {
4104 $this->mShowToc
= false;
4106 if ( isset( $this->mDoubleUnderscores
['hiddencat'] )
4107 && $this->mTitle
->getNamespace() == NS_CATEGORY
4109 $this->addTrackingCategory( 'hidden-category-category' );
4111 # (T10068) Allow control over whether robots index a page.
4112 # __INDEX__ always overrides __NOINDEX__, see T16899
4113 if ( isset( $this->mDoubleUnderscores
['noindex'] ) && $this->mTitle
->canUseNoindex() ) {
4114 $this->mOutput
->setIndexPolicy( 'noindex' );
4115 $this->addTrackingCategory( 'noindex-category' );
4117 if ( isset( $this->mDoubleUnderscores
['index'] ) && $this->mTitle
->canUseNoindex() ) {
4118 $this->mOutput
->setIndexPolicy( 'index' );
4119 $this->addTrackingCategory( 'index-category' );
4122 # Cache all double underscores in the database
4123 foreach ( $this->mDoubleUnderscores
as $key => $val ) {
4124 $this->mOutput
->setProperty( $key, '' );
4131 * @see ParserOutput::addTrackingCategory()
4132 * @param string $msg Message key
4133 * @return bool Whether the addition was successful
4135 public function addTrackingCategory( $msg ) {
4136 return $this->mOutput
->addTrackingCategory( $msg, $this->mTitle
);
4140 * This function accomplishes several tasks:
4141 * 1) Auto-number headings if that option is enabled
4142 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4143 * 3) Add a Table of contents on the top for users who have enabled the option
4144 * 4) Auto-anchor headings
4146 * It loops through all headlines, collects the necessary data, then splits up the
4147 * string and re-inserts the newly formatted headlines.
4149 * @param string $text
4150 * @param string $origText Original, untouched wikitext
4151 * @param bool $isMain
4152 * @return mixed|string
4155 public function formatHeadings( $text, $origText, $isMain = true ) {
4156 # Inhibit editsection links if requested in the page
4157 if ( isset( $this->mDoubleUnderscores
['noeditsection'] ) ) {
4158 $maybeShowEditLink = false;
4160 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4163 # Get all headlines for numbering them and adding funky stuff like [edit]
4164 # links - this is for later, but we need the number of headlines right now
4165 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4166 # be trimmed here since whitespace in HTML headings is significant.
4168 $numMatches = preg_match_all(
4169 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4174 # if there are fewer than 4 headlines in the article, do not show TOC
4175 # unless it's been explicitly enabled.
4176 $enoughToc = $this->mShowToc
&&
4177 ( ( $numMatches >= 4 ) ||
$this->mForceTocPosition
);
4179 # Allow user to stipulate that a page should have a "new section"
4180 # link added via __NEWSECTIONLINK__
4181 if ( isset( $this->mDoubleUnderscores
['newsectionlink'] ) ) {
4182 $this->mOutput
->setNewSection( true );
4185 # Allow user to remove the "new section"
4186 # link via __NONEWSECTIONLINK__
4187 if ( isset( $this->mDoubleUnderscores
['nonewsectionlink'] ) ) {
4188 $this->mOutput
->hideNewSection( true );
4191 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4192 # override above conditions and always show TOC above first header
4193 if ( isset( $this->mDoubleUnderscores
['forcetoc'] ) ) {
4194 $this->mShowToc
= true;
4202 # Ugh .. the TOC should have neat indentation levels which can be
4203 # passed to the skin functions. These are determined here
4207 $sublevelCount = [];
4213 $markerRegex = self
::MARKER_PREFIX
. "-h-(\d+)-" . self
::MARKER_SUFFIX
;
4214 $baseTitleText = $this->mTitle
->getPrefixedDBkey();
4215 $oldType = $this->mOutputType
;
4216 $this->setOutputType( self
::OT_WIKI
);
4217 $frame = $this->getPreprocessor()->newFrame();
4218 $root = $this->preprocessToDom( $origText );
4219 $node = $root->getFirstChild();
4224 $headlines = $numMatches !== false ?
$matches[3] : [];
4226 $maxTocLevel = $this->siteConfig
->get( 'MaxTocLevel' );
4227 foreach ( $headlines as $headline ) {
4228 $isTemplate = false;
4230 $sectionIndex = false;
4232 $markerMatches = [];
4233 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4234 $serial = $markerMatches[1];
4235 list( $titleText, $sectionIndex ) = $this->mHeadings
[$serial];
4236 $isTemplate = ( $titleText != $baseTitleText );
4237 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4241 $prevlevel = $level;
4243 $level = $matches[1][$headlineCount];
4245 if ( $level > $prevlevel ) {
4246 # Increase TOC level
4248 $sublevelCount[$toclevel] = 0;
4249 if ( $toclevel < $maxTocLevel ) {
4250 $prevtoclevel = $toclevel;
4251 $toc .= Linker
::tocIndent();
4254 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4255 # Decrease TOC level, find level to jump to
4257 for ( $i = $toclevel; $i > 0; $i-- ) {
4258 if ( $levelCount[$i] == $level ) {
4259 # Found last matching level
4262 } elseif ( $levelCount[$i] < $level ) {
4263 # Found first matching level below current level
4271 if ( $toclevel < $maxTocLevel ) {
4272 if ( $prevtoclevel < $maxTocLevel ) {
4273 # Unindent only if the previous toc level was shown :p
4274 $toc .= Linker
::tocUnindent( $prevtoclevel - $toclevel );
4275 $prevtoclevel = $toclevel;
4277 $toc .= Linker
::tocLineEnd();
4281 # No change in level, end TOC line
4282 if ( $toclevel < $maxTocLevel ) {
4283 $toc .= Linker
::tocLineEnd();
4287 $levelCount[$toclevel] = $level;
4289 # count number of headlines for each level
4290 $sublevelCount[$toclevel]++
;
4292 for ( $i = 1; $i <= $toclevel; $i++
) {
4293 if ( !empty( $sublevelCount[$i] ) ) {
4297 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4302 # The safe header is a version of the header text safe to use for links
4304 # Remove link placeholders by the link text.
4305 # <!--LINK number-->
4307 # link text with suffix
4308 # Do this before unstrip since link text can contain strip markers
4309 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4311 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4312 $safeHeadline = $this->mStripState
->unstripBoth( $safeHeadline );
4314 # Remove any <style> or <script> tags (T198618)
4315 $safeHeadline = preg_replace(
4316 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4321 # Strip out HTML (first regex removes any tag not allowed)
4323 # * <sup> and <sub> (T10393)
4327 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4328 # * <s> and <strike> (T35715)
4329 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4330 # to allow setting directionality in toc items.
4331 $tocline = preg_replace(
4333 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4334 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4340 # Strip '<span></span>', which is the result from the above if
4341 # <span id="foo"></span> is used to produce an additional anchor
4343 $tocline = str_replace( '<span></span>', '', $tocline );
4345 $tocline = trim( $tocline );
4347 # For the anchor, strip out HTML-y stuff period
4348 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4349 $safeHeadline = Sanitizer
::normalizeSectionNameWhitespace( $safeHeadline );
4351 # Save headline for section edit hint before it's escaped
4352 $headlineHint = $safeHeadline;
4354 # Decode HTML entities
4355 $safeHeadline = Sanitizer
::decodeCharReferences( $safeHeadline );
4357 $safeHeadline = self
::normalizeSectionName( $safeHeadline );
4359 $fallbackHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_FALLBACK
);
4360 $linkAnchor = Sanitizer
::escapeIdForLink( $safeHeadline );
4361 $safeHeadline = Sanitizer
::escapeIdForAttribute( $safeHeadline, Sanitizer
::ID_PRIMARY
);
4362 if ( $fallbackHeadline === $safeHeadline ) {
4363 # No reason to have both (in fact, we can't)
4364 $fallbackHeadline = false;
4367 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4368 # @todo FIXME: We may be changing them depending on the current locale.
4369 $arrayKey = strtolower( $safeHeadline );
4370 if ( $fallbackHeadline === false ) {
4371 $fallbackArrayKey = false;
4373 $fallbackArrayKey = strtolower( $fallbackHeadline );
4376 # Create the anchor for linking from the TOC to the section
4377 $anchor = $safeHeadline;
4378 $fallbackAnchor = $fallbackHeadline;
4379 if ( isset( $refers[$arrayKey] ) ) {
4380 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4381 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++
$i );
4383 $linkAnchor .= "_$i";
4384 $refers["${arrayKey}_$i"] = true;
4386 $refers[$arrayKey] = true;
4388 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4389 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4390 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++
$i );
4391 $fallbackAnchor .= "_$i";
4392 $refers["${fallbackArrayKey}_$i"] = true;
4394 $refers[$fallbackArrayKey] = true;
4397 # Don't number the heading if it is the only one (looks silly)
4398 if ( count( $matches[3] ) > 1 && $this->mOptions
->getNumberHeadings() ) {
4399 # the two are different if the line contains a link
4400 $headline = Html
::element(
4402 [ 'class' => 'mw-headline-number' ],
4404 ) . ' ' . $headline;
4407 if ( $enoughToc && ( !isset( $maxTocLevel ) ||
$toclevel < $maxTocLevel ) ) {
4408 $toc .= Linker
::tocLine( $linkAnchor, $tocline,
4409 $numbering, $toclevel, ( $isTemplate ?
false : $sectionIndex ) );
4412 # Add the section to the section tree
4413 # Find the DOM node for this header
4414 $noOffset = ( $isTemplate ||
$sectionIndex === false );
4415 while ( $node && !$noOffset ) {
4416 if ( $node->getName() === 'h' ) {
4417 $bits = $node->splitHeading();
4418 if ( $bits['i'] == $sectionIndex ) {
4422 $byteOffset +
= mb_strlen( $this->mStripState
->unstripBoth(
4423 $frame->expand( $node, PPFrame
::RECOVER_ORIG
) ) );
4424 $node = $node->getNextSibling();
4427 'toclevel' => $toclevel,
4430 'number' => $numbering,
4431 'index' => ( $isTemplate ?
'T-' : '' ) . $sectionIndex,
4432 'fromtitle' => $titleText,
4433 'byteoffset' => ( $noOffset ?
null : $byteOffset ),
4434 'anchor' => $anchor,
4437 # give headline the correct <h#> tag
4438 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4439 // Output edit section links as markers with styles that can be customized by skins
4440 if ( $isTemplate ) {
4441 # Put a T flag in the section identifier, to indicate to extractSections()
4442 # that sections inside <includeonly> should be counted.
4443 $editsectionPage = $titleText;
4444 $editsectionSection = "T-$sectionIndex";
4445 $editsectionContent = null;
4447 $editsectionPage = $this->mTitle
->getPrefixedText();
4448 $editsectionSection = $sectionIndex;
4449 $editsectionContent = $headlineHint;
4451 // We use a bit of pesudo-xml for editsection markers. The
4452 // language converter is run later on. Using a UNIQ style marker
4453 // leads to the converter screwing up the tokens when it
4454 // converts stuff. And trying to insert strip tags fails too. At
4455 // this point all real inputted tags have already been escaped,
4456 // so we don't have to worry about a user trying to input one of
4457 // these markers directly. We use a page and section attribute
4458 // to stop the language converter from converting these
4459 // important bits of data, but put the headline hint inside a
4460 // content block because the language converter is supposed to
4461 // be able to convert that piece of data.
4462 // Gets replaced with html in ParserOutput::getText
4463 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4464 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4465 if ( $editsectionContent !== null ) {
4466 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4473 $head[$headlineCount] = Linker
::makeHeadline( $level,
4474 $matches['attrib'][$headlineCount], $anchor, $headline,
4475 $editlink, $fallbackAnchor );
4480 $this->setOutputType( $oldType );
4482 # Never ever show TOC if no headers
4483 if ( $numVisible < 1 ) {
4488 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4489 $toc .= Linker
::tocUnindent( $prevtoclevel - 1 );
4491 $toc = Linker
::tocList( $toc, $this->mOptions
->getUserLangObj() );
4492 $this->mOutput
->setTOCHTML( $toc );
4493 $toc = self
::TOC_START
. $toc . self
::TOC_END
;
4497 $this->mOutput
->setSections( $tocraw );
4500 # split up and insert constructed headlines
4501 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4504 // build an array of document sections
4506 foreach ( $blocks as $block ) {
4507 // $head is zero-based, sections aren't.
4508 if ( empty( $head[$i - 1] ) ) {
4509 $sections[$i] = $block;
4511 $sections[$i] = $head[$i - 1] . $block;
4515 * Send a hook, one per section.
4516 * The idea here is to be able to make section-level DIVs, but to do so in a
4517 * lower-impact, more correct way than r50769
4520 * $section : the section number
4521 * &$sectionContent : ref to the content of the section
4522 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4524 Hooks
::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4529 if ( $enoughToc && $isMain && !$this->mForceTocPosition
) {
4530 // append the TOC at the beginning
4531 // Top anchor now in skin
4532 $sections[0] .= $toc . "\n";
4535 $full .= implode( '', $sections );
4537 if ( $this->mForceTocPosition
) {
4538 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4545 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4546 * conversion, substituting signatures, {{subst:}} templates, etc.
4548 * @param string $text The text to transform
4549 * @param Title $title The Title object for the current article
4550 * @param User $user The User object describing the current user
4551 * @param ParserOptions $options Parsing options
4552 * @param bool $clearState Whether to clear the parser state first
4553 * @return string The altered wiki markup
4555 public function preSaveTransform( $text, Title
$title, User
$user,
4556 ParserOptions
$options, $clearState = true
4558 if ( $clearState ) {
4559 $magicScopeVariable = $this->lock();
4561 $this->startParse( $title, $options, self
::OT_WIKI
, $clearState );
4562 $this->setUser( $user );
4564 // Strip U+0000 NULL (T159174)
4565 $text = str_replace( "\000", '', $text );
4567 // We still normalize line endings for backwards-compatibility
4568 // with other code that just calls PST, but this should already
4569 // be handled in TextContent subclasses
4570 $text = TextContent
::normalizeLineEndings( $text );
4572 if ( $options->getPreSaveTransform() ) {
4573 $text = $this->pstPass2( $text, $user );
4575 $text = $this->mStripState
->unstripBoth( $text );
4577 $this->setUser( null ); # Reset
4583 * Pre-save transform helper function
4585 * @param string $text
4590 private function pstPass2( $text, $user ) {
4591 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4592 # $this->contLang here in order to give everyone the same signature and use the default one
4593 # rather than the one selected in each user's preferences. (see also T14815)
4594 $ts = $this->mOptions
->getTimestamp();
4595 $timestamp = MWTimestamp
::getLocalInstance( $ts );
4596 $ts = $timestamp->format( 'YmdHis' );
4597 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4599 $d = $this->contLang
->timeanddate( $ts, false, false ) . " ($tzMsg)";
4601 # Variable replacement
4602 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4603 $text = $this->replaceVariables( $text );
4605 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4606 # which may corrupt this parser instance via its wfMessage()->text() call-
4609 if ( strpos( $text, '~~~' ) !== false ) {
4610 $sigText = $this->getUserSig( $user );
4611 $text = strtr( $text, [
4613 '~~~~' => "$sigText $d",
4616 # The main two signature forms used above are time-sensitive
4617 $this->mOutput
->setFlag( 'user-signature' );
4620 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4621 $tc = '[' . Title
::legalChars() . ']';
4622 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4624 // [[ns:page (context)|]]
4625 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4626 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4627 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4628 // [[ns:page (context), context|]] (using either single or double-width comma)
4629 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4630 // [[|page]] (reverse pipe trick: add context from page title)
4631 $p2 = "/\[\[\\|($tc+)]]/";
4633 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4634 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4635 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4636 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4638 $t = $this->mTitle
->getText();
4640 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4641 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4642 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4643 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4645 # if there's no context, don't bother duplicating the title
4646 $text = preg_replace( $p2, '[[\\1]]', $text );
4653 * Fetch the user's signature text, if any, and normalize to
4654 * validated, ready-to-insert wikitext.
4655 * If you have pre-fetched the nickname or the fancySig option, you can
4656 * specify them here to save a database query.
4657 * Do not reuse this parser instance after calling getUserSig(),
4658 * as it may have changed if it's the $wgParser.
4660 * @param User &$user
4661 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4662 * @param bool|null $fancySig whether the nicknname is the complete signature
4663 * or null to use default value
4666 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4667 $username = $user->getName();
4669 # If not given, retrieve from the user object.
4670 if ( $nickname === false ) {
4671 $nickname = $user->getOption( 'nickname' );
4674 if ( is_null( $fancySig ) ) {
4675 $fancySig = $user->getBoolOption( 'fancysig' );
4678 $nickname = $nickname == null ?
$username : $nickname;
4680 if ( mb_strlen( $nickname ) > $this->siteConfig
->get( 'MaxSigChars' ) ) {
4681 $nickname = $username;
4682 wfDebug( __METHOD__
. ": $username has overlong signature.\n" );
4683 } elseif ( $fancySig !== false ) {
4684 # Sig. might contain markup; validate this
4685 if ( $this->validateSig( $nickname ) !== false ) {
4686 # Validated; clean up (if needed) and return it
4687 return $this->cleanSig( $nickname, true );
4689 # Failed to validate; fall back to the default
4690 $nickname = $username;
4691 wfDebug( __METHOD__
. ": $username has bad XML tags in signature.\n" );
4695 # Make sure nickname doesnt get a sig in a sig
4696 $nickname = self
::cleanSigInSig( $nickname );
4698 # If we're still here, make it a link to the user page
4699 $userText = wfEscapeWikiText( $username );
4700 $nickText = wfEscapeWikiText( $nickname );
4701 $msgName = $user->isAnon() ?
'signature-anon' : 'signature';
4703 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4704 ->title( $this->getTitle() )->text();
4708 * Check that the user's signature contains no bad XML
4710 * @param string $text
4711 * @return string|bool An expanded string, or false if invalid.
4713 public function validateSig( $text ) {
4714 return Xml
::isWellFormedXmlFragment( $text ) ?
$text : false;
4718 * Clean up signature text
4720 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4721 * 2) Substitute all transclusions
4723 * @param string $text
4724 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4725 * @return string Signature text
4727 public function cleanSig( $text, $parsing = false ) {
4730 $magicScopeVariable = $this->lock();
4731 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PREPROCESS
, true );
4734 # Option to disable this feature
4735 if ( !$this->mOptions
->getCleanSignatures() ) {
4739 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4740 # => Move this logic to braceSubstitution()
4741 $substWord = $this->magicWordFactory
->get( 'subst' );
4742 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4743 $substText = '{{' . $substWord->getSynonym( 0 );
4745 $text = preg_replace( $substRegex, $substText, $text );
4746 $text = self
::cleanSigInSig( $text );
4747 $dom = $this->preprocessToDom( $text );
4748 $frame = $this->getPreprocessor()->newFrame();
4749 $text = $frame->expand( $dom );
4752 $text = $this->mStripState
->unstripBoth( $text );
4759 * Strip 3, 4 or 5 tildes out of signatures.
4761 * @param string $text
4762 * @return string Signature text with /~{3,5}/ removed
4764 public static function cleanSigInSig( $text ) {
4765 $text = preg_replace( '/~{3,5}/', '', $text );
4770 * Set up some variables which are usually set up in parse()
4771 * so that an external function can call some class members with confidence
4773 * @param Title|null $title
4774 * @param ParserOptions $options
4775 * @param int $outputType
4776 * @param bool $clearState
4778 public function startExternalParse( Title
$title = null, ParserOptions
$options,
4779 $outputType, $clearState = true
4781 $this->startParse( $title, $options, $outputType, $clearState );
4785 * @param Title|null $title
4786 * @param ParserOptions $options
4787 * @param int $outputType
4788 * @param bool $clearState
4790 private function startParse( Title
$title = null, ParserOptions
$options,
4791 $outputType, $clearState = true
4793 $this->setTitle( $title );
4794 $this->mOptions
= $options;
4795 $this->setOutputType( $outputType );
4796 if ( $clearState ) {
4797 $this->clearState();
4802 * Wrapper for preprocess()
4804 * @param string $text The text to preprocess
4805 * @param ParserOptions $options
4806 * @param Title|null $title Title object or null to use $wgTitle
4809 public function transformMsg( $text, $options, $title = null ) {
4810 static $executing = false;
4812 # Guard against infinite recursion
4823 $text = $this->preprocess( $text, $title, $options );
4830 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4831 * The callback should have the following form:
4832 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4834 * Transform and return $text. Use $parser for any required context, e.g. use
4835 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4837 * Hooks may return extended information by returning an array, of which the
4838 * first numbered element (index 0) must be the return string, and all other
4839 * entries are extracted into local variables within an internal function
4840 * in the Parser class.
4842 * This interface (introduced r61913) appears to be undocumented, but
4843 * 'markerType' is used by some core tag hooks to override which strip
4844 * array their results are placed in. **Use great caution if attempting
4845 * this interface, as it is not documented and injudicious use could smash
4846 * private variables.**
4848 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4849 * @param callable $callback The callback function (and object) to use for the tag
4850 * @throws MWException
4851 * @return callable|null The old value of the mTagHooks array associated with the hook
4853 public function setHook( $tag, callable
$callback ) {
4854 $tag = strtolower( $tag );
4855 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4856 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4858 $oldVal = $this->mTagHooks
[$tag] ??
null;
4859 $this->mTagHooks
[$tag] = $callback;
4860 if ( !in_array( $tag, $this->mStripList
) ) {
4861 $this->mStripList
[] = $tag;
4868 * As setHook(), but letting the contents be parsed.
4870 * Transparent tag hooks are like regular XML-style tag hooks, except they
4871 * operate late in the transformation sequence, on HTML instead of wikitext.
4873 * This is probably obsoleted by things dealing with parser frames?
4874 * The only extension currently using it is geoserver.
4877 * @todo better document or deprecate this
4879 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4880 * @param callable $callback The callback function (and object) to use for the tag
4881 * @throws MWException
4882 * @return callable|null The old value of the mTagHooks array associated with the hook
4884 public function setTransparentTagHook( $tag, callable
$callback ) {
4885 $tag = strtolower( $tag );
4886 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4887 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4889 $oldVal = $this->mTransparentTagHooks
[$tag] ??
null;
4890 $this->mTransparentTagHooks
[$tag] = $callback;
4896 * Remove all tag hooks
4898 public function clearTagHooks() {
4899 $this->mTagHooks
= [];
4900 $this->mFunctionTagHooks
= [];
4901 $this->mStripList
= $this->mDefaultStripList
;
4905 * Create a function, e.g. {{sum:1|2|3}}
4906 * The callback function should have the form:
4907 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4909 * Or with Parser::SFH_OBJECT_ARGS:
4910 * function myParserFunction( $parser, $frame, $args ) { ... }
4912 * The callback may either return the text result of the function, or an array with the text
4913 * in element 0, and a number of flags in the other elements. The names of the flags are
4914 * specified in the keys. Valid flags are:
4915 * found The text returned is valid, stop processing the template. This
4917 * nowiki Wiki markup in the return value should be escaped
4918 * isHTML The returned text is HTML, armour it against wikitext transformation
4920 * @param string $id The magic word ID
4921 * @param callable $callback The callback function (and object) to use
4922 * @param int $flags A combination of the following flags:
4923 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4925 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4926 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4927 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4928 * the arguments, and to control the way they are expanded.
4930 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4931 * arguments, for instance:
4932 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4934 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4935 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4936 * working if/when this is changed.
4938 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4941 * Please read the documentation in includes/parser/Preprocessor.php for more information
4942 * about the methods available in PPFrame and PPNode.
4944 * @throws MWException
4945 * @return string|callable The old callback function for this name, if any
4947 public function setFunctionHook( $id, callable
$callback, $flags = 0 ) {
4948 $oldVal = isset( $this->mFunctionHooks
[$id] ) ?
$this->mFunctionHooks
[$id][0] : null;
4949 $this->mFunctionHooks
[$id] = [ $callback, $flags ];
4951 # Add to function cache
4952 $mw = $this->magicWordFactory
->get( $id );
4954 throw new MWException( __METHOD__
. '() expecting a magic word identifier.' );
4957 $synonyms = $mw->getSynonyms();
4958 $sensitive = intval( $mw->isCaseSensitive() );
4960 foreach ( $synonyms as $syn ) {
4962 if ( !$sensitive ) {
4963 $syn = $this->contLang
->lc( $syn );
4966 if ( !( $flags & self
::SFH_NO_HASH
) ) {
4969 # Remove trailing colon
4970 if ( substr( $syn, -1, 1 ) === ':' ) {
4971 $syn = substr( $syn, 0, -1 );
4973 $this->mFunctionSynonyms
[$sensitive][$syn] = $id;
4979 * Get all registered function hook identifiers
4983 public function getFunctionHooks() {
4984 $this->firstCallInit();
4985 return array_keys( $this->mFunctionHooks
);
4989 * Create a tag function, e.g. "<test>some stuff</test>".
4990 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4991 * Unlike parser functions, their content is not preprocessed.
4992 * @param string $tag
4993 * @param callable $callback
4995 * @throws MWException
4998 public function setFunctionTagHook( $tag, callable
$callback, $flags ) {
4999 $tag = strtolower( $tag );
5000 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5001 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5003 $old = $this->mFunctionTagHooks
[$tag] ??
null;
5004 $this->mFunctionTagHooks
[$tag] = [ $callback, $flags ];
5006 if ( !in_array( $tag, $this->mStripList
) ) {
5007 $this->mStripList
[] = $tag;
5014 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5015 * Placeholders created in Linker::link()
5017 * @param string &$text
5018 * @param int $options
5020 public function replaceLinkHolders( &$text, $options = 0 ) {
5021 $this->mLinkHolders
->replace( $text );
5025 * Replace "<!--LINK-->" link placeholders with plain text of links
5026 * (not HTML-formatted).
5028 * @param string $text
5031 public function replaceLinkHoldersText( $text ) {
5032 return $this->mLinkHolders
->replaceText( $text );
5036 * Renders an image gallery from a text with one line per image.
5037 * text labels may be given by using |-style alternative text. E.g.
5038 * Image:one.jpg|The number "1"
5039 * Image:tree.jpg|A tree
5040 * given as text will return the HTML of a gallery with two images,
5041 * labeled 'The number "1"' and
5044 * @param string $text
5045 * @param array $params
5046 * @return string HTML
5048 public function renderImageGallery( $text, $params ) {
5050 if ( isset( $params['mode'] ) ) {
5051 $mode = $params['mode'];
5055 $ig = ImageGalleryBase
::factory( $mode );
5056 } catch ( Exception
$e ) {
5057 // If invalid type set, fallback to default.
5058 $ig = ImageGalleryBase
::factory( false );
5061 $ig->setContextTitle( $this->mTitle
);
5062 $ig->setShowBytes( false );
5063 $ig->setShowDimensions( false );
5064 $ig->setShowFilename( false );
5065 $ig->setParser( $this );
5066 $ig->setHideBadImages();
5067 $ig->setAttributes( Sanitizer
::validateTagAttributes( $params, 'ul' ) );
5069 if ( isset( $params['showfilename'] ) ) {
5070 $ig->setShowFilename( true );
5072 $ig->setShowFilename( false );
5074 if ( isset( $params['caption'] ) ) {
5075 // NOTE: We aren't passing a frame here or below. Frame info
5076 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5077 // See T107332#4030581
5078 $caption = $this->recursiveTagParse( $params['caption'] );
5079 $ig->setCaptionHtml( $caption );
5081 if ( isset( $params['perrow'] ) ) {
5082 $ig->setPerRow( $params['perrow'] );
5084 if ( isset( $params['widths'] ) ) {
5085 $ig->setWidths( $params['widths'] );
5087 if ( isset( $params['heights'] ) ) {
5088 $ig->setHeights( $params['heights'] );
5090 $ig->setAdditionalOptions( $params );
5092 // Avoid PHP 7.1 warning from passing $this by reference
5094 Hooks
::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5096 $lines = StringUtils
::explode( "\n", $text );
5097 foreach ( $lines as $line ) {
5098 # match lines like these:
5099 # Image:someimage.jpg|This is some image
5101 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5103 if ( count( $matches ) == 0 ) {
5107 if ( strpos( $matches[0], '%' ) !== false ) {
5108 $matches[1] = rawurldecode( $matches[1] );
5110 $title = Title
::newFromText( $matches[1], NS_FILE
);
5111 if ( is_null( $title ) ) {
5112 # Bogus title. Ignore these so we don't bomb out later.
5116 # We need to get what handler the file uses, to figure out parameters.
5117 # Note, a hook can overide the file name, and chose an entirely different
5118 # file (which potentially could be of a different type and have different handler).
5121 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5122 [ $this, $title, &$options, &$descQuery ] );
5123 # Don't register it now, as TraditionalImageGallery does that later.
5124 $file = $this->fetchFileNoRegister( $title, $options );
5125 $handler = $file ?
$file->getHandler() : false;
5128 'img_alt' => 'gallery-internal-alt',
5129 'img_link' => 'gallery-internal-link',
5132 $paramMap +
= $handler->getParamMap();
5133 // We don't want people to specify per-image widths.
5134 // Additionally the width parameter would need special casing anyhow.
5135 unset( $paramMap['img_width'] );
5138 $mwArray = $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5143 $handlerOptions = [];
5144 if ( isset( $matches[3] ) ) {
5145 // look for an |alt= definition while trying not to break existing
5146 // captions with multiple pipes (|) in it, until a more sensible grammar
5147 // is defined for images in galleries
5149 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5150 // splitting on '|' is a bit odd, and different from makeImage.
5151 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5152 // Protect LanguageConverter markup
5153 $parameterMatches = StringUtils
::delimiterExplode(
5154 '-{', '}-', '|', $matches[3], true /* nested */
5157 foreach ( $parameterMatches as $parameterMatch ) {
5158 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5160 $paramName = $paramMap[$magicName];
5162 switch ( $paramName ) {
5163 case 'gallery-internal-alt':
5164 $alt = $this->stripAltText( $match, false );
5166 case 'gallery-internal-link':
5167 $linkValue = $this->stripAltText( $match, false );
5168 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5169 // Result of LanguageConverter::markNoConversion
5170 // invoked on an external link.
5171 $linkValue = substr( $linkValue, 4, -2 );
5173 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5174 if ( $type === 'link-url' ) {
5176 $this->mOutput
->addExternalLink( $target );
5177 } elseif ( $type === 'link-title' ) {
5178 $link = $target->getLinkURL();
5179 $this->mOutput
->addLink( $target );
5183 // Must be a handler specific parameter.
5184 if ( $handler->validateParam( $paramName, $match ) ) {
5185 $handlerOptions[$paramName] = $match;
5187 // Guess not, consider it as caption.
5188 wfDebug( "$parameterMatch failed parameter validation\n" );
5189 $label = $parameterMatch;
5195 $label = $parameterMatch;
5200 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5202 $html = $ig->toHTML();
5203 Hooks
::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5208 * @param MediaHandler $handler
5211 public function getImageParams( $handler ) {
5213 $handlerClass = get_class( $handler );
5217 if ( !isset( $this->mImageParams
[$handlerClass] ) ) {
5218 # Initialise static lists
5219 static $internalParamNames = [
5220 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5221 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5222 'bottom', 'text-bottom' ],
5223 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5224 'upright', 'border', 'link', 'alt', 'class' ],
5226 static $internalParamMap;
5227 if ( !$internalParamMap ) {
5228 $internalParamMap = [];
5229 foreach ( $internalParamNames as $type => $names ) {
5230 foreach ( $names as $name ) {
5231 // For grep: img_left, img_right, img_center, img_none,
5232 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5233 // img_bottom, img_text_bottom,
5234 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5235 // img_border, img_link, img_alt, img_class
5236 $magicName = str_replace( '-', '_', "img_$name" );
5237 $internalParamMap[$magicName] = [ $type, $name ];
5242 # Add handler params
5243 $paramMap = $internalParamMap;
5245 $handlerParamMap = $handler->getParamMap();
5246 foreach ( $handlerParamMap as $magic => $paramName ) {
5247 $paramMap[$magic] = [ 'handler', $paramName ];
5250 $this->mImageParams
[$handlerClass] = $paramMap;
5251 $this->mImageParamsMagicArray
[$handlerClass] =
5252 $this->magicWordFactory
->newArray( array_keys( $paramMap ) );
5254 return [ $this->mImageParams
[$handlerClass], $this->mImageParamsMagicArray
[$handlerClass] ];
5258 * Parse image options text and use it to make an image
5260 * @param Title $title
5261 * @param string $options
5262 * @param LinkHolderArray|bool $holders
5263 * @return string HTML
5265 public function makeImage( $title, $options, $holders = false ) {
5266 # Check if the options text is of the form "options|alt text"
5268 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5269 # * left no resizing, just left align. label is used for alt= only
5270 # * right same, but right aligned
5271 # * none same, but not aligned
5272 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5273 # * center center the image
5274 # * frame Keep original image size, no magnify-button.
5275 # * framed Same as "frame"
5276 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5277 # * upright reduce width for upright images, rounded to full __0 px
5278 # * border draw a 1px border around the image
5279 # * alt Text for HTML alt attribute (defaults to empty)
5280 # * class Set a class for img node
5281 # * link Set the target of the image link. Can be external, interwiki, or local
5282 # vertical-align values (no % or length right now):
5292 # Protect LanguageConverter markup when splitting into parts
5293 $parts = StringUtils
::delimiterExplode(
5294 '-{', '}-', '|', $options, true /* allow nesting */
5297 # Give extensions a chance to select the file revision for us
5300 Hooks
::run( 'BeforeParserFetchFileAndTitle',
5301 [ $this, $title, &$options, &$descQuery ] );
5302 # Fetch and register the file (file title may be different via hooks)
5303 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5306 $handler = $file ?
$file->getHandler() : false;
5308 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5311 $this->addTrackingCategory( 'broken-file-category' );
5314 # Process the input parameters
5316 $params = [ 'frame' => [], 'handler' => [],
5317 'horizAlign' => [], 'vertAlign' => [] ];
5318 $seenformat = false;
5319 foreach ( $parts as $part ) {
5320 $part = trim( $part );
5321 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5323 if ( isset( $paramMap[$magicName] ) ) {
5324 list( $type, $paramName ) = $paramMap[$magicName];
5326 # Special case; width and height come in one variable together
5327 if ( $type === 'handler' && $paramName === 'width' ) {
5328 $parsedWidthParam = self
::parseWidthParam( $value );
5329 if ( isset( $parsedWidthParam['width'] ) ) {
5330 $width = $parsedWidthParam['width'];
5331 if ( $handler->validateParam( 'width', $width ) ) {
5332 $params[$type]['width'] = $width;
5336 if ( isset( $parsedWidthParam['height'] ) ) {
5337 $height = $parsedWidthParam['height'];
5338 if ( $handler->validateParam( 'height', $height ) ) {
5339 $params[$type]['height'] = $height;
5343 # else no validation -- T15436
5345 if ( $type === 'handler' ) {
5346 # Validate handler parameter
5347 $validated = $handler->validateParam( $paramName, $value );
5349 # Validate internal parameters
5350 switch ( $paramName ) {
5354 # @todo FIXME: Possibly check validity here for
5355 # manualthumb? downstream behavior seems odd with
5356 # missing manual thumbs.
5358 $value = $this->stripAltText( $value, $holders );
5361 list( $paramName, $value ) =
5362 $this->parseLinkParameter(
5363 $this->stripAltText( $value, $holders )
5367 if ( $paramName === 'no-link' ) {
5370 if ( $paramName === 'link-url' ) {
5371 if ( $this->mOptions
->getExternalLinkTarget() ) {
5372 $params[$type]['link-target'] = $this->mOptions
->getExternalLinkTarget();
5380 // use first appearing option, discard others.
5381 $validated = !$seenformat;
5385 # Most other things appear to be empty or numeric...
5386 $validated = ( $value === false ||
is_numeric( trim( $value ) ) );
5391 $params[$type][$paramName] = $value;
5395 if ( !$validated ) {
5400 # Process alignment parameters
5401 if ( $params['horizAlign'] ) {
5402 $params['frame']['align'] = key( $params['horizAlign'] );
5404 if ( $params['vertAlign'] ) {
5405 $params['frame']['valign'] = key( $params['vertAlign'] );
5408 $params['frame']['caption'] = $caption;
5410 # Will the image be presented in a frame, with the caption below?
5411 $imageIsFramed = isset( $params['frame']['frame'] )
5412 ||
isset( $params['frame']['framed'] )
5413 ||
isset( $params['frame']['thumbnail'] )
5414 ||
isset( $params['frame']['manualthumb'] );
5416 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5417 # came to also set the caption, ordinary text after the image -- which
5418 # makes no sense, because that just repeats the text multiple times in
5419 # screen readers. It *also* came to set the title attribute.
5420 # Now that we have an alt attribute, we should not set the alt text to
5421 # equal the caption: that's worse than useless, it just repeats the
5422 # text. This is the framed/thumbnail case. If there's no caption, we
5423 # use the unnamed parameter for alt text as well, just for the time be-
5424 # ing, if the unnamed param is set and the alt param is not.
5425 # For the future, we need to figure out if we want to tweak this more,
5426 # e.g., introducing a title= parameter for the title; ignoring the un-
5427 # named parameter entirely for images without a caption; adding an ex-
5428 # plicit caption= parameter and preserving the old magic unnamed para-
5430 if ( $imageIsFramed ) { # Framed image
5431 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5432 # No caption or alt text, add the filename as the alt text so
5433 # that screen readers at least get some description of the image
5434 $params['frame']['alt'] = $title->getText();
5436 # Do not set $params['frame']['title'] because tooltips don't make sense
5438 } else { # Inline image
5439 if ( !isset( $params['frame']['alt'] ) ) {
5440 # No alt text, use the "caption" for the alt text
5441 if ( $caption !== '' ) {
5442 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5444 # No caption, fall back to using the filename for the
5446 $params['frame']['alt'] = $title->getText();
5449 # Use the "caption" for the tooltip text
5450 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5452 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5454 Hooks
::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5456 # Linker does the rest
5457 $time = $options['time'] ??
false;
5458 $ret = Linker
::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5459 $time, $descQuery, $this->mOptions
->getThumbSize() );
5461 # Give the handler a chance to modify the parser object
5463 $handler->parserTransformHook( $this, $file );
5470 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5472 * Adds an entry to appropriate link tables.
5475 * @param string $value
5476 * @return array of `[ type, target ]`, where:
5477 * - `type` is one of:
5478 * - `null`: Given value is not a valid link target, use default
5479 * - `'no-link'`: Given value is empty, do not generate a link
5480 * - `'link-url'`: Given value is a valid external link
5481 * - `'link-title'`: Given value is a valid internal link
5483 * - When `type` is `null` or `'no-link'`: `false`
5484 * - When `type` is `'link-url'`: URL string corresponding to given value
5485 * - When `type` is `'link-title'`: Title object corresponding to given value
5487 public function parseLinkParameter( $value ) {
5488 $chars = self
::EXT_LINK_URL_CLASS
;
5489 $addr = self
::EXT_LINK_ADDR
;
5490 $prots = $this->mUrlProtocols
;
5493 if ( $value === '' ) {
5495 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5496 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5497 $this->mOutput
->addExternalLink( $value );
5502 $linkTitle = Title
::newFromText( $value );
5504 $this->mOutput
->addLink( $linkTitle );
5505 $type = 'link-title';
5506 $target = $linkTitle;
5509 return [ $type, $target ];
5513 * @param string $caption
5514 * @param LinkHolderArray|bool $holders
5515 * @return mixed|string
5517 protected function stripAltText( $caption, $holders ) {
5518 # Strip bad stuff out of the title (tooltip). We can't just use
5519 # replaceLinkHoldersText() here, because if this function is called
5520 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5522 $tooltip = $holders->replaceText( $caption );
5524 $tooltip = $this->replaceLinkHoldersText( $caption );
5527 # make sure there are no placeholders in thumbnail attributes
5528 # that are later expanded to html- so expand them now and
5530 $tooltip = $this->mStripState
->unstripBoth( $tooltip );
5531 # Compatibility hack! In HTML certain entity references not terminated
5532 # by a semicolon are decoded (but not if we're in an attribute; that's
5533 # how link URLs get away without properly escaping & in queries).
5534 # But wikitext has always required semicolon-termination of entities,
5535 # so encode & where needed to avoid decode of semicolon-less entities.
5537 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5538 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5539 $tooltip = preg_replace( "/
5540 & # 1. entity prefix
5541 (?= # 2. followed by:
5542 (?: # a. one of the legacy semicolon-less named entities
5543 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5544 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5545 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5546 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5547 U(?:acute|circ|grave|uml)|Yacute|
5548 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5549 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5550 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5551 frac(?:1(?:2|4)|34)|
5552 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5553 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5554 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5555 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5556 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5557 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5558 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5559 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5560 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5562 (?:[^;]|$)) # b. and not followed by a semicolon
5563 # S = study, for efficiency
5564 /Sx", '&', $tooltip );
5565 $tooltip = Sanitizer
::stripAllTags( $tooltip );
5571 * Set a flag in the output object indicating that the content is dynamic and
5572 * shouldn't be cached.
5573 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5575 public function disableCache() {
5576 wfDebug( "Parser output marked as uncacheable.\n" );
5577 if ( !$this->mOutput
) {
5578 throw new MWException( __METHOD__
.
5579 " can only be called when actually parsing something" );
5581 $this->mOutput
->updateCacheExpiry( 0 ); // new style, for consistency
5585 * Callback from the Sanitizer for expanding items found in HTML attribute
5586 * values, so they can be safely tested and escaped.
5588 * @param string &$text
5589 * @param bool|PPFrame $frame
5592 public function attributeStripCallback( &$text, $frame = false ) {
5593 $text = $this->replaceVariables( $text, $frame );
5594 $text = $this->mStripState
->unstripBoth( $text );
5603 public function getTags() {
5604 $this->firstCallInit();
5606 array_keys( $this->mTransparentTagHooks
),
5607 array_keys( $this->mTagHooks
),
5608 array_keys( $this->mFunctionTagHooks
)
5616 public function getFunctionSynonyms() {
5617 $this->firstCallInit();
5618 return $this->mFunctionSynonyms
;
5625 public function getUrlProtocols() {
5626 return $this->mUrlProtocols
;
5630 * Replace transparent tags in $text with the values given by the callbacks.
5632 * Transparent tag hooks are like regular XML-style tag hooks, except they
5633 * operate late in the transformation sequence, on HTML instead of wikitext.
5635 * @param string $text
5639 public function replaceTransparentTags( $text ) {
5641 $elements = array_keys( $this->mTransparentTagHooks
);
5642 $text = self
::extractTagsAndParams( $elements, $text, $matches );
5645 foreach ( $matches as $marker => $data ) {
5646 list( $element, $content, $params, $tag ) = $data;
5647 $tagName = strtolower( $element );
5648 if ( isset( $this->mTransparentTagHooks
[$tagName] ) ) {
5649 $output = call_user_func_array(
5650 $this->mTransparentTagHooks
[$tagName],
5651 [ $content, $params, $this ]
5656 $replacements[$marker] = $output;
5658 return strtr( $text, $replacements );
5662 * Break wikitext input into sections, and either pull or replace
5663 * some particular section's text.
5665 * External callers should use the getSection and replaceSection methods.
5667 * @param string $text Page wikitext
5668 * @param string|int $sectionId A section identifier string of the form:
5669 * "<flag1> - <flag2> - ... - <section number>"
5671 * Currently the only recognised flag is "T", which means the target section number
5672 * was derived during a template inclusion parse, in other words this is a template
5673 * section edit link. If no flags are given, it was an ordinary section edit link.
5674 * This flag is required to avoid a section numbering mismatch when a section is
5675 * enclosed by "<includeonly>" (T8563).
5677 * The section number 0 pulls the text before the first heading; other numbers will
5678 * pull the given section along with its lower-level subsections. If the section is
5679 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5681 * Section 0 is always considered to exist, even if it only contains the empty
5682 * string. If $text is the empty string and section 0 is replaced, $newText is
5685 * @param string $mode One of "get" or "replace"
5686 * @param string $newText Replacement text for section data.
5687 * @return string For "get", the extracted section text.
5688 * for "replace", the whole page with the section replaced.
5690 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5691 global $wgTitle; # not generally used but removes an ugly failure mode
5693 $magicScopeVariable = $this->lock();
5694 $this->startParse( $wgTitle, new ParserOptions
, self
::OT_PLAIN
, true );
5696 $frame = $this->getPreprocessor()->newFrame();
5698 # Process section extraction flags
5700 $sectionParts = explode( '-', $sectionId );
5701 $sectionIndex = array_pop( $sectionParts );
5702 foreach ( $sectionParts as $part ) {
5703 if ( $part === 'T' ) {
5704 $flags |
= self
::PTD_FOR_INCLUSION
;
5708 # Check for empty input
5709 if ( strval( $text ) === '' ) {
5710 # Only sections 0 and T-0 exist in an empty document
5711 if ( $sectionIndex == 0 ) {
5712 if ( $mode === 'get' ) {
5718 if ( $mode === 'get' ) {
5726 # Preprocess the text
5727 $root = $this->preprocessToDom( $text, $flags );
5729 # <h> nodes indicate section breaks
5730 # They can only occur at the top level, so we can find them by iterating the root's children
5731 $node = $root->getFirstChild();
5733 # Find the target section
5734 if ( $sectionIndex == 0 ) {
5735 # Section zero doesn't nest, level=big
5736 $targetLevel = 1000;
5739 if ( $node->getName() === 'h' ) {
5740 $bits = $node->splitHeading();
5741 if ( $bits['i'] == $sectionIndex ) {
5742 $targetLevel = $bits['level'];
5746 if ( $mode === 'replace' ) {
5747 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5749 $node = $node->getNextSibling();
5755 if ( $mode === 'get' ) {
5762 # Find the end of the section, including nested sections
5764 if ( $node->getName() === 'h' ) {
5765 $bits = $node->splitHeading();
5766 $curLevel = $bits['level'];
5767 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5771 if ( $mode === 'get' ) {
5772 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5774 $node = $node->getNextSibling();
5777 # Write out the remainder (in replace mode only)
5778 if ( $mode === 'replace' ) {
5779 # Output the replacement text
5780 # Add two newlines on -- trailing whitespace in $newText is conventionally
5781 # stripped by the editor, so we need both newlines to restore the paragraph gap
5782 # Only add trailing whitespace if there is newText
5783 if ( $newText != "" ) {
5784 $outText .= $newText . "\n\n";
5788 $outText .= $frame->expand( $node, PPFrame
::RECOVER_ORIG
);
5789 $node = $node->getNextSibling();
5793 if ( is_string( $outText ) ) {
5794 # Re-insert stripped tags
5795 $outText = rtrim( $this->mStripState
->unstripBoth( $outText ) );
5802 * This function returns the text of a section, specified by a number ($section).
5803 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5804 * the first section before any such heading (section 0).
5806 * If a section contains subsections, these are also returned.
5808 * @param string $text Text to look in
5809 * @param string|int $sectionId Section identifier as a number or string
5810 * (e.g. 0, 1 or 'T-1').
5811 * @param string $defaultText Default to return if section is not found
5813 * @return string Text of the requested section
5815 public function getSection( $text, $sectionId, $defaultText = '' ) {
5816 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5820 * This function returns $oldtext after the content of the section
5821 * specified by $section has been replaced with $text. If the target
5822 * section does not exist, $oldtext is returned unchanged.
5824 * @param string $oldText Former text of the article
5825 * @param string|int $sectionId Section identifier as a number or string
5826 * (e.g. 0, 1 or 'T-1').
5827 * @param string $newText Replacing text
5829 * @return string Modified text
5831 public function replaceSection( $oldText, $sectionId, $newText ) {
5832 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5836 * Get the ID of the revision we are parsing
5840 public function getRevisionId() {
5841 return $this->mRevisionId
;
5845 * Get the revision object for $this->mRevisionId
5847 * @return Revision|null Either a Revision object or null
5848 * @since 1.23 (public since 1.23)
5850 public function getRevisionObject() {
5851 if ( !is_null( $this->mRevisionObject
) ) {
5852 return $this->mRevisionObject
;
5855 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5856 // This is useful when parsing revision that has not yet been saved.
5857 // However, if we get back a saved revision even though we are in
5858 // preview mode, we'll have to ignore it, see below.
5859 // NOTE: This callback may be used to inject an OLD revision that was
5860 // already loaded, so "current" is a bit of a misnomer. We can't just
5861 // skip it if mRevisionId is set.
5862 $rev = call_user_func(
5863 $this->mOptions
->getCurrentRevisionCallback(), $this->getTitle(), $this
5866 if ( $this->mRevisionId
=== null && $rev && $rev->getId() ) {
5867 // We are in preview mode (mRevisionId is null), and the current revision callback
5868 // returned an existing revision. Ignore it and return null, it's probably the page's
5869 // current revision, which is not what we want here. Note that we do want to call the
5870 // callback to allow the unsaved revision to be injected here, e.g. for
5871 // self-transclusion previews.
5875 // If the parse is for a new revision, then the callback should have
5876 // already been set to force the object and should match mRevisionId.
5877 // If not, try to fetch by mRevisionId for sanity.
5878 if ( $this->mRevisionId
&& $rev && $rev->getId() != $this->mRevisionId
) {
5879 $rev = Revision
::newFromId( $this->mRevisionId
);
5882 $this->mRevisionObject
= $rev;
5884 return $this->mRevisionObject
;
5888 * Get the timestamp associated with the current revision, adjusted for
5889 * the default server-local timestamp
5892 public function getRevisionTimestamp() {
5893 if ( is_null( $this->mRevisionTimestamp
) ) {
5894 $revObject = $this->getRevisionObject();
5895 $timestamp = $revObject ?
$revObject->getTimestamp() : wfTimestampNow();
5897 # The cryptic '' timezone parameter tells to use the site-default
5898 # timezone offset instead of the user settings.
5899 # Since this value will be saved into the parser cache, served
5900 # to other users, and potentially even used inside links and such,
5901 # it needs to be consistent for all visitors.
5902 $this->mRevisionTimestamp
= $this->contLang
->userAdjust( $timestamp, '' );
5904 return $this->mRevisionTimestamp
;
5908 * Get the name of the user that edited the last revision
5910 * @return string User name
5912 public function getRevisionUser() {
5913 if ( is_null( $this->mRevisionUser
) ) {
5914 $revObject = $this->getRevisionObject();
5916 # if this template is subst: the revision id will be blank,
5917 # so just use the current user's name
5919 $this->mRevisionUser
= $revObject->getUserText();
5920 } elseif ( $this->ot
['wiki'] ||
$this->mOptions
->getIsPreview() ) {
5921 $this->mRevisionUser
= $this->getUser()->getName();
5924 return $this->mRevisionUser
;
5928 * Get the size of the revision
5930 * @return int|null Revision size
5932 public function getRevisionSize() {
5933 if ( is_null( $this->mRevisionSize
) ) {
5934 $revObject = $this->getRevisionObject();
5936 # if this variable is subst: the revision id will be blank,
5937 # so just use the parser input size, because the own substituation
5938 # will change the size.
5940 $this->mRevisionSize
= $revObject->getSize();
5942 $this->mRevisionSize
= $this->mInputSize
;
5945 return $this->mRevisionSize
;
5949 * Mutator for $mDefaultSort
5951 * @param string $sort New value
5953 public function setDefaultSort( $sort ) {
5954 $this->mDefaultSort
= $sort;
5955 $this->mOutput
->setProperty( 'defaultsort', $sort );
5959 * Accessor for $mDefaultSort
5960 * Will use the empty string if none is set.
5962 * This value is treated as a prefix, so the
5963 * empty string is equivalent to sorting by
5968 public function getDefaultSort() {
5969 if ( $this->mDefaultSort
!== false ) {
5970 return $this->mDefaultSort
;
5977 * Accessor for $mDefaultSort
5978 * Unlike getDefaultSort(), will return false if none is set
5980 * @return string|bool
5982 public function getCustomDefaultSort() {
5983 return $this->mDefaultSort
;
5986 private static function getSectionNameFromStrippedText( $text ) {
5987 $text = Sanitizer
::normalizeSectionNameWhitespace( $text );
5988 $text = Sanitizer
::decodeCharReferences( $text );
5989 $text = self
::normalizeSectionName( $text );
5993 private static function makeAnchor( $sectionName ) {
5994 return '#' . Sanitizer
::escapeIdForLink( $sectionName );
5997 private function makeLegacyAnchor( $sectionName ) {
5998 $fragmentMode = $this->siteConfig
->get( 'FragmentMode' );
5999 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6000 // ForAttribute() and ForLink() are the same for legacy encoding
6001 $id = Sanitizer
::escapeIdForAttribute( $sectionName, Sanitizer
::ID_FALLBACK
);
6003 $id = Sanitizer
::escapeIdForLink( $sectionName );
6010 * Try to guess the section anchor name based on a wikitext fragment
6011 * presumably extracted from a heading, for example "Header" from
6014 * @param string $text
6015 * @return string Anchor (starting with '#')
6017 public function guessSectionNameFromWikiText( $text ) {
6018 # Strip out wikitext links(they break the anchor)
6019 $text = $this->stripSectionName( $text );
6020 $sectionName = self
::getSectionNameFromStrippedText( $text );
6021 return self
::makeAnchor( $sectionName );
6025 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6026 * instead, if possible. For use in redirects, since various versions
6027 * of Microsoft browsers interpret Location: headers as something other
6028 * than UTF-8, resulting in breakage.
6030 * @param string $text The section name
6031 * @return string Anchor (starting with '#')
6033 public function guessLegacySectionNameFromWikiText( $text ) {
6034 # Strip out wikitext links(they break the anchor)
6035 $text = $this->stripSectionName( $text );
6036 $sectionName = self
::getSectionNameFromStrippedText( $text );
6037 return $this->makeLegacyAnchor( $sectionName );
6041 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6042 * @param string $text Section name (plain text)
6043 * @return string Anchor (starting with '#')
6045 public static function guessSectionNameFromStrippedText( $text ) {
6046 $sectionName = self
::getSectionNameFromStrippedText( $text );
6047 return self
::makeAnchor( $sectionName );
6051 * Apply the same normalization as code making links to this section would
6053 * @param string $text
6056 private static function normalizeSectionName( $text ) {
6057 # T90902: ensure the same normalization is applied for IDs as to links
6058 $titleParser = MediaWikiServices
::getInstance()->getTitleParser();
6061 $parts = $titleParser->splitTitleString( "#$text" );
6062 } catch ( MalformedTitleException
$ex ) {
6065 return $parts['fragment'];
6069 * Strips a text string of wikitext for use in a section anchor
6071 * Accepts a text string and then removes all wikitext from the
6072 * string and leaves only the resultant text (i.e. the result of
6073 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6074 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6075 * to create valid section anchors by mimicing the output of the
6076 * parser when headings are parsed.
6078 * @param string $text Text string to be stripped of wikitext
6079 * for use in a Section anchor
6080 * @return string Filtered text string
6082 public function stripSectionName( $text ) {
6083 # Strip internal link markup
6084 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6085 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6087 # Strip external link markup
6088 # @todo FIXME: Not tolerant to blank link text
6089 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6090 # on how many empty links there are on the page - need to figure that out.
6091 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols
. ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6093 # Parse wikitext quotes (italics & bold)
6094 $text = $this->doQuotes( $text );
6097 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
6102 * strip/replaceVariables/unstrip for preprocessor regression testing
6104 * @param string $text
6105 * @param Title $title
6106 * @param ParserOptions $options
6107 * @param int $outputType
6111 public function testSrvus( $text, Title
$title, ParserOptions
$options,
6112 $outputType = self
::OT_HTML
6114 $magicScopeVariable = $this->lock();
6115 $this->startParse( $title, $options, $outputType, true );
6117 $text = $this->replaceVariables( $text );
6118 $text = $this->mStripState
->unstripBoth( $text );
6119 $text = Sanitizer
::removeHTMLtags( $text );
6124 * @param string $text
6125 * @param Title $title
6126 * @param ParserOptions $options
6129 public function testPst( $text, Title
$title, ParserOptions
$options ) {
6130 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6134 * @param string $text
6135 * @param Title $title
6136 * @param ParserOptions $options
6139 public function testPreprocess( $text, Title
$title, ParserOptions
$options ) {
6140 return $this->testSrvus( $text, $title, $options, self
::OT_PREPROCESS
);
6144 * Call a callback function on all regions of the given text that are not
6145 * inside strip markers, and replace those regions with the return value
6146 * of the callback. For example, with input:
6150 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6151 * two strings will be replaced with the value returned by the callback in
6155 * @param callable $callback
6159 public function markerSkipCallback( $s, $callback ) {
6162 while ( $i < strlen( $s ) ) {
6163 $markerStart = strpos( $s, self
::MARKER_PREFIX
, $i );
6164 if ( $markerStart === false ) {
6165 $out .= call_user_func( $callback, substr( $s, $i ) );
6168 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6169 $markerEnd = strpos( $s, self
::MARKER_SUFFIX
, $markerStart );
6170 if ( $markerEnd === false ) {
6171 $out .= substr( $s, $markerStart );
6174 $markerEnd +
= strlen( self
::MARKER_SUFFIX
);
6175 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6184 * Remove any strip markers found in the given text.
6186 * @param string $text
6189 public function killMarkers( $text ) {
6190 return $this->mStripState
->killMarkers( $text );
6194 * Save the parser state required to convert the given half-parsed text to
6195 * HTML. "Half-parsed" in this context means the output of
6196 * recursiveTagParse() or internalParse(). This output has strip markers
6197 * from replaceVariables (extensionSubstitution() etc.), and link
6198 * placeholders from replaceLinkHolders().
6200 * Returns an array which can be serialized and stored persistently. This
6201 * array can later be loaded into another parser instance with
6202 * unserializeHalfParsedText(). The text can then be safely incorporated into
6203 * the return value of a parser hook.
6205 * @deprecated since 1.31
6206 * @param string $text
6210 public function serializeHalfParsedText( $text ) {
6211 wfDeprecated( __METHOD__
, '1.31' );
6214 'version' => self
::HALF_PARSED_VERSION
,
6215 'stripState' => $this->mStripState
->getSubState( $text ),
6216 'linkHolders' => $this->mLinkHolders
->getSubArray( $text )
6222 * Load the parser state given in the $data array, which is assumed to
6223 * have been generated by serializeHalfParsedText(). The text contents is
6224 * extracted from the array, and its markers are transformed into markers
6225 * appropriate for the current Parser instance. This transformed text is
6226 * returned, and can be safely included in the return value of a parser
6229 * If the $data array has been stored persistently, the caller should first
6230 * check whether it is still valid, by calling isValidHalfParsedText().
6232 * @deprecated since 1.31
6233 * @param array $data Serialized data
6234 * @throws MWException
6237 public function unserializeHalfParsedText( $data ) {
6238 wfDeprecated( __METHOD__
, '1.31' );
6239 if ( !isset( $data['version'] ) ||
$data['version'] != self
::HALF_PARSED_VERSION
) {
6240 throw new MWException( __METHOD__
. ': invalid version' );
6243 # First, extract the strip state.
6244 $texts = [ $data['text'] ];
6245 $texts = $this->mStripState
->merge( $data['stripState'], $texts );
6247 # Now renumber links
6248 $texts = $this->mLinkHolders
->mergeForeign( $data['linkHolders'], $texts );
6250 # Should be good to go.
6255 * Returns true if the given array, presumed to be generated by
6256 * serializeHalfParsedText(), is compatible with the current version of the
6259 * @deprecated since 1.31
6260 * @param array $data
6264 public function isValidHalfParsedText( $data ) {
6265 wfDeprecated( __METHOD__
, '1.31' );
6266 return isset( $data['version'] ) && $data['version'] == self
::HALF_PARSED_VERSION
;
6270 * Parsed a width param of imagelink like 300px or 200x300px
6272 * @param string $value
6273 * @param bool $parseHeight
6278 public static function parseWidthParam( $value, $parseHeight = true ) {
6279 $parsedWidthParam = [];
6280 if ( $value === '' ) {
6281 return $parsedWidthParam;
6284 # (T15500) In both cases (width/height and width only),
6285 # permit trailing "px" for backward compatibility.
6286 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6287 $width = intval( $m[1] );
6288 $height = intval( $m[2] );
6289 $parsedWidthParam['width'] = $width;
6290 $parsedWidthParam['height'] = $height;
6291 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6292 $width = intval( $value );
6293 $parsedWidthParam['width'] = $width;
6295 return $parsedWidthParam;
6299 * Lock the current instance of the parser.
6301 * This is meant to stop someone from calling the parser
6302 * recursively and messing up all the strip state.
6304 * @throws MWException If parser is in a parse
6305 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6307 protected function lock() {
6308 if ( $this->mInParse
) {
6309 throw new MWException( "Parser state cleared while parsing. "
6310 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse
);
6313 // Save the backtrace when locking, so that if some code tries locking again,
6314 // we can print the lock owner's backtrace for easier debugging
6316 $this->mInParse
= $e->getTraceAsString();
6318 $recursiveCheck = new ScopedCallback( function () {
6319 $this->mInParse
= false;
6322 return $recursiveCheck;
6326 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6328 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6329 * or if there is more than one <p/> tag in the input HTML.
6331 * @param string $html
6335 public static function stripOuterParagraph( $html ) {
6337 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6338 if ( strpos( $m[1], '</p>' ) === false ) {
6347 * Return this parser if it is not doing anything, otherwise
6348 * get a fresh parser. You can use this method by doing
6349 * $myParser = $wgParser->getFreshParser(), or more simply
6350 * $wgParser->getFreshParser()->parse( ... );
6351 * if you're unsure if $wgParser is safe to use.
6354 * @return Parser A parser object that is not parsing anything
6356 public function getFreshParser() {
6357 if ( $this->mInParse
) {
6358 return $this->factory
->create();
6365 * Set's up the PHP implementation of OOUI for use in this request
6366 * and instructs OutputPage to enable OOUI for itself.
6370 public function enableOOUI() {
6371 OutputPage
::setupOOUI();
6372 $this->mOutput
->setEnableOOUI( true );