DatabaseMysqlBase visibility cleanups
[mediawiki.git] / includes / parser / Parser.php
blob27c95358a09df3cb6e675e8bdf9e9717cb460670
1 <?php
2 /**
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
20 * @file
21 * @ingroup Parser
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
27 /**
28 * @defgroup Parser Parser
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
37 * There are seven main entry points into the Parser class:
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
54 * Globals used:
55 * object: $wgContLang
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
59 * @par Settings:
60 * $wgNamespacesWithSubpages
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
68 * @ingroup Parser
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
76 const VERSION = '1.6.4';
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
82 const HALF_PARSED_VERSION = 2;
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
92 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
93 # Simplified expression to match an IPv4 or IPv6 address, or
94 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
95 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
96 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
97 // @codingStandardsIgnoreStart Generic.Files.LineLength
98 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
99 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
100 // @codingStandardsIgnoreEnd
102 # Regular expression for a non-newline space
103 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105 # Flags for preprocessToDom
106 const PTD_FOR_INCLUSION = 1;
108 # Allowed values for $this->mOutputType
109 # Parameter to startExternalParse().
110 const OT_HTML = 1; # like parse()
111 const OT_WIKI = 2; # like preSaveTransform()
112 const OT_PREPROCESS = 3; # like preprocess()
113 const OT_MSG = 3;
114 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
117 * @var string Prefix and suffix for temporary replacement strings
118 * for the multipass parser.
120 * \x7f should never appear in input as it's disallowed in XML.
121 * Using it at the front also gives us a little extra robustness
122 * since it shouldn't match when butted up against identifier-like
123 * string constructs.
125 * Must not consist of all title characters, or else it will change
126 * the behavior of <nowiki> in a link.
128 * Must have a character that needs escaping in attributes, otherwise
129 * someone could put a strip marker in an attribute, to get around
130 * escaping quote marks, and break out of the attribute. Thus we add
131 * `'".
133 const MARKER_SUFFIX = "-QINU`\"'\x7f";
134 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
136 # Markers used for wrapping the table of contents
137 const TOC_START = '<mw:toc>';
138 const TOC_END = '</mw:toc>';
140 # Persistent:
141 public $mTagHooks = [];
142 public $mTransparentTagHooks = [];
143 public $mFunctionHooks = [];
144 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
145 public $mFunctionTagHooks = [];
146 public $mStripList = [];
147 public $mDefaultStripList = [];
148 public $mVarCache = [];
149 public $mImageParams = [];
150 public $mImageParamsMagicArray = [];
151 public $mMarkerIndex = 0;
152 public $mFirstCall = true;
154 # Initialised by initialiseVariables()
157 * @var MagicWordArray
159 public $mVariables;
162 * @var MagicWordArray
164 public $mSubstWords;
165 # Initialised in constructor
166 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
168 # Initialized in getPreprocessor()
169 /** @var Preprocessor */
170 public $mPreprocessor;
172 # Cleared with clearState():
174 * @var ParserOutput
176 public $mOutput;
177 public $mAutonumber;
180 * @var StripState
182 public $mStripState;
184 public $mIncludeCount;
186 * @var LinkHolderArray
188 public $mLinkHolders;
190 public $mLinkID;
191 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
192 public $mDefaultSort;
193 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
194 public $mExpensiveFunctionCount; # number of expensive parser function calls
195 public $mShowToc, $mForceTocPosition;
198 * @var User
200 public $mUser; # User object; only used when doing pre-save transform
202 # Temporary
203 # These are variables reset at least once per parse regardless of $clearState
206 * @var ParserOptions
208 public $mOptions;
211 * @var Title
213 public $mTitle; # Title context, used for self-link rendering and similar things
214 public $mOutputType; # Output type, one of the OT_xxx constants
215 public $ot; # Shortcut alias, see setOutputType()
216 public $mRevisionObject; # The revision object of the specified revision ID
217 public $mRevisionId; # ID to display in {{REVISIONID}} tags
218 public $mRevisionTimestamp; # The timestamp of the specified revision ID
219 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
220 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
221 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
222 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225 * @var string Deprecated accessor for the strip marker prefix.
226 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
228 public $mUniqPrefix = Parser::MARKER_PREFIX;
231 * @var array Array with the language name of each language link (i.e. the
232 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
233 * duplicate language links to the ParserOutput.
235 public $mLangLinkLanguages;
238 * @var MapCacheLRU|null
239 * @since 1.24
241 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
243 public $currentRevisionCache;
246 * @var bool Recursive call protection.
247 * This variable should be treated as if it were private.
249 public $mInParse = false;
251 /** @var SectionProfiler */
252 protected $mProfiler;
255 * @var LinkRenderer
257 protected $mLinkRenderer;
260 * @param array $conf
262 public function __construct( $conf = [] ) {
263 $this->mConf = $conf;
264 $this->mUrlProtocols = wfUrlProtocols();
265 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
266 self::EXT_LINK_ADDR .
267 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
268 if ( isset( $conf['preprocessorClass'] ) ) {
269 $this->mPreprocessorClass = $conf['preprocessorClass'];
270 } elseif ( defined( 'HPHP_VERSION' ) ) {
271 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
272 $this->mPreprocessorClass = 'Preprocessor_Hash';
273 } elseif ( extension_loaded( 'domxml' ) ) {
274 # PECL extension that conflicts with the core DOM extension (bug 13770)
275 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
276 $this->mPreprocessorClass = 'Preprocessor_Hash';
277 } elseif ( extension_loaded( 'dom' ) ) {
278 $this->mPreprocessorClass = 'Preprocessor_DOM';
279 } else {
280 $this->mPreprocessorClass = 'Preprocessor_Hash';
282 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
286 * Reduce memory usage to reduce the impact of circular references
288 public function __destruct() {
289 if ( isset( $this->mLinkHolders ) ) {
290 unset( $this->mLinkHolders );
292 foreach ( $this as $name => $value ) {
293 unset( $this->$name );
298 * Allow extensions to clean up when the parser is cloned
300 public function __clone() {
301 $this->mInParse = false;
303 // Bug 56226: When you create a reference "to" an object field, that
304 // makes the object field itself be a reference too (until the other
305 // reference goes out of scope). When cloning, any field that's a
306 // reference is copied as a reference in the new object. Both of these
307 // are defined PHP5 behaviors, as inconvenient as it is for us when old
308 // hooks from PHP4 days are passing fields by reference.
309 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
310 // Make a non-reference copy of the field, then rebind the field to
311 // reference the new copy.
312 $tmp = $this->$k;
313 $this->$k =& $tmp;
314 unset( $tmp );
317 Hooks::run( 'ParserCloned', [ $this ] );
321 * Do various kinds of initialisation on the first call of the parser
323 public function firstCallInit() {
324 if ( !$this->mFirstCall ) {
325 return;
327 $this->mFirstCall = false;
329 CoreParserFunctions::register( $this );
330 CoreTagHooks::register( $this );
331 $this->initialiseVariables();
333 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
337 * Clear Parser state
339 * @private
341 public function clearState() {
342 if ( $this->mFirstCall ) {
343 $this->firstCallInit();
345 $this->mOutput = new ParserOutput;
346 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
347 $this->mAutonumber = 0;
348 $this->mIncludeCount = [];
349 $this->mLinkHolders = new LinkHolderArray( $this );
350 $this->mLinkID = 0;
351 $this->mRevisionObject = $this->mRevisionTimestamp =
352 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
353 $this->mVarCache = [];
354 $this->mUser = null;
355 $this->mLangLinkLanguages = [];
356 $this->currentRevisionCache = null;
358 $this->mStripState = new StripState;
360 # Clear these on every parse, bug 4549
361 $this->mTplRedirCache = $this->mTplDomCache = [];
363 $this->mShowToc = true;
364 $this->mForceTocPosition = false;
365 $this->mIncludeSizes = [
366 'post-expand' => 0,
367 'arg' => 0,
369 $this->mPPNodeCount = 0;
370 $this->mGeneratedPPNodeCount = 0;
371 $this->mHighestExpansionDepth = 0;
372 $this->mDefaultSort = false;
373 $this->mHeadings = [];
374 $this->mDoubleUnderscores = [];
375 $this->mExpensiveFunctionCount = 0;
377 # Fix cloning
378 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
379 $this->mPreprocessor = null;
382 $this->mProfiler = new SectionProfiler();
384 Hooks::run( 'ParserClearState', [ &$this ] );
388 * Convert wikitext to HTML
389 * Do not call this function recursively.
391 * @param string $text Text we want to parse
392 * @param Title $title
393 * @param ParserOptions $options
394 * @param bool $linestart
395 * @param bool $clearState
396 * @param int $revid Number to pass in {{REVISIONID}}
397 * @return ParserOutput A ParserOutput
399 public function parse(
400 $text, Title $title, ParserOptions $options,
401 $linestart = true, $clearState = true, $revid = null
404 * First pass--just handle <nowiki> sections, pass the rest off
405 * to internalParse() which does all the real work.
408 global $wgShowHostnames;
410 if ( $clearState ) {
411 // We use U+007F DELETE to construct strip markers, so we have to make
412 // sure that this character does not occur in the input text.
413 $text = strtr( $text, "\x7f", "?" );
414 $magicScopeVariable = $this->lock();
417 $this->startParse( $title, $options, self::OT_HTML, $clearState );
419 $this->currentRevisionCache = null;
420 $this->mInputSize = strlen( $text );
421 if ( $this->mOptions->getEnableLimitReport() ) {
422 $this->mOutput->resetParseStartTime();
425 $oldRevisionId = $this->mRevisionId;
426 $oldRevisionObject = $this->mRevisionObject;
427 $oldRevisionTimestamp = $this->mRevisionTimestamp;
428 $oldRevisionUser = $this->mRevisionUser;
429 $oldRevisionSize = $this->mRevisionSize;
430 if ( $revid !== null ) {
431 $this->mRevisionId = $revid;
432 $this->mRevisionObject = null;
433 $this->mRevisionTimestamp = null;
434 $this->mRevisionUser = null;
435 $this->mRevisionSize = null;
438 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
439 # No more strip!
440 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
441 $text = $this->internalParse( $text );
442 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
444 $text = $this->internalParseHalfParsed( $text, true, $linestart );
447 * A converted title will be provided in the output object if title and
448 * content conversion are enabled, the article text does not contain
449 * a conversion-suppressing double-underscore tag, and no
450 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
451 * automatic link conversion.
453 if ( !( $options->getDisableTitleConversion()
454 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
455 || isset( $this->mDoubleUnderscores['notitleconvert'] )
456 || $this->mOutput->getDisplayTitle() !== false )
458 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
459 if ( $convruletitle ) {
460 $this->mOutput->setTitleText( $convruletitle );
461 } else {
462 $titleText = $this->getConverterLanguage()->convertTitle( $title );
463 $this->mOutput->setTitleText( $titleText );
467 # Done parsing! Compute runtime adaptive expiry if set
468 $this->mOutput->finalizeAdaptiveCacheExpiry();
470 # Warn if too many heavyweight parser functions were used
471 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
472 $this->limitationWarn( 'expensive-parserfunction',
473 $this->mExpensiveFunctionCount,
474 $this->mOptions->getExpensiveParserFunctionLimit()
478 # Information on include size limits, for the benefit of users who try to skirt them
479 if ( $this->mOptions->getEnableLimitReport() ) {
480 $max = $this->mOptions->getMaxIncludeSize();
482 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
483 if ( $cpuTime !== null ) {
484 $this->mOutput->setLimitReportData( 'limitreport-cputime',
485 sprintf( "%.3f", $cpuTime )
489 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
490 $this->mOutput->setLimitReportData( 'limitreport-walltime',
491 sprintf( "%.3f", $wallTime )
494 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
495 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
497 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
498 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
500 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
501 [ $this->mIncludeSizes['post-expand'], $max ]
503 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
504 [ $this->mIncludeSizes['arg'], $max ]
506 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
507 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
509 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
510 [ $this->mExpensiveFunctionCount,
511 $this->mOptions->getExpensiveParserFunctionLimit() ]
513 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
515 $limitReport = '';
516 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
517 if ( $limitReport != '' ) {
518 // Sanitize for comment. Note '‐' in the replacement is U+2010,
519 // which looks much like the problematic '-'.
520 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
521 $text .= "\n<!-- \nNewPP limit report\n$limitReport-->\n";
524 // Add on template profiling data in human/machine readable way
525 $dataByFunc = $this->mProfiler->getFunctionStats();
526 uasort( $dataByFunc, function ( $a, $b ) {
527 return $a['real'] < $b['real']; // descending order
528 } );
529 $profileReport = [];
530 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
531 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
532 $item['%real'], $item['real'], $item['calls'], $item['name'] );
534 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
536 // Add other cache related metadata
537 if ( $wgShowHostnames ) {
538 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
540 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
541 $this->mOutput->getCacheTime() );
542 $this->mOutput->setLimitReportData( 'cachereport-ttl',
543 $this->mOutput->getCacheExpiry() );
544 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
545 $this->mOutput->hasDynamicContent() );
547 if ( $this->mGeneratedPPNodeCount
548 > $this->mOptions->getMaxGeneratedPPNodeCount() / 10
550 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
551 $this->mTitle->getPrefixedDBkey() );
554 $this->mOutput->setText( $text );
556 $this->mRevisionId = $oldRevisionId;
557 $this->mRevisionObject = $oldRevisionObject;
558 $this->mRevisionTimestamp = $oldRevisionTimestamp;
559 $this->mRevisionUser = $oldRevisionUser;
560 $this->mRevisionSize = $oldRevisionSize;
561 $this->mInputSize = false;
562 $this->currentRevisionCache = null;
564 return $this->mOutput;
568 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
569 * can be called from an extension tag hook.
571 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
572 * instead, which means that lists and links have not been fully parsed yet,
573 * and strip markers are still present.
575 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
577 * Use this function if you're a parser tag hook and you want to parse
578 * wikitext before or after applying additional transformations, and you
579 * intend to *return the result as hook output*, which will cause it to go
580 * through the rest of parsing process automatically.
582 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
583 * $text are not expanded
585 * @param string $text Text extension wants to have parsed
586 * @param bool|PPFrame $frame The frame to use for expanding any template variables
587 * @return string UNSAFE half-parsed HTML
589 public function recursiveTagParse( $text, $frame = false ) {
590 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
591 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
592 $text = $this->internalParse( $text, false, $frame );
593 return $text;
597 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
598 * point can be called from an extension tag hook.
600 * The output of this function is fully-parsed HTML that is safe for output.
601 * If you're a parser tag hook, you might want to use recursiveTagParse()
602 * instead.
604 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
605 * $text are not expanded
607 * @since 1.25
609 * @param string $text Text extension wants to have parsed
610 * @param bool|PPFrame $frame The frame to use for expanding any template variables
611 * @return string Fully parsed HTML
613 public function recursiveTagParseFully( $text, $frame = false ) {
614 $text = $this->recursiveTagParse( $text, $frame );
615 $text = $this->internalParseHalfParsed( $text, false );
616 return $text;
620 * Expand templates and variables in the text, producing valid, static wikitext.
621 * Also removes comments.
622 * Do not call this function recursively.
623 * @param string $text
624 * @param Title $title
625 * @param ParserOptions $options
626 * @param int|null $revid
627 * @param bool|PPFrame $frame
628 * @return mixed|string
630 public function preprocess( $text, Title $title = null,
631 ParserOptions $options, $revid = null, $frame = false
633 $magicScopeVariable = $this->lock();
634 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
635 if ( $revid !== null ) {
636 $this->mRevisionId = $revid;
638 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
639 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
640 $text = $this->replaceVariables( $text, $frame );
641 $text = $this->mStripState->unstripBoth( $text );
642 return $text;
646 * Recursive parser entry point that can be called from an extension tag
647 * hook.
649 * @param string $text Text to be expanded
650 * @param bool|PPFrame $frame The frame to use for expanding any template variables
651 * @return string
652 * @since 1.19
654 public function recursivePreprocess( $text, $frame = false ) {
655 $text = $this->replaceVariables( $text, $frame );
656 $text = $this->mStripState->unstripBoth( $text );
657 return $text;
661 * Process the wikitext for the "?preload=" feature. (bug 5210)
663 * "<noinclude>", "<includeonly>" etc. are parsed as for template
664 * transclusion, comments, templates, arguments, tags hooks and parser
665 * functions are untouched.
667 * @param string $text
668 * @param Title $title
669 * @param ParserOptions $options
670 * @param array $params
671 * @return string
673 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
674 $msg = new RawMessage( $text );
675 $text = $msg->params( $params )->plain();
677 # Parser (re)initialisation
678 $magicScopeVariable = $this->lock();
679 $this->startParse( $title, $options, self::OT_PLAIN, true );
681 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
682 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
683 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
684 $text = $this->mStripState->unstripBoth( $text );
685 return $text;
689 * Get a random string
691 * @return string
692 * @deprecated since 1.26; use wfRandomString() instead.
694 public static function getRandomString() {
695 wfDeprecated( __METHOD__, '1.26' );
696 return wfRandomString( 16 );
700 * Set the current user.
701 * Should only be used when doing pre-save transform.
703 * @param User|null $user User object or null (to reset)
705 public function setUser( $user ) {
706 $this->mUser = $user;
710 * Accessor for mUniqPrefix.
712 * @return string
713 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
715 public function uniqPrefix() {
716 wfDeprecated( __METHOD__, '1.26' );
717 return self::MARKER_PREFIX;
721 * Set the context title
723 * @param Title $t
725 public function setTitle( $t ) {
726 if ( !$t ) {
727 $t = Title::newFromText( 'NO TITLE' );
730 if ( $t->hasFragment() ) {
731 # Strip the fragment to avoid various odd effects
732 $this->mTitle = $t->createFragmentTarget( '' );
733 } else {
734 $this->mTitle = $t;
739 * Accessor for the Title object
741 * @return Title
743 public function getTitle() {
744 return $this->mTitle;
748 * Accessor/mutator for the Title object
750 * @param Title $x Title object or null to just get the current one
751 * @return Title
753 public function Title( $x = null ) {
754 return wfSetVar( $this->mTitle, $x );
758 * Set the output type
760 * @param int $ot New value
762 public function setOutputType( $ot ) {
763 $this->mOutputType = $ot;
764 # Shortcut alias
765 $this->ot = [
766 'html' => $ot == self::OT_HTML,
767 'wiki' => $ot == self::OT_WIKI,
768 'pre' => $ot == self::OT_PREPROCESS,
769 'plain' => $ot == self::OT_PLAIN,
774 * Accessor/mutator for the output type
776 * @param int|null $x New value or null to just get the current one
777 * @return int
779 public function OutputType( $x = null ) {
780 return wfSetVar( $this->mOutputType, $x );
784 * Get the ParserOutput object
786 * @return ParserOutput
788 public function getOutput() {
789 return $this->mOutput;
793 * Get the ParserOptions object
795 * @return ParserOptions
797 public function getOptions() {
798 return $this->mOptions;
802 * Accessor/mutator for the ParserOptions object
804 * @param ParserOptions $x New value or null to just get the current one
805 * @return ParserOptions Current ParserOptions object
807 public function Options( $x = null ) {
808 return wfSetVar( $this->mOptions, $x );
812 * @return int
814 public function nextLinkID() {
815 return $this->mLinkID++;
819 * @param int $id
821 public function setLinkID( $id ) {
822 $this->mLinkID = $id;
826 * Get a language object for use in parser functions such as {{FORMATNUM:}}
827 * @return Language
829 public function getFunctionLang() {
830 return $this->getTargetLanguage();
834 * Get the target language for the content being parsed. This is usually the
835 * language that the content is in.
837 * @since 1.19
839 * @throws MWException
840 * @return Language
842 public function getTargetLanguage() {
843 $target = $this->mOptions->getTargetLanguage();
845 if ( $target !== null ) {
846 return $target;
847 } elseif ( $this->mOptions->getInterfaceMessage() ) {
848 return $this->mOptions->getUserLangObj();
849 } elseif ( is_null( $this->mTitle ) ) {
850 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
853 return $this->mTitle->getPageLanguage();
857 * Get the language object for language conversion
858 * @return Language|null
860 public function getConverterLanguage() {
861 return $this->getTargetLanguage();
865 * Get a User object either from $this->mUser, if set, or from the
866 * ParserOptions object otherwise
868 * @return User
870 public function getUser() {
871 if ( !is_null( $this->mUser ) ) {
872 return $this->mUser;
874 return $this->mOptions->getUser();
878 * Get a preprocessor object
880 * @return Preprocessor
882 public function getPreprocessor() {
883 if ( !isset( $this->mPreprocessor ) ) {
884 $class = $this->mPreprocessorClass;
885 $this->mPreprocessor = new $class( $this );
887 return $this->mPreprocessor;
891 * Get a LinkRenderer instance to make links with
893 * @since 1.28
894 * @return LinkRenderer
896 public function getLinkRenderer() {
897 if ( !$this->mLinkRenderer ) {
898 $this->mLinkRenderer = MediaWikiServices::getInstance()
899 ->getLinkRendererFactory()->create();
900 $this->mLinkRenderer->setStubThreshold(
901 $this->getOptions()->getStubThreshold()
905 return $this->mLinkRenderer;
909 * Replaces all occurrences of HTML-style comments and the given tags
910 * in the text with a random marker and returns the next text. The output
911 * parameter $matches will be an associative array filled with data in
912 * the form:
914 * @code
915 * 'UNIQ-xxxxx' => [
916 * 'element',
917 * 'tag content',
918 * [ 'param' => 'x' ],
919 * '<element param="x">tag content</element>' ]
920 * @endcode
922 * @param array $elements List of element names. Comments are always extracted.
923 * @param string $text Source text string.
924 * @param array $matches Out parameter, Array: extracted tags
925 * @param string|null $uniq_prefix
926 * @return string Stripped text
927 * @since 1.26 The uniq_prefix argument is deprecated.
929 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
930 if ( $uniq_prefix !== null ) {
931 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
933 static $n = 1;
934 $stripped = '';
935 $matches = [];
937 $taglist = implode( '|', $elements );
938 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
940 while ( $text != '' ) {
941 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
942 $stripped .= $p[0];
943 if ( count( $p ) < 5 ) {
944 break;
946 if ( count( $p ) > 5 ) {
947 # comment
948 $element = $p[4];
949 $attributes = '';
950 $close = '';
951 $inside = $p[5];
952 } else {
953 # tag
954 $element = $p[1];
955 $attributes = $p[2];
956 $close = $p[3];
957 $inside = $p[4];
960 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
961 $stripped .= $marker;
963 if ( $close === '/>' ) {
964 # Empty element tag, <tag />
965 $content = null;
966 $text = $inside;
967 $tail = null;
968 } else {
969 if ( $element === '!--' ) {
970 $end = '/(-->)/';
971 } else {
972 $end = "/(<\\/$element\\s*>)/i";
974 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
975 $content = $q[0];
976 if ( count( $q ) < 3 ) {
977 # No end tag -- let it run out to the end of the text.
978 $tail = '';
979 $text = '';
980 } else {
981 $tail = $q[1];
982 $text = $q[2];
986 $matches[$marker] = [ $element,
987 $content,
988 Sanitizer::decodeTagAttributes( $attributes ),
989 "<$element$attributes$close$content$tail" ];
991 return $stripped;
995 * Get a list of strippable XML-like elements
997 * @return array
999 public function getStripList() {
1000 return $this->mStripList;
1004 * Add an item to the strip state
1005 * Returns the unique tag which must be inserted into the stripped text
1006 * The tag will be replaced with the original text in unstrip()
1008 * @param string $text
1010 * @return string
1012 public function insertStripItem( $text ) {
1013 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1014 $this->mMarkerIndex++;
1015 $this->mStripState->addGeneral( $marker, $text );
1016 return $marker;
1020 * parse the wiki syntax used to render tables
1022 * @private
1023 * @param string $text
1024 * @return string
1026 public function doTableStuff( $text ) {
1028 $lines = StringUtils::explode( "\n", $text );
1029 $out = '';
1030 $td_history = []; # Is currently a td tag open?
1031 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1032 $tr_history = []; # Is currently a tr tag open?
1033 $tr_attributes = []; # history of tr attributes
1034 $has_opened_tr = []; # Did this table open a <tr> element?
1035 $indent_level = 0; # indent level of the table
1037 foreach ( $lines as $outLine ) {
1038 $line = trim( $outLine );
1040 if ( $line === '' ) { # empty line, go to next line
1041 $out .= $outLine . "\n";
1042 continue;
1045 $first_character = $line[0];
1046 $first_two = substr( $line, 0, 2 );
1047 $matches = [];
1049 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1050 # First check if we are starting a new table
1051 $indent_level = strlen( $matches[1] );
1053 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1054 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1056 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1057 array_push( $td_history, false );
1058 array_push( $last_tag_history, '' );
1059 array_push( $tr_history, false );
1060 array_push( $tr_attributes, '' );
1061 array_push( $has_opened_tr, false );
1062 } elseif ( count( $td_history ) == 0 ) {
1063 # Don't do any of the following
1064 $out .= $outLine . "\n";
1065 continue;
1066 } elseif ( $first_two === '|}' ) {
1067 # We are ending a table
1068 $line = '</table>' . substr( $line, 2 );
1069 $last_tag = array_pop( $last_tag_history );
1071 if ( !array_pop( $has_opened_tr ) ) {
1072 $line = "<tr><td></td></tr>{$line}";
1075 if ( array_pop( $tr_history ) ) {
1076 $line = "</tr>{$line}";
1079 if ( array_pop( $td_history ) ) {
1080 $line = "</{$last_tag}>{$line}";
1082 array_pop( $tr_attributes );
1083 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1084 } elseif ( $first_two === '|-' ) {
1085 # Now we have a table row
1086 $line = preg_replace( '#^\|-+#', '', $line );
1088 # Whats after the tag is now only attributes
1089 $attributes = $this->mStripState->unstripBoth( $line );
1090 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1091 array_pop( $tr_attributes );
1092 array_push( $tr_attributes, $attributes );
1094 $line = '';
1095 $last_tag = array_pop( $last_tag_history );
1096 array_pop( $has_opened_tr );
1097 array_push( $has_opened_tr, true );
1099 if ( array_pop( $tr_history ) ) {
1100 $line = '</tr>';
1103 if ( array_pop( $td_history ) ) {
1104 $line = "</{$last_tag}>{$line}";
1107 $outLine = $line;
1108 array_push( $tr_history, false );
1109 array_push( $td_history, false );
1110 array_push( $last_tag_history, '' );
1111 } elseif ( $first_character === '|'
1112 || $first_character === '!'
1113 || $first_two === '|+'
1115 # This might be cell elements, td, th or captions
1116 if ( $first_two === '|+' ) {
1117 $first_character = '+';
1118 $line = substr( $line, 2 );
1119 } else {
1120 $line = substr( $line, 1 );
1123 // Implies both are valid for table headings.
1124 if ( $first_character === '!' ) {
1125 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1128 # Split up multiple cells on the same line.
1129 # FIXME : This can result in improper nesting of tags processed
1130 # by earlier parser steps.
1131 $cells = explode( '||', $line );
1133 $outLine = '';
1135 # Loop through each table cell
1136 foreach ( $cells as $cell ) {
1137 $previous = '';
1138 if ( $first_character !== '+' ) {
1139 $tr_after = array_pop( $tr_attributes );
1140 if ( !array_pop( $tr_history ) ) {
1141 $previous = "<tr{$tr_after}>\n";
1143 array_push( $tr_history, true );
1144 array_push( $tr_attributes, '' );
1145 array_pop( $has_opened_tr );
1146 array_push( $has_opened_tr, true );
1149 $last_tag = array_pop( $last_tag_history );
1151 if ( array_pop( $td_history ) ) {
1152 $previous = "</{$last_tag}>\n{$previous}";
1155 if ( $first_character === '|' ) {
1156 $last_tag = 'td';
1157 } elseif ( $first_character === '!' ) {
1158 $last_tag = 'th';
1159 } elseif ( $first_character === '+' ) {
1160 $last_tag = 'caption';
1161 } else {
1162 $last_tag = '';
1165 array_push( $last_tag_history, $last_tag );
1167 # A cell could contain both parameters and data
1168 $cell_data = explode( '|', $cell, 2 );
1170 # Bug 553: Note that a '|' inside an invalid link should not
1171 # be mistaken as delimiting cell parameters
1172 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1173 $cell = "{$previous}<{$last_tag}>{$cell}";
1174 } elseif ( count( $cell_data ) == 1 ) {
1175 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1176 } else {
1177 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1178 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1179 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1182 $outLine .= $cell;
1183 array_push( $td_history, true );
1186 $out .= $outLine . "\n";
1189 # Closing open td, tr && table
1190 while ( count( $td_history ) > 0 ) {
1191 if ( array_pop( $td_history ) ) {
1192 $out .= "</td>\n";
1194 if ( array_pop( $tr_history ) ) {
1195 $out .= "</tr>\n";
1197 if ( !array_pop( $has_opened_tr ) ) {
1198 $out .= "<tr><td></td></tr>\n";
1201 $out .= "</table>\n";
1204 # Remove trailing line-ending (b/c)
1205 if ( substr( $out, -1 ) === "\n" ) {
1206 $out = substr( $out, 0, -1 );
1209 # special case: don't return empty table
1210 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1211 $out = '';
1214 return $out;
1218 * Helper function for parse() that transforms wiki markup into half-parsed
1219 * HTML. Only called for $mOutputType == self::OT_HTML.
1221 * @private
1223 * @param string $text The text to parse
1224 * @param bool $isMain Whether this is being called from the main parse() function
1225 * @param PPFrame|bool $frame A pre-processor frame
1227 * @return string
1229 public function internalParse( $text, $isMain = true, $frame = false ) {
1231 $origText = $text;
1233 # Hook to suspend the parser in this state
1234 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1235 return $text;
1238 # if $frame is provided, then use $frame for replacing any variables
1239 if ( $frame ) {
1240 # use frame depth to infer how include/noinclude tags should be handled
1241 # depth=0 means this is the top-level document; otherwise it's an included document
1242 if ( !$frame->depth ) {
1243 $flag = 0;
1244 } else {
1245 $flag = Parser::PTD_FOR_INCLUSION;
1247 $dom = $this->preprocessToDom( $text, $flag );
1248 $text = $frame->expand( $dom );
1249 } else {
1250 # if $frame is not provided, then use old-style replaceVariables
1251 $text = $this->replaceVariables( $text );
1254 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1255 $text = Sanitizer::removeHTMLtags(
1256 $text,
1257 [ &$this, 'attributeStripCallback' ],
1258 false,
1259 array_keys( $this->mTransparentTagHooks ),
1261 [ &$this, 'addTrackingCategory' ]
1263 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1265 # Tables need to come after variable replacement for things to work
1266 # properly; putting them before other transformations should keep
1267 # exciting things like link expansions from showing up in surprising
1268 # places.
1269 $text = $this->doTableStuff( $text );
1271 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1273 $text = $this->doDoubleUnderscore( $text );
1275 $text = $this->doHeadings( $text );
1276 $text = $this->replaceInternalLinks( $text );
1277 $text = $this->doAllQuotes( $text );
1278 $text = $this->replaceExternalLinks( $text );
1280 # replaceInternalLinks may sometimes leave behind
1281 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1282 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1284 $text = $this->doMagicLinks( $text );
1285 $text = $this->formatHeadings( $text, $origText, $isMain );
1287 return $text;
1291 * Helper function for parse() that transforms half-parsed HTML into fully
1292 * parsed HTML.
1294 * @param string $text
1295 * @param bool $isMain
1296 * @param bool $linestart
1297 * @return string
1299 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1300 $text = $this->mStripState->unstripGeneral( $text );
1302 if ( $isMain ) {
1303 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1306 # Clean up special characters, only run once, next-to-last before doBlockLevels
1307 $fixtags = [
1308 # french spaces, last one Guillemet-left
1309 # only if there is something before the space
1310 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1311 # french spaces, Guillemet-right
1312 '/(\\302\\253) /' => '\\1&#160;',
1313 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1315 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1317 $text = $this->doBlockLevels( $text, $linestart );
1319 $this->replaceLinkHolders( $text );
1322 * The input doesn't get language converted if
1323 * a) It's disabled
1324 * b) Content isn't converted
1325 * c) It's a conversion table
1326 * d) it is an interface message (which is in the user language)
1328 if ( !( $this->mOptions->getDisableContentConversion()
1329 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1331 if ( !$this->mOptions->getInterfaceMessage() ) {
1332 # The position of the convert() call should not be changed. it
1333 # assumes that the links are all replaced and the only thing left
1334 # is the <nowiki> mark.
1335 $text = $this->getConverterLanguage()->convert( $text );
1339 $text = $this->mStripState->unstripNoWiki( $text );
1341 if ( $isMain ) {
1342 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1345 $text = $this->replaceTransparentTags( $text );
1346 $text = $this->mStripState->unstripGeneral( $text );
1348 $text = Sanitizer::normalizeCharReferences( $text );
1350 if ( MWTidy::isEnabled() ) {
1351 if ( $this->mOptions->getTidy() ) {
1352 $text = MWTidy::tidy( $text );
1354 } else {
1355 # attempt to sanitize at least some nesting problems
1356 # (bug #2702 and quite a few others)
1357 $tidyregs = [
1358 # ''Something [http://www.cool.com cool''] -->
1359 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1360 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1361 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1362 # fix up an anchor inside another anchor, only
1363 # at least for a single single nested link (bug 3695)
1364 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1365 '\\1\\2</a>\\3</a>\\1\\4</a>',
1366 # fix div inside inline elements- doBlockLevels won't wrap a line which
1367 # contains a div, so fix it up here; replace
1368 # div with escaped text
1369 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1370 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1371 # remove empty italic or bold tag pairs, some
1372 # introduced by rules above
1373 '/<([bi])><\/\\1>/' => '',
1376 $text = preg_replace(
1377 array_keys( $tidyregs ),
1378 array_values( $tidyregs ),
1379 $text );
1382 if ( $isMain ) {
1383 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1386 return $text;
1390 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1391 * magic external links.
1393 * DML
1394 * @private
1396 * @param string $text
1398 * @return string
1400 public function doMagicLinks( $text ) {
1401 $prots = wfUrlProtocolsWithoutProtRel();
1402 $urlChar = self::EXT_LINK_URL_CLASS;
1403 $addr = self::EXT_LINK_ADDR;
1404 $space = self::SPACE_NOT_NL; # non-newline space
1405 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1406 $spaces = "$space++"; # possessive match of 1 or more spaces
1407 $text = preg_replace_callback(
1408 '!(?: # Start cases
1409 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1410 (<.*?>) | # m[2]: Skip stuff inside
1411 # HTML elements' . "
1412 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1413 # m[4]: Post-protocol path
1414 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1415 ([0-9]+)\b |
1416 \bISBN $spaces ( # m[6]: ISBN, capture number
1417 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1418 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1419 [0-9Xx] # check digit
1421 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1422 return $text;
1426 * @throws MWException
1427 * @param array $m
1428 * @return HTML|string
1430 public function magicLinkCallback( $m ) {
1431 if ( isset( $m[1] ) && $m[1] !== '' ) {
1432 # Skip anchor
1433 return $m[0];
1434 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1435 # Skip HTML element
1436 return $m[0];
1437 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1438 # Free external link
1439 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1440 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1441 # RFC or PMID
1442 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1443 if ( !$this->mOptions->getMagicRFCLinks() ) {
1444 return $m[0];
1446 $keyword = 'RFC';
1447 $urlmsg = 'rfcurl';
1448 $cssClass = 'mw-magiclink-rfc';
1449 $id = $m[5];
1450 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1451 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1452 return $m[0];
1454 $keyword = 'PMID';
1455 $urlmsg = 'pubmedurl';
1456 $cssClass = 'mw-magiclink-pmid';
1457 $id = $m[5];
1458 } else {
1459 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1460 substr( $m[0], 0, 20 ) . '"' );
1462 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1463 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1464 } elseif ( isset( $m[6] ) && $m[6] !== ''
1465 && $this->mOptions->getMagicISBNLinks()
1467 # ISBN
1468 $isbn = $m[6];
1469 $space = self::SPACE_NOT_NL; # non-newline space
1470 $isbn = preg_replace( "/$space/", ' ', $isbn );
1471 $num = strtr( $isbn, [
1472 '-' => '',
1473 ' ' => '',
1474 'x' => 'X',
1475 ] );
1476 return $this->getLinkRenderer()->makeKnownLink(
1477 SpecialPage::getTitleFor( 'Booksources', $num ),
1478 "ISBN $isbn",
1480 'class' => 'internal mw-magiclink-isbn',
1481 'title' => false // suppress title attribute
1484 } else {
1485 return $m[0];
1490 * Make a free external link, given a user-supplied URL
1492 * @param string $url
1493 * @param int $numPostProto
1494 * The number of characters after the protocol.
1495 * @return string HTML
1496 * @private
1498 public function makeFreeExternalLink( $url, $numPostProto ) {
1499 $trail = '';
1501 # The characters '<' and '>' (which were escaped by
1502 # removeHTMLtags()) should not be included in
1503 # URLs, per RFC 2396.
1504 # Make &nbsp; terminate a URL as well (bug T84937)
1505 $m2 = [];
1506 if ( preg_match(
1507 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1508 $url,
1509 $m2,
1510 PREG_OFFSET_CAPTURE
1511 ) ) {
1512 $trail = substr( $url, $m2[0][1] ) . $trail;
1513 $url = substr( $url, 0, $m2[0][1] );
1516 # Move trailing punctuation to $trail
1517 $sep = ',;\.:!?';
1518 # If there is no left bracket, then consider right brackets fair game too
1519 if ( strpos( $url, '(' ) === false ) {
1520 $sep .= ')';
1523 $urlRev = strrev( $url );
1524 $numSepChars = strspn( $urlRev, $sep );
1525 # Don't break a trailing HTML entity by moving the ; into $trail
1526 # This is in hot code, so use substr_compare to avoid having to
1527 # create a new string object for the comparison
1528 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1529 # more optimization: instead of running preg_match with a $
1530 # anchor, which can be slow, do the match on the reversed
1531 # string starting at the desired offset.
1532 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1533 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1534 $numSepChars--;
1537 if ( $numSepChars ) {
1538 $trail = substr( $url, -$numSepChars ) . $trail;
1539 $url = substr( $url, 0, -$numSepChars );
1542 # Verify that we still have a real URL after trail removal, and
1543 # not just lone protocol
1544 if ( strlen( $trail ) >= $numPostProto ) {
1545 return $url . $trail;
1548 $url = Sanitizer::cleanUrl( $url );
1550 # Is this an external image?
1551 $text = $this->maybeMakeExternalImage( $url );
1552 if ( $text === false ) {
1553 # Not an image, make a link
1554 $text = Linker::makeExternalLink( $url,
1555 $this->getConverterLanguage()->markNoConversion( $url, true ),
1556 true, 'free',
1557 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1558 # Register it in the output object...
1559 # Replace unnecessary URL escape codes with their equivalent characters
1560 $pasteurized = self::normalizeLinkUrl( $url );
1561 $this->mOutput->addExternalLink( $pasteurized );
1563 return $text . $trail;
1567 * Parse headers and return html
1569 * @private
1571 * @param string $text
1573 * @return string
1575 public function doHeadings( $text ) {
1576 for ( $i = 6; $i >= 1; --$i ) {
1577 $h = str_repeat( '=', $i );
1578 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1580 return $text;
1584 * Replace single quotes with HTML markup
1585 * @private
1587 * @param string $text
1589 * @return string The altered text
1591 public function doAllQuotes( $text ) {
1592 $outtext = '';
1593 $lines = StringUtils::explode( "\n", $text );
1594 foreach ( $lines as $line ) {
1595 $outtext .= $this->doQuotes( $line ) . "\n";
1597 $outtext = substr( $outtext, 0, -1 );
1598 return $outtext;
1602 * Helper function for doAllQuotes()
1604 * @param string $text
1606 * @return string
1608 public function doQuotes( $text ) {
1609 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1610 $countarr = count( $arr );
1611 if ( $countarr == 1 ) {
1612 return $text;
1615 // First, do some preliminary work. This may shift some apostrophes from
1616 // being mark-up to being text. It also counts the number of occurrences
1617 // of bold and italics mark-ups.
1618 $numbold = 0;
1619 $numitalics = 0;
1620 for ( $i = 1; $i < $countarr; $i += 2 ) {
1621 $thislen = strlen( $arr[$i] );
1622 // If there are ever four apostrophes, assume the first is supposed to
1623 // be text, and the remaining three constitute mark-up for bold text.
1624 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1625 if ( $thislen == 4 ) {
1626 $arr[$i - 1] .= "'";
1627 $arr[$i] = "'''";
1628 $thislen = 3;
1629 } elseif ( $thislen > 5 ) {
1630 // If there are more than 5 apostrophes in a row, assume they're all
1631 // text except for the last 5.
1632 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1633 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1634 $arr[$i] = "'''''";
1635 $thislen = 5;
1637 // Count the number of occurrences of bold and italics mark-ups.
1638 if ( $thislen == 2 ) {
1639 $numitalics++;
1640 } elseif ( $thislen == 3 ) {
1641 $numbold++;
1642 } elseif ( $thislen == 5 ) {
1643 $numitalics++;
1644 $numbold++;
1648 // If there is an odd number of both bold and italics, it is likely
1649 // that one of the bold ones was meant to be an apostrophe followed
1650 // by italics. Which one we cannot know for certain, but it is more
1651 // likely to be one that has a single-letter word before it.
1652 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1653 $firstsingleletterword = -1;
1654 $firstmultiletterword = -1;
1655 $firstspace = -1;
1656 for ( $i = 1; $i < $countarr; $i += 2 ) {
1657 if ( strlen( $arr[$i] ) == 3 ) {
1658 $x1 = substr( $arr[$i - 1], -1 );
1659 $x2 = substr( $arr[$i - 1], -2, 1 );
1660 if ( $x1 === ' ' ) {
1661 if ( $firstspace == -1 ) {
1662 $firstspace = $i;
1664 } elseif ( $x2 === ' ' ) {
1665 $firstsingleletterword = $i;
1666 // if $firstsingleletterword is set, we don't
1667 // look at the other options, so we can bail early.
1668 break;
1669 } else {
1670 if ( $firstmultiletterword == -1 ) {
1671 $firstmultiletterword = $i;
1677 // If there is a single-letter word, use it!
1678 if ( $firstsingleletterword > -1 ) {
1679 $arr[$firstsingleletterword] = "''";
1680 $arr[$firstsingleletterword - 1] .= "'";
1681 } elseif ( $firstmultiletterword > -1 ) {
1682 // If not, but there's a multi-letter word, use that one.
1683 $arr[$firstmultiletterword] = "''";
1684 $arr[$firstmultiletterword - 1] .= "'";
1685 } elseif ( $firstspace > -1 ) {
1686 // ... otherwise use the first one that has neither.
1687 // (notice that it is possible for all three to be -1 if, for example,
1688 // there is only one pentuple-apostrophe in the line)
1689 $arr[$firstspace] = "''";
1690 $arr[$firstspace - 1] .= "'";
1694 // Now let's actually convert our apostrophic mush to HTML!
1695 $output = '';
1696 $buffer = '';
1697 $state = '';
1698 $i = 0;
1699 foreach ( $arr as $r ) {
1700 if ( ( $i % 2 ) == 0 ) {
1701 if ( $state === 'both' ) {
1702 $buffer .= $r;
1703 } else {
1704 $output .= $r;
1706 } else {
1707 $thislen = strlen( $r );
1708 if ( $thislen == 2 ) {
1709 if ( $state === 'i' ) {
1710 $output .= '</i>';
1711 $state = '';
1712 } elseif ( $state === 'bi' ) {
1713 $output .= '</i>';
1714 $state = 'b';
1715 } elseif ( $state === 'ib' ) {
1716 $output .= '</b></i><b>';
1717 $state = 'b';
1718 } elseif ( $state === 'both' ) {
1719 $output .= '<b><i>' . $buffer . '</i>';
1720 $state = 'b';
1721 } else { // $state can be 'b' or ''
1722 $output .= '<i>';
1723 $state .= 'i';
1725 } elseif ( $thislen == 3 ) {
1726 if ( $state === 'b' ) {
1727 $output .= '</b>';
1728 $state = '';
1729 } elseif ( $state === 'bi' ) {
1730 $output .= '</i></b><i>';
1731 $state = 'i';
1732 } elseif ( $state === 'ib' ) {
1733 $output .= '</b>';
1734 $state = 'i';
1735 } elseif ( $state === 'both' ) {
1736 $output .= '<i><b>' . $buffer . '</b>';
1737 $state = 'i';
1738 } else { // $state can be 'i' or ''
1739 $output .= '<b>';
1740 $state .= 'b';
1742 } elseif ( $thislen == 5 ) {
1743 if ( $state === 'b' ) {
1744 $output .= '</b><i>';
1745 $state = 'i';
1746 } elseif ( $state === 'i' ) {
1747 $output .= '</i><b>';
1748 $state = 'b';
1749 } elseif ( $state === 'bi' ) {
1750 $output .= '</i></b>';
1751 $state = '';
1752 } elseif ( $state === 'ib' ) {
1753 $output .= '</b></i>';
1754 $state = '';
1755 } elseif ( $state === 'both' ) {
1756 $output .= '<i><b>' . $buffer . '</b></i>';
1757 $state = '';
1758 } else { // ($state == '')
1759 $buffer = '';
1760 $state = 'both';
1764 $i++;
1766 // Now close all remaining tags. Notice that the order is important.
1767 if ( $state === 'b' || $state === 'ib' ) {
1768 $output .= '</b>';
1770 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1771 $output .= '</i>';
1773 if ( $state === 'bi' ) {
1774 $output .= '</b>';
1776 // There might be lonely ''''', so make sure we have a buffer
1777 if ( $state === 'both' && $buffer ) {
1778 $output .= '<b><i>' . $buffer . '</i></b>';
1780 return $output;
1784 * Replace external links (REL)
1786 * Note: this is all very hackish and the order of execution matters a lot.
1787 * Make sure to run tests/parser/parserTests.php if you change this code.
1789 * @private
1791 * @param string $text
1793 * @throws MWException
1794 * @return string
1796 public function replaceExternalLinks( $text ) {
1798 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1799 if ( $bits === false ) {
1800 throw new MWException( "PCRE needs to be compiled with "
1801 . "--enable-unicode-properties in order for MediaWiki to function" );
1803 $s = array_shift( $bits );
1805 $i = 0;
1806 while ( $i < count( $bits ) ) {
1807 $url = $bits[$i++];
1808 $i++; // protocol
1809 $text = $bits[$i++];
1810 $trail = $bits[$i++];
1812 # The characters '<' and '>' (which were escaped by
1813 # removeHTMLtags()) should not be included in
1814 # URLs, per RFC 2396.
1815 $m2 = [];
1816 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1817 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1818 $url = substr( $url, 0, $m2[0][1] );
1821 # If the link text is an image URL, replace it with an <img> tag
1822 # This happened by accident in the original parser, but some people used it extensively
1823 $img = $this->maybeMakeExternalImage( $text );
1824 if ( $img !== false ) {
1825 $text = $img;
1828 $dtrail = '';
1830 # Set linktype for CSS - if URL==text, link is essentially free
1831 $linktype = ( $text === $url ) ? 'free' : 'text';
1833 # No link text, e.g. [http://domain.tld/some.link]
1834 if ( $text == '' ) {
1835 # Autonumber
1836 $langObj = $this->getTargetLanguage();
1837 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1838 $linktype = 'autonumber';
1839 } else {
1840 # Have link text, e.g. [http://domain.tld/some.link text]s
1841 # Check for trail
1842 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1845 $text = $this->getConverterLanguage()->markNoConversion( $text );
1847 $url = Sanitizer::cleanUrl( $url );
1849 # Use the encoded URL
1850 # This means that users can paste URLs directly into the text
1851 # Funny characters like ö aren't valid in URLs anyway
1852 # This was changed in August 2004
1853 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1854 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1856 # Register link in the output object.
1857 # Replace unnecessary URL escape codes with the referenced character
1858 # This prevents spammers from hiding links from the filters
1859 $pasteurized = self::normalizeLinkUrl( $url );
1860 $this->mOutput->addExternalLink( $pasteurized );
1863 return $s;
1867 * Get the rel attribute for a particular external link.
1869 * @since 1.21
1870 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1871 * nofollow if appropriate
1872 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1873 * @return string|null Rel attribute for $url
1875 public static function getExternalLinkRel( $url = false, $title = null ) {
1876 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1877 $ns = $title ? $title->getNamespace() : false;
1878 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1879 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1881 return 'nofollow';
1883 return null;
1887 * Get an associative array of additional HTML attributes appropriate for a
1888 * particular external link. This currently may include rel => nofollow
1889 * (depending on configuration, namespace, and the URL's domain) and/or a
1890 * target attribute (depending on configuration).
1892 * @param string $url URL to extract the domain from for rel =>
1893 * nofollow if appropriate
1894 * @return array Associative array of HTML attributes
1896 public function getExternalLinkAttribs( $url ) {
1897 $attribs = [];
1898 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1900 $target = $this->mOptions->getExternalLinkTarget();
1901 if ( $target ) {
1902 $attribs['target'] = $target;
1903 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1904 // T133507. New windows can navigate parent cross-origin.
1905 // Including noreferrer due to lacking browser
1906 // support of noopener. Eventually noreferrer should be removed.
1907 if ( $rel !== '' ) {
1908 $rel .= ' ';
1910 $rel .= 'noreferrer noopener';
1913 $attribs['rel'] = $rel;
1914 return $attribs;
1918 * Replace unusual escape codes in a URL with their equivalent characters
1920 * @deprecated since 1.24, use normalizeLinkUrl
1921 * @param string $url
1922 * @return string
1924 public static function replaceUnusualEscapes( $url ) {
1925 wfDeprecated( __METHOD__, '1.24' );
1926 return self::normalizeLinkUrl( $url );
1930 * Replace unusual escape codes in a URL with their equivalent characters
1932 * This generally follows the syntax defined in RFC 3986, with special
1933 * consideration for HTTP query strings.
1935 * @param string $url
1936 * @return string
1938 public static function normalizeLinkUrl( $url ) {
1939 # First, make sure unsafe characters are encoded
1940 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1941 function ( $m ) {
1942 return rawurlencode( $m[0] );
1944 $url
1947 $ret = '';
1948 $end = strlen( $url );
1950 # Fragment part - 'fragment'
1951 $start = strpos( $url, '#' );
1952 if ( $start !== false && $start < $end ) {
1953 $ret = self::normalizeUrlComponent(
1954 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1955 $end = $start;
1958 # Query part - 'query' minus &=+;
1959 $start = strpos( $url, '?' );
1960 if ( $start !== false && $start < $end ) {
1961 $ret = self::normalizeUrlComponent(
1962 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1963 $end = $start;
1966 # Scheme and path part - 'pchar'
1967 # (we assume no userinfo or encoded colons in the host)
1968 $ret = self::normalizeUrlComponent(
1969 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1971 return $ret;
1974 private static function normalizeUrlComponent( $component, $unsafe ) {
1975 $callback = function ( $matches ) use ( $unsafe ) {
1976 $char = urldecode( $matches[0] );
1977 $ord = ord( $char );
1978 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1979 # Unescape it
1980 return $char;
1981 } else {
1982 # Leave it escaped, but use uppercase for a-f
1983 return strtoupper( $matches[0] );
1986 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1990 * make an image if it's allowed, either through the global
1991 * option, through the exception, or through the on-wiki whitelist
1993 * @param string $url
1995 * @return string
1997 private function maybeMakeExternalImage( $url ) {
1998 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1999 $imagesexception = !empty( $imagesfrom );
2000 $text = false;
2001 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2002 if ( $imagesexception && is_array( $imagesfrom ) ) {
2003 $imagematch = false;
2004 foreach ( $imagesfrom as $match ) {
2005 if ( strpos( $url, $match ) === 0 ) {
2006 $imagematch = true;
2007 break;
2010 } elseif ( $imagesexception ) {
2011 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2012 } else {
2013 $imagematch = false;
2016 if ( $this->mOptions->getAllowExternalImages()
2017 || ( $imagesexception && $imagematch )
2019 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2020 # Image found
2021 $text = Linker::makeExternalImage( $url );
2024 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2025 && preg_match( self::EXT_IMAGE_REGEX, $url )
2027 $whitelist = explode(
2028 "\n",
2029 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2032 foreach ( $whitelist as $entry ) {
2033 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2034 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2035 continue;
2037 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2038 # Image matches a whitelist entry
2039 $text = Linker::makeExternalImage( $url );
2040 break;
2044 return $text;
2048 * Process [[ ]] wikilinks
2050 * @param string $s
2052 * @return string Processed text
2054 * @private
2056 public function replaceInternalLinks( $s ) {
2057 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2058 return $s;
2062 * Process [[ ]] wikilinks (RIL)
2063 * @param string $s
2064 * @throws MWException
2065 * @return LinkHolderArray
2067 * @private
2069 public function replaceInternalLinks2( &$s ) {
2070 global $wgExtraInterlanguageLinkPrefixes;
2072 static $tc = false, $e1, $e1_img;
2073 # the % is needed to support urlencoded titles as well
2074 if ( !$tc ) {
2075 $tc = Title::legalChars() . '#%';
2076 # Match a link having the form [[namespace:link|alternate]]trail
2077 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2078 # Match cases where there is no "]]", which might still be images
2079 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2082 $holders = new LinkHolderArray( $this );
2084 # split the entire text string on occurrences of [[
2085 $a = StringUtils::explode( '[[', ' ' . $s );
2086 # get the first element (all text up to first [[), and remove the space we added
2087 $s = $a->current();
2088 $a->next();
2089 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2090 $s = substr( $s, 1 );
2092 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2093 $e2 = null;
2094 if ( $useLinkPrefixExtension ) {
2095 # Match the end of a line for a word that's not followed by whitespace,
2096 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2097 global $wgContLang;
2098 $charset = $wgContLang->linkPrefixCharset();
2099 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2102 if ( is_null( $this->mTitle ) ) {
2103 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2105 $nottalk = !$this->mTitle->isTalkPage();
2107 if ( $useLinkPrefixExtension ) {
2108 $m = [];
2109 if ( preg_match( $e2, $s, $m ) ) {
2110 $first_prefix = $m[2];
2111 } else {
2112 $first_prefix = false;
2114 } else {
2115 $prefix = '';
2118 $useSubpages = $this->areSubpagesAllowed();
2120 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2121 # Loop for each link
2122 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2123 // @codingStandardsIgnoreEnd
2125 # Check for excessive memory usage
2126 if ( $holders->isBig() ) {
2127 # Too big
2128 # Do the existence check, replace the link holders and clear the array
2129 $holders->replace( $s );
2130 $holders->clear();
2133 if ( $useLinkPrefixExtension ) {
2134 if ( preg_match( $e2, $s, $m ) ) {
2135 $prefix = $m[2];
2136 $s = $m[1];
2137 } else {
2138 $prefix = '';
2140 # first link
2141 if ( $first_prefix ) {
2142 $prefix = $first_prefix;
2143 $first_prefix = false;
2147 $might_be_img = false;
2149 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2150 $text = $m[2];
2151 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2152 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2153 # the real problem is with the $e1 regex
2154 # See bug 1300.
2155 # Still some problems for cases where the ] is meant to be outside punctuation,
2156 # and no image is in sight. See bug 2095.
2157 if ( $text !== ''
2158 && substr( $m[3], 0, 1 ) === ']'
2159 && strpos( $text, '[' ) !== false
2161 $text .= ']'; # so that replaceExternalLinks($text) works later
2162 $m[3] = substr( $m[3], 1 );
2164 # fix up urlencoded title texts
2165 if ( strpos( $m[1], '%' ) !== false ) {
2166 # Should anchors '#' also be rejected?
2167 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2169 $trail = $m[3];
2170 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2171 # Invalid, but might be an image with a link in its caption
2172 $might_be_img = true;
2173 $text = $m[2];
2174 if ( strpos( $m[1], '%' ) !== false ) {
2175 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2177 $trail = "";
2178 } else { # Invalid form; output directly
2179 $s .= $prefix . '[[' . $line;
2180 continue;
2183 $origLink = $m[1];
2185 # Don't allow internal links to pages containing
2186 # PROTO: where PROTO is a valid URL protocol; these
2187 # should be external links.
2188 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2189 $s .= $prefix . '[[' . $line;
2190 continue;
2193 # Make subpage if necessary
2194 if ( $useSubpages ) {
2195 $link = $this->maybeDoSubpageLink( $origLink, $text );
2196 } else {
2197 $link = $origLink;
2200 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2201 if ( !$noforce ) {
2202 # Strip off leading ':'
2203 $link = substr( $link, 1 );
2206 $unstrip = $this->mStripState->unstripNoWiki( $link );
2207 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2208 if ( $nt === null ) {
2209 $s .= $prefix . '[[' . $line;
2210 continue;
2213 $ns = $nt->getNamespace();
2214 $iw = $nt->getInterwiki();
2216 if ( $might_be_img ) { # if this is actually an invalid link
2217 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2218 $found = false;
2219 while ( true ) {
2220 # look at the next 'line' to see if we can close it there
2221 $a->next();
2222 $next_line = $a->current();
2223 if ( $next_line === false || $next_line === null ) {
2224 break;
2226 $m = explode( ']]', $next_line, 3 );
2227 if ( count( $m ) == 3 ) {
2228 # the first ]] closes the inner link, the second the image
2229 $found = true;
2230 $text .= "[[{$m[0]}]]{$m[1]}";
2231 $trail = $m[2];
2232 break;
2233 } elseif ( count( $m ) == 2 ) {
2234 # if there's exactly one ]] that's fine, we'll keep looking
2235 $text .= "[[{$m[0]}]]{$m[1]}";
2236 } else {
2237 # if $next_line is invalid too, we need look no further
2238 $text .= '[[' . $next_line;
2239 break;
2242 if ( !$found ) {
2243 # we couldn't find the end of this imageLink, so output it raw
2244 # but don't ignore what might be perfectly normal links in the text we've examined
2245 $holders->merge( $this->replaceInternalLinks2( $text ) );
2246 $s .= "{$prefix}[[$link|$text";
2247 # note: no $trail, because without an end, there *is* no trail
2248 continue;
2250 } else { # it's not an image, so output it raw
2251 $s .= "{$prefix}[[$link|$text";
2252 # note: no $trail, because without an end, there *is* no trail
2253 continue;
2257 $wasblank = ( $text == '' );
2258 if ( $wasblank ) {
2259 $text = $link;
2260 } else {
2261 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2262 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2263 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2264 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2265 $text = $this->doQuotes( $text );
2268 # Link not escaped by : , create the various objects
2269 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2270 # Interwikis
2271 if (
2272 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2273 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2274 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2277 # Bug 24502: filter duplicates
2278 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2279 $this->mLangLinkLanguages[$iw] = true;
2280 $this->mOutput->addLanguageLink( $nt->getFullText() );
2283 $s = rtrim( $s . $prefix );
2284 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2285 continue;
2288 if ( $ns == NS_FILE ) {
2289 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2290 if ( $wasblank ) {
2291 # if no parameters were passed, $text
2292 # becomes something like "File:Foo.png",
2293 # which we don't want to pass on to the
2294 # image generator
2295 $text = '';
2296 } else {
2297 # recursively parse links inside the image caption
2298 # actually, this will parse them in any other parameters, too,
2299 # but it might be hard to fix that, and it doesn't matter ATM
2300 $text = $this->replaceExternalLinks( $text );
2301 $holders->merge( $this->replaceInternalLinks2( $text ) );
2303 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2304 $s .= $prefix . $this->armorLinks(
2305 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2306 continue;
2308 } elseif ( $ns == NS_CATEGORY ) {
2309 $s = rtrim( $s . "\n" ); # bug 87
2311 if ( $wasblank ) {
2312 $sortkey = $this->getDefaultSort();
2313 } else {
2314 $sortkey = $text;
2316 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2317 $sortkey = str_replace( "\n", '', $sortkey );
2318 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2319 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2322 * Strip the whitespace Category links produce, see bug 87
2324 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2326 continue;
2330 # Self-link checking. For some languages, variants of the title are checked in
2331 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2332 # for linking to a different variant.
2333 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2334 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2335 continue;
2338 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2339 # @todo FIXME: Should do batch file existence checks, see comment below
2340 if ( $ns == NS_MEDIA ) {
2341 # Give extensions a chance to select the file revision for us
2342 $options = [];
2343 $descQuery = false;
2344 Hooks::run( 'BeforeParserFetchFileAndTitle',
2345 [ $this, $nt, &$options, &$descQuery ] );
2346 # Fetch and register the file (file title may be different via hooks)
2347 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2348 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2349 $s .= $prefix . $this->armorLinks(
2350 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2351 continue;
2354 # Some titles, such as valid special pages or files in foreign repos, should
2355 # be shown as bluelinks even though they're not included in the page table
2356 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2357 # batch file existence checks for NS_FILE and NS_MEDIA
2358 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2359 $this->mOutput->addLink( $nt );
2360 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2361 } else {
2362 # Links will be added to the output link list after checking
2363 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2366 return $holders;
2370 * Render a forced-blue link inline; protect against double expansion of
2371 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2372 * Since this little disaster has to split off the trail text to avoid
2373 * breaking URLs in the following text without breaking trails on the
2374 * wiki links, it's been made into a horrible function.
2376 * @param Title $nt
2377 * @param string $text
2378 * @param string $trail
2379 * @param string $prefix
2380 * @return string HTML-wikitext mix oh yuck
2382 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2383 list( $inside, $trail ) = Linker::splitTrail( $trail );
2385 if ( $text == '' ) {
2386 $text = htmlspecialchars( $nt->getPrefixedText() );
2389 $link = $this->getLinkRenderer()->makeKnownLink(
2390 $nt, new HtmlArmor( "$prefix$text$inside" )
2393 return $this->armorLinks( $link ) . $trail;
2397 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2398 * going to go through further parsing steps before inline URL expansion.
2400 * Not needed quite as much as it used to be since free links are a bit
2401 * more sensible these days. But bracketed links are still an issue.
2403 * @param string $text More-or-less HTML
2404 * @return string Less-or-more HTML with NOPARSE bits
2406 public function armorLinks( $text ) {
2407 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2408 self::MARKER_PREFIX . "NOPARSE$1", $text );
2412 * Return true if subpage links should be expanded on this page.
2413 * @return bool
2415 public function areSubpagesAllowed() {
2416 # Some namespaces don't allow subpages
2417 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2421 * Handle link to subpage if necessary
2423 * @param string $target The source of the link
2424 * @param string &$text The link text, modified as necessary
2425 * @return string The full name of the link
2426 * @private
2428 public function maybeDoSubpageLink( $target, &$text ) {
2429 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2433 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2435 * @param string $text
2436 * @param bool $linestart Whether or not this is at the start of a line.
2437 * @private
2438 * @return string The lists rendered as HTML
2440 public function doBlockLevels( $text, $linestart ) {
2441 return BlockLevelPass::doBlockLevels( $text, $linestart );
2445 * Return value of a magic variable (like PAGENAME)
2447 * @private
2449 * @param int $index
2450 * @param bool|PPFrame $frame
2452 * @throws MWException
2453 * @return string
2455 public function getVariableValue( $index, $frame = false ) {
2456 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2457 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2459 if ( is_null( $this->mTitle ) ) {
2460 // If no title set, bad things are going to happen
2461 // later. Title should always be set since this
2462 // should only be called in the middle of a parse
2463 // operation (but the unit-tests do funky stuff)
2464 throw new MWException( __METHOD__ . ' Should only be '
2465 . ' called while parsing (no title set)' );
2469 * Some of these require message or data lookups and can be
2470 * expensive to check many times.
2472 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2473 if ( isset( $this->mVarCache[$index] ) ) {
2474 return $this->mVarCache[$index];
2478 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2479 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2481 $pageLang = $this->getFunctionLang();
2483 switch ( $index ) {
2484 case '!':
2485 $value = '|';
2486 break;
2487 case 'currentmonth':
2488 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2489 break;
2490 case 'currentmonth1':
2491 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2492 break;
2493 case 'currentmonthname':
2494 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2495 break;
2496 case 'currentmonthnamegen':
2497 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2498 break;
2499 case 'currentmonthabbrev':
2500 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2501 break;
2502 case 'currentday':
2503 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2504 break;
2505 case 'currentday2':
2506 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2507 break;
2508 case 'localmonth':
2509 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2510 break;
2511 case 'localmonth1':
2512 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2513 break;
2514 case 'localmonthname':
2515 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2516 break;
2517 case 'localmonthnamegen':
2518 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2519 break;
2520 case 'localmonthabbrev':
2521 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2522 break;
2523 case 'localday':
2524 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2525 break;
2526 case 'localday2':
2527 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2528 break;
2529 case 'pagename':
2530 $value = wfEscapeWikiText( $this->mTitle->getText() );
2531 break;
2532 case 'pagenamee':
2533 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2534 break;
2535 case 'fullpagename':
2536 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2537 break;
2538 case 'fullpagenamee':
2539 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2540 break;
2541 case 'subpagename':
2542 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2543 break;
2544 case 'subpagenamee':
2545 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2546 break;
2547 case 'rootpagename':
2548 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2549 break;
2550 case 'rootpagenamee':
2551 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2552 ' ',
2553 '_',
2554 $this->mTitle->getRootText()
2555 ) ) );
2556 break;
2557 case 'basepagename':
2558 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2559 break;
2560 case 'basepagenamee':
2561 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2562 ' ',
2563 '_',
2564 $this->mTitle->getBaseText()
2565 ) ) );
2566 break;
2567 case 'talkpagename':
2568 if ( $this->mTitle->canTalk() ) {
2569 $talkPage = $this->mTitle->getTalkPage();
2570 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2571 } else {
2572 $value = '';
2574 break;
2575 case 'talkpagenamee':
2576 if ( $this->mTitle->canTalk() ) {
2577 $talkPage = $this->mTitle->getTalkPage();
2578 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2579 } else {
2580 $value = '';
2582 break;
2583 case 'subjectpagename':
2584 $subjPage = $this->mTitle->getSubjectPage();
2585 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2586 break;
2587 case 'subjectpagenamee':
2588 $subjPage = $this->mTitle->getSubjectPage();
2589 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2590 break;
2591 case 'pageid': // requested in bug 23427
2592 $pageid = $this->getTitle()->getArticleID();
2593 if ( $pageid == 0 ) {
2594 # 0 means the page doesn't exist in the database,
2595 # which means the user is previewing a new page.
2596 # The vary-revision flag must be set, because the magic word
2597 # will have a different value once the page is saved.
2598 $this->mOutput->setFlag( 'vary-revision' );
2599 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2601 $value = $pageid ? $pageid : null;
2602 break;
2603 case 'revisionid':
2604 # Let the edit saving system know we should parse the page
2605 # *after* a revision ID has been assigned.
2606 $this->mOutput->setFlag( 'vary-revision-id' );
2607 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2608 $value = $this->mRevisionId;
2609 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2610 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2611 $this->mOutput->setSpeculativeRevIdUsed( $value );
2613 break;
2614 case 'revisionday':
2615 # Let the edit saving system know we should parse the page
2616 # *after* a revision ID has been assigned. This is for null edits.
2617 $this->mOutput->setFlag( 'vary-revision' );
2618 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2619 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2620 break;
2621 case 'revisionday2':
2622 # Let the edit saving system know we should parse the page
2623 # *after* a revision ID has been assigned. This is for null edits.
2624 $this->mOutput->setFlag( 'vary-revision' );
2625 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2626 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2627 break;
2628 case 'revisionmonth':
2629 # Let the edit saving system know we should parse the page
2630 # *after* a revision ID has been assigned. This is for null edits.
2631 $this->mOutput->setFlag( 'vary-revision' );
2632 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2633 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2634 break;
2635 case 'revisionmonth1':
2636 # Let the edit saving system know we should parse the page
2637 # *after* a revision ID has been assigned. This is for null edits.
2638 $this->mOutput->setFlag( 'vary-revision' );
2639 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2640 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2641 break;
2642 case 'revisionyear':
2643 # Let the edit saving system know we should parse the page
2644 # *after* a revision ID has been assigned. This is for null edits.
2645 $this->mOutput->setFlag( 'vary-revision' );
2646 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2647 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2648 break;
2649 case 'revisiontimestamp':
2650 # Let the edit saving system know we should parse the page
2651 # *after* a revision ID has been assigned. This is for null edits.
2652 $this->mOutput->setFlag( 'vary-revision' );
2653 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2654 $value = $this->getRevisionTimestamp();
2655 break;
2656 case 'revisionuser':
2657 # Let the edit saving system know we should parse the page
2658 # *after* a revision ID has been assigned for null edits.
2659 $this->mOutput->setFlag( 'vary-user' );
2660 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2661 $value = $this->getRevisionUser();
2662 break;
2663 case 'revisionsize':
2664 $value = $this->getRevisionSize();
2665 break;
2666 case 'namespace':
2667 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2668 break;
2669 case 'namespacee':
2670 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2671 break;
2672 case 'namespacenumber':
2673 $value = $this->mTitle->getNamespace();
2674 break;
2675 case 'talkspace':
2676 $value = $this->mTitle->canTalk()
2677 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2678 : '';
2679 break;
2680 case 'talkspacee':
2681 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2682 break;
2683 case 'subjectspace':
2684 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2685 break;
2686 case 'subjectspacee':
2687 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2688 break;
2689 case 'currentdayname':
2690 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2691 break;
2692 case 'currentyear':
2693 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2694 break;
2695 case 'currenttime':
2696 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2697 break;
2698 case 'currenthour':
2699 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2700 break;
2701 case 'currentweek':
2702 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2703 # int to remove the padding
2704 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2705 break;
2706 case 'currentdow':
2707 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2708 break;
2709 case 'localdayname':
2710 $value = $pageLang->getWeekdayName(
2711 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2713 break;
2714 case 'localyear':
2715 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2716 break;
2717 case 'localtime':
2718 $value = $pageLang->time(
2719 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2720 false,
2721 false
2723 break;
2724 case 'localhour':
2725 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2726 break;
2727 case 'localweek':
2728 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2729 # int to remove the padding
2730 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2731 break;
2732 case 'localdow':
2733 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2734 break;
2735 case 'numberofarticles':
2736 $value = $pageLang->formatNum( SiteStats::articles() );
2737 break;
2738 case 'numberoffiles':
2739 $value = $pageLang->formatNum( SiteStats::images() );
2740 break;
2741 case 'numberofusers':
2742 $value = $pageLang->formatNum( SiteStats::users() );
2743 break;
2744 case 'numberofactiveusers':
2745 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2746 break;
2747 case 'numberofpages':
2748 $value = $pageLang->formatNum( SiteStats::pages() );
2749 break;
2750 case 'numberofadmins':
2751 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2752 break;
2753 case 'numberofedits':
2754 $value = $pageLang->formatNum( SiteStats::edits() );
2755 break;
2756 case 'currenttimestamp':
2757 $value = wfTimestamp( TS_MW, $ts );
2758 break;
2759 case 'localtimestamp':
2760 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2761 break;
2762 case 'currentversion':
2763 $value = SpecialVersion::getVersion();
2764 break;
2765 case 'articlepath':
2766 return $wgArticlePath;
2767 case 'sitename':
2768 return $wgSitename;
2769 case 'server':
2770 return $wgServer;
2771 case 'servername':
2772 return $wgServerName;
2773 case 'scriptpath':
2774 return $wgScriptPath;
2775 case 'stylepath':
2776 return $wgStylePath;
2777 case 'directionmark':
2778 return $pageLang->getDirMark();
2779 case 'contentlanguage':
2780 global $wgLanguageCode;
2781 return $wgLanguageCode;
2782 case 'cascadingsources':
2783 $value = CoreParserFunctions::cascadingsources( $this );
2784 break;
2785 default:
2786 $ret = null;
2787 Hooks::run(
2788 'ParserGetVariableValueSwitch',
2789 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2792 return $ret;
2795 if ( $index ) {
2796 $this->mVarCache[$index] = $value;
2799 return $value;
2803 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2805 * @private
2807 public function initialiseVariables() {
2808 $variableIDs = MagicWord::getVariableIDs();
2809 $substIDs = MagicWord::getSubstIDs();
2811 $this->mVariables = new MagicWordArray( $variableIDs );
2812 $this->mSubstWords = new MagicWordArray( $substIDs );
2816 * Preprocess some wikitext and return the document tree.
2817 * This is the ghost of replace_variables().
2819 * @param string $text The text to parse
2820 * @param int $flags Bitwise combination of:
2821 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2822 * included. Default is to assume a direct page view.
2824 * The generated DOM tree must depend only on the input text and the flags.
2825 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2827 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2828 * change in the DOM tree for a given text, must be passed through the section identifier
2829 * in the section edit link and thus back to extractSections().
2831 * The output of this function is currently only cached in process memory, but a persistent
2832 * cache may be implemented at a later date which takes further advantage of these strict
2833 * dependency requirements.
2835 * @return PPNode
2837 public function preprocessToDom( $text, $flags = 0 ) {
2838 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2839 return $dom;
2843 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2845 * @param string $s
2847 * @return array
2849 public static function splitWhitespace( $s ) {
2850 $ltrimmed = ltrim( $s );
2851 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2852 $trimmed = rtrim( $ltrimmed );
2853 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2854 if ( $diff > 0 ) {
2855 $w2 = substr( $ltrimmed, -$diff );
2856 } else {
2857 $w2 = '';
2859 return [ $w1, $trimmed, $w2 ];
2863 * Replace magic variables, templates, and template arguments
2864 * with the appropriate text. Templates are substituted recursively,
2865 * taking care to avoid infinite loops.
2867 * Note that the substitution depends on value of $mOutputType:
2868 * self::OT_WIKI: only {{subst:}} templates
2869 * self::OT_PREPROCESS: templates but not extension tags
2870 * self::OT_HTML: all templates and extension tags
2872 * @param string $text The text to transform
2873 * @param bool|PPFrame $frame Object describing the arguments passed to the
2874 * template. Arguments may also be provided as an associative array, as
2875 * was the usual case before MW1.12. Providing arguments this way may be
2876 * useful for extensions wishing to perform variable replacement
2877 * explicitly.
2878 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2879 * double-brace expansion.
2880 * @return string
2882 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2883 # Is there any text? Also, Prevent too big inclusions!
2884 $textSize = strlen( $text );
2885 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2886 return $text;
2889 if ( $frame === false ) {
2890 $frame = $this->getPreprocessor()->newFrame();
2891 } elseif ( !( $frame instanceof PPFrame ) ) {
2892 wfDebug( __METHOD__ . " called using plain parameters instead of "
2893 . "a PPFrame instance. Creating custom frame.\n" );
2894 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2897 $dom = $this->preprocessToDom( $text );
2898 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2899 $text = $frame->expand( $dom, $flags );
2901 return $text;
2905 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2907 * @param array $args
2909 * @return array
2911 public static function createAssocArgs( $args ) {
2912 $assocArgs = [];
2913 $index = 1;
2914 foreach ( $args as $arg ) {
2915 $eqpos = strpos( $arg, '=' );
2916 if ( $eqpos === false ) {
2917 $assocArgs[$index++] = $arg;
2918 } else {
2919 $name = trim( substr( $arg, 0, $eqpos ) );
2920 $value = trim( substr( $arg, $eqpos + 1 ) );
2921 if ( $value === false ) {
2922 $value = '';
2924 if ( $name !== false ) {
2925 $assocArgs[$name] = $value;
2930 return $assocArgs;
2934 * Warn the user when a parser limitation is reached
2935 * Will warn at most once the user per limitation type
2937 * The results are shown during preview and run through the Parser (See EditPage.php)
2939 * @param string $limitationType Should be one of:
2940 * 'expensive-parserfunction' (corresponding messages:
2941 * 'expensive-parserfunction-warning',
2942 * 'expensive-parserfunction-category')
2943 * 'post-expand-template-argument' (corresponding messages:
2944 * 'post-expand-template-argument-warning',
2945 * 'post-expand-template-argument-category')
2946 * 'post-expand-template-inclusion' (corresponding messages:
2947 * 'post-expand-template-inclusion-warning',
2948 * 'post-expand-template-inclusion-category')
2949 * 'node-count-exceeded' (corresponding messages:
2950 * 'node-count-exceeded-warning',
2951 * 'node-count-exceeded-category')
2952 * 'expansion-depth-exceeded' (corresponding messages:
2953 * 'expansion-depth-exceeded-warning',
2954 * 'expansion-depth-exceeded-category')
2955 * @param string|int|null $current Current value
2956 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2957 * exceeded, provide the values (optional)
2959 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2960 # does no harm if $current and $max are present but are unnecessary for the message
2961 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2962 # only during preview, and that would split the parser cache unnecessarily.
2963 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2964 ->text();
2965 $this->mOutput->addWarning( $warning );
2966 $this->addTrackingCategory( "$limitationType-category" );
2970 * Return the text of a template, after recursively
2971 * replacing any variables or templates within the template.
2973 * @param array $piece The parts of the template
2974 * $piece['title']: the title, i.e. the part before the |
2975 * $piece['parts']: the parameter array
2976 * $piece['lineStart']: whether the brace was at the start of a line
2977 * @param PPFrame $frame The current frame, contains template arguments
2978 * @throws Exception
2979 * @return string The text of the template
2981 public function braceSubstitution( $piece, $frame ) {
2983 // Flags
2985 // $text has been filled
2986 $found = false;
2987 // wiki markup in $text should be escaped
2988 $nowiki = false;
2989 // $text is HTML, armour it against wikitext transformation
2990 $isHTML = false;
2991 // Force interwiki transclusion to be done in raw mode not rendered
2992 $forceRawInterwiki = false;
2993 // $text is a DOM node needing expansion in a child frame
2994 $isChildObj = false;
2995 // $text is a DOM node needing expansion in the current frame
2996 $isLocalObj = false;
2998 # Title object, where $text came from
2999 $title = false;
3001 # $part1 is the bit before the first |, and must contain only title characters.
3002 # Various prefixes will be stripped from it later.
3003 $titleWithSpaces = $frame->expand( $piece['title'] );
3004 $part1 = trim( $titleWithSpaces );
3005 $titleText = false;
3007 # Original title text preserved for various purposes
3008 $originalTitle = $part1;
3010 # $args is a list of argument nodes, starting from index 0, not including $part1
3011 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3012 # below won't work b/c this $args isn't an object
3013 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3015 $profileSection = null; // profile templates
3017 # SUBST
3018 if ( !$found ) {
3019 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3021 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3022 # Decide whether to expand template or keep wikitext as-is.
3023 if ( $this->ot['wiki'] ) {
3024 if ( $substMatch === false ) {
3025 $literal = true; # literal when in PST with no prefix
3026 } else {
3027 $literal = false; # expand when in PST with subst: or safesubst:
3029 } else {
3030 if ( $substMatch == 'subst' ) {
3031 $literal = true; # literal when not in PST with plain subst:
3032 } else {
3033 $literal = false; # expand when not in PST with safesubst: or no prefix
3036 if ( $literal ) {
3037 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3038 $isLocalObj = true;
3039 $found = true;
3043 # Variables
3044 if ( !$found && $args->getLength() == 0 ) {
3045 $id = $this->mVariables->matchStartToEnd( $part1 );
3046 if ( $id !== false ) {
3047 $text = $this->getVariableValue( $id, $frame );
3048 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3049 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3051 $found = true;
3055 # MSG, MSGNW and RAW
3056 if ( !$found ) {
3057 # Check for MSGNW:
3058 $mwMsgnw = MagicWord::get( 'msgnw' );
3059 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3060 $nowiki = true;
3061 } else {
3062 # Remove obsolete MSG:
3063 $mwMsg = MagicWord::get( 'msg' );
3064 $mwMsg->matchStartAndRemove( $part1 );
3067 # Check for RAW:
3068 $mwRaw = MagicWord::get( 'raw' );
3069 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3070 $forceRawInterwiki = true;
3074 # Parser functions
3075 if ( !$found ) {
3076 $colonPos = strpos( $part1, ':' );
3077 if ( $colonPos !== false ) {
3078 $func = substr( $part1, 0, $colonPos );
3079 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3080 $argsLength = $args->getLength();
3081 for ( $i = 0; $i < $argsLength; $i++ ) {
3082 $funcArgs[] = $args->item( $i );
3084 try {
3085 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3086 } catch ( Exception $ex ) {
3087 throw $ex;
3090 # The interface for parser functions allows for extracting
3091 # flags into the local scope. Extract any forwarded flags
3092 # here.
3093 extract( $result );
3097 # Finish mangling title and then check for loops.
3098 # Set $title to a Title object and $titleText to the PDBK
3099 if ( !$found ) {
3100 $ns = NS_TEMPLATE;
3101 # Split the title into page and subpage
3102 $subpage = '';
3103 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3104 if ( $part1 !== $relative ) {
3105 $part1 = $relative;
3106 $ns = $this->mTitle->getNamespace();
3108 $title = Title::newFromText( $part1, $ns );
3109 if ( $title ) {
3110 $titleText = $title->getPrefixedText();
3111 # Check for language variants if the template is not found
3112 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3113 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3115 # Do recursion depth check
3116 $limit = $this->mOptions->getMaxTemplateDepth();
3117 if ( $frame->depth >= $limit ) {
3118 $found = true;
3119 $text = '<span class="error">'
3120 . wfMessage( 'parser-template-recursion-depth-warning' )
3121 ->numParams( $limit )->inContentLanguage()->text()
3122 . '</span>';
3127 # Load from database
3128 if ( !$found && $title ) {
3129 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3130 if ( !$title->isExternal() ) {
3131 if ( $title->isSpecialPage()
3132 && $this->mOptions->getAllowSpecialInclusion()
3133 && $this->ot['html']
3135 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3136 // Pass the template arguments as URL parameters.
3137 // "uselang" will have no effect since the Language object
3138 // is forced to the one defined in ParserOptions.
3139 $pageArgs = [];
3140 $argsLength = $args->getLength();
3141 for ( $i = 0; $i < $argsLength; $i++ ) {
3142 $bits = $args->item( $i )->splitArg();
3143 if ( strval( $bits['index'] ) === '' ) {
3144 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3145 $value = trim( $frame->expand( $bits['value'] ) );
3146 $pageArgs[$name] = $value;
3150 // Create a new context to execute the special page
3151 $context = new RequestContext;
3152 $context->setTitle( $title );
3153 $context->setRequest( new FauxRequest( $pageArgs ) );
3154 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3155 $context->setUser( $this->getUser() );
3156 } else {
3157 // If this page is cached, then we better not be per user.
3158 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3160 $context->setLanguage( $this->mOptions->getUserLangObj() );
3161 $ret = SpecialPageFactory::capturePath(
3162 $title, $context, $this->getLinkRenderer() );
3163 if ( $ret ) {
3164 $text = $context->getOutput()->getHTML();
3165 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3166 $found = true;
3167 $isHTML = true;
3168 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3169 $this->mOutput->updateRuntimeAdaptiveExpiry(
3170 $specialPage->maxIncludeCacheTime()
3174 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3175 $found = false; # access denied
3176 wfDebug( __METHOD__ . ": template inclusion denied for " .
3177 $title->getPrefixedDBkey() . "\n" );
3178 } else {
3179 list( $text, $title ) = $this->getTemplateDom( $title );
3180 if ( $text !== false ) {
3181 $found = true;
3182 $isChildObj = true;
3186 # If the title is valid but undisplayable, make a link to it
3187 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3188 $text = "[[:$titleText]]";
3189 $found = true;
3191 } elseif ( $title->isTrans() ) {
3192 # Interwiki transclusion
3193 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3194 $text = $this->interwikiTransclude( $title, 'render' );
3195 $isHTML = true;
3196 } else {
3197 $text = $this->interwikiTransclude( $title, 'raw' );
3198 # Preprocess it like a template
3199 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3200 $isChildObj = true;
3202 $found = true;
3205 # Do infinite loop check
3206 # This has to be done after redirect resolution to avoid infinite loops via redirects
3207 if ( !$frame->loopCheck( $title ) ) {
3208 $found = true;
3209 $text = '<span class="error">'
3210 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3211 . '</span>';
3212 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3216 # If we haven't found text to substitute by now, we're done
3217 # Recover the source wikitext and return it
3218 if ( !$found ) {
3219 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3220 if ( $profileSection ) {
3221 $this->mProfiler->scopedProfileOut( $profileSection );
3223 return [ 'object' => $text ];
3226 # Expand DOM-style return values in a child frame
3227 if ( $isChildObj ) {
3228 # Clean up argument array
3229 $newFrame = $frame->newChild( $args, $title );
3231 if ( $nowiki ) {
3232 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3233 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3234 # Expansion is eligible for the empty-frame cache
3235 $text = $newFrame->cachedExpand( $titleText, $text );
3236 } else {
3237 # Uncached expansion
3238 $text = $newFrame->expand( $text );
3241 if ( $isLocalObj && $nowiki ) {
3242 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3243 $isLocalObj = false;
3246 if ( $profileSection ) {
3247 $this->mProfiler->scopedProfileOut( $profileSection );
3250 # Replace raw HTML by a placeholder
3251 if ( $isHTML ) {
3252 $text = $this->insertStripItem( $text );
3253 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3254 # Escape nowiki-style return values
3255 $text = wfEscapeWikiText( $text );
3256 } elseif ( is_string( $text )
3257 && !$piece['lineStart']
3258 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3260 # Bug 529: if the template begins with a table or block-level
3261 # element, it should be treated as beginning a new line.
3262 # This behavior is somewhat controversial.
3263 $text = "\n" . $text;
3266 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3267 # Error, oversize inclusion
3268 if ( $titleText !== false ) {
3269 # Make a working, properly escaped link if possible (bug 23588)
3270 $text = "[[:$titleText]]";
3271 } else {
3272 # This will probably not be a working link, but at least it may
3273 # provide some hint of where the problem is
3274 preg_replace( '/^:/', '', $originalTitle );
3275 $text = "[[:$originalTitle]]";
3277 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3278 . 'post-expand include size too large -->' );
3279 $this->limitationWarn( 'post-expand-template-inclusion' );
3282 if ( $isLocalObj ) {
3283 $ret = [ 'object' => $text ];
3284 } else {
3285 $ret = [ 'text' => $text ];
3288 return $ret;
3292 * Call a parser function and return an array with text and flags.
3294 * The returned array will always contain a boolean 'found', indicating
3295 * whether the parser function was found or not. It may also contain the
3296 * following:
3297 * text: string|object, resulting wikitext or PP DOM object
3298 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3299 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3300 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3301 * nowiki: bool, wiki markup in $text should be escaped
3303 * @since 1.21
3304 * @param PPFrame $frame The current frame, contains template arguments
3305 * @param string $function Function name
3306 * @param array $args Arguments to the function
3307 * @throws MWException
3308 * @return array
3310 public function callParserFunction( $frame, $function, array $args = [] ) {
3311 global $wgContLang;
3313 # Case sensitive functions
3314 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3315 $function = $this->mFunctionSynonyms[1][$function];
3316 } else {
3317 # Case insensitive functions
3318 $function = $wgContLang->lc( $function );
3319 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3320 $function = $this->mFunctionSynonyms[0][$function];
3321 } else {
3322 return [ 'found' => false ];
3326 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3328 # Workaround for PHP bug 35229 and similar
3329 if ( !is_callable( $callback ) ) {
3330 throw new MWException( "Tag hook for $function is not callable\n" );
3333 $allArgs = [ &$this ];
3334 if ( $flags & self::SFH_OBJECT_ARGS ) {
3335 # Convert arguments to PPNodes and collect for appending to $allArgs
3336 $funcArgs = [];
3337 foreach ( $args as $k => $v ) {
3338 if ( $v instanceof PPNode || $k === 0 ) {
3339 $funcArgs[] = $v;
3340 } else {
3341 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3345 # Add a frame parameter, and pass the arguments as an array
3346 $allArgs[] = $frame;
3347 $allArgs[] = $funcArgs;
3348 } else {
3349 # Convert arguments to plain text and append to $allArgs
3350 foreach ( $args as $k => $v ) {
3351 if ( $v instanceof PPNode ) {
3352 $allArgs[] = trim( $frame->expand( $v ) );
3353 } elseif ( is_int( $k ) && $k >= 0 ) {
3354 $allArgs[] = trim( $v );
3355 } else {
3356 $allArgs[] = trim( "$k=$v" );
3361 $result = call_user_func_array( $callback, $allArgs );
3363 # The interface for function hooks allows them to return a wikitext
3364 # string or an array containing the string and any flags. This mungs
3365 # things around to match what this method should return.
3366 if ( !is_array( $result ) ) {
3367 $result =[
3368 'found' => true,
3369 'text' => $result,
3371 } else {
3372 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3373 $result['text'] = $result[0];
3375 unset( $result[0] );
3376 $result += [
3377 'found' => true,
3381 $noparse = true;
3382 $preprocessFlags = 0;
3383 if ( isset( $result['noparse'] ) ) {
3384 $noparse = $result['noparse'];
3386 if ( isset( $result['preprocessFlags'] ) ) {
3387 $preprocessFlags = $result['preprocessFlags'];
3390 if ( !$noparse ) {
3391 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3392 $result['isChildObj'] = true;
3395 return $result;
3399 * Get the semi-parsed DOM representation of a template with a given title,
3400 * and its redirect destination title. Cached.
3402 * @param Title $title
3404 * @return array
3406 public function getTemplateDom( $title ) {
3407 $cacheTitle = $title;
3408 $titleText = $title->getPrefixedDBkey();
3410 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3411 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3412 $title = Title::makeTitle( $ns, $dbk );
3413 $titleText = $title->getPrefixedDBkey();
3415 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3416 return [ $this->mTplDomCache[$titleText], $title ];
3419 # Cache miss, go to the database
3420 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3422 if ( $text === false ) {
3423 $this->mTplDomCache[$titleText] = false;
3424 return [ false, $title ];
3427 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3428 $this->mTplDomCache[$titleText] = $dom;
3430 if ( !$title->equals( $cacheTitle ) ) {
3431 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3432 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3435 return [ $dom, $title ];
3439 * Fetch the current revision of a given title. Note that the revision
3440 * (and even the title) may not exist in the database, so everything
3441 * contributing to the output of the parser should use this method
3442 * where possible, rather than getting the revisions themselves. This
3443 * method also caches its results, so using it benefits performance.
3445 * @since 1.24
3446 * @param Title $title
3447 * @return Revision
3449 public function fetchCurrentRevisionOfTitle( $title ) {
3450 $cacheKey = $title->getPrefixedDBkey();
3451 if ( !$this->currentRevisionCache ) {
3452 $this->currentRevisionCache = new MapCacheLRU( 100 );
3454 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3455 $this->currentRevisionCache->set( $cacheKey,
3456 // Defaults to Parser::statelessFetchRevision()
3457 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3460 return $this->currentRevisionCache->get( $cacheKey );
3464 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3465 * without passing them on to it.
3467 * @since 1.24
3468 * @param Title $title
3469 * @param Parser|bool $parser
3470 * @return Revision|bool False if missing
3472 public static function statelessFetchRevision( Title $title, $parser = false ) {
3473 $pageId = $title->getArticleID();
3474 $revId = $title->getLatestRevID();
3476 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3477 if ( $rev ) {
3478 $rev->setTitle( $title );
3481 return $rev;
3485 * Fetch the unparsed text of a template and register a reference to it.
3486 * @param Title $title
3487 * @return array ( string or false, Title )
3489 public function fetchTemplateAndTitle( $title ) {
3490 // Defaults to Parser::statelessFetchTemplate()
3491 $templateCb = $this->mOptions->getTemplateCallback();
3492 $stuff = call_user_func( $templateCb, $title, $this );
3493 // We use U+007F DELETE to distinguish strip markers from regular text.
3494 $text = $stuff['text'];
3495 if ( is_string( $stuff['text'] ) ) {
3496 $text = strtr( $text, "\x7f", "?" );
3498 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3499 if ( isset( $stuff['deps'] ) ) {
3500 foreach ( $stuff['deps'] as $dep ) {
3501 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3502 if ( $dep['title']->equals( $this->getTitle() ) ) {
3503 // If we transclude ourselves, the final result
3504 // will change based on the new version of the page
3505 $this->mOutput->setFlag( 'vary-revision' );
3509 return [ $text, $finalTitle ];
3513 * Fetch the unparsed text of a template and register a reference to it.
3514 * @param Title $title
3515 * @return string|bool
3517 public function fetchTemplate( $title ) {
3518 return $this->fetchTemplateAndTitle( $title )[0];
3522 * Static function to get a template
3523 * Can be overridden via ParserOptions::setTemplateCallback().
3525 * @param Title $title
3526 * @param bool|Parser $parser
3528 * @return array
3530 public static function statelessFetchTemplate( $title, $parser = false ) {
3531 $text = $skip = false;
3532 $finalTitle = $title;
3533 $deps = [];
3535 # Loop to fetch the article, with up to 1 redirect
3536 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3537 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3538 // @codingStandardsIgnoreEnd
3539 # Give extensions a chance to select the revision instead
3540 $id = false; # Assume current
3541 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3542 [ $parser, $title, &$skip, &$id ] );
3544 if ( $skip ) {
3545 $text = false;
3546 $deps[] = [
3547 'title' => $title,
3548 'page_id' => $title->getArticleID(),
3549 'rev_id' => null
3551 break;
3553 # Get the revision
3554 if ( $id ) {
3555 $rev = Revision::newFromId( $id );
3556 } elseif ( $parser ) {
3557 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3558 } else {
3559 $rev = Revision::newFromTitle( $title );
3561 $rev_id = $rev ? $rev->getId() : 0;
3562 # If there is no current revision, there is no page
3563 if ( $id === false && !$rev ) {
3564 $linkCache = LinkCache::singleton();
3565 $linkCache->addBadLinkObj( $title );
3568 $deps[] = [
3569 'title' => $title,
3570 'page_id' => $title->getArticleID(),
3571 'rev_id' => $rev_id ];
3572 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3573 # We fetched a rev from a different title; register it too...
3574 $deps[] = [
3575 'title' => $rev->getTitle(),
3576 'page_id' => $rev->getPage(),
3577 'rev_id' => $rev_id ];
3580 if ( $rev ) {
3581 $content = $rev->getContent();
3582 $text = $content ? $content->getWikitextForTransclusion() : null;
3584 if ( $text === false || $text === null ) {
3585 $text = false;
3586 break;
3588 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3589 global $wgContLang;
3590 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3591 if ( !$message->exists() ) {
3592 $text = false;
3593 break;
3595 $content = $message->content();
3596 $text = $message->plain();
3597 } else {
3598 break;
3600 if ( !$content ) {
3601 break;
3603 # Redirect?
3604 $finalTitle = $title;
3605 $title = $content->getRedirectTarget();
3607 return [
3608 'text' => $text,
3609 'finalTitle' => $finalTitle,
3610 'deps' => $deps ];
3614 * Fetch a file and its title and register a reference to it.
3615 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3616 * @param Title $title
3617 * @param array $options Array of options to RepoGroup::findFile
3618 * @return File|bool
3620 public function fetchFile( $title, $options = [] ) {
3621 return $this->fetchFileAndTitle( $title, $options )[0];
3625 * Fetch a file and its title and register a reference to it.
3626 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3627 * @param Title $title
3628 * @param array $options Array of options to RepoGroup::findFile
3629 * @return array ( File or false, Title of file )
3631 public function fetchFileAndTitle( $title, $options = [] ) {
3632 $file = $this->fetchFileNoRegister( $title, $options );
3634 $time = $file ? $file->getTimestamp() : false;
3635 $sha1 = $file ? $file->getSha1() : false;
3636 # Register the file as a dependency...
3637 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3638 if ( $file && !$title->equals( $file->getTitle() ) ) {
3639 # Update fetched file title
3640 $title = $file->getTitle();
3641 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3643 return [ $file, $title ];
3647 * Helper function for fetchFileAndTitle.
3649 * Also useful if you need to fetch a file but not use it yet,
3650 * for example to get the file's handler.
3652 * @param Title $title
3653 * @param array $options Array of options to RepoGroup::findFile
3654 * @return File|bool
3656 protected function fetchFileNoRegister( $title, $options = [] ) {
3657 if ( isset( $options['broken'] ) ) {
3658 $file = false; // broken thumbnail forced by hook
3659 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3660 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3661 } else { // get by (name,timestamp)
3662 $file = wfFindFile( $title, $options );
3664 return $file;
3668 * Transclude an interwiki link.
3670 * @param Title $title
3671 * @param string $action
3673 * @return string
3675 public function interwikiTransclude( $title, $action ) {
3676 global $wgEnableScaryTranscluding;
3678 if ( !$wgEnableScaryTranscluding ) {
3679 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3682 $url = $title->getFullURL( [ 'action' => $action ] );
3684 if ( strlen( $url ) > 255 ) {
3685 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3687 return $this->fetchScaryTemplateMaybeFromCache( $url );
3691 * @param string $url
3692 * @return mixed|string
3694 public function fetchScaryTemplateMaybeFromCache( $url ) {
3695 global $wgTranscludeCacheExpiry;
3696 $dbr = wfGetDB( DB_REPLICA );
3697 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3698 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3699 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3700 if ( $obj ) {
3701 return $obj->tc_contents;
3704 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3705 $status = $req->execute(); // Status object
3706 if ( $status->isOK() ) {
3707 $text = $req->getContent();
3708 } elseif ( $req->getStatus() != 200 ) {
3709 // Though we failed to fetch the content, this status is useless.
3710 return wfMessage( 'scarytranscludefailed-httpstatus' )
3711 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3712 } else {
3713 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3716 $dbw = wfGetDB( DB_MASTER );
3717 $dbw->replace( 'transcache', [ 'tc_url' ], [
3718 'tc_url' => $url,
3719 'tc_time' => $dbw->timestamp( time() ),
3720 'tc_contents' => $text
3721 ] );
3722 return $text;
3726 * Triple brace replacement -- used for template arguments
3727 * @private
3729 * @param array $piece
3730 * @param PPFrame $frame
3732 * @return array
3734 public function argSubstitution( $piece, $frame ) {
3736 $error = false;
3737 $parts = $piece['parts'];
3738 $nameWithSpaces = $frame->expand( $piece['title'] );
3739 $argName = trim( $nameWithSpaces );
3740 $object = false;
3741 $text = $frame->getArgument( $argName );
3742 if ( $text === false && $parts->getLength() > 0
3743 && ( $this->ot['html']
3744 || $this->ot['pre']
3745 || ( $this->ot['wiki'] && $frame->isTemplate() )
3748 # No match in frame, use the supplied default
3749 $object = $parts->item( 0 )->getChildren();
3751 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3752 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3753 $this->limitationWarn( 'post-expand-template-argument' );
3756 if ( $text === false && $object === false ) {
3757 # No match anywhere
3758 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3760 if ( $error !== false ) {
3761 $text .= $error;
3763 if ( $object !== false ) {
3764 $ret = [ 'object' => $object ];
3765 } else {
3766 $ret = [ 'text' => $text ];
3769 return $ret;
3773 * Return the text to be used for a given extension tag.
3774 * This is the ghost of strip().
3776 * @param array $params Associative array of parameters:
3777 * name PPNode for the tag name
3778 * attr PPNode for unparsed text where tag attributes are thought to be
3779 * attributes Optional associative array of parsed attributes
3780 * inner Contents of extension element
3781 * noClose Original text did not have a close tag
3782 * @param PPFrame $frame
3784 * @throws MWException
3785 * @return string
3787 public function extensionSubstitution( $params, $frame ) {
3788 static $errorStr = '<span class="error">';
3789 static $errorLen = 20;
3791 $name = $frame->expand( $params['name'] );
3792 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3793 // Probably expansion depth or node count exceeded. Just punt the
3794 // error up.
3795 return $name;
3798 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3799 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3800 // See above
3801 return $attrText;
3804 // We can't safely check if the expansion for $content resulted in an
3805 // error, because the content could happen to be the error string
3806 // (T149622).
3807 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3809 $marker = self::MARKER_PREFIX . "-$name-"
3810 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3812 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3813 ( $this->ot['html'] || $this->ot['pre'] );
3814 if ( $isFunctionTag ) {
3815 $markerType = 'none';
3816 } else {
3817 $markerType = 'general';
3819 if ( $this->ot['html'] || $isFunctionTag ) {
3820 $name = strtolower( $name );
3821 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3822 if ( isset( $params['attributes'] ) ) {
3823 $attributes = $attributes + $params['attributes'];
3826 if ( isset( $this->mTagHooks[$name] ) ) {
3827 # Workaround for PHP bug 35229 and similar
3828 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3829 throw new MWException( "Tag hook for $name is not callable\n" );
3831 $output = call_user_func_array( $this->mTagHooks[$name],
3832 [ $content, $attributes, $this, $frame ] );
3833 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3834 list( $callback, ) = $this->mFunctionTagHooks[$name];
3835 if ( !is_callable( $callback ) ) {
3836 throw new MWException( "Tag hook for $name is not callable\n" );
3839 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3840 } else {
3841 $output = '<span class="error">Invalid tag extension name: ' .
3842 htmlspecialchars( $name ) . '</span>';
3845 if ( is_array( $output ) ) {
3846 # Extract flags to local scope (to override $markerType)
3847 $flags = $output;
3848 $output = $flags[0];
3849 unset( $flags[0] );
3850 extract( $flags );
3852 } else {
3853 if ( is_null( $attrText ) ) {
3854 $attrText = '';
3856 if ( isset( $params['attributes'] ) ) {
3857 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3858 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3859 htmlspecialchars( $attrValue ) . '"';
3862 if ( $content === null ) {
3863 $output = "<$name$attrText/>";
3864 } else {
3865 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3866 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3867 // See above
3868 return $close;
3870 $output = "<$name$attrText>$content$close";
3874 if ( $markerType === 'none' ) {
3875 return $output;
3876 } elseif ( $markerType === 'nowiki' ) {
3877 $this->mStripState->addNoWiki( $marker, $output );
3878 } elseif ( $markerType === 'general' ) {
3879 $this->mStripState->addGeneral( $marker, $output );
3880 } else {
3881 throw new MWException( __METHOD__ . ': invalid marker type' );
3883 return $marker;
3887 * Increment an include size counter
3889 * @param string $type The type of expansion
3890 * @param int $size The size of the text
3891 * @return bool False if this inclusion would take it over the maximum, true otherwise
3893 public function incrementIncludeSize( $type, $size ) {
3894 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3895 return false;
3896 } else {
3897 $this->mIncludeSizes[$type] += $size;
3898 return true;
3903 * Increment the expensive function count
3905 * @return bool False if the limit has been exceeded
3907 public function incrementExpensiveFunctionCount() {
3908 $this->mExpensiveFunctionCount++;
3909 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3913 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3914 * Fills $this->mDoubleUnderscores, returns the modified text
3916 * @param string $text
3918 * @return string
3920 public function doDoubleUnderscore( $text ) {
3922 # The position of __TOC__ needs to be recorded
3923 $mw = MagicWord::get( 'toc' );
3924 if ( $mw->match( $text ) ) {
3925 $this->mShowToc = true;
3926 $this->mForceTocPosition = true;
3928 # Set a placeholder. At the end we'll fill it in with the TOC.
3929 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3931 # Only keep the first one.
3932 $text = $mw->replace( '', $text );
3935 # Now match and remove the rest of them
3936 $mwa = MagicWord::getDoubleUnderscoreArray();
3937 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3939 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3940 $this->mOutput->mNoGallery = true;
3942 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3943 $this->mShowToc = false;
3945 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3946 && $this->mTitle->getNamespace() == NS_CATEGORY
3948 $this->addTrackingCategory( 'hidden-category-category' );
3950 # (bug 8068) Allow control over whether robots index a page.
3951 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3952 # is not desirable, the last one on the page should win.
3953 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3954 $this->mOutput->setIndexPolicy( 'noindex' );
3955 $this->addTrackingCategory( 'noindex-category' );
3957 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3958 $this->mOutput->setIndexPolicy( 'index' );
3959 $this->addTrackingCategory( 'index-category' );
3962 # Cache all double underscores in the database
3963 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3964 $this->mOutput->setProperty( $key, '' );
3967 return $text;
3971 * @see ParserOutput::addTrackingCategory()
3972 * @param string $msg Message key
3973 * @return bool Whether the addition was successful
3975 public function addTrackingCategory( $msg ) {
3976 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3980 * This function accomplishes several tasks:
3981 * 1) Auto-number headings if that option is enabled
3982 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3983 * 3) Add a Table of contents on the top for users who have enabled the option
3984 * 4) Auto-anchor headings
3986 * It loops through all headlines, collects the necessary data, then splits up the
3987 * string and re-inserts the newly formatted headlines.
3989 * @param string $text
3990 * @param string $origText Original, untouched wikitext
3991 * @param bool $isMain
3992 * @return mixed|string
3993 * @private
3995 public function formatHeadings( $text, $origText, $isMain = true ) {
3996 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3998 # Inhibit editsection links if requested in the page
3999 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4000 $maybeShowEditLink = $showEditLink = false;
4001 } else {
4002 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4003 $showEditLink = $this->mOptions->getEditSection();
4005 if ( $showEditLink ) {
4006 $this->mOutput->setEditSectionTokens( true );
4009 # Get all headlines for numbering them and adding funky stuff like [edit]
4010 # links - this is for later, but we need the number of headlines right now
4011 $matches = [];
4012 $numMatches = preg_match_all(
4013 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4014 $text,
4015 $matches
4018 # if there are fewer than 4 headlines in the article, do not show TOC
4019 # unless it's been explicitly enabled.
4020 $enoughToc = $this->mShowToc &&
4021 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4023 # Allow user to stipulate that a page should have a "new section"
4024 # link added via __NEWSECTIONLINK__
4025 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4026 $this->mOutput->setNewSection( true );
4029 # Allow user to remove the "new section"
4030 # link via __NONEWSECTIONLINK__
4031 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4032 $this->mOutput->hideNewSection( true );
4035 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4036 # override above conditions and always show TOC above first header
4037 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4038 $this->mShowToc = true;
4039 $enoughToc = true;
4042 # headline counter
4043 $headlineCount = 0;
4044 $numVisible = 0;
4046 # Ugh .. the TOC should have neat indentation levels which can be
4047 # passed to the skin functions. These are determined here
4048 $toc = '';
4049 $full = '';
4050 $head = [];
4051 $sublevelCount = [];
4052 $levelCount = [];
4053 $level = 0;
4054 $prevlevel = 0;
4055 $toclevel = 0;
4056 $prevtoclevel = 0;
4057 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4058 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4059 $oldType = $this->mOutputType;
4060 $this->setOutputType( self::OT_WIKI );
4061 $frame = $this->getPreprocessor()->newFrame();
4062 $root = $this->preprocessToDom( $origText );
4063 $node = $root->getFirstChild();
4064 $byteOffset = 0;
4065 $tocraw = [];
4066 $refers = [];
4068 $headlines = $numMatches !== false ? $matches[3] : [];
4070 foreach ( $headlines as $headline ) {
4071 $isTemplate = false;
4072 $titleText = false;
4073 $sectionIndex = false;
4074 $numbering = '';
4075 $markerMatches = [];
4076 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4077 $serial = $markerMatches[1];
4078 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4079 $isTemplate = ( $titleText != $baseTitleText );
4080 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4083 if ( $toclevel ) {
4084 $prevlevel = $level;
4086 $level = $matches[1][$headlineCount];
4088 if ( $level > $prevlevel ) {
4089 # Increase TOC level
4090 $toclevel++;
4091 $sublevelCount[$toclevel] = 0;
4092 if ( $toclevel < $wgMaxTocLevel ) {
4093 $prevtoclevel = $toclevel;
4094 $toc .= Linker::tocIndent();
4095 $numVisible++;
4097 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4098 # Decrease TOC level, find level to jump to
4100 for ( $i = $toclevel; $i > 0; $i-- ) {
4101 if ( $levelCount[$i] == $level ) {
4102 # Found last matching level
4103 $toclevel = $i;
4104 break;
4105 } elseif ( $levelCount[$i] < $level ) {
4106 # Found first matching level below current level
4107 $toclevel = $i + 1;
4108 break;
4111 if ( $i == 0 ) {
4112 $toclevel = 1;
4114 if ( $toclevel < $wgMaxTocLevel ) {
4115 if ( $prevtoclevel < $wgMaxTocLevel ) {
4116 # Unindent only if the previous toc level was shown :p
4117 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4118 $prevtoclevel = $toclevel;
4119 } else {
4120 $toc .= Linker::tocLineEnd();
4123 } else {
4124 # No change in level, end TOC line
4125 if ( $toclevel < $wgMaxTocLevel ) {
4126 $toc .= Linker::tocLineEnd();
4130 $levelCount[$toclevel] = $level;
4132 # count number of headlines for each level
4133 $sublevelCount[$toclevel]++;
4134 $dot = 0;
4135 for ( $i = 1; $i <= $toclevel; $i++ ) {
4136 if ( !empty( $sublevelCount[$i] ) ) {
4137 if ( $dot ) {
4138 $numbering .= '.';
4140 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4141 $dot = 1;
4145 # The safe header is a version of the header text safe to use for links
4147 # Remove link placeholders by the link text.
4148 # <!--LINK number-->
4149 # turns into
4150 # link text with suffix
4151 # Do this before unstrip since link text can contain strip markers
4152 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4154 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4155 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4157 # Strip out HTML (first regex removes any tag not allowed)
4158 # Allowed tags are:
4159 # * <sup> and <sub> (bug 8393)
4160 # * <i> (bug 26375)
4161 # * <b> (r105284)
4162 # * <bdi> (bug 72884)
4163 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4164 # * <s> and <strike> (T35715)
4165 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4166 # to allow setting directionality in toc items.
4167 $tocline = preg_replace(
4169 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4170 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4172 [ '', '<$1>' ],
4173 $safeHeadline
4176 # Strip '<span></span>', which is the result from the above if
4177 # <span id="foo"></span> is used to produce an additional anchor
4178 # for a section.
4179 $tocline = str_replace( '<span></span>', '', $tocline );
4181 $tocline = trim( $tocline );
4183 # For the anchor, strip out HTML-y stuff period
4184 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4185 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4187 # Save headline for section edit hint before it's escaped
4188 $headlineHint = $safeHeadline;
4190 if ( $wgExperimentalHtmlIds ) {
4191 # For reverse compatibility, provide an id that's
4192 # HTML4-compatible, like we used to.
4193 # It may be worth noting, academically, that it's possible for
4194 # the legacy anchor to conflict with a non-legacy headline
4195 # anchor on the page. In this case likely the "correct" thing
4196 # would be to either drop the legacy anchors or make sure
4197 # they're numbered first. However, this would require people
4198 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4199 # manually, so let's not bother worrying about it.
4200 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4201 [ 'noninitial', 'legacy' ] );
4202 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4204 if ( $legacyHeadline == $safeHeadline ) {
4205 # No reason to have both (in fact, we can't)
4206 $legacyHeadline = false;
4208 } else {
4209 $legacyHeadline = false;
4210 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4211 'noninitial' );
4214 # HTML names must be case-insensitively unique (bug 10721).
4215 # This does not apply to Unicode characters per
4216 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4217 # @todo FIXME: We may be changing them depending on the current locale.
4218 $arrayKey = strtolower( $safeHeadline );
4219 if ( $legacyHeadline === false ) {
4220 $legacyArrayKey = false;
4221 } else {
4222 $legacyArrayKey = strtolower( $legacyHeadline );
4225 # Create the anchor for linking from the TOC to the section
4226 $anchor = $safeHeadline;
4227 $legacyAnchor = $legacyHeadline;
4228 if ( isset( $refers[$arrayKey] ) ) {
4229 // @codingStandardsIgnoreStart
4230 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4231 // @codingStandardsIgnoreEnd
4232 $anchor .= "_$i";
4233 $refers["${arrayKey}_$i"] = true;
4234 } else {
4235 $refers[$arrayKey] = true;
4237 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4238 // @codingStandardsIgnoreStart
4239 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4240 // @codingStandardsIgnoreEnd
4241 $legacyAnchor .= "_$i";
4242 $refers["${legacyArrayKey}_$i"] = true;
4243 } else {
4244 $refers[$legacyArrayKey] = true;
4247 # Don't number the heading if it is the only one (looks silly)
4248 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4249 # the two are different if the line contains a link
4250 $headline = Html::element(
4251 'span',
4252 [ 'class' => 'mw-headline-number' ],
4253 $numbering
4254 ) . ' ' . $headline;
4257 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4258 $toc .= Linker::tocLine( $anchor, $tocline,
4259 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4262 # Add the section to the section tree
4263 # Find the DOM node for this header
4264 $noOffset = ( $isTemplate || $sectionIndex === false );
4265 while ( $node && !$noOffset ) {
4266 if ( $node->getName() === 'h' ) {
4267 $bits = $node->splitHeading();
4268 if ( $bits['i'] == $sectionIndex ) {
4269 break;
4272 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4273 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4274 $node = $node->getNextSibling();
4276 $tocraw[] = [
4277 'toclevel' => $toclevel,
4278 'level' => $level,
4279 'line' => $tocline,
4280 'number' => $numbering,
4281 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4282 'fromtitle' => $titleText,
4283 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4284 'anchor' => $anchor,
4287 # give headline the correct <h#> tag
4288 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4289 // Output edit section links as markers with styles that can be customized by skins
4290 if ( $isTemplate ) {
4291 # Put a T flag in the section identifier, to indicate to extractSections()
4292 # that sections inside <includeonly> should be counted.
4293 $editsectionPage = $titleText;
4294 $editsectionSection = "T-$sectionIndex";
4295 $editsectionContent = null;
4296 } else {
4297 $editsectionPage = $this->mTitle->getPrefixedText();
4298 $editsectionSection = $sectionIndex;
4299 $editsectionContent = $headlineHint;
4301 // We use a bit of pesudo-xml for editsection markers. The
4302 // language converter is run later on. Using a UNIQ style marker
4303 // leads to the converter screwing up the tokens when it
4304 // converts stuff. And trying to insert strip tags fails too. At
4305 // this point all real inputted tags have already been escaped,
4306 // so we don't have to worry about a user trying to input one of
4307 // these markers directly. We use a page and section attribute
4308 // to stop the language converter from converting these
4309 // important bits of data, but put the headline hint inside a
4310 // content block because the language converter is supposed to
4311 // be able to convert that piece of data.
4312 // Gets replaced with html in ParserOutput::getText
4313 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4314 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4315 if ( $editsectionContent !== null ) {
4316 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4317 } else {
4318 $editlink .= '/>';
4320 } else {
4321 $editlink = '';
4323 $head[$headlineCount] = Linker::makeHeadline( $level,
4324 $matches['attrib'][$headlineCount], $anchor, $headline,
4325 $editlink, $legacyAnchor );
4327 $headlineCount++;
4330 $this->setOutputType( $oldType );
4332 # Never ever show TOC if no headers
4333 if ( $numVisible < 1 ) {
4334 $enoughToc = false;
4337 if ( $enoughToc ) {
4338 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4339 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4341 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4342 $this->mOutput->setTOCHTML( $toc );
4343 $toc = self::TOC_START . $toc . self::TOC_END;
4344 $this->mOutput->addModules( 'mediawiki.toc' );
4347 if ( $isMain ) {
4348 $this->mOutput->setSections( $tocraw );
4351 # split up and insert constructed headlines
4352 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4353 $i = 0;
4355 // build an array of document sections
4356 $sections = [];
4357 foreach ( $blocks as $block ) {
4358 // $head is zero-based, sections aren't.
4359 if ( empty( $head[$i - 1] ) ) {
4360 $sections[$i] = $block;
4361 } else {
4362 $sections[$i] = $head[$i - 1] . $block;
4366 * Send a hook, one per section.
4367 * The idea here is to be able to make section-level DIVs, but to do so in a
4368 * lower-impact, more correct way than r50769
4370 * $this : caller
4371 * $section : the section number
4372 * &$sectionContent : ref to the content of the section
4373 * $showEditLinks : boolean describing whether this section has an edit link
4375 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4377 $i++;
4380 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4381 // append the TOC at the beginning
4382 // Top anchor now in skin
4383 $sections[0] = $sections[0] . $toc . "\n";
4386 $full .= implode( '', $sections );
4388 if ( $this->mForceTocPosition ) {
4389 return str_replace( '<!--MWTOC-->', $toc, $full );
4390 } else {
4391 return $full;
4396 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4397 * conversion, substituting signatures, {{subst:}} templates, etc.
4399 * @param string $text The text to transform
4400 * @param Title $title The Title object for the current article
4401 * @param User $user The User object describing the current user
4402 * @param ParserOptions $options Parsing options
4403 * @param bool $clearState Whether to clear the parser state first
4404 * @return string The altered wiki markup
4406 public function preSaveTransform( $text, Title $title, User $user,
4407 ParserOptions $options, $clearState = true
4409 if ( $clearState ) {
4410 $magicScopeVariable = $this->lock();
4412 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4413 $this->setUser( $user );
4415 // We still normalize line endings for backwards-compatibility
4416 // with other code that just calls PST, but this should already
4417 // be handled in TextContent subclasses
4418 $text = TextContent::normalizeLineEndings( $text );
4420 if ( $options->getPreSaveTransform() ) {
4421 $text = $this->pstPass2( $text, $user );
4423 $text = $this->mStripState->unstripBoth( $text );
4425 $this->setUser( null ); # Reset
4427 return $text;
4431 * Pre-save transform helper function
4433 * @param string $text
4434 * @param User $user
4436 * @return string
4438 private function pstPass2( $text, $user ) {
4439 global $wgContLang;
4441 # Note: This is the timestamp saved as hardcoded wikitext to
4442 # the database, we use $wgContLang here in order to give
4443 # everyone the same signature and use the default one rather
4444 # than the one selected in each user's preferences.
4445 # (see also bug 12815)
4446 $ts = $this->mOptions->getTimestamp();
4447 $timestamp = MWTimestamp::getLocalInstance( $ts );
4448 $ts = $timestamp->format( 'YmdHis' );
4449 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4451 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4453 # Variable replacement
4454 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4455 $text = $this->replaceVariables( $text );
4457 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4458 # which may corrupt this parser instance via its wfMessage()->text() call-
4460 # Signatures
4461 $sigText = $this->getUserSig( $user );
4462 $text = strtr( $text, [
4463 '~~~~~' => $d,
4464 '~~~~' => "$sigText $d",
4465 '~~~' => $sigText
4466 ] );
4468 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4469 $tc = '[' . Title::legalChars() . ']';
4470 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4472 // [[ns:page (context)|]]
4473 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4474 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4475 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4476 // [[ns:page (context), context|]] (using either single or double-width comma)
4477 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4478 // [[|page]] (reverse pipe trick: add context from page title)
4479 $p2 = "/\[\[\\|($tc+)]]/";
4481 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4482 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4483 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4484 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4486 $t = $this->mTitle->getText();
4487 $m = [];
4488 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4489 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4490 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4491 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4492 } else {
4493 # if there's no context, don't bother duplicating the title
4494 $text = preg_replace( $p2, '[[\\1]]', $text );
4497 return $text;
4501 * Fetch the user's signature text, if any, and normalize to
4502 * validated, ready-to-insert wikitext.
4503 * If you have pre-fetched the nickname or the fancySig option, you can
4504 * specify them here to save a database query.
4505 * Do not reuse this parser instance after calling getUserSig(),
4506 * as it may have changed if it's the $wgParser.
4508 * @param User $user
4509 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4510 * @param bool|null $fancySig whether the nicknname is the complete signature
4511 * or null to use default value
4512 * @return string
4514 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4515 global $wgMaxSigChars;
4517 $username = $user->getName();
4519 # If not given, retrieve from the user object.
4520 if ( $nickname === false ) {
4521 $nickname = $user->getOption( 'nickname' );
4524 if ( is_null( $fancySig ) ) {
4525 $fancySig = $user->getBoolOption( 'fancysig' );
4528 $nickname = $nickname == null ? $username : $nickname;
4530 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4531 $nickname = $username;
4532 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4533 } elseif ( $fancySig !== false ) {
4534 # Sig. might contain markup; validate this
4535 if ( $this->validateSig( $nickname ) !== false ) {
4536 # Validated; clean up (if needed) and return it
4537 return $this->cleanSig( $nickname, true );
4538 } else {
4539 # Failed to validate; fall back to the default
4540 $nickname = $username;
4541 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4545 # Make sure nickname doesnt get a sig in a sig
4546 $nickname = self::cleanSigInSig( $nickname );
4548 # If we're still here, make it a link to the user page
4549 $userText = wfEscapeWikiText( $username );
4550 $nickText = wfEscapeWikiText( $nickname );
4551 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4553 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4554 ->title( $this->getTitle() )->text();
4558 * Check that the user's signature contains no bad XML
4560 * @param string $text
4561 * @return string|bool An expanded string, or false if invalid.
4563 public function validateSig( $text ) {
4564 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4568 * Clean up signature text
4570 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4571 * 2) Substitute all transclusions
4573 * @param string $text
4574 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4575 * @return string Signature text
4577 public function cleanSig( $text, $parsing = false ) {
4578 if ( !$parsing ) {
4579 global $wgTitle;
4580 $magicScopeVariable = $this->lock();
4581 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4584 # Option to disable this feature
4585 if ( !$this->mOptions->getCleanSignatures() ) {
4586 return $text;
4589 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4590 # => Move this logic to braceSubstitution()
4591 $substWord = MagicWord::get( 'subst' );
4592 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4593 $substText = '{{' . $substWord->getSynonym( 0 );
4595 $text = preg_replace( $substRegex, $substText, $text );
4596 $text = self::cleanSigInSig( $text );
4597 $dom = $this->preprocessToDom( $text );
4598 $frame = $this->getPreprocessor()->newFrame();
4599 $text = $frame->expand( $dom );
4601 if ( !$parsing ) {
4602 $text = $this->mStripState->unstripBoth( $text );
4605 return $text;
4609 * Strip 3, 4 or 5 tildes out of signatures.
4611 * @param string $text
4612 * @return string Signature text with /~{3,5}/ removed
4614 public static function cleanSigInSig( $text ) {
4615 $text = preg_replace( '/~{3,5}/', '', $text );
4616 return $text;
4620 * Set up some variables which are usually set up in parse()
4621 * so that an external function can call some class members with confidence
4623 * @param Title|null $title
4624 * @param ParserOptions $options
4625 * @param int $outputType
4626 * @param bool $clearState
4628 public function startExternalParse( Title $title = null, ParserOptions $options,
4629 $outputType, $clearState = true
4631 $this->startParse( $title, $options, $outputType, $clearState );
4635 * @param Title|null $title
4636 * @param ParserOptions $options
4637 * @param int $outputType
4638 * @param bool $clearState
4640 private function startParse( Title $title = null, ParserOptions $options,
4641 $outputType, $clearState = true
4643 $this->setTitle( $title );
4644 $this->mOptions = $options;
4645 $this->setOutputType( $outputType );
4646 if ( $clearState ) {
4647 $this->clearState();
4652 * Wrapper for preprocess()
4654 * @param string $text The text to preprocess
4655 * @param ParserOptions $options Options
4656 * @param Title|null $title Title object or null to use $wgTitle
4657 * @return string
4659 public function transformMsg( $text, $options, $title = null ) {
4660 static $executing = false;
4662 # Guard against infinite recursion
4663 if ( $executing ) {
4664 return $text;
4666 $executing = true;
4668 if ( !$title ) {
4669 global $wgTitle;
4670 $title = $wgTitle;
4673 $text = $this->preprocess( $text, $title, $options );
4675 $executing = false;
4676 return $text;
4680 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4681 * The callback should have the following form:
4682 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4684 * Transform and return $text. Use $parser for any required context, e.g. use
4685 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4687 * Hooks may return extended information by returning an array, of which the
4688 * first numbered element (index 0) must be the return string, and all other
4689 * entries are extracted into local variables within an internal function
4690 * in the Parser class.
4692 * This interface (introduced r61913) appears to be undocumented, but
4693 * 'markerType' is used by some core tag hooks to override which strip
4694 * array their results are placed in. **Use great caution if attempting
4695 * this interface, as it is not documented and injudicious use could smash
4696 * private variables.**
4698 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4699 * @param callable $callback The callback function (and object) to use for the tag
4700 * @throws MWException
4701 * @return callable|null The old value of the mTagHooks array associated with the hook
4703 public function setHook( $tag, $callback ) {
4704 $tag = strtolower( $tag );
4705 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4706 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4708 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4709 $this->mTagHooks[$tag] = $callback;
4710 if ( !in_array( $tag, $this->mStripList ) ) {
4711 $this->mStripList[] = $tag;
4714 return $oldVal;
4718 * As setHook(), but letting the contents be parsed.
4720 * Transparent tag hooks are like regular XML-style tag hooks, except they
4721 * operate late in the transformation sequence, on HTML instead of wikitext.
4723 * This is probably obsoleted by things dealing with parser frames?
4724 * The only extension currently using it is geoserver.
4726 * @since 1.10
4727 * @todo better document or deprecate this
4729 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4730 * @param callable $callback The callback function (and object) to use for the tag
4731 * @throws MWException
4732 * @return callable|null The old value of the mTagHooks array associated with the hook
4734 public function setTransparentTagHook( $tag, $callback ) {
4735 $tag = strtolower( $tag );
4736 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4737 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4739 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4740 $this->mTransparentTagHooks[$tag] = $callback;
4742 return $oldVal;
4746 * Remove all tag hooks
4748 public function clearTagHooks() {
4749 $this->mTagHooks = [];
4750 $this->mFunctionTagHooks = [];
4751 $this->mStripList = $this->mDefaultStripList;
4755 * Create a function, e.g. {{sum:1|2|3}}
4756 * The callback function should have the form:
4757 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4759 * Or with Parser::SFH_OBJECT_ARGS:
4760 * function myParserFunction( $parser, $frame, $args ) { ... }
4762 * The callback may either return the text result of the function, or an array with the text
4763 * in element 0, and a number of flags in the other elements. The names of the flags are
4764 * specified in the keys. Valid flags are:
4765 * found The text returned is valid, stop processing the template. This
4766 * is on by default.
4767 * nowiki Wiki markup in the return value should be escaped
4768 * isHTML The returned text is HTML, armour it against wikitext transformation
4770 * @param string $id The magic word ID
4771 * @param callable $callback The callback function (and object) to use
4772 * @param int $flags A combination of the following flags:
4773 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4775 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4776 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4777 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4778 * the arguments, and to control the way they are expanded.
4780 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4781 * arguments, for instance:
4782 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4784 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4785 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4786 * working if/when this is changed.
4788 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4789 * expansion.
4791 * Please read the documentation in includes/parser/Preprocessor.php for more information
4792 * about the methods available in PPFrame and PPNode.
4794 * @throws MWException
4795 * @return string|callable The old callback function for this name, if any
4797 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4798 global $wgContLang;
4800 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4801 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4803 # Add to function cache
4804 $mw = MagicWord::get( $id );
4805 if ( !$mw ) {
4806 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4809 $synonyms = $mw->getSynonyms();
4810 $sensitive = intval( $mw->isCaseSensitive() );
4812 foreach ( $synonyms as $syn ) {
4813 # Case
4814 if ( !$sensitive ) {
4815 $syn = $wgContLang->lc( $syn );
4817 # Add leading hash
4818 if ( !( $flags & self::SFH_NO_HASH ) ) {
4819 $syn = '#' . $syn;
4821 # Remove trailing colon
4822 if ( substr( $syn, -1, 1 ) === ':' ) {
4823 $syn = substr( $syn, 0, -1 );
4825 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4827 return $oldVal;
4831 * Get all registered function hook identifiers
4833 * @return array
4835 public function getFunctionHooks() {
4836 return array_keys( $this->mFunctionHooks );
4840 * Create a tag function, e.g. "<test>some stuff</test>".
4841 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4842 * Unlike parser functions, their content is not preprocessed.
4843 * @param string $tag
4844 * @param callable $callback
4845 * @param int $flags
4846 * @throws MWException
4847 * @return null
4849 public function setFunctionTagHook( $tag, $callback, $flags ) {
4850 $tag = strtolower( $tag );
4851 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4852 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4854 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4855 $this->mFunctionTagHooks[$tag] : null;
4856 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4858 if ( !in_array( $tag, $this->mStripList ) ) {
4859 $this->mStripList[] = $tag;
4862 return $old;
4866 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4867 * Placeholders created in Linker::link()
4869 * @param string $text
4870 * @param int $options
4872 public function replaceLinkHolders( &$text, $options = 0 ) {
4873 $this->mLinkHolders->replace( $text );
4877 * Replace "<!--LINK-->" link placeholders with plain text of links
4878 * (not HTML-formatted).
4880 * @param string $text
4881 * @return string
4883 public function replaceLinkHoldersText( $text ) {
4884 return $this->mLinkHolders->replaceText( $text );
4888 * Renders an image gallery from a text with one line per image.
4889 * text labels may be given by using |-style alternative text. E.g.
4890 * Image:one.jpg|The number "1"
4891 * Image:tree.jpg|A tree
4892 * given as text will return the HTML of a gallery with two images,
4893 * labeled 'The number "1"' and
4894 * 'A tree'.
4896 * @param string $text
4897 * @param array $params
4898 * @return string HTML
4900 public function renderImageGallery( $text, $params ) {
4902 $mode = false;
4903 if ( isset( $params['mode'] ) ) {
4904 $mode = $params['mode'];
4907 try {
4908 $ig = ImageGalleryBase::factory( $mode );
4909 } catch ( Exception $e ) {
4910 // If invalid type set, fallback to default.
4911 $ig = ImageGalleryBase::factory( false );
4914 $ig->setContextTitle( $this->mTitle );
4915 $ig->setShowBytes( false );
4916 $ig->setShowFilename( false );
4917 $ig->setParser( $this );
4918 $ig->setHideBadImages();
4919 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4921 if ( isset( $params['showfilename'] ) ) {
4922 $ig->setShowFilename( true );
4923 } else {
4924 $ig->setShowFilename( false );
4926 if ( isset( $params['caption'] ) ) {
4927 $caption = $params['caption'];
4928 $caption = htmlspecialchars( $caption );
4929 $caption = $this->replaceInternalLinks( $caption );
4930 $ig->setCaptionHtml( $caption );
4932 if ( isset( $params['perrow'] ) ) {
4933 $ig->setPerRow( $params['perrow'] );
4935 if ( isset( $params['widths'] ) ) {
4936 $ig->setWidths( $params['widths'] );
4938 if ( isset( $params['heights'] ) ) {
4939 $ig->setHeights( $params['heights'] );
4941 $ig->setAdditionalOptions( $params );
4943 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4945 $lines = StringUtils::explode( "\n", $text );
4946 foreach ( $lines as $line ) {
4947 # match lines like these:
4948 # Image:someimage.jpg|This is some image
4949 $matches = [];
4950 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4951 # Skip empty lines
4952 if ( count( $matches ) == 0 ) {
4953 continue;
4956 if ( strpos( $matches[0], '%' ) !== false ) {
4957 $matches[1] = rawurldecode( $matches[1] );
4959 $title = Title::newFromText( $matches[1], NS_FILE );
4960 if ( is_null( $title ) ) {
4961 # Bogus title. Ignore these so we don't bomb out later.
4962 continue;
4965 # We need to get what handler the file uses, to figure out parameters.
4966 # Note, a hook can overide the file name, and chose an entirely different
4967 # file (which potentially could be of a different type and have different handler).
4968 $options = [];
4969 $descQuery = false;
4970 Hooks::run( 'BeforeParserFetchFileAndTitle',
4971 [ $this, $title, &$options, &$descQuery ] );
4972 # Don't register it now, as ImageGallery does that later.
4973 $file = $this->fetchFileNoRegister( $title, $options );
4974 $handler = $file ? $file->getHandler() : false;
4976 $paramMap = [
4977 'img_alt' => 'gallery-internal-alt',
4978 'img_link' => 'gallery-internal-link',
4980 if ( $handler ) {
4981 $paramMap = $paramMap + $handler->getParamMap();
4982 // We don't want people to specify per-image widths.
4983 // Additionally the width parameter would need special casing anyhow.
4984 unset( $paramMap['img_width'] );
4987 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4989 $label = '';
4990 $alt = '';
4991 $link = '';
4992 $handlerOptions = [];
4993 if ( isset( $matches[3] ) ) {
4994 // look for an |alt= definition while trying not to break existing
4995 // captions with multiple pipes (|) in it, until a more sensible grammar
4996 // is defined for images in galleries
4998 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4999 // splitting on '|' is a bit odd, and different from makeImage.
5000 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5001 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5003 foreach ( $parameterMatches as $parameterMatch ) {
5004 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5005 if ( $magicName ) {
5006 $paramName = $paramMap[$magicName];
5008 switch ( $paramName ) {
5009 case 'gallery-internal-alt':
5010 $alt = $this->stripAltText( $match, false );
5011 break;
5012 case 'gallery-internal-link':
5013 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5014 $chars = self::EXT_LINK_URL_CLASS;
5015 $addr = self::EXT_LINK_ADDR;
5016 $prots = $this->mUrlProtocols;
5017 // check to see if link matches an absolute url, if not then it must be a wiki link.
5018 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5019 $link = $linkValue;
5020 } else {
5021 $localLinkTitle = Title::newFromText( $linkValue );
5022 if ( $localLinkTitle !== null ) {
5023 $link = $localLinkTitle->getLinkURL();
5026 break;
5027 default:
5028 // Must be a handler specific parameter.
5029 if ( $handler->validateParam( $paramName, $match ) ) {
5030 $handlerOptions[$paramName] = $match;
5031 } else {
5032 // Guess not, consider it as caption.
5033 wfDebug( "$parameterMatch failed parameter validation\n" );
5034 $label = '|' . $parameterMatch;
5038 } else {
5039 // Last pipe wins.
5040 $label = '|' . $parameterMatch;
5043 // Remove the pipe.
5044 $label = substr( $label, 1 );
5047 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5049 $html = $ig->toHTML();
5050 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5051 return $html;
5055 * @param MediaHandler $handler
5056 * @return array
5058 public function getImageParams( $handler ) {
5059 if ( $handler ) {
5060 $handlerClass = get_class( $handler );
5061 } else {
5062 $handlerClass = '';
5064 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5065 # Initialise static lists
5066 static $internalParamNames = [
5067 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5068 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5069 'bottom', 'text-bottom' ],
5070 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5071 'upright', 'border', 'link', 'alt', 'class' ],
5073 static $internalParamMap;
5074 if ( !$internalParamMap ) {
5075 $internalParamMap = [];
5076 foreach ( $internalParamNames as $type => $names ) {
5077 foreach ( $names as $name ) {
5078 $magicName = str_replace( '-', '_', "img_$name" );
5079 $internalParamMap[$magicName] = [ $type, $name ];
5084 # Add handler params
5085 $paramMap = $internalParamMap;
5086 if ( $handler ) {
5087 $handlerParamMap = $handler->getParamMap();
5088 foreach ( $handlerParamMap as $magic => $paramName ) {
5089 $paramMap[$magic] = [ 'handler', $paramName ];
5092 $this->mImageParams[$handlerClass] = $paramMap;
5093 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5095 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5099 * Parse image options text and use it to make an image
5101 * @param Title $title
5102 * @param string $options
5103 * @param LinkHolderArray|bool $holders
5104 * @return string HTML
5106 public function makeImage( $title, $options, $holders = false ) {
5107 # Check if the options text is of the form "options|alt text"
5108 # Options are:
5109 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5110 # * left no resizing, just left align. label is used for alt= only
5111 # * right same, but right aligned
5112 # * none same, but not aligned
5113 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5114 # * center center the image
5115 # * frame Keep original image size, no magnify-button.
5116 # * framed Same as "frame"
5117 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5118 # * upright reduce width for upright images, rounded to full __0 px
5119 # * border draw a 1px border around the image
5120 # * alt Text for HTML alt attribute (defaults to empty)
5121 # * class Set a class for img node
5122 # * link Set the target of the image link. Can be external, interwiki, or local
5123 # vertical-align values (no % or length right now):
5124 # * baseline
5125 # * sub
5126 # * super
5127 # * top
5128 # * text-top
5129 # * middle
5130 # * bottom
5131 # * text-bottom
5133 $parts = StringUtils::explode( "|", $options );
5135 # Give extensions a chance to select the file revision for us
5136 $options = [];
5137 $descQuery = false;
5138 Hooks::run( 'BeforeParserFetchFileAndTitle',
5139 [ $this, $title, &$options, &$descQuery ] );
5140 # Fetch and register the file (file title may be different via hooks)
5141 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5143 # Get parameter map
5144 $handler = $file ? $file->getHandler() : false;
5146 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5148 if ( !$file ) {
5149 $this->addTrackingCategory( 'broken-file-category' );
5152 # Process the input parameters
5153 $caption = '';
5154 $params = [ 'frame' => [], 'handler' => [],
5155 'horizAlign' => [], 'vertAlign' => [] ];
5156 $seenformat = false;
5157 foreach ( $parts as $part ) {
5158 $part = trim( $part );
5159 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5160 $validated = false;
5161 if ( isset( $paramMap[$magicName] ) ) {
5162 list( $type, $paramName ) = $paramMap[$magicName];
5164 # Special case; width and height come in one variable together
5165 if ( $type === 'handler' && $paramName === 'width' ) {
5166 $parsedWidthParam = $this->parseWidthParam( $value );
5167 if ( isset( $parsedWidthParam['width'] ) ) {
5168 $width = $parsedWidthParam['width'];
5169 if ( $handler->validateParam( 'width', $width ) ) {
5170 $params[$type]['width'] = $width;
5171 $validated = true;
5174 if ( isset( $parsedWidthParam['height'] ) ) {
5175 $height = $parsedWidthParam['height'];
5176 if ( $handler->validateParam( 'height', $height ) ) {
5177 $params[$type]['height'] = $height;
5178 $validated = true;
5181 # else no validation -- bug 13436
5182 } else {
5183 if ( $type === 'handler' ) {
5184 # Validate handler parameter
5185 $validated = $handler->validateParam( $paramName, $value );
5186 } else {
5187 # Validate internal parameters
5188 switch ( $paramName ) {
5189 case 'manualthumb':
5190 case 'alt':
5191 case 'class':
5192 # @todo FIXME: Possibly check validity here for
5193 # manualthumb? downstream behavior seems odd with
5194 # missing manual thumbs.
5195 $validated = true;
5196 $value = $this->stripAltText( $value, $holders );
5197 break;
5198 case 'link':
5199 $chars = self::EXT_LINK_URL_CLASS;
5200 $addr = self::EXT_LINK_ADDR;
5201 $prots = $this->mUrlProtocols;
5202 if ( $value === '' ) {
5203 $paramName = 'no-link';
5204 $value = true;
5205 $validated = true;
5206 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5207 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5208 $paramName = 'link-url';
5209 $this->mOutput->addExternalLink( $value );
5210 if ( $this->mOptions->getExternalLinkTarget() ) {
5211 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5213 $validated = true;
5215 } else {
5216 $linkTitle = Title::newFromText( $value );
5217 if ( $linkTitle ) {
5218 $paramName = 'link-title';
5219 $value = $linkTitle;
5220 $this->mOutput->addLink( $linkTitle );
5221 $validated = true;
5224 break;
5225 case 'frameless':
5226 case 'framed':
5227 case 'thumbnail':
5228 // use first appearing option, discard others.
5229 $validated = ! $seenformat;
5230 $seenformat = true;
5231 break;
5232 default:
5233 # Most other things appear to be empty or numeric...
5234 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5238 if ( $validated ) {
5239 $params[$type][$paramName] = $value;
5243 if ( !$validated ) {
5244 $caption = $part;
5248 # Process alignment parameters
5249 if ( $params['horizAlign'] ) {
5250 $params['frame']['align'] = key( $params['horizAlign'] );
5252 if ( $params['vertAlign'] ) {
5253 $params['frame']['valign'] = key( $params['vertAlign'] );
5256 $params['frame']['caption'] = $caption;
5258 # Will the image be presented in a frame, with the caption below?
5259 $imageIsFramed = isset( $params['frame']['frame'] )
5260 || isset( $params['frame']['framed'] )
5261 || isset( $params['frame']['thumbnail'] )
5262 || isset( $params['frame']['manualthumb'] );
5264 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5265 # came to also set the caption, ordinary text after the image -- which
5266 # makes no sense, because that just repeats the text multiple times in
5267 # screen readers. It *also* came to set the title attribute.
5268 # Now that we have an alt attribute, we should not set the alt text to
5269 # equal the caption: that's worse than useless, it just repeats the
5270 # text. This is the framed/thumbnail case. If there's no caption, we
5271 # use the unnamed parameter for alt text as well, just for the time be-
5272 # ing, if the unnamed param is set and the alt param is not.
5273 # For the future, we need to figure out if we want to tweak this more,
5274 # e.g., introducing a title= parameter for the title; ignoring the un-
5275 # named parameter entirely for images without a caption; adding an ex-
5276 # plicit caption= parameter and preserving the old magic unnamed para-
5277 # meter for BC; ...
5278 if ( $imageIsFramed ) { # Framed image
5279 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5280 # No caption or alt text, add the filename as the alt text so
5281 # that screen readers at least get some description of the image
5282 $params['frame']['alt'] = $title->getText();
5284 # Do not set $params['frame']['title'] because tooltips don't make sense
5285 # for framed images
5286 } else { # Inline image
5287 if ( !isset( $params['frame']['alt'] ) ) {
5288 # No alt text, use the "caption" for the alt text
5289 if ( $caption !== '' ) {
5290 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5291 } else {
5292 # No caption, fall back to using the filename for the
5293 # alt text
5294 $params['frame']['alt'] = $title->getText();
5297 # Use the "caption" for the tooltip text
5298 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5301 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5303 # Linker does the rest
5304 $time = isset( $options['time'] ) ? $options['time'] : false;
5305 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5306 $time, $descQuery, $this->mOptions->getThumbSize() );
5308 # Give the handler a chance to modify the parser object
5309 if ( $handler ) {
5310 $handler->parserTransformHook( $this, $file );
5313 return $ret;
5317 * @param string $caption
5318 * @param LinkHolderArray|bool $holders
5319 * @return mixed|string
5321 protected function stripAltText( $caption, $holders ) {
5322 # Strip bad stuff out of the title (tooltip). We can't just use
5323 # replaceLinkHoldersText() here, because if this function is called
5324 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5325 if ( $holders ) {
5326 $tooltip = $holders->replaceText( $caption );
5327 } else {
5328 $tooltip = $this->replaceLinkHoldersText( $caption );
5331 # make sure there are no placeholders in thumbnail attributes
5332 # that are later expanded to html- so expand them now and
5333 # remove the tags
5334 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5335 $tooltip = Sanitizer::stripAllTags( $tooltip );
5337 return $tooltip;
5341 * Set a flag in the output object indicating that the content is dynamic and
5342 * shouldn't be cached.
5343 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5345 public function disableCache() {
5346 wfDebug( "Parser output marked as uncacheable.\n" );
5347 if ( !$this->mOutput ) {
5348 throw new MWException( __METHOD__ .
5349 " can only be called when actually parsing something" );
5351 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5355 * Callback from the Sanitizer for expanding items found in HTML attribute
5356 * values, so they can be safely tested and escaped.
5358 * @param string $text
5359 * @param bool|PPFrame $frame
5360 * @return string
5362 public function attributeStripCallback( &$text, $frame = false ) {
5363 $text = $this->replaceVariables( $text, $frame );
5364 $text = $this->mStripState->unstripBoth( $text );
5365 return $text;
5369 * Accessor
5371 * @return array
5373 public function getTags() {
5374 return array_merge(
5375 array_keys( $this->mTransparentTagHooks ),
5376 array_keys( $this->mTagHooks ),
5377 array_keys( $this->mFunctionTagHooks )
5382 * Replace transparent tags in $text with the values given by the callbacks.
5384 * Transparent tag hooks are like regular XML-style tag hooks, except they
5385 * operate late in the transformation sequence, on HTML instead of wikitext.
5387 * @param string $text
5389 * @return string
5391 public function replaceTransparentTags( $text ) {
5392 $matches = [];
5393 $elements = array_keys( $this->mTransparentTagHooks );
5394 $text = self::extractTagsAndParams( $elements, $text, $matches );
5395 $replacements = [];
5397 foreach ( $matches as $marker => $data ) {
5398 list( $element, $content, $params, $tag ) = $data;
5399 $tagName = strtolower( $element );
5400 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5401 $output = call_user_func_array(
5402 $this->mTransparentTagHooks[$tagName],
5403 [ $content, $params, $this ]
5405 } else {
5406 $output = $tag;
5408 $replacements[$marker] = $output;
5410 return strtr( $text, $replacements );
5414 * Break wikitext input into sections, and either pull or replace
5415 * some particular section's text.
5417 * External callers should use the getSection and replaceSection methods.
5419 * @param string $text Page wikitext
5420 * @param string|number $sectionId A section identifier string of the form:
5421 * "<flag1> - <flag2> - ... - <section number>"
5423 * Currently the only recognised flag is "T", which means the target section number
5424 * was derived during a template inclusion parse, in other words this is a template
5425 * section edit link. If no flags are given, it was an ordinary section edit link.
5426 * This flag is required to avoid a section numbering mismatch when a section is
5427 * enclosed by "<includeonly>" (bug 6563).
5429 * The section number 0 pulls the text before the first heading; other numbers will
5430 * pull the given section along with its lower-level subsections. If the section is
5431 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5433 * Section 0 is always considered to exist, even if it only contains the empty
5434 * string. If $text is the empty string and section 0 is replaced, $newText is
5435 * returned.
5437 * @param string $mode One of "get" or "replace"
5438 * @param string $newText Replacement text for section data.
5439 * @return string For "get", the extracted section text.
5440 * for "replace", the whole page with the section replaced.
5442 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5443 global $wgTitle; # not generally used but removes an ugly failure mode
5445 $magicScopeVariable = $this->lock();
5446 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5447 $outText = '';
5448 $frame = $this->getPreprocessor()->newFrame();
5450 # Process section extraction flags
5451 $flags = 0;
5452 $sectionParts = explode( '-', $sectionId );
5453 $sectionIndex = array_pop( $sectionParts );
5454 foreach ( $sectionParts as $part ) {
5455 if ( $part === 'T' ) {
5456 $flags |= self::PTD_FOR_INCLUSION;
5460 # Check for empty input
5461 if ( strval( $text ) === '' ) {
5462 # Only sections 0 and T-0 exist in an empty document
5463 if ( $sectionIndex == 0 ) {
5464 if ( $mode === 'get' ) {
5465 return '';
5466 } else {
5467 return $newText;
5469 } else {
5470 if ( $mode === 'get' ) {
5471 return $newText;
5472 } else {
5473 return $text;
5478 # Preprocess the text
5479 $root = $this->preprocessToDom( $text, $flags );
5481 # <h> nodes indicate section breaks
5482 # They can only occur at the top level, so we can find them by iterating the root's children
5483 $node = $root->getFirstChild();
5485 # Find the target section
5486 if ( $sectionIndex == 0 ) {
5487 # Section zero doesn't nest, level=big
5488 $targetLevel = 1000;
5489 } else {
5490 while ( $node ) {
5491 if ( $node->getName() === 'h' ) {
5492 $bits = $node->splitHeading();
5493 if ( $bits['i'] == $sectionIndex ) {
5494 $targetLevel = $bits['level'];
5495 break;
5498 if ( $mode === 'replace' ) {
5499 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5501 $node = $node->getNextSibling();
5505 if ( !$node ) {
5506 # Not found
5507 if ( $mode === 'get' ) {
5508 return $newText;
5509 } else {
5510 return $text;
5514 # Find the end of the section, including nested sections
5515 do {
5516 if ( $node->getName() === 'h' ) {
5517 $bits = $node->splitHeading();
5518 $curLevel = $bits['level'];
5519 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5520 break;
5523 if ( $mode === 'get' ) {
5524 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5526 $node = $node->getNextSibling();
5527 } while ( $node );
5529 # Write out the remainder (in replace mode only)
5530 if ( $mode === 'replace' ) {
5531 # Output the replacement text
5532 # Add two newlines on -- trailing whitespace in $newText is conventionally
5533 # stripped by the editor, so we need both newlines to restore the paragraph gap
5534 # Only add trailing whitespace if there is newText
5535 if ( $newText != "" ) {
5536 $outText .= $newText . "\n\n";
5539 while ( $node ) {
5540 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5541 $node = $node->getNextSibling();
5545 if ( is_string( $outText ) ) {
5546 # Re-insert stripped tags
5547 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5550 return $outText;
5554 * This function returns the text of a section, specified by a number ($section).
5555 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5556 * the first section before any such heading (section 0).
5558 * If a section contains subsections, these are also returned.
5560 * @param string $text Text to look in
5561 * @param string|number $sectionId Section identifier as a number or string
5562 * (e.g. 0, 1 or 'T-1').
5563 * @param string $defaultText Default to return if section is not found
5565 * @return string Text of the requested section
5567 public function getSection( $text, $sectionId, $defaultText = '' ) {
5568 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5572 * This function returns $oldtext after the content of the section
5573 * specified by $section has been replaced with $text. If the target
5574 * section does not exist, $oldtext is returned unchanged.
5576 * @param string $oldText Former text of the article
5577 * @param string|number $sectionId Section identifier as a number or string
5578 * (e.g. 0, 1 or 'T-1').
5579 * @param string $newText Replacing text
5581 * @return string Modified text
5583 public function replaceSection( $oldText, $sectionId, $newText ) {
5584 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5588 * Get the ID of the revision we are parsing
5590 * @return int|null
5592 public function getRevisionId() {
5593 return $this->mRevisionId;
5597 * Get the revision object for $this->mRevisionId
5599 * @return Revision|null Either a Revision object or null
5600 * @since 1.23 (public since 1.23)
5602 public function getRevisionObject() {
5603 if ( !is_null( $this->mRevisionObject ) ) {
5604 return $this->mRevisionObject;
5606 if ( is_null( $this->mRevisionId ) ) {
5607 return null;
5610 $rev = call_user_func(
5611 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5614 # If the parse is for a new revision, then the callback should have
5615 # already been set to force the object and should match mRevisionId.
5616 # If not, try to fetch by mRevisionId for sanity.
5617 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5618 $rev = Revision::newFromId( $this->mRevisionId );
5621 $this->mRevisionObject = $rev;
5623 return $this->mRevisionObject;
5627 * Get the timestamp associated with the current revision, adjusted for
5628 * the default server-local timestamp
5629 * @return string
5631 public function getRevisionTimestamp() {
5632 if ( is_null( $this->mRevisionTimestamp ) ) {
5633 global $wgContLang;
5635 $revObject = $this->getRevisionObject();
5636 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5638 # The cryptic '' timezone parameter tells to use the site-default
5639 # timezone offset instead of the user settings.
5640 # Since this value will be saved into the parser cache, served
5641 # to other users, and potentially even used inside links and such,
5642 # it needs to be consistent for all visitors.
5643 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5646 return $this->mRevisionTimestamp;
5650 * Get the name of the user that edited the last revision
5652 * @return string User name
5654 public function getRevisionUser() {
5655 if ( is_null( $this->mRevisionUser ) ) {
5656 $revObject = $this->getRevisionObject();
5658 # if this template is subst: the revision id will be blank,
5659 # so just use the current user's name
5660 if ( $revObject ) {
5661 $this->mRevisionUser = $revObject->getUserText();
5662 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5663 $this->mRevisionUser = $this->getUser()->getName();
5666 return $this->mRevisionUser;
5670 * Get the size of the revision
5672 * @return int|null Revision size
5674 public function getRevisionSize() {
5675 if ( is_null( $this->mRevisionSize ) ) {
5676 $revObject = $this->getRevisionObject();
5678 # if this variable is subst: the revision id will be blank,
5679 # so just use the parser input size, because the own substituation
5680 # will change the size.
5681 if ( $revObject ) {
5682 $this->mRevisionSize = $revObject->getSize();
5683 } else {
5684 $this->mRevisionSize = $this->mInputSize;
5687 return $this->mRevisionSize;
5691 * Mutator for $mDefaultSort
5693 * @param string $sort New value
5695 public function setDefaultSort( $sort ) {
5696 $this->mDefaultSort = $sort;
5697 $this->mOutput->setProperty( 'defaultsort', $sort );
5701 * Accessor for $mDefaultSort
5702 * Will use the empty string if none is set.
5704 * This value is treated as a prefix, so the
5705 * empty string is equivalent to sorting by
5706 * page name.
5708 * @return string
5710 public function getDefaultSort() {
5711 if ( $this->mDefaultSort !== false ) {
5712 return $this->mDefaultSort;
5713 } else {
5714 return '';
5719 * Accessor for $mDefaultSort
5720 * Unlike getDefaultSort(), will return false if none is set
5722 * @return string|bool
5724 public function getCustomDefaultSort() {
5725 return $this->mDefaultSort;
5729 * Try to guess the section anchor name based on a wikitext fragment
5730 * presumably extracted from a heading, for example "Header" from
5731 * "== Header ==".
5733 * @param string $text
5735 * @return string
5737 public function guessSectionNameFromWikiText( $text ) {
5738 # Strip out wikitext links(they break the anchor)
5739 $text = $this->stripSectionName( $text );
5740 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5741 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5745 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5746 * instead. For use in redirects, since IE6 interprets Redirect: headers
5747 * as something other than UTF-8 (apparently?), resulting in breakage.
5749 * @param string $text The section name
5750 * @return string An anchor
5752 public function guessLegacySectionNameFromWikiText( $text ) {
5753 # Strip out wikitext links(they break the anchor)
5754 $text = $this->stripSectionName( $text );
5755 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5756 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5760 * Strips a text string of wikitext for use in a section anchor
5762 * Accepts a text string and then removes all wikitext from the
5763 * string and leaves only the resultant text (i.e. the result of
5764 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5765 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5766 * to create valid section anchors by mimicing the output of the
5767 * parser when headings are parsed.
5769 * @param string $text Text string to be stripped of wikitext
5770 * for use in a Section anchor
5771 * @return string Filtered text string
5773 public function stripSectionName( $text ) {
5774 # Strip internal link markup
5775 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5776 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5778 # Strip external link markup
5779 # @todo FIXME: Not tolerant to blank link text
5780 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5781 # on how many empty links there are on the page - need to figure that out.
5782 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5784 # Parse wikitext quotes (italics & bold)
5785 $text = $this->doQuotes( $text );
5787 # Strip HTML tags
5788 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5789 return $text;
5793 * strip/replaceVariables/unstrip for preprocessor regression testing
5795 * @param string $text
5796 * @param Title $title
5797 * @param ParserOptions $options
5798 * @param int $outputType
5800 * @return string
5802 public function testSrvus( $text, Title $title, ParserOptions $options,
5803 $outputType = self::OT_HTML
5805 $magicScopeVariable = $this->lock();
5806 $this->startParse( $title, $options, $outputType, true );
5808 $text = $this->replaceVariables( $text );
5809 $text = $this->mStripState->unstripBoth( $text );
5810 $text = Sanitizer::removeHTMLtags( $text );
5811 return $text;
5815 * @param string $text
5816 * @param Title $title
5817 * @param ParserOptions $options
5818 * @return string
5820 public function testPst( $text, Title $title, ParserOptions $options ) {
5821 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5825 * @param string $text
5826 * @param Title $title
5827 * @param ParserOptions $options
5828 * @return string
5830 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5831 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5835 * Call a callback function on all regions of the given text that are not
5836 * inside strip markers, and replace those regions with the return value
5837 * of the callback. For example, with input:
5839 * aaa<MARKER>bbb
5841 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5842 * two strings will be replaced with the value returned by the callback in
5843 * each case.
5845 * @param string $s
5846 * @param callable $callback
5848 * @return string
5850 public function markerSkipCallback( $s, $callback ) {
5851 $i = 0;
5852 $out = '';
5853 while ( $i < strlen( $s ) ) {
5854 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5855 if ( $markerStart === false ) {
5856 $out .= call_user_func( $callback, substr( $s, $i ) );
5857 break;
5858 } else {
5859 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5860 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5861 if ( $markerEnd === false ) {
5862 $out .= substr( $s, $markerStart );
5863 break;
5864 } else {
5865 $markerEnd += strlen( self::MARKER_SUFFIX );
5866 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5867 $i = $markerEnd;
5871 return $out;
5875 * Remove any strip markers found in the given text.
5877 * @param string $text Input string
5878 * @return string
5880 public function killMarkers( $text ) {
5881 return $this->mStripState->killMarkers( $text );
5885 * Save the parser state required to convert the given half-parsed text to
5886 * HTML. "Half-parsed" in this context means the output of
5887 * recursiveTagParse() or internalParse(). This output has strip markers
5888 * from replaceVariables (extensionSubstitution() etc.), and link
5889 * placeholders from replaceLinkHolders().
5891 * Returns an array which can be serialized and stored persistently. This
5892 * array can later be loaded into another parser instance with
5893 * unserializeHalfParsedText(). The text can then be safely incorporated into
5894 * the return value of a parser hook.
5896 * @param string $text
5898 * @return array
5900 public function serializeHalfParsedText( $text ) {
5901 $data = [
5902 'text' => $text,
5903 'version' => self::HALF_PARSED_VERSION,
5904 'stripState' => $this->mStripState->getSubState( $text ),
5905 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5907 return $data;
5911 * Load the parser state given in the $data array, which is assumed to
5912 * have been generated by serializeHalfParsedText(). The text contents is
5913 * extracted from the array, and its markers are transformed into markers
5914 * appropriate for the current Parser instance. This transformed text is
5915 * returned, and can be safely included in the return value of a parser
5916 * hook.
5918 * If the $data array has been stored persistently, the caller should first
5919 * check whether it is still valid, by calling isValidHalfParsedText().
5921 * @param array $data Serialized data
5922 * @throws MWException
5923 * @return string
5925 public function unserializeHalfParsedText( $data ) {
5926 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5927 throw new MWException( __METHOD__ . ': invalid version' );
5930 # First, extract the strip state.
5931 $texts = [ $data['text'] ];
5932 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5934 # Now renumber links
5935 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5937 # Should be good to go.
5938 return $texts[0];
5942 * Returns true if the given array, presumed to be generated by
5943 * serializeHalfParsedText(), is compatible with the current version of the
5944 * parser.
5946 * @param array $data
5948 * @return bool
5950 public function isValidHalfParsedText( $data ) {
5951 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5955 * Parsed a width param of imagelink like 300px or 200x300px
5957 * @param string $value
5959 * @return array
5960 * @since 1.20
5962 public function parseWidthParam( $value ) {
5963 $parsedWidthParam = [];
5964 if ( $value === '' ) {
5965 return $parsedWidthParam;
5967 $m = [];
5968 # (bug 13500) In both cases (width/height and width only),
5969 # permit trailing "px" for backward compatibility.
5970 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5971 $width = intval( $m[1] );
5972 $height = intval( $m[2] );
5973 $parsedWidthParam['width'] = $width;
5974 $parsedWidthParam['height'] = $height;
5975 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5976 $width = intval( $value );
5977 $parsedWidthParam['width'] = $width;
5979 return $parsedWidthParam;
5983 * Lock the current instance of the parser.
5985 * This is meant to stop someone from calling the parser
5986 * recursively and messing up all the strip state.
5988 * @throws MWException If parser is in a parse
5989 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5991 protected function lock() {
5992 if ( $this->mInParse ) {
5993 throw new MWException( "Parser state cleared while parsing. "
5994 . "Did you call Parser::parse recursively?" );
5996 $this->mInParse = true;
5998 $recursiveCheck = new ScopedCallback( function() {
5999 $this->mInParse = false;
6000 } );
6002 return $recursiveCheck;
6006 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6008 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6009 * or if there is more than one <p/> tag in the input HTML.
6011 * @param string $html
6012 * @return string
6013 * @since 1.24
6015 public static function stripOuterParagraph( $html ) {
6016 $m = [];
6017 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6018 if ( strpos( $m[1], '</p>' ) === false ) {
6019 $html = $m[1];
6023 return $html;
6027 * Return this parser if it is not doing anything, otherwise
6028 * get a fresh parser. You can use this method by doing
6029 * $myParser = $wgParser->getFreshParser(), or more simply
6030 * $wgParser->getFreshParser()->parse( ... );
6031 * if you're unsure if $wgParser is safe to use.
6033 * @since 1.24
6034 * @return Parser A parser object that is not parsing anything
6036 public function getFreshParser() {
6037 global $wgParserConf;
6038 if ( $this->mInParse ) {
6039 return new $wgParserConf['class']( $wgParserConf );
6040 } else {
6041 return $this;
6046 * Set's up the PHP implementation of OOUI for use in this request
6047 * and instructs OutputPage to enable OOUI for itself.
6049 * @since 1.26
6051 public function enableOOUI() {
6052 OutputPage::setupOOUI();
6053 $this->mOutput->setEnableOOUI( true );