dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Make use of the list() feature where it makes sense
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
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.
9 *
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.
14 *
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
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use MediaWiki\Special\SpecialPageFactory;
26 use Wikimedia\ScopedCallback;
27
28 /**
29 * @defgroup Parser Parser
30 */
31
32 /**
33 * PHP Parser - Processes wiki markup (which uses a more user-friendly
34 * syntax, such as "[[link]]" for making links), and provides a one-way
35 * transformation of that wiki markup it into (X)HTML output / markup
36 * (which in turn the browser understands, and can display).
37 *
38 * There are seven main entry points into the Parser class:
39 *
40 * - Parser::parse()
41 * produces HTML output
42 * - Parser::preSaveTransform()
43 * produces altered wiki markup
44 * - Parser::preprocess()
45 * removes HTML comments and expands templates
46 * - Parser::cleanSig() and Parser::cleanSigInSig()
47 * cleans a signature before saving it to preferences
48 * - Parser::getSection()
49 * return the content of a section from an article for section editing
50 * - Parser::replaceSection()
51 * replaces a section by number inside an article
52 * - Parser::getPreloadText()
53 * removes <noinclude> sections and <includeonly> tags
54 *
55 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
56 *
57 * @par Settings:
58 * $wgNamespacesWithSubpages
59 *
60 * @par Settings only within ParserOptions:
61 * $wgAllowExternalImages
62 * $wgAllowSpecialInclusion
63 * $wgInterwikiMagic
64 * $wgMaxArticleSize
65 *
66 * @ingroup Parser
67 */
68 class Parser {
69 /**
70 * Update this version number when the ParserOutput format
71 * changes in an incompatible way, so the parser cache
72 * can automatically discard old data.
73 */
74 const VERSION = '1.6.4';
75
76 /**
77 * Update this version number when the output of serialiseHalfParsedText()
78 * changes in an incompatible way
79 */
80 const HALF_PARSED_VERSION = 2;
81
82 # Flags for Parser::setFunctionHook
83 const SFH_NO_HASH = 1;
84 const SFH_OBJECT_ARGS = 2;
85
86 # Constants needed for external link processing
87 # Everything except bracket, space, or control characters
88 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
89 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
90 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
91 # uses to replace invalid HTML characters.
92 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
93 # Simplified expression to match an IPv4 or IPv6 address, or
94 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
95 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
96 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
97 // phpcs:ignore Generic.Files.LineLength
98 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
99 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
100
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 /**
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
118 *
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
122 * string constructs.
123 *
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
126 *
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
130 * `'".
131 */
132 const MARKER_SUFFIX = "-QINU`\"'\x7f";
133 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
134
135 # Markers used for wrapping the table of contents
136 const TOC_START = '<mw:toc>';
137 const TOC_END = '</mw:toc>';
138
139 /** @var int Assume that no output will later be saved this many seconds after parsing */
140 const MAX_TTS = 900;
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 /**
155 * @var bool Whether firstCallInit still needs to be called
156 */
157 public $mFirstCall = true;
158
159 # Initialised by initialiseVariables()
160
161 /**
162 * @var MagicWordArray
163 */
164 public $mVariables;
165
166 /**
167 * @var MagicWordArray
168 */
169 public $mSubstWords;
170 # Initialised in constructor
171 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
172
173 # Initialized in getPreprocessor()
174 /** @var Preprocessor */
175 public $mPreprocessor;
176
177 # Cleared with clearState():
178 /**
179 * @var ParserOutput
180 */
181 public $mOutput;
182 public $mAutonumber;
183
184 /**
185 * @var StripState
186 */
187 public $mStripState;
188
189 public $mIncludeCount;
190 /**
191 * @var LinkHolderArray
192 */
193 public $mLinkHolders;
194
195 public $mLinkID;
196 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
197 public $mDefaultSort;
198 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
199 public $mExpensiveFunctionCount; # number of expensive parser function calls
200 public $mShowToc, $mForceTocPosition;
201
202 /**
203 * @var User
204 */
205 public $mUser; # User object; only used when doing pre-save transform
206
207 # Temporary
208 # These are variables reset at least once per parse regardless of $clearState
209
210 /**
211 * @var ParserOptions
212 */
213 public $mOptions;
214
215 /**
216 * @var Title
217 */
218 public $mTitle; # Title context, used for self-link rendering and similar things
219 public $mOutputType; # Output type, one of the OT_xxx constants
220 public $ot; # Shortcut alias, see setOutputType()
221 public $mRevisionObject; # The revision object of the specified revision ID
222 public $mRevisionId; # ID to display in {{REVISIONID}} tags
223 public $mRevisionTimestamp; # The timestamp of the specified revision ID
224 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
225 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
226 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
227 public $mInputSize = false; # For {{PAGESIZE}} on current page.
228
229 /**
230 * @var string Deprecated accessor for the strip marker prefix.
231 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
232 */
233 public $mUniqPrefix = self::MARKER_PREFIX;
234
235 /**
236 * @var array Array with the language name of each language link (i.e. the
237 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
238 * duplicate language links to the ParserOutput.
239 */
240 public $mLangLinkLanguages;
241
242 /**
243 * @var MapCacheLRU|null
244 * @since 1.24
245 *
246 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
247 */
248 public $currentRevisionCache;
249
250 /**
251 * @var bool|string Recursive call protection.
252 * This variable should be treated as if it were private.
253 */
254 public $mInParse = false;
255
256 /** @var SectionProfiler */
257 protected $mProfiler;
258
259 /**
260 * @var LinkRenderer
261 */
262 protected $mLinkRenderer;
263
264 /** @var MagicWordFactory */
265 private $magicWordFactory;
266
267 /** @var Language */
268 private $contLang;
269
270 /** @var ParserFactory */
271 private $factory;
272
273 /** @var SpecialPageFactory */
274 private $specialPageFactory;
275
276 /** @var Config */
277 private $siteConfig;
278
279 /**
280 * @param array $parserConf See $wgParserConf documentation
281 * @param MagicWordFactory|null $magicWordFactory
282 * @param Language|null $contLang Content language
283 * @param ParserFactory|null $factory
284 * @param string|null $urlProtocols As returned from wfUrlProtocols()
285 * @param SpecialPageFactory|null $spFactory
286 * @param Config|null $siteConfig
287 */
288 public function __construct(
289 array $parserConf = [], MagicWordFactory $magicWordFactory = null,
290 Language $contLang = null, ParserFactory $factory = null, $urlProtocols = null,
291 SpecialPageFactory $spFactory = null, Config $siteConfig = null
292 ) {
293 $this->mConf = $parserConf;
294 $this->mUrlProtocols = $urlProtocols ?? wfUrlProtocols();
295 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
296 self::EXT_LINK_ADDR .
297 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
298 if ( isset( $parserConf['preprocessorClass'] ) ) {
299 $this->mPreprocessorClass = $parserConf['preprocessorClass'];
300 } elseif ( wfIsHHVM() ) {
301 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
302 $this->mPreprocessorClass = Preprocessor_Hash::class;
303 } elseif ( extension_loaded( 'domxml' ) ) {
304 # PECL extension that conflicts with the core DOM extension (T15770)
305 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
306 $this->mPreprocessorClass = Preprocessor_Hash::class;
307 } elseif ( extension_loaded( 'dom' ) ) {
308 $this->mPreprocessorClass = Preprocessor_DOM::class;
309 } else {
310 $this->mPreprocessorClass = Preprocessor_Hash::class;
311 }
312 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
313
314 $services = MediaWikiServices::getInstance();
315 $this->magicWordFactory = $magicWordFactory ??
316 $services->getMagicWordFactory();
317
318 $this->contLang = $contLang ?? $services->getContentLanguage();
319
320 $this->factory = $factory ?? $services->getParserFactory();
321 $this->specialPageFactory = $spFactory ?? $services->getSpecialPageFactory();
322 $this->siteConfig = $siteConfig ?? MediaWikiServices::getInstance()->getMainConfig();
323 }
324
325 /**
326 * Reduce memory usage to reduce the impact of circular references
327 */
328 public function __destruct() {
329 if ( isset( $this->mLinkHolders ) ) {
330 unset( $this->mLinkHolders );
331 }
332 foreach ( $this as $name => $value ) {
333 unset( $this->$name );
334 }
335 }
336
337 /**
338 * Allow extensions to clean up when the parser is cloned
339 */
340 public function __clone() {
341 $this->mInParse = false;
342
343 // T58226: When you create a reference "to" an object field, that
344 // makes the object field itself be a reference too (until the other
345 // reference goes out of scope). When cloning, any field that's a
346 // reference is copied as a reference in the new object. Both of these
347 // are defined PHP5 behaviors, as inconvenient as it is for us when old
348 // hooks from PHP4 days are passing fields by reference.
349 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
350 // Make a non-reference copy of the field, then rebind the field to
351 // reference the new copy.
352 $tmp = $this->$k;
353 $this->$k =& $tmp;
354 unset( $tmp );
355 }
356
357 Hooks::run( 'ParserCloned', [ $this ] );
358 }
359
360 /**
361 * Do various kinds of initialisation on the first call of the parser
362 */
363 public function firstCallInit() {
364 if ( !$this->mFirstCall ) {
365 return;
366 }
367 $this->mFirstCall = false;
368
369 CoreParserFunctions::register( $this );
370 CoreTagHooks::register( $this );
371 $this->initialiseVariables();
372
373 // Avoid PHP 7.1 warning from passing $this by reference
374 $parser = $this;
375 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
376 }
377
378 /**
379 * Clear Parser state
380 *
381 * @private
382 */
383 public function clearState() {
384 $this->firstCallInit();
385 $this->mOutput = new ParserOutput;
386 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
387 $this->mAutonumber = 0;
388 $this->mIncludeCount = [];
389 $this->mLinkHolders = new LinkHolderArray( $this );
390 $this->mLinkID = 0;
391 $this->mRevisionObject = $this->mRevisionTimestamp =
392 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
393 $this->mVarCache = [];
394 $this->mUser = null;
395 $this->mLangLinkLanguages = [];
396 $this->currentRevisionCache = null;
397
398 $this->mStripState = new StripState( $this );
399
400 # Clear these on every parse, T6549
401 $this->mTplRedirCache = $this->mTplDomCache = [];
402
403 $this->mShowToc = true;
404 $this->mForceTocPosition = false;
405 $this->mIncludeSizes = [
406 'post-expand' => 0,
407 'arg' => 0,
408 ];
409 $this->mPPNodeCount = 0;
410 $this->mGeneratedPPNodeCount = 0;
411 $this->mHighestExpansionDepth = 0;
412 $this->mDefaultSort = false;
413 $this->mHeadings = [];
414 $this->mDoubleUnderscores = [];
415 $this->mExpensiveFunctionCount = 0;
416
417 # Fix cloning
418 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
419 $this->mPreprocessor = null;
420 }
421
422 $this->mProfiler = new SectionProfiler();
423
424 // Avoid PHP 7.1 warning from passing $this by reference
425 $parser = $this;
426 Hooks::run( 'ParserClearState', [ &$parser ] );
427 }
428
429 /**
430 * Convert wikitext to HTML
431 * Do not call this function recursively.
432 *
433 * @param string $text Text we want to parse
434 * @param-taint $text escapes_htmlnoent
435 * @param Title $title
436 * @param ParserOptions $options
437 * @param bool $linestart
438 * @param bool $clearState
439 * @param int|null $revid Number to pass in {{REVISIONID}}
440 * @return ParserOutput A ParserOutput
441 * @return-taint escaped
442 */
443 public function parse(
444 $text, Title $title, ParserOptions $options,
445 $linestart = true, $clearState = true, $revid = null
446 ) {
447 if ( $clearState ) {
448 // We use U+007F DELETE to construct strip markers, so we have to make
449 // sure that this character does not occur in the input text.
450 $text = strtr( $text, "\x7f", "?" );
451 $magicScopeVariable = $this->lock();
452 }
453 // Strip U+0000 NULL (T159174)
454 $text = str_replace( "\000", '', $text );
455
456 $this->startParse( $title, $options, self::OT_HTML, $clearState );
457
458 $this->currentRevisionCache = null;
459 $this->mInputSize = strlen( $text );
460 if ( $this->mOptions->getEnableLimitReport() ) {
461 $this->mOutput->resetParseStartTime();
462 }
463
464 $oldRevisionId = $this->mRevisionId;
465 $oldRevisionObject = $this->mRevisionObject;
466 $oldRevisionTimestamp = $this->mRevisionTimestamp;
467 $oldRevisionUser = $this->mRevisionUser;
468 $oldRevisionSize = $this->mRevisionSize;
469 if ( $revid !== null ) {
470 $this->mRevisionId = $revid;
471 $this->mRevisionObject = null;
472 $this->mRevisionTimestamp = null;
473 $this->mRevisionUser = null;
474 $this->mRevisionSize = null;
475 }
476
477 // Avoid PHP 7.1 warning from passing $this by reference
478 $parser = $this;
479 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
480 # No more strip!
481 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
482 $text = $this->internalParse( $text );
483 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
484
485 $text = $this->internalParseHalfParsed( $text, true, $linestart );
486
487 /**
488 * A converted title will be provided in the output object if title and
489 * content conversion are enabled, the article text does not contain
490 * a conversion-suppressing double-underscore tag, and no
491 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
492 * automatic link conversion.
493 */
494 if ( !( $options->getDisableTitleConversion()
495 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
496 || isset( $this->mDoubleUnderscores['notitleconvert'] )
497 || $this->mOutput->getDisplayTitle() !== false )
498 ) {
499 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
500 if ( $convruletitle ) {
501 $this->mOutput->setTitleText( $convruletitle );
502 } else {
503 $titleText = $this->getTargetLanguage()->convertTitle( $title );
504 $this->mOutput->setTitleText( $titleText );
505 }
506 }
507
508 # Compute runtime adaptive expiry if set
509 $this->mOutput->finalizeAdaptiveCacheExpiry();
510
511 # Warn if too many heavyweight parser functions were used
512 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
513 $this->limitationWarn( 'expensive-parserfunction',
514 $this->mExpensiveFunctionCount,
515 $this->mOptions->getExpensiveParserFunctionLimit()
516 );
517 }
518
519 # Information on limits, for the benefit of users who try to skirt them
520 if ( $this->mOptions->getEnableLimitReport() ) {
521 $text .= $this->makeLimitReport();
522 }
523
524 # Wrap non-interface parser output in a <div> so it can be targeted
525 # with CSS (T37247)
526 $class = $this->mOptions->getWrapOutputClass();
527 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
528 $this->mOutput->addWrapperDivClass( $class );
529 }
530
531 $this->mOutput->setText( $text );
532
533 $this->mRevisionId = $oldRevisionId;
534 $this->mRevisionObject = $oldRevisionObject;
535 $this->mRevisionTimestamp = $oldRevisionTimestamp;
536 $this->mRevisionUser = $oldRevisionUser;
537 $this->mRevisionSize = $oldRevisionSize;
538 $this->mInputSize = false;
539 $this->currentRevisionCache = null;
540
541 return $this->mOutput;
542 }
543
544 /**
545 * Set the limit report data in the current ParserOutput, and return the
546 * limit report HTML comment.
547 *
548 * @return string
549 */
550 protected function makeLimitReport() {
551 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
552
553 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
554 if ( $cpuTime !== null ) {
555 $this->mOutput->setLimitReportData( 'limitreport-cputime',
556 sprintf( "%.3f", $cpuTime )
557 );
558 }
559
560 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
561 $this->mOutput->setLimitReportData( 'limitreport-walltime',
562 sprintf( "%.3f", $wallTime )
563 );
564
565 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
566 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
567 );
568 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
569 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
570 );
571 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
572 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
573 );
574 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
575 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
576 );
577 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
578 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
579 );
580 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
581 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
582 );
583
584 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
585 $this->mOutput->setLimitReportData( $key, $value );
586 }
587
588 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
589
590 $limitReport = "NewPP limit report\n";
591 if ( $this->siteConfig->get( 'ShowHostnames' ) ) {
592 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
593 }
594 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
595 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
596 $limitReport .= 'Dynamic content: ' .
597 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
598 "\n";
599
600 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
601 if ( Hooks::run( 'ParserLimitReportFormat',
602 [ $key, &$value, &$limitReport, false, false ]
603 ) ) {
604 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
605 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
606 ->inLanguage( 'en' )->useDatabase( false );
607 if ( !$valueMsg->exists() ) {
608 $valueMsg = new RawMessage( '$1' );
609 }
610 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
611 $valueMsg->params( $value );
612 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
613 }
614 }
615 }
616 // Since we're not really outputting HTML, decode the entities and
617 // then re-encode the things that need hiding inside HTML comments.
618 $limitReport = htmlspecialchars_decode( $limitReport );
619
620 // Sanitize for comment. Note '‐' in the replacement is U+2010,
621 // which looks much like the problematic '-'.
622 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
623 $text = "\n<!-- \n$limitReport-->\n";
624
625 // Add on template profiling data in human/machine readable way
626 $dataByFunc = $this->mProfiler->getFunctionStats();
627 uasort( $dataByFunc, function ( $a, $b ) {
628 return $b['real'] <=> $a['real']; // descending order
629 } );
630 $profileReport = [];
631 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
632 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
633 $item['%real'], $item['real'], $item['calls'],
634 htmlspecialchars( $item['name'] ) );
635 }
636 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
637 $text .= implode( "\n", $profileReport ) . "\n-->\n";
638
639 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
640
641 // Add other cache related metadata
642 if ( $this->siteConfig->get( 'ShowHostnames' ) ) {
643 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
644 }
645 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
646 $this->mOutput->getCacheTime() );
647 $this->mOutput->setLimitReportData( 'cachereport-ttl',
648 $this->mOutput->getCacheExpiry() );
649 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
650 $this->mOutput->hasDynamicContent() );
651
652 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
653 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
654 $this->mTitle->getPrefixedDBkey() );
655 }
656 return $text;
657 }
658
659 /**
660 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
661 * can be called from an extension tag hook.
662 *
663 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
664 * instead, which means that lists and links have not been fully parsed yet,
665 * and strip markers are still present.
666 *
667 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
668 *
669 * Use this function if you're a parser tag hook and you want to parse
670 * wikitext before or after applying additional transformations, and you
671 * intend to *return the result as hook output*, which will cause it to go
672 * through the rest of parsing process automatically.
673 *
674 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
675 * $text are not expanded
676 *
677 * @param string $text Text extension wants to have parsed
678 * @param-taint $text escapes_htmlnoent
679 * @param bool|PPFrame $frame The frame to use for expanding any template variables
680 * @return string UNSAFE half-parsed HTML
681 * @return-taint escaped
682 */
683 public function recursiveTagParse( $text, $frame = false ) {
684 // Avoid PHP 7.1 warning from passing $this by reference
685 $parser = $this;
686 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
687 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
688 $text = $this->internalParse( $text, false, $frame );
689 return $text;
690 }
691
692 /**
693 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
694 * point can be called from an extension tag hook.
695 *
696 * The output of this function is fully-parsed HTML that is safe for output.
697 * If you're a parser tag hook, you might want to use recursiveTagParse()
698 * instead.
699 *
700 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
701 * $text are not expanded
702 *
703 * @since 1.25
704 *
705 * @param string $text Text extension wants to have parsed
706 * @param-taint $text escapes_htmlnoent
707 * @param bool|PPFrame $frame The frame to use for expanding any template variables
708 * @return string Fully parsed HTML
709 * @return-taint escaped
710 */
711 public function recursiveTagParseFully( $text, $frame = false ) {
712 $text = $this->recursiveTagParse( $text, $frame );
713 $text = $this->internalParseHalfParsed( $text, false );
714 return $text;
715 }
716
717 /**
718 * Expand templates and variables in the text, producing valid, static wikitext.
719 * Also removes comments.
720 * Do not call this function recursively.
721 * @param string $text
722 * @param Title|null $title
723 * @param ParserOptions $options
724 * @param int|null $revid
725 * @param bool|PPFrame $frame
726 * @return mixed|string
727 */
728 public function preprocess( $text, Title $title = null,
729 ParserOptions $options, $revid = null, $frame = false
730 ) {
731 $magicScopeVariable = $this->lock();
732 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
733 if ( $revid !== null ) {
734 $this->mRevisionId = $revid;
735 }
736 // Avoid PHP 7.1 warning from passing $this by reference
737 $parser = $this;
738 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
739 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
740 $text = $this->replaceVariables( $text, $frame );
741 $text = $this->mStripState->unstripBoth( $text );
742 return $text;
743 }
744
745 /**
746 * Recursive parser entry point that can be called from an extension tag
747 * hook.
748 *
749 * @param string $text Text to be expanded
750 * @param bool|PPFrame $frame The frame to use for expanding any template variables
751 * @return string
752 * @since 1.19
753 */
754 public function recursivePreprocess( $text, $frame = false ) {
755 $text = $this->replaceVariables( $text, $frame );
756 $text = $this->mStripState->unstripBoth( $text );
757 return $text;
758 }
759
760 /**
761 * Process the wikitext for the "?preload=" feature. (T7210)
762 *
763 * "<noinclude>", "<includeonly>" etc. are parsed as for template
764 * transclusion, comments, templates, arguments, tags hooks and parser
765 * functions are untouched.
766 *
767 * @param string $text
768 * @param Title $title
769 * @param ParserOptions $options
770 * @param array $params
771 * @return string
772 */
773 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
774 $msg = new RawMessage( $text );
775 $text = $msg->params( $params )->plain();
776
777 # Parser (re)initialisation
778 $magicScopeVariable = $this->lock();
779 $this->startParse( $title, $options, self::OT_PLAIN, true );
780
781 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
782 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
783 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
784 $text = $this->mStripState->unstripBoth( $text );
785 return $text;
786 }
787
788 /**
789 * Set the current user.
790 * Should only be used when doing pre-save transform.
791 *
792 * @param User|null $user User object or null (to reset)
793 */
794 public function setUser( $user ) {
795 $this->mUser = $user;
796 }
797
798 /**
799 * Set the context title
800 *
801 * @param Title $t
802 */
803 public function setTitle( $t ) {
804 if ( !$t ) {
805 $t = Title::newFromText( 'NO TITLE' );
806 }
807
808 if ( $t->hasFragment() ) {
809 # Strip the fragment to avoid various odd effects
810 $this->mTitle = $t->createFragmentTarget( '' );
811 } else {
812 $this->mTitle = $t;
813 }
814 }
815
816 /**
817 * Accessor for the Title object
818 *
819 * @return Title
820 */
821 public function getTitle() {
822 return $this->mTitle;
823 }
824
825 /**
826 * Accessor/mutator for the Title object
827 *
828 * @param Title|null $x Title object or null to just get the current one
829 * @return Title
830 */
831 public function Title( $x = null ) {
832 return wfSetVar( $this->mTitle, $x );
833 }
834
835 /**
836 * Set the output type
837 *
838 * @param int $ot New value
839 */
840 public function setOutputType( $ot ) {
841 $this->mOutputType = $ot;
842 # Shortcut alias
843 $this->ot = [
844 'html' => $ot == self::OT_HTML,
845 'wiki' => $ot == self::OT_WIKI,
846 'pre' => $ot == self::OT_PREPROCESS,
847 'plain' => $ot == self::OT_PLAIN,
848 ];
849 }
850
851 /**
852 * Accessor/mutator for the output type
853 *
854 * @param int|null $x New value or null to just get the current one
855 * @return int
856 */
857 public function OutputType( $x = null ) {
858 return wfSetVar( $this->mOutputType, $x );
859 }
860
861 /**
862 * Get the ParserOutput object
863 *
864 * @return ParserOutput
865 */
866 public function getOutput() {
867 return $this->mOutput;
868 }
869
870 /**
871 * Get the ParserOptions object
872 *
873 * @return ParserOptions
874 */
875 public function getOptions() {
876 return $this->mOptions;
877 }
878
879 /**
880 * Accessor/mutator for the ParserOptions object
881 *
882 * @param ParserOptions|null $x New value or null to just get the current one
883 * @return ParserOptions Current ParserOptions object
884 */
885 public function Options( $x = null ) {
886 return wfSetVar( $this->mOptions, $x );
887 }
888
889 /**
890 * @return int
891 */
892 public function nextLinkID() {
893 return $this->mLinkID++;
894 }
895
896 /**
897 * @param int $id
898 */
899 public function setLinkID( $id ) {
900 $this->mLinkID = $id;
901 }
902
903 /**
904 * Get a language object for use in parser functions such as {{FORMATNUM:}}
905 * @return Language
906 */
907 public function getFunctionLang() {
908 return $this->getTargetLanguage();
909 }
910
911 /**
912 * Get the target language for the content being parsed. This is usually the
913 * language that the content is in.
914 *
915 * @since 1.19
916 *
917 * @throws MWException
918 * @return Language
919 */
920 public function getTargetLanguage() {
921 $target = $this->mOptions->getTargetLanguage();
922
923 if ( $target !== null ) {
924 return $target;
925 } elseif ( $this->mOptions->getInterfaceMessage() ) {
926 return $this->mOptions->getUserLangObj();
927 } elseif ( is_null( $this->mTitle ) ) {
928 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
929 }
930
931 return $this->mTitle->getPageLanguage();
932 }
933
934 /**
935 * Get the language object for language conversion
936 * @deprecated since 1.32, just use getTargetLanguage()
937 * @return Language|null
938 */
939 public function getConverterLanguage() {
940 return $this->getTargetLanguage();
941 }
942
943 /**
944 * Get a User object either from $this->mUser, if set, or from the
945 * ParserOptions object otherwise
946 *
947 * @return User
948 */
949 public function getUser() {
950 if ( !is_null( $this->mUser ) ) {
951 return $this->mUser;
952 }
953 return $this->mOptions->getUser();
954 }
955
956 /**
957 * Get a preprocessor object
958 *
959 * @return Preprocessor
960 */
961 public function getPreprocessor() {
962 if ( !isset( $this->mPreprocessor ) ) {
963 $class = $this->mPreprocessorClass;
964 $this->mPreprocessor = new $class( $this );
965 }
966 return $this->mPreprocessor;
967 }
968
969 /**
970 * Get a LinkRenderer instance to make links with
971 *
972 * @since 1.28
973 * @return LinkRenderer
974 */
975 public function getLinkRenderer() {
976 if ( !$this->mLinkRenderer ) {
977 $this->mLinkRenderer = MediaWikiServices::getInstance()
978 ->getLinkRendererFactory()->create();
979 $this->mLinkRenderer->setStubThreshold(
980 $this->getOptions()->getStubThreshold()
981 );
982 }
983
984 return $this->mLinkRenderer;
985 }
986
987 /**
988 * Get the MagicWordFactory that this Parser is using
989 *
990 * @since 1.32
991 * @return MagicWordFactory
992 */
993 public function getMagicWordFactory() {
994 return $this->magicWordFactory;
995 }
996
997 /**
998 * Get the content language that this Parser is using
999 *
1000 * @since 1.32
1001 * @return Language
1002 */
1003 public function getContentLanguage() {
1004 return $this->contLang;
1005 }
1006
1007 /**
1008 * Replaces all occurrences of HTML-style comments and the given tags
1009 * in the text with a random marker and returns the next text. The output
1010 * parameter $matches will be an associative array filled with data in
1011 * the form:
1012 *
1013 * @code
1014 * 'UNIQ-xxxxx' => [
1015 * 'element',
1016 * 'tag content',
1017 * [ 'param' => 'x' ],
1018 * '<element param="x">tag content</element>' ]
1019 * @endcode
1020 *
1021 * @param array $elements List of element names. Comments are always extracted.
1022 * @param string $text Source text string.
1023 * @param array &$matches Out parameter, Array: extracted tags
1024 * @return string Stripped text
1025 */
1026 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1027 static $n = 1;
1028 $stripped = '';
1029 $matches = [];
1030
1031 $taglist = implode( '|', $elements );
1032 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1033
1034 while ( $text != '' ) {
1035 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
1036 $stripped .= $p[0];
1037 if ( count( $p ) < 5 ) {
1038 break;
1039 }
1040 if ( count( $p ) > 5 ) {
1041 # comment
1042 $element = $p[4];
1043 $attributes = '';
1044 $close = '';
1045 $inside = $p[5];
1046 } else {
1047 # tag
1048 list( , $element, $attributes, $close, $inside ) = $p;
1049 }
1050
1051 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1052 $stripped .= $marker;
1053
1054 if ( $close === '/>' ) {
1055 # Empty element tag, <tag />
1056 $content = null;
1057 $text = $inside;
1058 $tail = null;
1059 } else {
1060 if ( $element === '!--' ) {
1061 $end = '/(-->)/';
1062 } else {
1063 $end = "/(<\\/$element\\s*>)/i";
1064 }
1065 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1066 $content = $q[0];
1067 if ( count( $q ) < 3 ) {
1068 # No end tag -- let it run out to the end of the text.
1069 $tail = '';
1070 $text = '';
1071 } else {
1072 list( , $tail, $text ) = $q;
1073 }
1074 }
1075
1076 $matches[$marker] = [ $element,
1077 $content,
1078 Sanitizer::decodeTagAttributes( $attributes ),
1079 "<$element$attributes$close$content$tail" ];
1080 }
1081 return $stripped;
1082 }
1083
1084 /**
1085 * Get a list of strippable XML-like elements
1086 *
1087 * @return array
1088 */
1089 public function getStripList() {
1090 return $this->mStripList;
1091 }
1092
1093 /**
1094 * Add an item to the strip state
1095 * Returns the unique tag which must be inserted into the stripped text
1096 * The tag will be replaced with the original text in unstrip()
1097 *
1098 * @param string $text
1099 *
1100 * @return string
1101 */
1102 public function insertStripItem( $text ) {
1103 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1104 $this->mMarkerIndex++;
1105 $this->mStripState->addGeneral( $marker, $text );
1106 return $marker;
1107 }
1108
1109 /**
1110 * parse the wiki syntax used to render tables
1111 *
1112 * @private
1113 * @param string $text
1114 * @return string
1115 */
1116 public function doTableStuff( $text ) {
1117 $lines = StringUtils::explode( "\n", $text );
1118 $out = '';
1119 $td_history = []; # Is currently a td tag open?
1120 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1121 $tr_history = []; # Is currently a tr tag open?
1122 $tr_attributes = []; # history of tr attributes
1123 $has_opened_tr = []; # Did this table open a <tr> element?
1124 $indent_level = 0; # indent level of the table
1125
1126 foreach ( $lines as $outLine ) {
1127 $line = trim( $outLine );
1128
1129 if ( $line === '' ) { # empty line, go to next line
1130 $out .= $outLine . "\n";
1131 continue;
1132 }
1133
1134 $first_character = $line[0];
1135 $first_two = substr( $line, 0, 2 );
1136 $matches = [];
1137
1138 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1139 # First check if we are starting a new table
1140 $indent_level = strlen( $matches[1] );
1141
1142 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1143 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1144
1145 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1146 array_push( $td_history, false );
1147 array_push( $last_tag_history, '' );
1148 array_push( $tr_history, false );
1149 array_push( $tr_attributes, '' );
1150 array_push( $has_opened_tr, false );
1151 } elseif ( count( $td_history ) == 0 ) {
1152 # Don't do any of the following
1153 $out .= $outLine . "\n";
1154 continue;
1155 } elseif ( $first_two === '|}' ) {
1156 # We are ending a table
1157 $line = '</table>' . substr( $line, 2 );
1158 $last_tag = array_pop( $last_tag_history );
1159
1160 if ( !array_pop( $has_opened_tr ) ) {
1161 $line = "<tr><td></td></tr>{$line}";
1162 }
1163
1164 if ( array_pop( $tr_history ) ) {
1165 $line = "</tr>{$line}";
1166 }
1167
1168 if ( array_pop( $td_history ) ) {
1169 $line = "</{$last_tag}>{$line}";
1170 }
1171 array_pop( $tr_attributes );
1172 if ( $indent_level > 0 ) {
1173 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1174 } else {
1175 $outLine = $line;
1176 }
1177 } elseif ( $first_two === '|-' ) {
1178 # Now we have a table row
1179 $line = preg_replace( '#^\|-+#', '', $line );
1180
1181 # Whats after the tag is now only attributes
1182 $attributes = $this->mStripState->unstripBoth( $line );
1183 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1184 array_pop( $tr_attributes );
1185 array_push( $tr_attributes, $attributes );
1186
1187 $line = '';
1188 $last_tag = array_pop( $last_tag_history );
1189 array_pop( $has_opened_tr );
1190 array_push( $has_opened_tr, true );
1191
1192 if ( array_pop( $tr_history ) ) {
1193 $line = '</tr>';
1194 }
1195
1196 if ( array_pop( $td_history ) ) {
1197 $line = "</{$last_tag}>{$line}";
1198 }
1199
1200 $outLine = $line;
1201 array_push( $tr_history, false );
1202 array_push( $td_history, false );
1203 array_push( $last_tag_history, '' );
1204 } elseif ( $first_character === '|'
1205 || $first_character === '!'
1206 || $first_two === '|+'
1207 ) {
1208 # This might be cell elements, td, th or captions
1209 if ( $first_two === '|+' ) {
1210 $first_character = '+';
1211 $line = substr( $line, 2 );
1212 } else {
1213 $line = substr( $line, 1 );
1214 }
1215
1216 // Implies both are valid for table headings.
1217 if ( $first_character === '!' ) {
1218 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1219 }
1220
1221 # Split up multiple cells on the same line.
1222 # FIXME : This can result in improper nesting of tags processed
1223 # by earlier parser steps.
1224 $cells = explode( '||', $line );
1225
1226 $outLine = '';
1227
1228 # Loop through each table cell
1229 foreach ( $cells as $cell ) {
1230 $previous = '';
1231 if ( $first_character !== '+' ) {
1232 $tr_after = array_pop( $tr_attributes );
1233 if ( !array_pop( $tr_history ) ) {
1234 $previous = "<tr{$tr_after}>\n";
1235 }
1236 array_push( $tr_history, true );
1237 array_push( $tr_attributes, '' );
1238 array_pop( $has_opened_tr );
1239 array_push( $has_opened_tr, true );
1240 }
1241
1242 $last_tag = array_pop( $last_tag_history );
1243
1244 if ( array_pop( $td_history ) ) {
1245 $previous = "</{$last_tag}>\n{$previous}";
1246 }
1247
1248 if ( $first_character === '|' ) {
1249 $last_tag = 'td';
1250 } elseif ( $first_character === '!' ) {
1251 $last_tag = 'th';
1252 } elseif ( $first_character === '+' ) {
1253 $last_tag = 'caption';
1254 } else {
1255 $last_tag = '';
1256 }
1257
1258 array_push( $last_tag_history, $last_tag );
1259
1260 # A cell could contain both parameters and data
1261 $cell_data = explode( '|', $cell, 2 );
1262
1263 # T2553: Note that a '|' inside an invalid link should not
1264 # be mistaken as delimiting cell parameters
1265 # Bug T153140: Neither should language converter markup.
1266 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1267 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1268 } elseif ( count( $cell_data ) == 1 ) {
1269 // Whitespace in cells is trimmed
1270 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1271 } else {
1272 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1273 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1274 // Whitespace in cells is trimmed
1275 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1276 }
1277
1278 $outLine .= $cell;
1279 array_push( $td_history, true );
1280 }
1281 }
1282 $out .= $outLine . "\n";
1283 }
1284
1285 # Closing open td, tr && table
1286 while ( count( $td_history ) > 0 ) {
1287 if ( array_pop( $td_history ) ) {
1288 $out .= "</td>\n";
1289 }
1290 if ( array_pop( $tr_history ) ) {
1291 $out .= "</tr>\n";
1292 }
1293 if ( !array_pop( $has_opened_tr ) ) {
1294 $out .= "<tr><td></td></tr>\n";
1295 }
1296
1297 $out .= "</table>\n";
1298 }
1299
1300 # Remove trailing line-ending (b/c)
1301 if ( substr( $out, -1 ) === "\n" ) {
1302 $out = substr( $out, 0, -1 );
1303 }
1304
1305 # special case: don't return empty table
1306 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1307 $out = '';
1308 }
1309
1310 return $out;
1311 }
1312
1313 /**
1314 * Helper function for parse() that transforms wiki markup into half-parsed
1315 * HTML. Only called for $mOutputType == self::OT_HTML.
1316 *
1317 * @private
1318 *
1319 * @param string $text The text to parse
1320 * @param-taint $text escapes_html
1321 * @param bool $isMain Whether this is being called from the main parse() function
1322 * @param PPFrame|bool $frame A pre-processor frame
1323 *
1324 * @return string
1325 */
1326 public function internalParse( $text, $isMain = true, $frame = false ) {
1327 $origText = $text;
1328
1329 // Avoid PHP 7.1 warning from passing $this by reference
1330 $parser = $this;
1331
1332 # Hook to suspend the parser in this state
1333 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1334 return $text;
1335 }
1336
1337 # if $frame is provided, then use $frame for replacing any variables
1338 if ( $frame ) {
1339 # use frame depth to infer how include/noinclude tags should be handled
1340 # depth=0 means this is the top-level document; otherwise it's an included document
1341 if ( !$frame->depth ) {
1342 $flag = 0;
1343 } else {
1344 $flag = self::PTD_FOR_INCLUSION;
1345 }
1346 $dom = $this->preprocessToDom( $text, $flag );
1347 $text = $frame->expand( $dom );
1348 } else {
1349 # if $frame is not provided, then use old-style replaceVariables
1350 $text = $this->replaceVariables( $text );
1351 }
1352
1353 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1354 $text = Sanitizer::removeHTMLtags(
1355 $text,
1356 [ $this, 'attributeStripCallback' ],
1357 false,
1358 array_keys( $this->mTransparentTagHooks ),
1359 [],
1360 [ $this, 'addTrackingCategory' ]
1361 );
1362 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1363
1364 # Tables need to come after variable replacement for things to work
1365 # properly; putting them before other transformations should keep
1366 # exciting things like link expansions from showing up in surprising
1367 # places.
1368 $text = $this->doTableStuff( $text );
1369
1370 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1371
1372 $text = $this->doDoubleUnderscore( $text );
1373
1374 $text = $this->doHeadings( $text );
1375 $text = $this->replaceInternalLinks( $text );
1376 $text = $this->doAllQuotes( $text );
1377 $text = $this->replaceExternalLinks( $text );
1378
1379 # replaceInternalLinks may sometimes leave behind
1380 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1381 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1382
1383 $text = $this->doMagicLinks( $text );
1384 $text = $this->formatHeadings( $text, $origText, $isMain );
1385
1386 return $text;
1387 }
1388
1389 /**
1390 * Helper function for parse() that transforms half-parsed HTML into fully
1391 * parsed HTML.
1392 *
1393 * @param string $text
1394 * @param bool $isMain
1395 * @param bool $linestart
1396 * @return string
1397 */
1398 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1399 $text = $this->mStripState->unstripGeneral( $text );
1400
1401 // Avoid PHP 7.1 warning from passing $this by reference
1402 $parser = $this;
1403
1404 if ( $isMain ) {
1405 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1406 }
1407
1408 # Clean up special characters, only run once, next-to-last before doBlockLevels
1409 $text = Sanitizer::armorFrenchSpaces( $text );
1410
1411 $text = $this->doBlockLevels( $text, $linestart );
1412
1413 $this->replaceLinkHolders( $text );
1414
1415 /**
1416 * The input doesn't get language converted if
1417 * a) It's disabled
1418 * b) Content isn't converted
1419 * c) It's a conversion table
1420 * d) it is an interface message (which is in the user language)
1421 */
1422 if ( !( $this->mOptions->getDisableContentConversion()
1423 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1424 ) {
1425 if ( !$this->mOptions->getInterfaceMessage() ) {
1426 # The position of the convert() call should not be changed. it
1427 # assumes that the links are all replaced and the only thing left
1428 # is the <nowiki> mark.
1429 $text = $this->getTargetLanguage()->convert( $text );
1430 }
1431 }
1432
1433 $text = $this->mStripState->unstripNoWiki( $text );
1434
1435 if ( $isMain ) {
1436 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1437 }
1438
1439 $text = $this->replaceTransparentTags( $text );
1440 $text = $this->mStripState->unstripGeneral( $text );
1441
1442 $text = Sanitizer::normalizeCharReferences( $text );
1443
1444 if ( MWTidy::isEnabled() ) {
1445 if ( $this->mOptions->getTidy() ) {
1446 $text = MWTidy::tidy( $text );
1447 }
1448 } else {
1449 # attempt to sanitize at least some nesting problems
1450 # (T4702 and quite a few others)
1451 # This code path is buggy and deprecated!
1452 wfDeprecated( 'disabling tidy', '1.33' );
1453 $tidyregs = [
1454 # ''Something [http://www.cool.com cool''] -->
1455 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1456 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1457 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1458 # fix up an anchor inside another anchor, only
1459 # at least for a single single nested link (T5695)
1460 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1461 '\\1\\2</a>\\3</a>\\1\\4</a>',
1462 # fix div inside inline elements- doBlockLevels won't wrap a line which
1463 # contains a div, so fix it up here; replace
1464 # div with escaped text
1465 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1466 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1467 # remove empty italic or bold tag pairs, some
1468 # introduced by rules above
1469 '/<([bi])><\/\\1>/' => '',
1470 ];
1471
1472 $text = preg_replace(
1473 array_keys( $tidyregs ),
1474 array_values( $tidyregs ),
1475 $text );
1476 }
1477
1478 if ( $isMain ) {
1479 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1480 }
1481
1482 return $text;
1483 }
1484
1485 /**
1486 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1487 * magic external links.
1488 *
1489 * DML
1490 * @private
1491 *
1492 * @param string $text
1493 *
1494 * @return string
1495 */
1496 public function doMagicLinks( $text ) {
1497 $prots = wfUrlProtocolsWithoutProtRel();
1498 $urlChar = self::EXT_LINK_URL_CLASS;
1499 $addr = self::EXT_LINK_ADDR;
1500 $space = self::SPACE_NOT_NL; # non-newline space
1501 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1502 $spaces = "$space++"; # possessive match of 1 or more spaces
1503 $text = preg_replace_callback(
1504 '!(?: # Start cases
1505 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1506 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1507 (\b # m[3]: Free external links
1508 (?i:$prots)
1509 ($addr$urlChar*) # m[4]: Post-protocol path
1510 ) |
1511 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1512 ([0-9]+)\b |
1513 \bISBN $spaces ( # m[6]: ISBN, capture number
1514 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1515 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1516 [0-9Xx] # check digit
1517 )\b
1518 )!xu", [ $this, 'magicLinkCallback' ], $text );
1519 return $text;
1520 }
1521
1522 /**
1523 * @throws MWException
1524 * @param array $m
1525 * @return string HTML
1526 */
1527 public function magicLinkCallback( $m ) {
1528 if ( isset( $m[1] ) && $m[1] !== '' ) {
1529 # Skip anchor
1530 return $m[0];
1531 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1532 # Skip HTML element
1533 return $m[0];
1534 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1535 # Free external link
1536 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1537 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1538 # RFC or PMID
1539 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1540 if ( !$this->mOptions->getMagicRFCLinks() ) {
1541 return $m[0];
1542 }
1543 $keyword = 'RFC';
1544 $urlmsg = 'rfcurl';
1545 $cssClass = 'mw-magiclink-rfc';
1546 $trackingCat = 'magiclink-tracking-rfc';
1547 $id = $m[5];
1548 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1549 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1550 return $m[0];
1551 }
1552 $keyword = 'PMID';
1553 $urlmsg = 'pubmedurl';
1554 $cssClass = 'mw-magiclink-pmid';
1555 $trackingCat = 'magiclink-tracking-pmid';
1556 $id = $m[5];
1557 } else {
1558 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1559 substr( $m[0], 0, 20 ) . '"' );
1560 }
1561 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1562 $this->addTrackingCategory( $trackingCat );
1563 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1564 } elseif ( isset( $m[6] ) && $m[6] !== ''
1565 && $this->mOptions->getMagicISBNLinks()
1566 ) {
1567 # ISBN
1568 $isbn = $m[6];
1569 $space = self::SPACE_NOT_NL; # non-newline space
1570 $isbn = preg_replace( "/$space/", ' ', $isbn );
1571 $num = strtr( $isbn, [
1572 '-' => '',
1573 ' ' => '',
1574 'x' => 'X',
1575 ] );
1576 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1577 return $this->getLinkRenderer()->makeKnownLink(
1578 SpecialPage::getTitleFor( 'Booksources', $num ),
1579 "ISBN $isbn",
1580 [
1581 'class' => 'internal mw-magiclink-isbn',
1582 'title' => false // suppress title attribute
1583 ]
1584 );
1585 } else {
1586 return $m[0];
1587 }
1588 }
1589
1590 /**
1591 * Make a free external link, given a user-supplied URL
1592 *
1593 * @param string $url
1594 * @param int $numPostProto
1595 * The number of characters after the protocol.
1596 * @return string HTML
1597 * @private
1598 */
1599 public function makeFreeExternalLink( $url, $numPostProto ) {
1600 $trail = '';
1601
1602 # The characters '<' and '>' (which were escaped by
1603 # removeHTMLtags()) should not be included in
1604 # URLs, per RFC 2396.
1605 # Make &nbsp; terminate a URL as well (bug T84937)
1606 $m2 = [];
1607 if ( preg_match(
1608 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1609 $url,
1610 $m2,
1611 PREG_OFFSET_CAPTURE
1612 ) ) {
1613 $trail = substr( $url, $m2[0][1] ) . $trail;
1614 $url = substr( $url, 0, $m2[0][1] );
1615 }
1616
1617 # Move trailing punctuation to $trail
1618 $sep = ',;\.:!?';
1619 # If there is no left bracket, then consider right brackets fair game too
1620 if ( strpos( $url, '(' ) === false ) {
1621 $sep .= ')';
1622 }
1623
1624 $urlRev = strrev( $url );
1625 $numSepChars = strspn( $urlRev, $sep );
1626 # Don't break a trailing HTML entity by moving the ; into $trail
1627 # This is in hot code, so use substr_compare to avoid having to
1628 # create a new string object for the comparison
1629 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1630 # more optimization: instead of running preg_match with a $
1631 # anchor, which can be slow, do the match on the reversed
1632 # string starting at the desired offset.
1633 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1634 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1635 $numSepChars--;
1636 }
1637 }
1638 if ( $numSepChars ) {
1639 $trail = substr( $url, -$numSepChars ) . $trail;
1640 $url = substr( $url, 0, -$numSepChars );
1641 }
1642
1643 # Verify that we still have a real URL after trail removal, and
1644 # not just lone protocol
1645 if ( strlen( $trail ) >= $numPostProto ) {
1646 return $url . $trail;
1647 }
1648
1649 $url = Sanitizer::cleanUrl( $url );
1650
1651 # Is this an external image?
1652 $text = $this->maybeMakeExternalImage( $url );
1653 if ( $text === false ) {
1654 # Not an image, make a link
1655 $text = Linker::makeExternalLink( $url,
1656 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1657 true, 'free',
1658 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1659 # Register it in the output object...
1660 $this->mOutput->addExternalLink( $url );
1661 }
1662 return $text . $trail;
1663 }
1664
1665 /**
1666 * Parse headers and return html
1667 *
1668 * @private
1669 *
1670 * @param string $text
1671 *
1672 * @return string
1673 */
1674 public function doHeadings( $text ) {
1675 for ( $i = 6; $i >= 1; --$i ) {
1676 $h = str_repeat( '=', $i );
1677 // Trim non-newline whitespace from headings
1678 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1679 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1680 }
1681 return $text;
1682 }
1683
1684 /**
1685 * Replace single quotes with HTML markup
1686 * @private
1687 *
1688 * @param string $text
1689 *
1690 * @return string The altered text
1691 */
1692 public function doAllQuotes( $text ) {
1693 $outtext = '';
1694 $lines = StringUtils::explode( "\n", $text );
1695 foreach ( $lines as $line ) {
1696 $outtext .= $this->doQuotes( $line ) . "\n";
1697 }
1698 $outtext = substr( $outtext, 0, -1 );
1699 return $outtext;
1700 }
1701
1702 /**
1703 * Helper function for doAllQuotes()
1704 *
1705 * @param string $text
1706 *
1707 * @return string
1708 */
1709 public function doQuotes( $text ) {
1710 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1711 $countarr = count( $arr );
1712 if ( $countarr == 1 ) {
1713 return $text;
1714 }
1715
1716 // First, do some preliminary work. This may shift some apostrophes from
1717 // being mark-up to being text. It also counts the number of occurrences
1718 // of bold and italics mark-ups.
1719 $numbold = 0;
1720 $numitalics = 0;
1721 for ( $i = 1; $i < $countarr; $i += 2 ) {
1722 $thislen = strlen( $arr[$i] );
1723 // If there are ever four apostrophes, assume the first is supposed to
1724 // be text, and the remaining three constitute mark-up for bold text.
1725 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1726 if ( $thislen == 4 ) {
1727 $arr[$i - 1] .= "'";
1728 $arr[$i] = "'''";
1729 $thislen = 3;
1730 } elseif ( $thislen > 5 ) {
1731 // If there are more than 5 apostrophes in a row, assume they're all
1732 // text except for the last 5.
1733 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1734 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1735 $arr[$i] = "'''''";
1736 $thislen = 5;
1737 }
1738 // Count the number of occurrences of bold and italics mark-ups.
1739 if ( $thislen == 2 ) {
1740 $numitalics++;
1741 } elseif ( $thislen == 3 ) {
1742 $numbold++;
1743 } elseif ( $thislen == 5 ) {
1744 $numitalics++;
1745 $numbold++;
1746 }
1747 }
1748
1749 // If there is an odd number of both bold and italics, it is likely
1750 // that one of the bold ones was meant to be an apostrophe followed
1751 // by italics. Which one we cannot know for certain, but it is more
1752 // likely to be one that has a single-letter word before it.
1753 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1754 $firstsingleletterword = -1;
1755 $firstmultiletterword = -1;
1756 $firstspace = -1;
1757 for ( $i = 1; $i < $countarr; $i += 2 ) {
1758 if ( strlen( $arr[$i] ) == 3 ) {
1759 $x1 = substr( $arr[$i - 1], -1 );
1760 $x2 = substr( $arr[$i - 1], -2, 1 );
1761 if ( $x1 === ' ' ) {
1762 if ( $firstspace == -1 ) {
1763 $firstspace = $i;
1764 }
1765 } elseif ( $x2 === ' ' ) {
1766 $firstsingleletterword = $i;
1767 // if $firstsingleletterword is set, we don't
1768 // look at the other options, so we can bail early.
1769 break;
1770 } else {
1771 if ( $firstmultiletterword == -1 ) {
1772 $firstmultiletterword = $i;
1773 }
1774 }
1775 }
1776 }
1777
1778 // If there is a single-letter word, use it!
1779 if ( $firstsingleletterword > -1 ) {
1780 $arr[$firstsingleletterword] = "''";
1781 $arr[$firstsingleletterword - 1] .= "'";
1782 } elseif ( $firstmultiletterword > -1 ) {
1783 // If not, but there's a multi-letter word, use that one.
1784 $arr[$firstmultiletterword] = "''";
1785 $arr[$firstmultiletterword - 1] .= "'";
1786 } elseif ( $firstspace > -1 ) {
1787 // ... otherwise use the first one that has neither.
1788 // (notice that it is possible for all three to be -1 if, for example,
1789 // there is only one pentuple-apostrophe in the line)
1790 $arr[$firstspace] = "''";
1791 $arr[$firstspace - 1] .= "'";
1792 }
1793 }
1794
1795 // Now let's actually convert our apostrophic mush to HTML!
1796 $output = '';
1797 $buffer = '';
1798 $state = '';
1799 $i = 0;
1800 foreach ( $arr as $r ) {
1801 if ( ( $i % 2 ) == 0 ) {
1802 if ( $state === 'both' ) {
1803 $buffer .= $r;
1804 } else {
1805 $output .= $r;
1806 }
1807 } else {
1808 $thislen = strlen( $r );
1809 if ( $thislen == 2 ) {
1810 if ( $state === 'i' ) {
1811 $output .= '</i>';
1812 $state = '';
1813 } elseif ( $state === 'bi' ) {
1814 $output .= '</i>';
1815 $state = 'b';
1816 } elseif ( $state === 'ib' ) {
1817 $output .= '</b></i><b>';
1818 $state = 'b';
1819 } elseif ( $state === 'both' ) {
1820 $output .= '<b><i>' . $buffer . '</i>';
1821 $state = 'b';
1822 } else { // $state can be 'b' or ''
1823 $output .= '<i>';
1824 $state .= 'i';
1825 }
1826 } elseif ( $thislen == 3 ) {
1827 if ( $state === 'b' ) {
1828 $output .= '</b>';
1829 $state = '';
1830 } elseif ( $state === 'bi' ) {
1831 $output .= '</i></b><i>';
1832 $state = 'i';
1833 } elseif ( $state === 'ib' ) {
1834 $output .= '</b>';
1835 $state = 'i';
1836 } elseif ( $state === 'both' ) {
1837 $output .= '<i><b>' . $buffer . '</b>';
1838 $state = 'i';
1839 } else { // $state can be 'i' or ''
1840 $output .= '<b>';
1841 $state .= 'b';
1842 }
1843 } elseif ( $thislen == 5 ) {
1844 if ( $state === 'b' ) {
1845 $output .= '</b><i>';
1846 $state = 'i';
1847 } elseif ( $state === 'i' ) {
1848 $output .= '</i><b>';
1849 $state = 'b';
1850 } elseif ( $state === 'bi' ) {
1851 $output .= '</i></b>';
1852 $state = '';
1853 } elseif ( $state === 'ib' ) {
1854 $output .= '</b></i>';
1855 $state = '';
1856 } elseif ( $state === 'both' ) {
1857 $output .= '<i><b>' . $buffer . '</b></i>';
1858 $state = '';
1859 } else { // ($state == '')
1860 $buffer = '';
1861 $state = 'both';
1862 }
1863 }
1864 }
1865 $i++;
1866 }
1867 // Now close all remaining tags. Notice that the order is important.
1868 if ( $state === 'b' || $state === 'ib' ) {
1869 $output .= '</b>';
1870 }
1871 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1872 $output .= '</i>';
1873 }
1874 if ( $state === 'bi' ) {
1875 $output .= '</b>';
1876 }
1877 // There might be lonely ''''', so make sure we have a buffer
1878 if ( $state === 'both' && $buffer ) {
1879 $output .= '<b><i>' . $buffer . '</i></b>';
1880 }
1881 return $output;
1882 }
1883
1884 /**
1885 * Replace external links (REL)
1886 *
1887 * Note: this is all very hackish and the order of execution matters a lot.
1888 * Make sure to run tests/parser/parserTests.php if you change this code.
1889 *
1890 * @private
1891 *
1892 * @param string $text
1893 *
1894 * @throws MWException
1895 * @return string
1896 */
1897 public function replaceExternalLinks( $text ) {
1898 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1899 if ( $bits === false ) {
1900 throw new MWException( "PCRE needs to be compiled with "
1901 . "--enable-unicode-properties in order for MediaWiki to function" );
1902 }
1903 $s = array_shift( $bits );
1904
1905 $i = 0;
1906 while ( $i < count( $bits ) ) {
1907 $url = $bits[$i++];
1908 $i++; // protocol
1909 $text = $bits[$i++];
1910 $trail = $bits[$i++];
1911
1912 # The characters '<' and '>' (which were escaped by
1913 # removeHTMLtags()) should not be included in
1914 # URLs, per RFC 2396.
1915 $m2 = [];
1916 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1917 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1918 $url = substr( $url, 0, $m2[0][1] );
1919 }
1920
1921 # If the link text is an image URL, replace it with an <img> tag
1922 # This happened by accident in the original parser, but some people used it extensively
1923 $img = $this->maybeMakeExternalImage( $text );
1924 if ( $img !== false ) {
1925 $text = $img;
1926 }
1927
1928 $dtrail = '';
1929
1930 # Set linktype for CSS
1931 $linktype = 'text';
1932
1933 # No link text, e.g. [http://domain.tld/some.link]
1934 if ( $text == '' ) {
1935 # Autonumber
1936 $langObj = $this->getTargetLanguage();
1937 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1938 $linktype = 'autonumber';
1939 } else {
1940 # Have link text, e.g. [http://domain.tld/some.link text]s
1941 # Check for trail
1942 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1943 }
1944
1945 // Excluding protocol-relative URLs may avoid many false positives.
1946 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1947 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1948 }
1949
1950 $url = Sanitizer::cleanUrl( $url );
1951
1952 # Use the encoded URL
1953 # This means that users can paste URLs directly into the text
1954 # Funny characters like ö aren't valid in URLs anyway
1955 # This was changed in August 2004
1956 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1957 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1958
1959 # Register link in the output object.
1960 $this->mOutput->addExternalLink( $url );
1961 }
1962
1963 return $s;
1964 }
1965
1966 /**
1967 * Get the rel attribute for a particular external link.
1968 *
1969 * @since 1.21
1970 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1971 * nofollow if appropriate
1972 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1973 * @return string|null Rel attribute for $url
1974 */
1975 public static function getExternalLinkRel( $url = false, $title = null ) {
1976 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1977 $ns = $title ? $title->getNamespace() : false;
1978 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1979 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1980 ) {
1981 return 'nofollow';
1982 }
1983 return null;
1984 }
1985
1986 /**
1987 * Get an associative array of additional HTML attributes appropriate for a
1988 * particular external link. This currently may include rel => nofollow
1989 * (depending on configuration, namespace, and the URL's domain) and/or a
1990 * target attribute (depending on configuration).
1991 *
1992 * @param string $url URL to extract the domain from for rel =>
1993 * nofollow if appropriate
1994 * @return array Associative array of HTML attributes
1995 */
1996 public function getExternalLinkAttribs( $url ) {
1997 $attribs = [];
1998 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1999
2000 $target = $this->mOptions->getExternalLinkTarget();
2001 if ( $target ) {
2002 $attribs['target'] = $target;
2003 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2004 // T133507. New windows can navigate parent cross-origin.
2005 // Including noreferrer due to lacking browser
2006 // support of noopener. Eventually noreferrer should be removed.
2007 if ( $rel !== '' ) {
2008 $rel .= ' ';
2009 }
2010 $rel .= 'noreferrer noopener';
2011 }
2012 }
2013 $attribs['rel'] = $rel;
2014 return $attribs;
2015 }
2016
2017 /**
2018 * Replace unusual escape codes in a URL with their equivalent characters
2019 *
2020 * This generally follows the syntax defined in RFC 3986, with special
2021 * consideration for HTTP query strings.
2022 *
2023 * @param string $url
2024 * @return string
2025 */
2026 public static function normalizeLinkUrl( $url ) {
2027 # Test for RFC 3986 IPv6 syntax
2028 $scheme = '[a-z][a-z0-9+.-]*:';
2029 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2030 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2031 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2032 IP::isValid( rawurldecode( $m[1] ) )
2033 ) {
2034 $isIPv6 = rawurldecode( $m[1] );
2035 } else {
2036 $isIPv6 = false;
2037 }
2038
2039 # Make sure unsafe characters are encoded
2040 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2041 function ( $m ) {
2042 return rawurlencode( $m[0] );
2043 },
2044 $url
2045 );
2046
2047 $ret = '';
2048 $end = strlen( $url );
2049
2050 # Fragment part - 'fragment'
2051 $start = strpos( $url, '#' );
2052 if ( $start !== false && $start < $end ) {
2053 $ret = self::normalizeUrlComponent(
2054 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2055 $end = $start;
2056 }
2057
2058 # Query part - 'query' minus &=+;
2059 $start = strpos( $url, '?' );
2060 if ( $start !== false && $start < $end ) {
2061 $ret = self::normalizeUrlComponent(
2062 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2063 $end = $start;
2064 }
2065
2066 # Scheme and path part - 'pchar'
2067 # (we assume no userinfo or encoded colons in the host)
2068 $ret = self::normalizeUrlComponent(
2069 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2070
2071 # Fix IPv6 syntax
2072 if ( $isIPv6 !== false ) {
2073 $ipv6Host = "%5B({$isIPv6})%5D";
2074 $ret = preg_replace(
2075 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2076 "$1[$2]",
2077 $ret
2078 );
2079 }
2080
2081 return $ret;
2082 }
2083
2084 private static function normalizeUrlComponent( $component, $unsafe ) {
2085 $callback = function ( $matches ) use ( $unsafe ) {
2086 $char = urldecode( $matches[0] );
2087 $ord = ord( $char );
2088 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2089 # Unescape it
2090 return $char;
2091 } else {
2092 # Leave it escaped, but use uppercase for a-f
2093 return strtoupper( $matches[0] );
2094 }
2095 };
2096 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2097 }
2098
2099 /**
2100 * make an image if it's allowed, either through the global
2101 * option, through the exception, or through the on-wiki whitelist
2102 *
2103 * @param string $url
2104 *
2105 * @return string
2106 */
2107 private function maybeMakeExternalImage( $url ) {
2108 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2109 $imagesexception = !empty( $imagesfrom );
2110 $text = false;
2111 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2112 if ( $imagesexception && is_array( $imagesfrom ) ) {
2113 $imagematch = false;
2114 foreach ( $imagesfrom as $match ) {
2115 if ( strpos( $url, $match ) === 0 ) {
2116 $imagematch = true;
2117 break;
2118 }
2119 }
2120 } elseif ( $imagesexception ) {
2121 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2122 } else {
2123 $imagematch = false;
2124 }
2125
2126 if ( $this->mOptions->getAllowExternalImages()
2127 || ( $imagesexception && $imagematch )
2128 ) {
2129 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2130 # Image found
2131 $text = Linker::makeExternalImage( $url );
2132 }
2133 }
2134 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2135 && preg_match( self::EXT_IMAGE_REGEX, $url )
2136 ) {
2137 $whitelist = explode(
2138 "\n",
2139 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2140 );
2141
2142 foreach ( $whitelist as $entry ) {
2143 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2144 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2145 continue;
2146 }
2147 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2148 # Image matches a whitelist entry
2149 $text = Linker::makeExternalImage( $url );
2150 break;
2151 }
2152 }
2153 }
2154 return $text;
2155 }
2156
2157 /**
2158 * Process [[ ]] wikilinks
2159 *
2160 * @param string $s
2161 *
2162 * @return string Processed text
2163 *
2164 * @private
2165 */
2166 public function replaceInternalLinks( $s ) {
2167 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2168 return $s;
2169 }
2170
2171 /**
2172 * Process [[ ]] wikilinks (RIL)
2173 * @param string &$s
2174 * @throws MWException
2175 * @return LinkHolderArray
2176 *
2177 * @private
2178 */
2179 public function replaceInternalLinks2( &$s ) {
2180 static $tc = false, $e1, $e1_img;
2181 # the % is needed to support urlencoded titles as well
2182 if ( !$tc ) {
2183 $tc = Title::legalChars() . '#%';
2184 # Match a link having the form [[namespace:link|alternate]]trail
2185 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2186 # Match cases where there is no "]]", which might still be images
2187 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2188 }
2189
2190 $holders = new LinkHolderArray( $this );
2191
2192 # split the entire text string on occurrences of [[
2193 $a = StringUtils::explode( '[[', ' ' . $s );
2194 # get the first element (all text up to first [[), and remove the space we added
2195 $s = $a->current();
2196 $a->next();
2197 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2198 $s = substr( $s, 1 );
2199
2200 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2201 $e2 = null;
2202 if ( $useLinkPrefixExtension ) {
2203 # Match the end of a line for a word that's not followed by whitespace,
2204 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2205 $charset = $this->contLang->linkPrefixCharset();
2206 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2207 }
2208
2209 if ( is_null( $this->mTitle ) ) {
2210 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2211 }
2212 $nottalk = !$this->mTitle->isTalkPage();
2213
2214 if ( $useLinkPrefixExtension ) {
2215 $m = [];
2216 if ( preg_match( $e2, $s, $m ) ) {
2217 $first_prefix = $m[2];
2218 } else {
2219 $first_prefix = false;
2220 }
2221 } else {
2222 $prefix = '';
2223 }
2224
2225 $useSubpages = $this->areSubpagesAllowed();
2226
2227 # Loop for each link
2228 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2229 # Check for excessive memory usage
2230 if ( $holders->isBig() ) {
2231 # Too big
2232 # Do the existence check, replace the link holders and clear the array
2233 $holders->replace( $s );
2234 $holders->clear();
2235 }
2236
2237 if ( $useLinkPrefixExtension ) {
2238 if ( preg_match( $e2, $s, $m ) ) {
2239 list( , $s, $prefix ) = $m;
2240 } else {
2241 $prefix = '';
2242 }
2243 # first link
2244 if ( $first_prefix ) {
2245 $prefix = $first_prefix;
2246 $first_prefix = false;
2247 }
2248 }
2249
2250 $might_be_img = false;
2251
2252 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2253 $text = $m[2];
2254 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2255 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2256 # the real problem is with the $e1 regex
2257 # See T1500.
2258 # Still some problems for cases where the ] is meant to be outside punctuation,
2259 # and no image is in sight. See T4095.
2260 if ( $text !== ''
2261 && substr( $m[3], 0, 1 ) === ']'
2262 && strpos( $text, '[' ) !== false
2263 ) {
2264 $text .= ']'; # so that replaceExternalLinks($text) works later
2265 $m[3] = substr( $m[3], 1 );
2266 }
2267 # fix up urlencoded title texts
2268 if ( strpos( $m[1], '%' ) !== false ) {
2269 # Should anchors '#' also be rejected?
2270 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2271 }
2272 $trail = $m[3];
2273 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2274 # Invalid, but might be an image with a link in its caption
2275 $might_be_img = true;
2276 $text = $m[2];
2277 if ( strpos( $m[1], '%' ) !== false ) {
2278 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2279 }
2280 $trail = "";
2281 } else { # Invalid form; output directly
2282 $s .= $prefix . '[[' . $line;
2283 continue;
2284 }
2285
2286 $origLink = ltrim( $m[1], ' ' );
2287
2288 # Don't allow internal links to pages containing
2289 # PROTO: where PROTO is a valid URL protocol; these
2290 # should be external links.
2291 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2292 $s .= $prefix . '[[' . $line;
2293 continue;
2294 }
2295
2296 # Make subpage if necessary
2297 if ( $useSubpages ) {
2298 $link = $this->maybeDoSubpageLink( $origLink, $text );
2299 } else {
2300 $link = $origLink;
2301 }
2302
2303 // \x7f isn't a default legal title char, so most likely strip
2304 // markers will force us into the "invalid form" path above. But,
2305 // just in case, let's assert that xmlish tags aren't valid in
2306 // the title position.
2307 $unstrip = $this->mStripState->killMarkers( $link );
2308 $noMarkers = ( $unstrip === $link );
2309
2310 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2311 if ( $nt === null ) {
2312 $s .= $prefix . '[[' . $line;
2313 continue;
2314 }
2315
2316 $ns = $nt->getNamespace();
2317 $iw = $nt->getInterwiki();
2318
2319 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2320
2321 if ( $might_be_img ) { # if this is actually an invalid link
2322 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2323 $found = false;
2324 while ( true ) {
2325 # look at the next 'line' to see if we can close it there
2326 $a->next();
2327 $next_line = $a->current();
2328 if ( $next_line === false || $next_line === null ) {
2329 break;
2330 }
2331 $m = explode( ']]', $next_line, 3 );
2332 if ( count( $m ) == 3 ) {
2333 # the first ]] closes the inner link, the second the image
2334 $found = true;
2335 $text .= "[[{$m[0]}]]{$m[1]}";
2336 $trail = $m[2];
2337 break;
2338 } elseif ( count( $m ) == 2 ) {
2339 # if there's exactly one ]] that's fine, we'll keep looking
2340 $text .= "[[{$m[0]}]]{$m[1]}";
2341 } else {
2342 # if $next_line is invalid too, we need look no further
2343 $text .= '[[' . $next_line;
2344 break;
2345 }
2346 }
2347 if ( !$found ) {
2348 # we couldn't find the end of this imageLink, so output it raw
2349 # but don't ignore what might be perfectly normal links in the text we've examined
2350 $holders->merge( $this->replaceInternalLinks2( $text ) );
2351 $s .= "{$prefix}[[$link|$text";
2352 # note: no $trail, because without an end, there *is* no trail
2353 continue;
2354 }
2355 } else { # it's not an image, so output it raw
2356 $s .= "{$prefix}[[$link|$text";
2357 # note: no $trail, because without an end, there *is* no trail
2358 continue;
2359 }
2360 }
2361
2362 $wasblank = ( $text == '' );
2363 if ( $wasblank ) {
2364 $text = $link;
2365 if ( !$noforce ) {
2366 # Strip off leading ':'
2367 $text = substr( $text, 1 );
2368 }
2369 } else {
2370 # T6598 madness. Handle the quotes only if they come from the alternate part
2371 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2372 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2373 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2374 $text = $this->doQuotes( $text );
2375 }
2376
2377 # Link not escaped by : , create the various objects
2378 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2379 # Interwikis
2380 if (
2381 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2382 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2383 in_array( $iw, $this->siteConfig->get( 'ExtraInterlanguageLinkPrefixes' ) )
2384 )
2385 ) {
2386 # T26502: filter duplicates
2387 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2388 $this->mLangLinkLanguages[$iw] = true;
2389 $this->mOutput->addLanguageLink( $nt->getFullText() );
2390 }
2391
2392 /**
2393 * Strip the whitespace interwiki links produce, see T10897
2394 */
2395 $s = rtrim( $s . $prefix ) . $trail; # T175416
2396 continue;
2397 }
2398
2399 if ( $ns == NS_FILE ) {
2400 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2401 if ( $wasblank ) {
2402 # if no parameters were passed, $text
2403 # becomes something like "File:Foo.png",
2404 # which we don't want to pass on to the
2405 # image generator
2406 $text = '';
2407 } else {
2408 # recursively parse links inside the image caption
2409 # actually, this will parse them in any other parameters, too,
2410 # but it might be hard to fix that, and it doesn't matter ATM
2411 $text = $this->replaceExternalLinks( $text );
2412 $holders->merge( $this->replaceInternalLinks2( $text ) );
2413 }
2414 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2415 $s .= $prefix . $this->armorLinks(
2416 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2417 continue;
2418 }
2419 } elseif ( $ns == NS_CATEGORY ) {
2420 /**
2421 * Strip the whitespace Category links produce, see T2087
2422 */
2423 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2424
2425 if ( $wasblank ) {
2426 $sortkey = $this->getDefaultSort();
2427 } else {
2428 $sortkey = $text;
2429 }
2430 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2431 $sortkey = str_replace( "\n", '', $sortkey );
2432 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2433 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2434
2435 continue;
2436 }
2437 }
2438
2439 # Self-link checking. For some languages, variants of the title are checked in
2440 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2441 # for linking to a different variant.
2442 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2443 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2444 continue;
2445 }
2446
2447 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2448 # @todo FIXME: Should do batch file existence checks, see comment below
2449 if ( $ns == NS_MEDIA ) {
2450 # Give extensions a chance to select the file revision for us
2451 $options = [];
2452 $descQuery = false;
2453 Hooks::run( 'BeforeParserFetchFileAndTitle',
2454 [ $this, $nt, &$options, &$descQuery ] );
2455 # Fetch and register the file (file title may be different via hooks)
2456 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2457 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2458 $s .= $prefix . $this->armorLinks(
2459 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2460 continue;
2461 }
2462
2463 # Some titles, such as valid special pages or files in foreign repos, should
2464 # be shown as bluelinks even though they're not included in the page table
2465 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2466 # batch file existence checks for NS_FILE and NS_MEDIA
2467 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2468 $this->mOutput->addLink( $nt );
2469 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2470 } else {
2471 # Links will be added to the output link list after checking
2472 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2473 }
2474 }
2475 return $holders;
2476 }
2477
2478 /**
2479 * Render a forced-blue link inline; protect against double expansion of
2480 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2481 * Since this little disaster has to split off the trail text to avoid
2482 * breaking URLs in the following text without breaking trails on the
2483 * wiki links, it's been made into a horrible function.
2484 *
2485 * @param Title $nt
2486 * @param string $text
2487 * @param string $trail
2488 * @param string $prefix
2489 * @return string HTML-wikitext mix oh yuck
2490 */
2491 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2492 list( $inside, $trail ) = Linker::splitTrail( $trail );
2493
2494 if ( $text == '' ) {
2495 $text = htmlspecialchars( $nt->getPrefixedText() );
2496 }
2497
2498 $link = $this->getLinkRenderer()->makeKnownLink(
2499 $nt, new HtmlArmor( "$prefix$text$inside" )
2500 );
2501
2502 return $this->armorLinks( $link ) . $trail;
2503 }
2504
2505 /**
2506 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2507 * going to go through further parsing steps before inline URL expansion.
2508 *
2509 * Not needed quite as much as it used to be since free links are a bit
2510 * more sensible these days. But bracketed links are still an issue.
2511 *
2512 * @param string $text More-or-less HTML
2513 * @return string Less-or-more HTML with NOPARSE bits
2514 */
2515 public function armorLinks( $text ) {
2516 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2517 self::MARKER_PREFIX . "NOPARSE$1", $text );
2518 }
2519
2520 /**
2521 * Return true if subpage links should be expanded on this page.
2522 * @return bool
2523 */
2524 public function areSubpagesAllowed() {
2525 # Some namespaces don't allow subpages
2526 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2527 }
2528
2529 /**
2530 * Handle link to subpage if necessary
2531 *
2532 * @param string $target The source of the link
2533 * @param string &$text The link text, modified as necessary
2534 * @return string The full name of the link
2535 * @private
2536 */
2537 public function maybeDoSubpageLink( $target, &$text ) {
2538 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2539 }
2540
2541 /**
2542 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2543 *
2544 * @param string $text
2545 * @param bool $linestart Whether or not this is at the start of a line.
2546 * @private
2547 * @return string The lists rendered as HTML
2548 */
2549 public function doBlockLevels( $text, $linestart ) {
2550 return BlockLevelPass::doBlockLevels( $text, $linestart );
2551 }
2552
2553 /**
2554 * Return value of a magic variable (like PAGENAME)
2555 *
2556 * @private
2557 *
2558 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2559 * @param bool|PPFrame $frame
2560 *
2561 * @throws MWException
2562 * @return string
2563 */
2564 public function getVariableValue( $index, $frame = false ) {
2565 if ( is_null( $this->mTitle ) ) {
2566 // If no title set, bad things are going to happen
2567 // later. Title should always be set since this
2568 // should only be called in the middle of a parse
2569 // operation (but the unit-tests do funky stuff)
2570 throw new MWException( __METHOD__ . ' Should only be '
2571 . ' called while parsing (no title set)' );
2572 }
2573
2574 // Avoid PHP 7.1 warning from passing $this by reference
2575 $parser = $this;
2576
2577 /**
2578 * Some of these require message or data lookups and can be
2579 * expensive to check many times.
2580 */
2581 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2582 if ( isset( $this->mVarCache[$index] ) ) {
2583 return $this->mVarCache[$index];
2584 }
2585 }
2586
2587 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2588 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2589
2590 $pageLang = $this->getFunctionLang();
2591
2592 switch ( $index ) {
2593 case '!':
2594 $value = '|';
2595 break;
2596 case 'currentmonth':
2597 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2598 break;
2599 case 'currentmonth1':
2600 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2601 break;
2602 case 'currentmonthname':
2603 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2604 break;
2605 case 'currentmonthnamegen':
2606 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2607 break;
2608 case 'currentmonthabbrev':
2609 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2610 break;
2611 case 'currentday':
2612 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2613 break;
2614 case 'currentday2':
2615 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2616 break;
2617 case 'localmonth':
2618 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2619 break;
2620 case 'localmonth1':
2621 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2622 break;
2623 case 'localmonthname':
2624 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2625 break;
2626 case 'localmonthnamegen':
2627 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2628 break;
2629 case 'localmonthabbrev':
2630 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2631 break;
2632 case 'localday':
2633 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2634 break;
2635 case 'localday2':
2636 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2637 break;
2638 case 'pagename':
2639 $value = wfEscapeWikiText( $this->mTitle->getText() );
2640 break;
2641 case 'pagenamee':
2642 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2643 break;
2644 case 'fullpagename':
2645 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2646 break;
2647 case 'fullpagenamee':
2648 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2649 break;
2650 case 'subpagename':
2651 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2652 break;
2653 case 'subpagenamee':
2654 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2655 break;
2656 case 'rootpagename':
2657 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2658 break;
2659 case 'rootpagenamee':
2660 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2661 ' ',
2662 '_',
2663 $this->mTitle->getRootText()
2664 ) ) );
2665 break;
2666 case 'basepagename':
2667 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2668 break;
2669 case 'basepagenamee':
2670 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2671 ' ',
2672 '_',
2673 $this->mTitle->getBaseText()
2674 ) ) );
2675 break;
2676 case 'talkpagename':
2677 if ( $this->mTitle->canHaveTalkPage() ) {
2678 $talkPage = $this->mTitle->getTalkPage();
2679 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2680 } else {
2681 $value = '';
2682 }
2683 break;
2684 case 'talkpagenamee':
2685 if ( $this->mTitle->canHaveTalkPage() ) {
2686 $talkPage = $this->mTitle->getTalkPage();
2687 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2688 } else {
2689 $value = '';
2690 }
2691 break;
2692 case 'subjectpagename':
2693 $subjPage = $this->mTitle->getSubjectPage();
2694 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2695 break;
2696 case 'subjectpagenamee':
2697 $subjPage = $this->mTitle->getSubjectPage();
2698 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2699 break;
2700 case 'pageid': // requested in T25427
2701 $pageid = $this->getTitle()->getArticleID();
2702 if ( $pageid == 0 ) {
2703 # 0 means the page doesn't exist in the database,
2704 # which means the user is previewing a new page.
2705 # The vary-revision flag must be set, because the magic word
2706 # will have a different value once the page is saved.
2707 $this->mOutput->setFlag( 'vary-revision' );
2708 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2709 }
2710 $value = $pageid ?: null;
2711 break;
2712 case 'revisionid':
2713 # Let the edit saving system know we should parse the page
2714 # *after* a revision ID has been assigned.
2715 $this->mOutput->setFlag( 'vary-revision-id' );
2716 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2717 $value = $this->mRevisionId;
2718
2719 if ( !$value ) {
2720 $rev = $this->getRevisionObject();
2721 if ( $rev ) {
2722 $value = $rev->getId();
2723 }
2724 }
2725
2726 if ( !$value ) {
2727 $value = $this->mOptions->getSpeculativeRevId();
2728 if ( $value ) {
2729 $this->mOutput->setSpeculativeRevIdUsed( $value );
2730 }
2731 }
2732 break;
2733 case 'revisionday':
2734 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2735 break;
2736 case 'revisionday2':
2737 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2738 break;
2739 case 'revisionmonth':
2740 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2741 break;
2742 case 'revisionmonth1':
2743 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2744 break;
2745 case 'revisionyear':
2746 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2747 break;
2748 case 'revisiontimestamp':
2749 # Let the edit saving system know we should parse the page
2750 # *after* a revision ID has been assigned. This is for null edits.
2751 $this->mOutput->setFlag( 'vary-revision' );
2752 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2753 $value = $this->getRevisionTimestamp();
2754 break;
2755 case 'revisionuser':
2756 # Let the edit saving system know we should parse the page
2757 # *after* a revision ID has been assigned for null edits.
2758 $this->mOutput->setFlag( 'vary-user' );
2759 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2760 $value = $this->getRevisionUser();
2761 break;
2762 case 'revisionsize':
2763 $value = $this->getRevisionSize();
2764 break;
2765 case 'namespace':
2766 $value = str_replace( '_', ' ',
2767 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2768 break;
2769 case 'namespacee':
2770 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2771 break;
2772 case 'namespacenumber':
2773 $value = $this->mTitle->getNamespace();
2774 break;
2775 case 'talkspace':
2776 $value = $this->mTitle->canHaveTalkPage()
2777 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2778 : '';
2779 break;
2780 case 'talkspacee':
2781 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2782 break;
2783 case 'subjectspace':
2784 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2785 break;
2786 case 'subjectspacee':
2787 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2788 break;
2789 case 'currentdayname':
2790 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2791 break;
2792 case 'currentyear':
2793 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2794 break;
2795 case 'currenttime':
2796 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2797 break;
2798 case 'currenthour':
2799 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2800 break;
2801 case 'currentweek':
2802 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2803 # int to remove the padding
2804 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2805 break;
2806 case 'currentdow':
2807 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2808 break;
2809 case 'localdayname':
2810 $value = $pageLang->getWeekdayName(
2811 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2812 );
2813 break;
2814 case 'localyear':
2815 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2816 break;
2817 case 'localtime':
2818 $value = $pageLang->time(
2819 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2820 false,
2821 false
2822 );
2823 break;
2824 case 'localhour':
2825 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2826 break;
2827 case 'localweek':
2828 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2829 # int to remove the padding
2830 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2831 break;
2832 case 'localdow':
2833 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2834 break;
2835 case 'numberofarticles':
2836 $value = $pageLang->formatNum( SiteStats::articles() );
2837 break;
2838 case 'numberoffiles':
2839 $value = $pageLang->formatNum( SiteStats::images() );
2840 break;
2841 case 'numberofusers':
2842 $value = $pageLang->formatNum( SiteStats::users() );
2843 break;
2844 case 'numberofactiveusers':
2845 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2846 break;
2847 case 'numberofpages':
2848 $value = $pageLang->formatNum( SiteStats::pages() );
2849 break;
2850 case 'numberofadmins':
2851 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2852 break;
2853 case 'numberofedits':
2854 $value = $pageLang->formatNum( SiteStats::edits() );
2855 break;
2856 case 'currenttimestamp':
2857 $value = wfTimestamp( TS_MW, $ts );
2858 break;
2859 case 'localtimestamp':
2860 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2861 break;
2862 case 'currentversion':
2863 $value = SpecialVersion::getVersion();
2864 break;
2865 case 'articlepath':
2866 return $this->siteConfig->get( 'ArticlePath' );
2867 case 'sitename':
2868 return $this->siteConfig->get( 'Sitename' );
2869 case 'server':
2870 return $this->siteConfig->get( 'Server' );
2871 case 'servername':
2872 return $this->siteConfig->get( 'ServerName' );
2873 case 'scriptpath':
2874 return $this->siteConfig->get( 'ScriptPath' );
2875 case 'stylepath':
2876 return $this->siteConfig->get( 'StylePath' );
2877 case 'directionmark':
2878 return $pageLang->getDirMark();
2879 case 'contentlanguage':
2880 return $this->siteConfig->get( 'LanguageCode' );
2881 case 'pagelanguage':
2882 $value = $pageLang->getCode();
2883 break;
2884 case 'cascadingsources':
2885 $value = CoreParserFunctions::cascadingsources( $this );
2886 break;
2887 default:
2888 $ret = null;
2889 Hooks::run(
2890 'ParserGetVariableValueSwitch',
2891 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2892 );
2893
2894 return $ret;
2895 }
2896
2897 if ( $index ) {
2898 $this->mVarCache[$index] = $value;
2899 }
2900
2901 return $value;
2902 }
2903
2904 /**
2905 * @param int $start
2906 * @param int $len
2907 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2908 * @param string $variable Parser variable name
2909 * @return string
2910 */
2911 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2912 # Get the timezone-adjusted timestamp to be used for this revision
2913 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2914 # Possibly set vary-revision if there is not yet an associated revision
2915 if ( !$this->getRevisionObject() ) {
2916 # Get the timezone-adjusted timestamp $mtts seconds in the future
2917 $resThen = substr(
2918 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2919 $start,
2920 $len
2921 );
2922
2923 if ( $resNow !== $resThen ) {
2924 # Let the edit saving system know we should parse the page
2925 # *after* a revision ID has been assigned. This is for null edits.
2926 $this->mOutput->setFlag( 'vary-revision' );
2927 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2928 }
2929 }
2930
2931 return $resNow;
2932 }
2933
2934 /**
2935 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2936 *
2937 * @private
2938 */
2939 public function initialiseVariables() {
2940 $variableIDs = $this->magicWordFactory->getVariableIDs();
2941 $substIDs = $this->magicWordFactory->getSubstIDs();
2942
2943 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
2944 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
2945 }
2946
2947 /**
2948 * Preprocess some wikitext and return the document tree.
2949 * This is the ghost of replace_variables().
2950 *
2951 * @param string $text The text to parse
2952 * @param int $flags Bitwise combination of:
2953 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2954 * included. Default is to assume a direct page view.
2955 *
2956 * The generated DOM tree must depend only on the input text and the flags.
2957 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2958 *
2959 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2960 * change in the DOM tree for a given text, must be passed through the section identifier
2961 * in the section edit link and thus back to extractSections().
2962 *
2963 * The output of this function is currently only cached in process memory, but a persistent
2964 * cache may be implemented at a later date which takes further advantage of these strict
2965 * dependency requirements.
2966 *
2967 * @return PPNode
2968 */
2969 public function preprocessToDom( $text, $flags = 0 ) {
2970 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2971 return $dom;
2972 }
2973
2974 /**
2975 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2976 *
2977 * @param string $s
2978 *
2979 * @return array
2980 */
2981 public static function splitWhitespace( $s ) {
2982 $ltrimmed = ltrim( $s );
2983 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2984 $trimmed = rtrim( $ltrimmed );
2985 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2986 if ( $diff > 0 ) {
2987 $w2 = substr( $ltrimmed, -$diff );
2988 } else {
2989 $w2 = '';
2990 }
2991 return [ $w1, $trimmed, $w2 ];
2992 }
2993
2994 /**
2995 * Replace magic variables, templates, and template arguments
2996 * with the appropriate text. Templates are substituted recursively,
2997 * taking care to avoid infinite loops.
2998 *
2999 * Note that the substitution depends on value of $mOutputType:
3000 * self::OT_WIKI: only {{subst:}} templates
3001 * self::OT_PREPROCESS: templates but not extension tags
3002 * self::OT_HTML: all templates and extension tags
3003 *
3004 * @param string $text The text to transform
3005 * @param bool|PPFrame $frame Object describing the arguments passed to the
3006 * template. Arguments may also be provided as an associative array, as
3007 * was the usual case before MW1.12. Providing arguments this way may be
3008 * useful for extensions wishing to perform variable replacement
3009 * explicitly.
3010 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3011 * double-brace expansion.
3012 * @return string
3013 */
3014 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3015 # Is there any text? Also, Prevent too big inclusions!
3016 $textSize = strlen( $text );
3017 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3018 return $text;
3019 }
3020
3021 if ( $frame === false ) {
3022 $frame = $this->getPreprocessor()->newFrame();
3023 } elseif ( !( $frame instanceof PPFrame ) ) {
3024 wfDebug( __METHOD__ . " called using plain parameters instead of "
3025 . "a PPFrame instance. Creating custom frame.\n" );
3026 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3027 }
3028
3029 $dom = $this->preprocessToDom( $text );
3030 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3031 $text = $frame->expand( $dom, $flags );
3032
3033 return $text;
3034 }
3035
3036 /**
3037 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3038 *
3039 * @param array $args
3040 *
3041 * @return array
3042 */
3043 public static function createAssocArgs( $args ) {
3044 $assocArgs = [];
3045 $index = 1;
3046 foreach ( $args as $arg ) {
3047 $eqpos = strpos( $arg, '=' );
3048 if ( $eqpos === false ) {
3049 $assocArgs[$index++] = $arg;
3050 } else {
3051 $name = trim( substr( $arg, 0, $eqpos ) );
3052 $value = trim( substr( $arg, $eqpos + 1 ) );
3053 if ( $value === false ) {
3054 $value = '';
3055 }
3056 if ( $name !== false ) {
3057 $assocArgs[$name] = $value;
3058 }
3059 }
3060 }
3061
3062 return $assocArgs;
3063 }
3064
3065 /**
3066 * Warn the user when a parser limitation is reached
3067 * Will warn at most once the user per limitation type
3068 *
3069 * The results are shown during preview and run through the Parser (See EditPage.php)
3070 *
3071 * @param string $limitationType Should be one of:
3072 * 'expensive-parserfunction' (corresponding messages:
3073 * 'expensive-parserfunction-warning',
3074 * 'expensive-parserfunction-category')
3075 * 'post-expand-template-argument' (corresponding messages:
3076 * 'post-expand-template-argument-warning',
3077 * 'post-expand-template-argument-category')
3078 * 'post-expand-template-inclusion' (corresponding messages:
3079 * 'post-expand-template-inclusion-warning',
3080 * 'post-expand-template-inclusion-category')
3081 * 'node-count-exceeded' (corresponding messages:
3082 * 'node-count-exceeded-warning',
3083 * 'node-count-exceeded-category')
3084 * 'expansion-depth-exceeded' (corresponding messages:
3085 * 'expansion-depth-exceeded-warning',
3086 * 'expansion-depth-exceeded-category')
3087 * @param string|int|null $current Current value
3088 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3089 * exceeded, provide the values (optional)
3090 */
3091 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3092 # does no harm if $current and $max are present but are unnecessary for the message
3093 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3094 # only during preview, and that would split the parser cache unnecessarily.
3095 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3096 ->text();
3097 $this->mOutput->addWarning( $warning );
3098 $this->addTrackingCategory( "$limitationType-category" );
3099 }
3100
3101 /**
3102 * Return the text of a template, after recursively
3103 * replacing any variables or templates within the template.
3104 *
3105 * @param array $piece The parts of the template
3106 * $piece['title']: the title, i.e. the part before the |
3107 * $piece['parts']: the parameter array
3108 * $piece['lineStart']: whether the brace was at the start of a line
3109 * @param PPFrame $frame The current frame, contains template arguments
3110 * @throws Exception
3111 * @return string The text of the template
3112 */
3113 public function braceSubstitution( $piece, $frame ) {
3114 // Flags
3115
3116 // $text has been filled
3117 $found = false;
3118 // wiki markup in $text should be escaped
3119 $nowiki = false;
3120 // $text is HTML, armour it against wikitext transformation
3121 $isHTML = false;
3122 // Force interwiki transclusion to be done in raw mode not rendered
3123 $forceRawInterwiki = false;
3124 // $text is a DOM node needing expansion in a child frame
3125 $isChildObj = false;
3126 // $text is a DOM node needing expansion in the current frame
3127 $isLocalObj = false;
3128
3129 # Title object, where $text came from
3130 $title = false;
3131
3132 # $part1 is the bit before the first |, and must contain only title characters.
3133 # Various prefixes will be stripped from it later.
3134 $titleWithSpaces = $frame->expand( $piece['title'] );
3135 $part1 = trim( $titleWithSpaces );
3136 $titleText = false;
3137
3138 # Original title text preserved for various purposes
3139 $originalTitle = $part1;
3140
3141 # $args is a list of argument nodes, starting from index 0, not including $part1
3142 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3143 # below won't work b/c this $args isn't an object
3144 $args = ( $piece['parts'] == null ) ? [] : $piece['parts'];
3145
3146 $profileSection = null; // profile templates
3147
3148 # SUBST
3149 if ( !$found ) {
3150 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3151
3152 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3153 # Decide whether to expand template or keep wikitext as-is.
3154 if ( $this->ot['wiki'] ) {
3155 if ( $substMatch === false ) {
3156 $literal = true; # literal when in PST with no prefix
3157 } else {
3158 $literal = false; # expand when in PST with subst: or safesubst:
3159 }
3160 } else {
3161 if ( $substMatch == 'subst' ) {
3162 $literal = true; # literal when not in PST with plain subst:
3163 } else {
3164 $literal = false; # expand when not in PST with safesubst: or no prefix
3165 }
3166 }
3167 if ( $literal ) {
3168 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3169 $isLocalObj = true;
3170 $found = true;
3171 }
3172 }
3173
3174 # Variables
3175 if ( !$found && $args->getLength() == 0 ) {
3176 $id = $this->mVariables->matchStartToEnd( $part1 );
3177 if ( $id !== false ) {
3178 $text = $this->getVariableValue( $id, $frame );
3179 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3180 $this->mOutput->updateCacheExpiry(
3181 $this->magicWordFactory->getCacheTTL( $id ) );
3182 }
3183 $found = true;
3184 }
3185 }
3186
3187 # MSG, MSGNW and RAW
3188 if ( !$found ) {
3189 # Check for MSGNW:
3190 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3191 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3192 $nowiki = true;
3193 } else {
3194 # Remove obsolete MSG:
3195 $mwMsg = $this->magicWordFactory->get( 'msg' );
3196 $mwMsg->matchStartAndRemove( $part1 );
3197 }
3198
3199 # Check for RAW:
3200 $mwRaw = $this->magicWordFactory->get( 'raw' );
3201 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3202 $forceRawInterwiki = true;
3203 }
3204 }
3205
3206 # Parser functions
3207 if ( !$found ) {
3208 $colonPos = strpos( $part1, ':' );
3209 if ( $colonPos !== false ) {
3210 $func = substr( $part1, 0, $colonPos );
3211 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3212 $argsLength = $args->getLength();
3213 for ( $i = 0; $i < $argsLength; $i++ ) {
3214 $funcArgs[] = $args->item( $i );
3215 }
3216
3217 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3218
3219 // Extract any forwarded flags
3220 if ( isset( $result['title'] ) ) {
3221 $title = $result['title'];
3222 }
3223 if ( isset( $result['found'] ) ) {
3224 $found = $result['found'];
3225 }
3226 if ( array_key_exists( 'text', $result ) ) {
3227 // a string or null
3228 $text = $result['text'];
3229 }
3230 if ( isset( $result['nowiki'] ) ) {
3231 $nowiki = $result['nowiki'];
3232 }
3233 if ( isset( $result['isHTML'] ) ) {
3234 $isHTML = $result['isHTML'];
3235 }
3236 if ( isset( $result['forceRawInterwiki'] ) ) {
3237 $forceRawInterwiki = $result['forceRawInterwiki'];
3238 }
3239 if ( isset( $result['isChildObj'] ) ) {
3240 $isChildObj = $result['isChildObj'];
3241 }
3242 if ( isset( $result['isLocalObj'] ) ) {
3243 $isLocalObj = $result['isLocalObj'];
3244 }
3245 }
3246 }
3247
3248 # Finish mangling title and then check for loops.
3249 # Set $title to a Title object and $titleText to the PDBK
3250 if ( !$found ) {
3251 $ns = NS_TEMPLATE;
3252 # Split the title into page and subpage
3253 $subpage = '';
3254 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3255 if ( $part1 !== $relative ) {
3256 $part1 = $relative;
3257 $ns = $this->mTitle->getNamespace();
3258 }
3259 $title = Title::newFromText( $part1, $ns );
3260 if ( $title ) {
3261 $titleText = $title->getPrefixedText();
3262 # Check for language variants if the template is not found
3263 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3264 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3265 }
3266 # Do recursion depth check
3267 $limit = $this->mOptions->getMaxTemplateDepth();
3268 if ( $frame->depth >= $limit ) {
3269 $found = true;
3270 $text = '<span class="error">'
3271 . wfMessage( 'parser-template-recursion-depth-warning' )
3272 ->numParams( $limit )->inContentLanguage()->text()
3273 . '</span>';
3274 }
3275 }
3276 }
3277
3278 # Load from database
3279 if ( !$found && $title ) {
3280 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3281 if ( !$title->isExternal() ) {
3282 if ( $title->isSpecialPage()
3283 && $this->mOptions->getAllowSpecialInclusion()
3284 && $this->ot['html']
3285 ) {
3286 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3287 // Pass the template arguments as URL parameters.
3288 // "uselang" will have no effect since the Language object
3289 // is forced to the one defined in ParserOptions.
3290 $pageArgs = [];
3291 $argsLength = $args->getLength();
3292 for ( $i = 0; $i < $argsLength; $i++ ) {
3293 $bits = $args->item( $i )->splitArg();
3294 if ( strval( $bits['index'] ) === '' ) {
3295 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3296 $value = trim( $frame->expand( $bits['value'] ) );
3297 $pageArgs[$name] = $value;
3298 }
3299 }
3300
3301 // Create a new context to execute the special page
3302 $context = new RequestContext;
3303 $context->setTitle( $title );
3304 $context->setRequest( new FauxRequest( $pageArgs ) );
3305 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3306 $context->setUser( $this->getUser() );
3307 } else {
3308 // If this page is cached, then we better not be per user.
3309 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3310 }
3311 $context->setLanguage( $this->mOptions->getUserLangObj() );
3312 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3313 if ( $ret ) {
3314 $text = $context->getOutput()->getHTML();
3315 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3316 $found = true;
3317 $isHTML = true;
3318 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3319 $this->mOutput->updateRuntimeAdaptiveExpiry(
3320 $specialPage->maxIncludeCacheTime()
3321 );
3322 }
3323 }
3324 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3325 $found = false; # access denied
3326 wfDebug( __METHOD__ . ": template inclusion denied for " .
3327 $title->getPrefixedDBkey() . "\n" );
3328 } else {
3329 list( $text, $title ) = $this->getTemplateDom( $title );
3330 if ( $text !== false ) {
3331 $found = true;
3332 $isChildObj = true;
3333 }
3334 }
3335
3336 # If the title is valid but undisplayable, make a link to it
3337 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3338 $text = "[[:$titleText]]";
3339 $found = true;
3340 }
3341 } elseif ( $title->isTrans() ) {
3342 # Interwiki transclusion
3343 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3344 $text = $this->interwikiTransclude( $title, 'render' );
3345 $isHTML = true;
3346 } else {
3347 $text = $this->interwikiTransclude( $title, 'raw' );
3348 # Preprocess it like a template
3349 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3350 $isChildObj = true;
3351 }
3352 $found = true;
3353 }
3354
3355 # Do infinite loop check
3356 # This has to be done after redirect resolution to avoid infinite loops via redirects
3357 if ( !$frame->loopCheck( $title ) ) {
3358 $found = true;
3359 $text = '<span class="error">'
3360 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3361 . '</span>';
3362 $this->addTrackingCategory( 'template-loop-category' );
3363 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3364 wfEscapeWikiText( $titleText ) )->text() );
3365 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3366 }
3367 }
3368
3369 # If we haven't found text to substitute by now, we're done
3370 # Recover the source wikitext and return it
3371 if ( !$found ) {
3372 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3373 if ( $profileSection ) {
3374 $this->mProfiler->scopedProfileOut( $profileSection );
3375 }
3376 return [ 'object' => $text ];
3377 }
3378
3379 # Expand DOM-style return values in a child frame
3380 if ( $isChildObj ) {
3381 # Clean up argument array
3382 $newFrame = $frame->newChild( $args, $title );
3383
3384 if ( $nowiki ) {
3385 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3386 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3387 # Expansion is eligible for the empty-frame cache
3388 $text = $newFrame->cachedExpand( $titleText, $text );
3389 } else {
3390 # Uncached expansion
3391 $text = $newFrame->expand( $text );
3392 }
3393 }
3394 if ( $isLocalObj && $nowiki ) {
3395 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3396 $isLocalObj = false;
3397 }
3398
3399 if ( $profileSection ) {
3400 $this->mProfiler->scopedProfileOut( $profileSection );
3401 }
3402
3403 # Replace raw HTML by a placeholder
3404 if ( $isHTML ) {
3405 $text = $this->insertStripItem( $text );
3406 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3407 # Escape nowiki-style return values
3408 $text = wfEscapeWikiText( $text );
3409 } elseif ( is_string( $text )
3410 && !$piece['lineStart']
3411 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3412 ) {
3413 # T2529: if the template begins with a table or block-level
3414 # element, it should be treated as beginning a new line.
3415 # This behavior is somewhat controversial.
3416 $text = "\n" . $text;
3417 }
3418
3419 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3420 # Error, oversize inclusion
3421 if ( $titleText !== false ) {
3422 # Make a working, properly escaped link if possible (T25588)
3423 $text = "[[:$titleText]]";
3424 } else {
3425 # This will probably not be a working link, but at least it may
3426 # provide some hint of where the problem is
3427 preg_replace( '/^:/', '', $originalTitle );
3428 $text = "[[:$originalTitle]]";
3429 }
3430 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3431 . 'post-expand include size too large -->' );
3432 $this->limitationWarn( 'post-expand-template-inclusion' );
3433 }
3434
3435 if ( $isLocalObj ) {
3436 $ret = [ 'object' => $text ];
3437 } else {
3438 $ret = [ 'text' => $text ];
3439 }
3440
3441 return $ret;
3442 }
3443
3444 /**
3445 * Call a parser function and return an array with text and flags.
3446 *
3447 * The returned array will always contain a boolean 'found', indicating
3448 * whether the parser function was found or not. It may also contain the
3449 * following:
3450 * text: string|object, resulting wikitext or PP DOM object
3451 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3452 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3453 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3454 * nowiki: bool, wiki markup in $text should be escaped
3455 *
3456 * @since 1.21
3457 * @param PPFrame $frame The current frame, contains template arguments
3458 * @param string $function Function name
3459 * @param array $args Arguments to the function
3460 * @throws MWException
3461 * @return array
3462 */
3463 public function callParserFunction( $frame, $function, array $args = [] ) {
3464 # Case sensitive functions
3465 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3466 $function = $this->mFunctionSynonyms[1][$function];
3467 } else {
3468 # Case insensitive functions
3469 $function = $this->contLang->lc( $function );
3470 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3471 $function = $this->mFunctionSynonyms[0][$function];
3472 } else {
3473 return [ 'found' => false ];
3474 }
3475 }
3476
3477 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3478
3479 // Avoid PHP 7.1 warning from passing $this by reference
3480 $parser = $this;
3481
3482 $allArgs = [ &$parser ];
3483 if ( $flags & self::SFH_OBJECT_ARGS ) {
3484 # Convert arguments to PPNodes and collect for appending to $allArgs
3485 $funcArgs = [];
3486 foreach ( $args as $k => $v ) {
3487 if ( $v instanceof PPNode || $k === 0 ) {
3488 $funcArgs[] = $v;
3489 } else {
3490 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3491 }
3492 }
3493
3494 # Add a frame parameter, and pass the arguments as an array
3495 $allArgs[] = $frame;
3496 $allArgs[] = $funcArgs;
3497 } else {
3498 # Convert arguments to plain text and append to $allArgs
3499 foreach ( $args as $k => $v ) {
3500 if ( $v instanceof PPNode ) {
3501 $allArgs[] = trim( $frame->expand( $v ) );
3502 } elseif ( is_int( $k ) && $k >= 0 ) {
3503 $allArgs[] = trim( $v );
3504 } else {
3505 $allArgs[] = trim( "$k=$v" );
3506 }
3507 }
3508 }
3509
3510 $result = $callback( ...$allArgs );
3511
3512 # The interface for function hooks allows them to return a wikitext
3513 # string or an array containing the string and any flags. This mungs
3514 # things around to match what this method should return.
3515 if ( !is_array( $result ) ) {
3516 $result = [
3517 'found' => true,
3518 'text' => $result,
3519 ];
3520 } else {
3521 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3522 $result['text'] = $result[0];
3523 }
3524 unset( $result[0] );
3525 $result += [
3526 'found' => true,
3527 ];
3528 }
3529
3530 $noparse = true;
3531 $preprocessFlags = 0;
3532 if ( isset( $result['noparse'] ) ) {
3533 $noparse = $result['noparse'];
3534 }
3535 if ( isset( $result['preprocessFlags'] ) ) {
3536 $preprocessFlags = $result['preprocessFlags'];
3537 }
3538
3539 if ( !$noparse ) {
3540 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3541 $result['isChildObj'] = true;
3542 }
3543
3544 return $result;
3545 }
3546
3547 /**
3548 * Get the semi-parsed DOM representation of a template with a given title,
3549 * and its redirect destination title. Cached.
3550 *
3551 * @param Title $title
3552 *
3553 * @return array
3554 */
3555 public function getTemplateDom( $title ) {
3556 $cacheTitle = $title;
3557 $titleText = $title->getPrefixedDBkey();
3558
3559 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3560 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3561 $title = Title::makeTitle( $ns, $dbk );
3562 $titleText = $title->getPrefixedDBkey();
3563 }
3564 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3565 return [ $this->mTplDomCache[$titleText], $title ];
3566 }
3567
3568 # Cache miss, go to the database
3569 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3570
3571 if ( $text === false ) {
3572 $this->mTplDomCache[$titleText] = false;
3573 return [ false, $title ];
3574 }
3575
3576 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3577 $this->mTplDomCache[$titleText] = $dom;
3578
3579 if ( !$title->equals( $cacheTitle ) ) {
3580 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3581 [ $title->getNamespace(), $title->getDBkey() ];
3582 }
3583
3584 return [ $dom, $title ];
3585 }
3586
3587 /**
3588 * Fetch the current revision of a given title. Note that the revision
3589 * (and even the title) may not exist in the database, so everything
3590 * contributing to the output of the parser should use this method
3591 * where possible, rather than getting the revisions themselves. This
3592 * method also caches its results, so using it benefits performance.
3593 *
3594 * @since 1.24
3595 * @param Title $title
3596 * @return Revision
3597 */
3598 public function fetchCurrentRevisionOfTitle( $title ) {
3599 $cacheKey = $title->getPrefixedDBkey();
3600 if ( !$this->currentRevisionCache ) {
3601 $this->currentRevisionCache = new MapCacheLRU( 100 );
3602 }
3603 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3604 $this->currentRevisionCache->set( $cacheKey,
3605 // Defaults to Parser::statelessFetchRevision()
3606 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3607 );
3608 }
3609 return $this->currentRevisionCache->get( $cacheKey );
3610 }
3611
3612 /**
3613 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3614 * without passing them on to it.
3615 *
3616 * @since 1.24
3617 * @param Title $title
3618 * @param Parser|bool $parser
3619 * @return Revision|bool False if missing
3620 */
3621 public static function statelessFetchRevision( Title $title, $parser = false ) {
3622 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3623
3624 return $rev;
3625 }
3626
3627 /**
3628 * Fetch the unparsed text of a template and register a reference to it.
3629 * @param Title $title
3630 * @return array ( string or false, Title )
3631 */
3632 public function fetchTemplateAndTitle( $title ) {
3633 // Defaults to Parser::statelessFetchTemplate()
3634 $templateCb = $this->mOptions->getTemplateCallback();
3635 $stuff = call_user_func( $templateCb, $title, $this );
3636 // We use U+007F DELETE to distinguish strip markers from regular text.
3637 $text = $stuff['text'];
3638 if ( is_string( $stuff['text'] ) ) {
3639 $text = strtr( $text, "\x7f", "?" );
3640 }
3641 $finalTitle = $stuff['finalTitle'] ?? $title;
3642 if ( isset( $stuff['deps'] ) ) {
3643 foreach ( $stuff['deps'] as $dep ) {
3644 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3645 if ( $dep['title']->equals( $this->getTitle() ) ) {
3646 // If we transclude ourselves, the final result
3647 // will change based on the new version of the page
3648 $this->mOutput->setFlag( 'vary-revision' );
3649 }
3650 }
3651 }
3652 return [ $text, $finalTitle ];
3653 }
3654
3655 /**
3656 * Fetch the unparsed text of a template and register a reference to it.
3657 * @param Title $title
3658 * @return string|bool
3659 */
3660 public function fetchTemplate( $title ) {
3661 return $this->fetchTemplateAndTitle( $title )[0];
3662 }
3663
3664 /**
3665 * Static function to get a template
3666 * Can be overridden via ParserOptions::setTemplateCallback().
3667 *
3668 * @param Title $title
3669 * @param bool|Parser $parser
3670 *
3671 * @return array
3672 */
3673 public static function statelessFetchTemplate( $title, $parser = false ) {
3674 $text = $skip = false;
3675 $finalTitle = $title;
3676 $deps = [];
3677
3678 # Loop to fetch the article, with up to 1 redirect
3679 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3680 # Give extensions a chance to select the revision instead
3681 $id = false; # Assume current
3682 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3683 [ $parser, $title, &$skip, &$id ] );
3684
3685 if ( $skip ) {
3686 $text = false;
3687 $deps[] = [
3688 'title' => $title,
3689 'page_id' => $title->getArticleID(),
3690 'rev_id' => null
3691 ];
3692 break;
3693 }
3694 # Get the revision
3695 if ( $id ) {
3696 $rev = Revision::newFromId( $id );
3697 } elseif ( $parser ) {
3698 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3699 } else {
3700 $rev = Revision::newFromTitle( $title );
3701 }
3702 $rev_id = $rev ? $rev->getId() : 0;
3703 # If there is no current revision, there is no page
3704 if ( $id === false && !$rev ) {
3705 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3706 $linkCache->addBadLinkObj( $title );
3707 }
3708
3709 $deps[] = [
3710 'title' => $title,
3711 'page_id' => $title->getArticleID(),
3712 'rev_id' => $rev_id ];
3713 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3714 # We fetched a rev from a different title; register it too...
3715 $deps[] = [
3716 'title' => $rev->getTitle(),
3717 'page_id' => $rev->getPage(),
3718 'rev_id' => $rev_id ];
3719 }
3720
3721 if ( $rev ) {
3722 $content = $rev->getContent();
3723 $text = $content ? $content->getWikitextForTransclusion() : null;
3724
3725 Hooks::run( 'ParserFetchTemplate',
3726 [ $parser, $title, $rev, &$text, &$deps ] );
3727
3728 if ( $text === false || $text === null ) {
3729 $text = false;
3730 break;
3731 }
3732 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3733 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
3734 lcfirst( $title->getText() ) )->inContentLanguage();
3735 if ( !$message->exists() ) {
3736 $text = false;
3737 break;
3738 }
3739 $content = $message->content();
3740 $text = $message->plain();
3741 } else {
3742 break;
3743 }
3744 if ( !$content ) {
3745 break;
3746 }
3747 # Redirect?
3748 $finalTitle = $title;
3749 $title = $content->getRedirectTarget();
3750 }
3751 return [
3752 'text' => $text,
3753 'finalTitle' => $finalTitle,
3754 'deps' => $deps ];
3755 }
3756
3757 /**
3758 * Fetch a file and its title and register a reference to it.
3759 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3760 * @param Title $title
3761 * @param array $options Array of options to RepoGroup::findFile
3762 * @return File|bool
3763 * @deprecated since 1.32, use fetchFileAndTitle instead
3764 */
3765 public function fetchFile( $title, $options = [] ) {
3766 wfDeprecated( __METHOD__, '1.32' );
3767 return $this->fetchFileAndTitle( $title, $options )[0];
3768 }
3769
3770 /**
3771 * Fetch a file and its title and register a reference to it.
3772 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3773 * @param Title $title
3774 * @param array $options Array of options to RepoGroup::findFile
3775 * @return array ( File or false, Title of file )
3776 */
3777 public function fetchFileAndTitle( $title, $options = [] ) {
3778 $file = $this->fetchFileNoRegister( $title, $options );
3779
3780 $time = $file ? $file->getTimestamp() : false;
3781 $sha1 = $file ? $file->getSha1() : false;
3782 # Register the file as a dependency...
3783 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3784 if ( $file && !$title->equals( $file->getTitle() ) ) {
3785 # Update fetched file title
3786 $title = $file->getTitle();
3787 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3788 }
3789 return [ $file, $title ];
3790 }
3791
3792 /**
3793 * Helper function for fetchFileAndTitle.
3794 *
3795 * Also useful if you need to fetch a file but not use it yet,
3796 * for example to get the file's handler.
3797 *
3798 * @param Title $title
3799 * @param array $options Array of options to RepoGroup::findFile
3800 * @return File|bool
3801 */
3802 protected function fetchFileNoRegister( $title, $options = [] ) {
3803 if ( isset( $options['broken'] ) ) {
3804 $file = false; // broken thumbnail forced by hook
3805 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3806 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3807 } else { // get by (name,timestamp)
3808 $file = wfFindFile( $title, $options );
3809 }
3810 return $file;
3811 }
3812
3813 /**
3814 * Transclude an interwiki link.
3815 *
3816 * @param Title $title
3817 * @param string $action Usually one of (raw, render)
3818 *
3819 * @return string
3820 */
3821 public function interwikiTransclude( $title, $action ) {
3822 if ( !$this->siteConfig->get( 'EnableScaryTranscluding' ) ) {
3823 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3824 }
3825
3826 $url = $title->getFullURL( [ 'action' => $action ] );
3827 if ( strlen( $url ) > 1024 ) {
3828 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3829 }
3830
3831 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3832
3833 $fname = __METHOD__;
3834 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3835
3836 $data = $cache->getWithSetCallback(
3837 $cache->makeGlobalKey(
3838 'interwiki-transclude',
3839 ( $wikiId !== false ) ? $wikiId : 'external',
3840 sha1( $url )
3841 ),
3842 $this->siteConfig->get( 'TranscludeCacheExpiry' ),
3843 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3844 $req = MWHttpRequest::factory( $url, [], $fname );
3845
3846 $status = $req->execute(); // Status object
3847 if ( !$status->isOK() ) {
3848 $ttl = $cache::TTL_UNCACHEABLE;
3849 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3850 $ttl = min( $cache::TTL_LAGGED, $ttl );
3851 }
3852
3853 return [
3854 'text' => $status->isOK() ? $req->getContent() : null,
3855 'code' => $req->getStatus()
3856 ];
3857 },
3858 [
3859 'checkKeys' => ( $wikiId !== false )
3860 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3861 : [],
3862 'pcGroup' => 'interwiki-transclude:5',
3863 'pcTTL' => $cache::TTL_PROC_LONG
3864 ]
3865 );
3866
3867 if ( is_string( $data['text'] ) ) {
3868 $text = $data['text'];
3869 } elseif ( $data['code'] != 200 ) {
3870 // Though we failed to fetch the content, this status is useless.
3871 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3872 ->params( $url, $data['code'] )->inContentLanguage()->text();
3873 } else {
3874 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3875 }
3876
3877 return $text;
3878 }
3879
3880 /**
3881 * Triple brace replacement -- used for template arguments
3882 * @private
3883 *
3884 * @param array $piece
3885 * @param PPFrame $frame
3886 *
3887 * @return array
3888 */
3889 public function argSubstitution( $piece, $frame ) {
3890 $error = false;
3891 $parts = $piece['parts'];
3892 $nameWithSpaces = $frame->expand( $piece['title'] );
3893 $argName = trim( $nameWithSpaces );
3894 $object = false;
3895 $text = $frame->getArgument( $argName );
3896 if ( $text === false && $parts->getLength() > 0
3897 && ( $this->ot['html']
3898 || $this->ot['pre']
3899 || ( $this->ot['wiki'] && $frame->isTemplate() )
3900 )
3901 ) {
3902 # No match in frame, use the supplied default
3903 $object = $parts->item( 0 )->getChildren();
3904 }
3905 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3906 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3907 $this->limitationWarn( 'post-expand-template-argument' );
3908 }
3909
3910 if ( $text === false && $object === false ) {
3911 # No match anywhere
3912 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3913 }
3914 if ( $error !== false ) {
3915 $text .= $error;
3916 }
3917 if ( $object !== false ) {
3918 $ret = [ 'object' => $object ];
3919 } else {
3920 $ret = [ 'text' => $text ];
3921 }
3922
3923 return $ret;
3924 }
3925
3926 /**
3927 * Return the text to be used for a given extension tag.
3928 * This is the ghost of strip().
3929 *
3930 * @param array $params Associative array of parameters:
3931 * name PPNode for the tag name
3932 * attr PPNode for unparsed text where tag attributes are thought to be
3933 * attributes Optional associative array of parsed attributes
3934 * inner Contents of extension element
3935 * noClose Original text did not have a close tag
3936 * @param PPFrame $frame
3937 *
3938 * @throws MWException
3939 * @return string
3940 */
3941 public function extensionSubstitution( $params, $frame ) {
3942 static $errorStr = '<span class="error">';
3943 static $errorLen = 20;
3944
3945 $name = $frame->expand( $params['name'] );
3946 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3947 // Probably expansion depth or node count exceeded. Just punt the
3948 // error up.
3949 return $name;
3950 }
3951
3952 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3953 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3954 // See above
3955 return $attrText;
3956 }
3957
3958 // We can't safely check if the expansion for $content resulted in an
3959 // error, because the content could happen to be the error string
3960 // (T149622).
3961 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3962
3963 $marker = self::MARKER_PREFIX . "-$name-"
3964 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3965
3966 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3967 ( $this->ot['html'] || $this->ot['pre'] );
3968 if ( $isFunctionTag ) {
3969 $markerType = 'none';
3970 } else {
3971 $markerType = 'general';
3972 }
3973 if ( $this->ot['html'] || $isFunctionTag ) {
3974 $name = strtolower( $name );
3975 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3976 if ( isset( $params['attributes'] ) ) {
3977 $attributes += $params['attributes'];
3978 }
3979
3980 if ( isset( $this->mTagHooks[$name] ) ) {
3981 $output = call_user_func_array( $this->mTagHooks[$name],
3982 [ $content, $attributes, $this, $frame ] );
3983 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3984 list( $callback, ) = $this->mFunctionTagHooks[$name];
3985
3986 // Avoid PHP 7.1 warning from passing $this by reference
3987 $parser = $this;
3988 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3989 } else {
3990 $output = '<span class="error">Invalid tag extension name: ' .
3991 htmlspecialchars( $name ) . '</span>';
3992 }
3993
3994 if ( is_array( $output ) ) {
3995 // Extract flags
3996 $flags = $output;
3997 $output = $flags[0];
3998 if ( isset( $flags['markerType'] ) ) {
3999 $markerType = $flags['markerType'];
4000 }
4001 }
4002 } else {
4003 if ( is_null( $attrText ) ) {
4004 $attrText = '';
4005 }
4006 if ( isset( $params['attributes'] ) ) {
4007 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4008 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4009 htmlspecialchars( $attrValue ) . '"';
4010 }
4011 }
4012 if ( $content === null ) {
4013 $output = "<$name$attrText/>";
4014 } else {
4015 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4016 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4017 // See above
4018 return $close;
4019 }
4020 $output = "<$name$attrText>$content$close";
4021 }
4022 }
4023
4024 if ( $markerType === 'none' ) {
4025 return $output;
4026 } elseif ( $markerType === 'nowiki' ) {
4027 $this->mStripState->addNoWiki( $marker, $output );
4028 } elseif ( $markerType === 'general' ) {
4029 $this->mStripState->addGeneral( $marker, $output );
4030 } else {
4031 throw new MWException( __METHOD__ . ': invalid marker type' );
4032 }
4033 return $marker;
4034 }
4035
4036 /**
4037 * Increment an include size counter
4038 *
4039 * @param string $type The type of expansion
4040 * @param int $size The size of the text
4041 * @return bool False if this inclusion would take it over the maximum, true otherwise
4042 */
4043 public function incrementIncludeSize( $type, $size ) {
4044 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4045 return false;
4046 } else {
4047 $this->mIncludeSizes[$type] += $size;
4048 return true;
4049 }
4050 }
4051
4052 /**
4053 * Increment the expensive function count
4054 *
4055 * @return bool False if the limit has been exceeded
4056 */
4057 public function incrementExpensiveFunctionCount() {
4058 $this->mExpensiveFunctionCount++;
4059 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4060 }
4061
4062 /**
4063 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4064 * Fills $this->mDoubleUnderscores, returns the modified text
4065 *
4066 * @param string $text
4067 *
4068 * @return string
4069 */
4070 public function doDoubleUnderscore( $text ) {
4071 # The position of __TOC__ needs to be recorded
4072 $mw = $this->magicWordFactory->get( 'toc' );
4073 if ( $mw->match( $text ) ) {
4074 $this->mShowToc = true;
4075 $this->mForceTocPosition = true;
4076
4077 # Set a placeholder. At the end we'll fill it in with the TOC.
4078 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4079
4080 # Only keep the first one.
4081 $text = $mw->replace( '', $text );
4082 }
4083
4084 # Now match and remove the rest of them
4085 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4086 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4087
4088 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4089 $this->mOutput->mNoGallery = true;
4090 }
4091 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4092 $this->mShowToc = false;
4093 }
4094 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4095 && $this->mTitle->getNamespace() == NS_CATEGORY
4096 ) {
4097 $this->addTrackingCategory( 'hidden-category-category' );
4098 }
4099 # (T10068) Allow control over whether robots index a page.
4100 # __INDEX__ always overrides __NOINDEX__, see T16899
4101 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4102 $this->mOutput->setIndexPolicy( 'noindex' );
4103 $this->addTrackingCategory( 'noindex-category' );
4104 }
4105 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4106 $this->mOutput->setIndexPolicy( 'index' );
4107 $this->addTrackingCategory( 'index-category' );
4108 }
4109
4110 # Cache all double underscores in the database
4111 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4112 $this->mOutput->setProperty( $key, '' );
4113 }
4114
4115 return $text;
4116 }
4117
4118 /**
4119 * @see ParserOutput::addTrackingCategory()
4120 * @param string $msg Message key
4121 * @return bool Whether the addition was successful
4122 */
4123 public function addTrackingCategory( $msg ) {
4124 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4125 }
4126
4127 /**
4128 * This function accomplishes several tasks:
4129 * 1) Auto-number headings if that option is enabled
4130 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4131 * 3) Add a Table of contents on the top for users who have enabled the option
4132 * 4) Auto-anchor headings
4133 *
4134 * It loops through all headlines, collects the necessary data, then splits up the
4135 * string and re-inserts the newly formatted headlines.
4136 *
4137 * @param string $text
4138 * @param string $origText Original, untouched wikitext
4139 * @param bool $isMain
4140 * @return mixed|string
4141 * @private
4142 */
4143 public function formatHeadings( $text, $origText, $isMain = true ) {
4144 # Inhibit editsection links if requested in the page
4145 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4146 $maybeShowEditLink = false;
4147 } else {
4148 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4149 }
4150
4151 # Get all headlines for numbering them and adding funky stuff like [edit]
4152 # links - this is for later, but we need the number of headlines right now
4153 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4154 # be trimmed here since whitespace in HTML headings is significant.
4155 $matches = [];
4156 $numMatches = preg_match_all(
4157 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4158 $text,
4159 $matches
4160 );
4161
4162 # if there are fewer than 4 headlines in the article, do not show TOC
4163 # unless it's been explicitly enabled.
4164 $enoughToc = $this->mShowToc &&
4165 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4166
4167 # Allow user to stipulate that a page should have a "new section"
4168 # link added via __NEWSECTIONLINK__
4169 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4170 $this->mOutput->setNewSection( true );
4171 }
4172
4173 # Allow user to remove the "new section"
4174 # link via __NONEWSECTIONLINK__
4175 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4176 $this->mOutput->hideNewSection( true );
4177 }
4178
4179 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4180 # override above conditions and always show TOC above first header
4181 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4182 $this->mShowToc = true;
4183 $enoughToc = true;
4184 }
4185
4186 # headline counter
4187 $headlineCount = 0;
4188 $numVisible = 0;
4189
4190 # Ugh .. the TOC should have neat indentation levels which can be
4191 # passed to the skin functions. These are determined here
4192 $toc = '';
4193 $full = '';
4194 $head = [];
4195 $sublevelCount = [];
4196 $levelCount = [];
4197 $level = 0;
4198 $prevlevel = 0;
4199 $toclevel = 0;
4200 $prevtoclevel = 0;
4201 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4202 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4203 $oldType = $this->mOutputType;
4204 $this->setOutputType( self::OT_WIKI );
4205 $frame = $this->getPreprocessor()->newFrame();
4206 $root = $this->preprocessToDom( $origText );
4207 $node = $root->getFirstChild();
4208 $byteOffset = 0;
4209 $tocraw = [];
4210 $refers = [];
4211
4212 $headlines = $numMatches !== false ? $matches[3] : [];
4213
4214 $maxTocLevel = $this->siteConfig->get( 'MaxTocLevel' );
4215 foreach ( $headlines as $headline ) {
4216 $isTemplate = false;
4217 $titleText = false;
4218 $sectionIndex = false;
4219 $numbering = '';
4220 $markerMatches = [];
4221 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4222 $serial = $markerMatches[1];
4223 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4224 $isTemplate = ( $titleText != $baseTitleText );
4225 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4226 }
4227
4228 if ( $toclevel ) {
4229 $prevlevel = $level;
4230 }
4231 $level = $matches[1][$headlineCount];
4232
4233 if ( $level > $prevlevel ) {
4234 # Increase TOC level
4235 $toclevel++;
4236 $sublevelCount[$toclevel] = 0;
4237 if ( $toclevel < $maxTocLevel ) {
4238 $prevtoclevel = $toclevel;
4239 $toc .= Linker::tocIndent();
4240 $numVisible++;
4241 }
4242 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4243 # Decrease TOC level, find level to jump to
4244
4245 for ( $i = $toclevel; $i > 0; $i-- ) {
4246 if ( $levelCount[$i] == $level ) {
4247 # Found last matching level
4248 $toclevel = $i;
4249 break;
4250 } elseif ( $levelCount[$i] < $level ) {
4251 # Found first matching level below current level
4252 $toclevel = $i + 1;
4253 break;
4254 }
4255 }
4256 if ( $i == 0 ) {
4257 $toclevel = 1;
4258 }
4259 if ( $toclevel < $maxTocLevel ) {
4260 if ( $prevtoclevel < $maxTocLevel ) {
4261 # Unindent only if the previous toc level was shown :p
4262 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4263 $prevtoclevel = $toclevel;
4264 } else {
4265 $toc .= Linker::tocLineEnd();
4266 }
4267 }
4268 } else {
4269 # No change in level, end TOC line
4270 if ( $toclevel < $maxTocLevel ) {
4271 $toc .= Linker::tocLineEnd();
4272 }
4273 }
4274
4275 $levelCount[$toclevel] = $level;
4276
4277 # count number of headlines for each level
4278 $sublevelCount[$toclevel]++;
4279 $dot = 0;
4280 for ( $i = 1; $i <= $toclevel; $i++ ) {
4281 if ( !empty( $sublevelCount[$i] ) ) {
4282 if ( $dot ) {
4283 $numbering .= '.';
4284 }
4285 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4286 $dot = 1;
4287 }
4288 }
4289
4290 # The safe header is a version of the header text safe to use for links
4291
4292 # Remove link placeholders by the link text.
4293 # <!--LINK number-->
4294 # turns into
4295 # link text with suffix
4296 # Do this before unstrip since link text can contain strip markers
4297 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4298
4299 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4300 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4301
4302 # Remove any <style> or <script> tags (T198618)
4303 $safeHeadline = preg_replace(
4304 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4305 '',
4306 $safeHeadline
4307 );
4308
4309 # Strip out HTML (first regex removes any tag not allowed)
4310 # Allowed tags are:
4311 # * <sup> and <sub> (T10393)
4312 # * <i> (T28375)
4313 # * <b> (r105284)
4314 # * <bdi> (T74884)
4315 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4316 # * <s> and <strike> (T35715)
4317 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4318 # to allow setting directionality in toc items.
4319 $tocline = preg_replace(
4320 [
4321 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4322 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4323 ],
4324 [ '', '<$1>' ],
4325 $safeHeadline
4326 );
4327
4328 # Strip '<span></span>', which is the result from the above if
4329 # <span id="foo"></span> is used to produce an additional anchor
4330 # for a section.
4331 $tocline = str_replace( '<span></span>', '', $tocline );
4332
4333 $tocline = trim( $tocline );
4334
4335 # For the anchor, strip out HTML-y stuff period
4336 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4337 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4338
4339 # Save headline for section edit hint before it's escaped
4340 $headlineHint = $safeHeadline;
4341
4342 # Decode HTML entities
4343 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4344
4345 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4346
4347 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4348 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4349 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4350 if ( $fallbackHeadline === $safeHeadline ) {
4351 # No reason to have both (in fact, we can't)
4352 $fallbackHeadline = false;
4353 }
4354
4355 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4356 # @todo FIXME: We may be changing them depending on the current locale.
4357 $arrayKey = strtolower( $safeHeadline );
4358 if ( $fallbackHeadline === false ) {
4359 $fallbackArrayKey = false;
4360 } else {
4361 $fallbackArrayKey = strtolower( $fallbackHeadline );
4362 }
4363
4364 # Create the anchor for linking from the TOC to the section
4365 $anchor = $safeHeadline;
4366 $fallbackAnchor = $fallbackHeadline;
4367 if ( isset( $refers[$arrayKey] ) ) {
4368 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4369 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4370 $anchor .= "_$i";
4371 $linkAnchor .= "_$i";
4372 $refers["${arrayKey}_$i"] = true;
4373 } else {
4374 $refers[$arrayKey] = true;
4375 }
4376 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4377 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4378 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4379 $fallbackAnchor .= "_$i";
4380 $refers["${fallbackArrayKey}_$i"] = true;
4381 } else {
4382 $refers[$fallbackArrayKey] = true;
4383 }
4384
4385 # Don't number the heading if it is the only one (looks silly)
4386 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4387 # the two are different if the line contains a link
4388 $headline = Html::element(
4389 'span',
4390 [ 'class' => 'mw-headline-number' ],
4391 $numbering
4392 ) . ' ' . $headline;
4393 }
4394
4395 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4396 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4397 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4398 }
4399
4400 # Add the section to the section tree
4401 # Find the DOM node for this header
4402 $noOffset = ( $isTemplate || $sectionIndex === false );
4403 while ( $node && !$noOffset ) {
4404 if ( $node->getName() === 'h' ) {
4405 $bits = $node->splitHeading();
4406 if ( $bits['i'] == $sectionIndex ) {
4407 break;
4408 }
4409 }
4410 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4411 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4412 $node = $node->getNextSibling();
4413 }
4414 $tocraw[] = [
4415 'toclevel' => $toclevel,
4416 'level' => $level,
4417 'line' => $tocline,
4418 'number' => $numbering,
4419 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4420 'fromtitle' => $titleText,
4421 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4422 'anchor' => $anchor,
4423 ];
4424
4425 # give headline the correct <h#> tag
4426 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4427 // Output edit section links as markers with styles that can be customized by skins
4428 if ( $isTemplate ) {
4429 # Put a T flag in the section identifier, to indicate to extractSections()
4430 # that sections inside <includeonly> should be counted.
4431 $editsectionPage = $titleText;
4432 $editsectionSection = "T-$sectionIndex";
4433 $editsectionContent = null;
4434 } else {
4435 $editsectionPage = $this->mTitle->getPrefixedText();
4436 $editsectionSection = $sectionIndex;
4437 $editsectionContent = $headlineHint;
4438 }
4439 // We use a bit of pesudo-xml for editsection markers. The
4440 // language converter is run later on. Using a UNIQ style marker
4441 // leads to the converter screwing up the tokens when it
4442 // converts stuff. And trying to insert strip tags fails too. At
4443 // this point all real inputted tags have already been escaped,
4444 // so we don't have to worry about a user trying to input one of
4445 // these markers directly. We use a page and section attribute
4446 // to stop the language converter from converting these
4447 // important bits of data, but put the headline hint inside a
4448 // content block because the language converter is supposed to
4449 // be able to convert that piece of data.
4450 // Gets replaced with html in ParserOutput::getText
4451 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4452 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4453 if ( $editsectionContent !== null ) {
4454 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4455 } else {
4456 $editlink .= '/>';
4457 }
4458 } else {
4459 $editlink = '';
4460 }
4461 $head[$headlineCount] = Linker::makeHeadline( $level,
4462 $matches['attrib'][$headlineCount], $anchor, $headline,
4463 $editlink, $fallbackAnchor );
4464
4465 $headlineCount++;
4466 }
4467
4468 $this->setOutputType( $oldType );
4469
4470 # Never ever show TOC if no headers
4471 if ( $numVisible < 1 ) {
4472 $enoughToc = false;
4473 }
4474
4475 if ( $enoughToc ) {
4476 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4477 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4478 }
4479 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4480 $this->mOutput->setTOCHTML( $toc );
4481 $toc = self::TOC_START . $toc . self::TOC_END;
4482 }
4483
4484 if ( $isMain ) {
4485 $this->mOutput->setSections( $tocraw );
4486 }
4487
4488 # split up and insert constructed headlines
4489 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4490 $i = 0;
4491
4492 // build an array of document sections
4493 $sections = [];
4494 foreach ( $blocks as $block ) {
4495 // $head is zero-based, sections aren't.
4496 if ( empty( $head[$i - 1] ) ) {
4497 $sections[$i] = $block;
4498 } else {
4499 $sections[$i] = $head[$i - 1] . $block;
4500 }
4501
4502 /**
4503 * Send a hook, one per section.
4504 * The idea here is to be able to make section-level DIVs, but to do so in a
4505 * lower-impact, more correct way than r50769
4506 *
4507 * $this : caller
4508 * $section : the section number
4509 * &$sectionContent : ref to the content of the section
4510 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4511 */
4512 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4513
4514 $i++;
4515 }
4516
4517 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4518 // append the TOC at the beginning
4519 // Top anchor now in skin
4520 $sections[0] .= $toc . "\n";
4521 }
4522
4523 $full .= implode( '', $sections );
4524
4525 if ( $this->mForceTocPosition ) {
4526 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4527 } else {
4528 return $full;
4529 }
4530 }
4531
4532 /**
4533 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4534 * conversion, substituting signatures, {{subst:}} templates, etc.
4535 *
4536 * @param string $text The text to transform
4537 * @param Title $title The Title object for the current article
4538 * @param User $user The User object describing the current user
4539 * @param ParserOptions $options Parsing options
4540 * @param bool $clearState Whether to clear the parser state first
4541 * @return string The altered wiki markup
4542 */
4543 public function preSaveTransform( $text, Title $title, User $user,
4544 ParserOptions $options, $clearState = true
4545 ) {
4546 if ( $clearState ) {
4547 $magicScopeVariable = $this->lock();
4548 }
4549 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4550 $this->setUser( $user );
4551
4552 // Strip U+0000 NULL (T159174)
4553 $text = str_replace( "\000", '', $text );
4554
4555 // We still normalize line endings for backwards-compatibility
4556 // with other code that just calls PST, but this should already
4557 // be handled in TextContent subclasses
4558 $text = TextContent::normalizeLineEndings( $text );
4559
4560 if ( $options->getPreSaveTransform() ) {
4561 $text = $this->pstPass2( $text, $user );
4562 }
4563 $text = $this->mStripState->unstripBoth( $text );
4564
4565 $this->setUser( null ); # Reset
4566
4567 return $text;
4568 }
4569
4570 /**
4571 * Pre-save transform helper function
4572 *
4573 * @param string $text
4574 * @param User $user
4575 *
4576 * @return string
4577 */
4578 private function pstPass2( $text, $user ) {
4579 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4580 # $this->contLang here in order to give everyone the same signature and use the default one
4581 # rather than the one selected in each user's preferences. (see also T14815)
4582 $ts = $this->mOptions->getTimestamp();
4583 $timestamp = MWTimestamp::getLocalInstance( $ts );
4584 $ts = $timestamp->format( 'YmdHis' );
4585 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4586
4587 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4588
4589 # Variable replacement
4590 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4591 $text = $this->replaceVariables( $text );
4592
4593 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4594 # which may corrupt this parser instance via its wfMessage()->text() call-
4595
4596 # Signatures
4597 if ( strpos( $text, '~~~' ) !== false ) {
4598 $sigText = $this->getUserSig( $user );
4599 $text = strtr( $text, [
4600 '~~~~~' => $d,
4601 '~~~~' => "$sigText $d",
4602 '~~~' => $sigText
4603 ] );
4604 # The main two signature forms used above are time-sensitive
4605 $this->mOutput->setFlag( 'user-signature' );
4606 }
4607
4608 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4609 $tc = '[' . Title::legalChars() . ']';
4610 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4611
4612 // [[ns:page (context)|]]
4613 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4614 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4615 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4616 // [[ns:page (context), context|]] (using either single or double-width comma)
4617 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4618 // [[|page]] (reverse pipe trick: add context from page title)
4619 $p2 = "/\[\[\\|($tc+)]]/";
4620
4621 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4622 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4623 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4624 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4625
4626 $t = $this->mTitle->getText();
4627 $m = [];
4628 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4629 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4630 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4631 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4632 } else {
4633 # if there's no context, don't bother duplicating the title
4634 $text = preg_replace( $p2, '[[\\1]]', $text );
4635 }
4636
4637 return $text;
4638 }
4639
4640 /**
4641 * Fetch the user's signature text, if any, and normalize to
4642 * validated, ready-to-insert wikitext.
4643 * If you have pre-fetched the nickname or the fancySig option, you can
4644 * specify them here to save a database query.
4645 * Do not reuse this parser instance after calling getUserSig(),
4646 * as it may have changed if it's the $wgParser.
4647 *
4648 * @param User &$user
4649 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4650 * @param bool|null $fancySig whether the nicknname is the complete signature
4651 * or null to use default value
4652 * @return string
4653 */
4654 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4655 $username = $user->getName();
4656
4657 # If not given, retrieve from the user object.
4658 if ( $nickname === false ) {
4659 $nickname = $user->getOption( 'nickname' );
4660 }
4661
4662 if ( is_null( $fancySig ) ) {
4663 $fancySig = $user->getBoolOption( 'fancysig' );
4664 }
4665
4666 $nickname = $nickname == null ? $username : $nickname;
4667
4668 if ( mb_strlen( $nickname ) > $this->siteConfig->get( 'MaxSigChars' ) ) {
4669 $nickname = $username;
4670 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4671 } elseif ( $fancySig !== false ) {
4672 # Sig. might contain markup; validate this
4673 if ( $this->validateSig( $nickname ) !== false ) {
4674 # Validated; clean up (if needed) and return it
4675 return $this->cleanSig( $nickname, true );
4676 } else {
4677 # Failed to validate; fall back to the default
4678 $nickname = $username;
4679 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4680 }
4681 }
4682
4683 # Make sure nickname doesnt get a sig in a sig
4684 $nickname = self::cleanSigInSig( $nickname );
4685
4686 # If we're still here, make it a link to the user page
4687 $userText = wfEscapeWikiText( $username );
4688 $nickText = wfEscapeWikiText( $nickname );
4689 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4690
4691 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4692 ->title( $this->getTitle() )->text();
4693 }
4694
4695 /**
4696 * Check that the user's signature contains no bad XML
4697 *
4698 * @param string $text
4699 * @return string|bool An expanded string, or false if invalid.
4700 */
4701 public function validateSig( $text ) {
4702 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4703 }
4704
4705 /**
4706 * Clean up signature text
4707 *
4708 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4709 * 2) Substitute all transclusions
4710 *
4711 * @param string $text
4712 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4713 * @return string Signature text
4714 */
4715 public function cleanSig( $text, $parsing = false ) {
4716 if ( !$parsing ) {
4717 global $wgTitle;
4718 $magicScopeVariable = $this->lock();
4719 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4720 }
4721
4722 # Option to disable this feature
4723 if ( !$this->mOptions->getCleanSignatures() ) {
4724 return $text;
4725 }
4726
4727 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4728 # => Move this logic to braceSubstitution()
4729 $substWord = $this->magicWordFactory->get( 'subst' );
4730 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4731 $substText = '{{' . $substWord->getSynonym( 0 );
4732
4733 $text = preg_replace( $substRegex, $substText, $text );
4734 $text = self::cleanSigInSig( $text );
4735 $dom = $this->preprocessToDom( $text );
4736 $frame = $this->getPreprocessor()->newFrame();
4737 $text = $frame->expand( $dom );
4738
4739 if ( !$parsing ) {
4740 $text = $this->mStripState->unstripBoth( $text );
4741 }
4742
4743 return $text;
4744 }
4745
4746 /**
4747 * Strip 3, 4 or 5 tildes out of signatures.
4748 *
4749 * @param string $text
4750 * @return string Signature text with /~{3,5}/ removed
4751 */
4752 public static function cleanSigInSig( $text ) {
4753 $text = preg_replace( '/~{3,5}/', '', $text );
4754 return $text;
4755 }
4756
4757 /**
4758 * Set up some variables which are usually set up in parse()
4759 * so that an external function can call some class members with confidence
4760 *
4761 * @param Title|null $title
4762 * @param ParserOptions $options
4763 * @param int $outputType
4764 * @param bool $clearState
4765 */
4766 public function startExternalParse( Title $title = null, ParserOptions $options,
4767 $outputType, $clearState = true
4768 ) {
4769 $this->startParse( $title, $options, $outputType, $clearState );
4770 }
4771
4772 /**
4773 * @param Title|null $title
4774 * @param ParserOptions $options
4775 * @param int $outputType
4776 * @param bool $clearState
4777 */
4778 private function startParse( Title $title = null, ParserOptions $options,
4779 $outputType, $clearState = true
4780 ) {
4781 $this->setTitle( $title );
4782 $this->mOptions = $options;
4783 $this->setOutputType( $outputType );
4784 if ( $clearState ) {
4785 $this->clearState();
4786 }
4787 }
4788
4789 /**
4790 * Wrapper for preprocess()
4791 *
4792 * @param string $text The text to preprocess
4793 * @param ParserOptions $options
4794 * @param Title|null $title Title object or null to use $wgTitle
4795 * @return string
4796 */
4797 public function transformMsg( $text, $options, $title = null ) {
4798 static $executing = false;
4799
4800 # Guard against infinite recursion
4801 if ( $executing ) {
4802 return $text;
4803 }
4804 $executing = true;
4805
4806 if ( !$title ) {
4807 global $wgTitle;
4808 $title = $wgTitle;
4809 }
4810
4811 $text = $this->preprocess( $text, $title, $options );
4812
4813 $executing = false;
4814 return $text;
4815 }
4816
4817 /**
4818 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4819 * The callback should have the following form:
4820 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4821 *
4822 * Transform and return $text. Use $parser for any required context, e.g. use
4823 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4824 *
4825 * Hooks may return extended information by returning an array, of which the
4826 * first numbered element (index 0) must be the return string, and all other
4827 * entries are extracted into local variables within an internal function
4828 * in the Parser class.
4829 *
4830 * This interface (introduced r61913) appears to be undocumented, but
4831 * 'markerType' is used by some core tag hooks to override which strip
4832 * array their results are placed in. **Use great caution if attempting
4833 * this interface, as it is not documented and injudicious use could smash
4834 * private variables.**
4835 *
4836 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4837 * @param callable $callback The callback function (and object) to use for the tag
4838 * @throws MWException
4839 * @return callable|null The old value of the mTagHooks array associated with the hook
4840 */
4841 public function setHook( $tag, callable $callback ) {
4842 $tag = strtolower( $tag );
4843 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4844 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4845 }
4846 $oldVal = $this->mTagHooks[$tag] ?? null;
4847 $this->mTagHooks[$tag] = $callback;
4848 if ( !in_array( $tag, $this->mStripList ) ) {
4849 $this->mStripList[] = $tag;
4850 }
4851
4852 return $oldVal;
4853 }
4854
4855 /**
4856 * As setHook(), but letting the contents be parsed.
4857 *
4858 * Transparent tag hooks are like regular XML-style tag hooks, except they
4859 * operate late in the transformation sequence, on HTML instead of wikitext.
4860 *
4861 * This is probably obsoleted by things dealing with parser frames?
4862 * The only extension currently using it is geoserver.
4863 *
4864 * @since 1.10
4865 * @todo better document or deprecate this
4866 *
4867 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4868 * @param callable $callback The callback function (and object) to use for the tag
4869 * @throws MWException
4870 * @return callable|null The old value of the mTagHooks array associated with the hook
4871 */
4872 public function setTransparentTagHook( $tag, callable $callback ) {
4873 $tag = strtolower( $tag );
4874 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4875 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4876 }
4877 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4878 $this->mTransparentTagHooks[$tag] = $callback;
4879
4880 return $oldVal;
4881 }
4882
4883 /**
4884 * Remove all tag hooks
4885 */
4886 public function clearTagHooks() {
4887 $this->mTagHooks = [];
4888 $this->mFunctionTagHooks = [];
4889 $this->mStripList = $this->mDefaultStripList;
4890 }
4891
4892 /**
4893 * Create a function, e.g. {{sum:1|2|3}}
4894 * The callback function should have the form:
4895 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4896 *
4897 * Or with Parser::SFH_OBJECT_ARGS:
4898 * function myParserFunction( $parser, $frame, $args ) { ... }
4899 *
4900 * The callback may either return the text result of the function, or an array with the text
4901 * in element 0, and a number of flags in the other elements. The names of the flags are
4902 * specified in the keys. Valid flags are:
4903 * found The text returned is valid, stop processing the template. This
4904 * is on by default.
4905 * nowiki Wiki markup in the return value should be escaped
4906 * isHTML The returned text is HTML, armour it against wikitext transformation
4907 *
4908 * @param string $id The magic word ID
4909 * @param callable $callback The callback function (and object) to use
4910 * @param int $flags A combination of the following flags:
4911 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4912 *
4913 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4914 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4915 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4916 * the arguments, and to control the way they are expanded.
4917 *
4918 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4919 * arguments, for instance:
4920 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4921 *
4922 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4923 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4924 * working if/when this is changed.
4925 *
4926 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4927 * expansion.
4928 *
4929 * Please read the documentation in includes/parser/Preprocessor.php for more information
4930 * about the methods available in PPFrame and PPNode.
4931 *
4932 * @throws MWException
4933 * @return string|callable The old callback function for this name, if any
4934 */
4935 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4936 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4937 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4938
4939 # Add to function cache
4940 $mw = $this->magicWordFactory->get( $id );
4941 if ( !$mw ) {
4942 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4943 }
4944
4945 $synonyms = $mw->getSynonyms();
4946 $sensitive = intval( $mw->isCaseSensitive() );
4947
4948 foreach ( $synonyms as $syn ) {
4949 # Case
4950 if ( !$sensitive ) {
4951 $syn = $this->contLang->lc( $syn );
4952 }
4953 # Add leading hash
4954 if ( !( $flags & self::SFH_NO_HASH ) ) {
4955 $syn = '#' . $syn;
4956 }
4957 # Remove trailing colon
4958 if ( substr( $syn, -1, 1 ) === ':' ) {
4959 $syn = substr( $syn, 0, -1 );
4960 }
4961 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4962 }
4963 return $oldVal;
4964 }
4965
4966 /**
4967 * Get all registered function hook identifiers
4968 *
4969 * @return array
4970 */
4971 public function getFunctionHooks() {
4972 $this->firstCallInit();
4973 return array_keys( $this->mFunctionHooks );
4974 }
4975
4976 /**
4977 * Create a tag function, e.g. "<test>some stuff</test>".
4978 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4979 * Unlike parser functions, their content is not preprocessed.
4980 * @param string $tag
4981 * @param callable $callback
4982 * @param int $flags
4983 * @throws MWException
4984 * @return null
4985 */
4986 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4987 $tag = strtolower( $tag );
4988 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4989 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4990 }
4991 $old = $this->mFunctionTagHooks[$tag] ?? null;
4992 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4993
4994 if ( !in_array( $tag, $this->mStripList ) ) {
4995 $this->mStripList[] = $tag;
4996 }
4997
4998 return $old;
4999 }
5000
5001 /**
5002 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5003 * Placeholders created in Linker::link()
5004 *
5005 * @param string &$text
5006 * @param int $options
5007 */
5008 public function replaceLinkHolders( &$text, $options = 0 ) {
5009 $this->mLinkHolders->replace( $text );
5010 }
5011
5012 /**
5013 * Replace "<!--LINK-->" link placeholders with plain text of links
5014 * (not HTML-formatted).
5015 *
5016 * @param string $text
5017 * @return string
5018 */
5019 public function replaceLinkHoldersText( $text ) {
5020 return $this->mLinkHolders->replaceText( $text );
5021 }
5022
5023 /**
5024 * Renders an image gallery from a text with one line per image.
5025 * text labels may be given by using |-style alternative text. E.g.
5026 * Image:one.jpg|The number "1"
5027 * Image:tree.jpg|A tree
5028 * given as text will return the HTML of a gallery with two images,
5029 * labeled 'The number "1"' and
5030 * 'A tree'.
5031 *
5032 * @param string $text
5033 * @param array $params
5034 * @return string HTML
5035 */
5036 public function renderImageGallery( $text, $params ) {
5037 $mode = false;
5038 if ( isset( $params['mode'] ) ) {
5039 $mode = $params['mode'];
5040 }
5041
5042 try {
5043 $ig = ImageGalleryBase::factory( $mode );
5044 } catch ( Exception $e ) {
5045 // If invalid type set, fallback to default.
5046 $ig = ImageGalleryBase::factory( false );
5047 }
5048
5049 $ig->setContextTitle( $this->mTitle );
5050 $ig->setShowBytes( false );
5051 $ig->setShowDimensions( false );
5052 $ig->setShowFilename( false );
5053 $ig->setParser( $this );
5054 $ig->setHideBadImages();
5055 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5056
5057 if ( isset( $params['showfilename'] ) ) {
5058 $ig->setShowFilename( true );
5059 } else {
5060 $ig->setShowFilename( false );
5061 }
5062 if ( isset( $params['caption'] ) ) {
5063 // NOTE: We aren't passing a frame here or below. Frame info
5064 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5065 // See T107332#4030581
5066 $caption = $this->recursiveTagParse( $params['caption'] );
5067 $ig->setCaptionHtml( $caption );
5068 }
5069 if ( isset( $params['perrow'] ) ) {
5070 $ig->setPerRow( $params['perrow'] );
5071 }
5072 if ( isset( $params['widths'] ) ) {
5073 $ig->setWidths( $params['widths'] );
5074 }
5075 if ( isset( $params['heights'] ) ) {
5076 $ig->setHeights( $params['heights'] );
5077 }
5078 $ig->setAdditionalOptions( $params );
5079
5080 // Avoid PHP 7.1 warning from passing $this by reference
5081 $parser = $this;
5082 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5083
5084 $lines = StringUtils::explode( "\n", $text );
5085 foreach ( $lines as $line ) {
5086 # match lines like these:
5087 # Image:someimage.jpg|This is some image
5088 $matches = [];
5089 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5090 # Skip empty lines
5091 if ( count( $matches ) == 0 ) {
5092 continue;
5093 }
5094
5095 if ( strpos( $matches[0], '%' ) !== false ) {
5096 $matches[1] = rawurldecode( $matches[1] );
5097 }
5098 $title = Title::newFromText( $matches[1], NS_FILE );
5099 if ( is_null( $title ) ) {
5100 # Bogus title. Ignore these so we don't bomb out later.
5101 continue;
5102 }
5103
5104 # We need to get what handler the file uses, to figure out parameters.
5105 # Note, a hook can overide the file name, and chose an entirely different
5106 # file (which potentially could be of a different type and have different handler).
5107 $options = [];
5108 $descQuery = false;
5109 Hooks::run( 'BeforeParserFetchFileAndTitle',
5110 [ $this, $title, &$options, &$descQuery ] );
5111 # Don't register it now, as TraditionalImageGallery does that later.
5112 $file = $this->fetchFileNoRegister( $title, $options );
5113 $handler = $file ? $file->getHandler() : false;
5114
5115 $paramMap = [
5116 'img_alt' => 'gallery-internal-alt',
5117 'img_link' => 'gallery-internal-link',
5118 ];
5119 if ( $handler ) {
5120 $paramMap += $handler->getParamMap();
5121 // We don't want people to specify per-image widths.
5122 // Additionally the width parameter would need special casing anyhow.
5123 unset( $paramMap['img_width'] );
5124 }
5125
5126 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5127
5128 $label = '';
5129 $alt = '';
5130 $link = '';
5131 $handlerOptions = [];
5132 if ( isset( $matches[3] ) ) {
5133 // look for an |alt= definition while trying not to break existing
5134 // captions with multiple pipes (|) in it, until a more sensible grammar
5135 // is defined for images in galleries
5136
5137 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5138 // splitting on '|' is a bit odd, and different from makeImage.
5139 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5140 // Protect LanguageConverter markup
5141 $parameterMatches = StringUtils::delimiterExplode(
5142 '-{', '}-', '|', $matches[3], true /* nested */
5143 );
5144
5145 foreach ( $parameterMatches as $parameterMatch ) {
5146 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5147 if ( $magicName ) {
5148 $paramName = $paramMap[$magicName];
5149
5150 switch ( $paramName ) {
5151 case 'gallery-internal-alt':
5152 $alt = $this->stripAltText( $match, false );
5153 break;
5154 case 'gallery-internal-link':
5155 $linkValue = $this->stripAltText( $match, false );
5156 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5157 // Result of LanguageConverter::markNoConversion
5158 // invoked on an external link.
5159 $linkValue = substr( $linkValue, 4, -2 );
5160 }
5161 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5162 if ( $type === 'link-url' ) {
5163 $link = $target;
5164 $this->mOutput->addExternalLink( $target );
5165 } elseif ( $type === 'link-title' ) {
5166 $link = $target->getLinkURL();
5167 $this->mOutput->addLink( $target );
5168 }
5169 break;
5170 default:
5171 // Must be a handler specific parameter.
5172 if ( $handler->validateParam( $paramName, $match ) ) {
5173 $handlerOptions[$paramName] = $match;
5174 } else {
5175 // Guess not, consider it as caption.
5176 wfDebug( "$parameterMatch failed parameter validation\n" );
5177 $label = $parameterMatch;
5178 }
5179 }
5180
5181 } else {
5182 // Last pipe wins.
5183 $label = $parameterMatch;
5184 }
5185 }
5186 }
5187
5188 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5189 }
5190 $html = $ig->toHTML();
5191 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5192 return $html;
5193 }
5194
5195 /**
5196 * @param MediaHandler $handler
5197 * @return array
5198 */
5199 public function getImageParams( $handler ) {
5200 if ( $handler ) {
5201 $handlerClass = get_class( $handler );
5202 } else {
5203 $handlerClass = '';
5204 }
5205 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5206 # Initialise static lists
5207 static $internalParamNames = [
5208 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5209 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5210 'bottom', 'text-bottom' ],
5211 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5212 'upright', 'border', 'link', 'alt', 'class' ],
5213 ];
5214 static $internalParamMap;
5215 if ( !$internalParamMap ) {
5216 $internalParamMap = [];
5217 foreach ( $internalParamNames as $type => $names ) {
5218 foreach ( $names as $name ) {
5219 // For grep: img_left, img_right, img_center, img_none,
5220 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5221 // img_bottom, img_text_bottom,
5222 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5223 // img_border, img_link, img_alt, img_class
5224 $magicName = str_replace( '-', '_', "img_$name" );
5225 $internalParamMap[$magicName] = [ $type, $name ];
5226 }
5227 }
5228 }
5229
5230 # Add handler params
5231 $paramMap = $internalParamMap;
5232 if ( $handler ) {
5233 $handlerParamMap = $handler->getParamMap();
5234 foreach ( $handlerParamMap as $magic => $paramName ) {
5235 $paramMap[$magic] = [ 'handler', $paramName ];
5236 }
5237 }
5238 $this->mImageParams[$handlerClass] = $paramMap;
5239 $this->mImageParamsMagicArray[$handlerClass] =
5240 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5241 }
5242 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5243 }
5244
5245 /**
5246 * Parse image options text and use it to make an image
5247 *
5248 * @param Title $title
5249 * @param string $options
5250 * @param LinkHolderArray|bool $holders
5251 * @return string HTML
5252 */
5253 public function makeImage( $title, $options, $holders = false ) {
5254 # Check if the options text is of the form "options|alt text"
5255 # Options are:
5256 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5257 # * left no resizing, just left align. label is used for alt= only
5258 # * right same, but right aligned
5259 # * none same, but not aligned
5260 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5261 # * center center the image
5262 # * frame Keep original image size, no magnify-button.
5263 # * framed Same as "frame"
5264 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5265 # * upright reduce width for upright images, rounded to full __0 px
5266 # * border draw a 1px border around the image
5267 # * alt Text for HTML alt attribute (defaults to empty)
5268 # * class Set a class for img node
5269 # * link Set the target of the image link. Can be external, interwiki, or local
5270 # vertical-align values (no % or length right now):
5271 # * baseline
5272 # * sub
5273 # * super
5274 # * top
5275 # * text-top
5276 # * middle
5277 # * bottom
5278 # * text-bottom
5279
5280 # Protect LanguageConverter markup when splitting into parts
5281 $parts = StringUtils::delimiterExplode(
5282 '-{', '}-', '|', $options, true /* allow nesting */
5283 );
5284
5285 # Give extensions a chance to select the file revision for us
5286 $options = [];
5287 $descQuery = false;
5288 Hooks::run( 'BeforeParserFetchFileAndTitle',
5289 [ $this, $title, &$options, &$descQuery ] );
5290 # Fetch and register the file (file title may be different via hooks)
5291 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5292
5293 # Get parameter map
5294 $handler = $file ? $file->getHandler() : false;
5295
5296 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5297
5298 if ( !$file ) {
5299 $this->addTrackingCategory( 'broken-file-category' );
5300 }
5301
5302 # Process the input parameters
5303 $caption = '';
5304 $params = [ 'frame' => [], 'handler' => [],
5305 'horizAlign' => [], 'vertAlign' => [] ];
5306 $seenformat = false;
5307 foreach ( $parts as $part ) {
5308 $part = trim( $part );
5309 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5310 $validated = false;
5311 if ( isset( $paramMap[$magicName] ) ) {
5312 list( $type, $paramName ) = $paramMap[$magicName];
5313
5314 # Special case; width and height come in one variable together
5315 if ( $type === 'handler' && $paramName === 'width' ) {
5316 $parsedWidthParam = self::parseWidthParam( $value );
5317 if ( isset( $parsedWidthParam['width'] ) ) {
5318 $width = $parsedWidthParam['width'];
5319 if ( $handler->validateParam( 'width', $width ) ) {
5320 $params[$type]['width'] = $width;
5321 $validated = true;
5322 }
5323 }
5324 if ( isset( $parsedWidthParam['height'] ) ) {
5325 $height = $parsedWidthParam['height'];
5326 if ( $handler->validateParam( 'height', $height ) ) {
5327 $params[$type]['height'] = $height;
5328 $validated = true;
5329 }
5330 }
5331 # else no validation -- T15436
5332 } else {
5333 if ( $type === 'handler' ) {
5334 # Validate handler parameter
5335 $validated = $handler->validateParam( $paramName, $value );
5336 } else {
5337 # Validate internal parameters
5338 switch ( $paramName ) {
5339 case 'manualthumb':
5340 case 'alt':
5341 case 'class':
5342 # @todo FIXME: Possibly check validity here for
5343 # manualthumb? downstream behavior seems odd with
5344 # missing manual thumbs.
5345 $validated = true;
5346 $value = $this->stripAltText( $value, $holders );
5347 break;
5348 case 'link':
5349 list( $paramName, $value ) =
5350 $this->parseLinkParameter(
5351 $this->stripAltText( $value, $holders )
5352 );
5353 if ( $paramName ) {
5354 $validated = true;
5355 if ( $paramName === 'no-link' ) {
5356 $value = true;
5357 }
5358 if ( $paramName === 'link-url' ) {
5359 if ( $this->mOptions->getExternalLinkTarget() ) {
5360 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5361 }
5362 }
5363 }
5364 break;
5365 case 'frameless':
5366 case 'framed':
5367 case 'thumbnail':
5368 // use first appearing option, discard others.
5369 $validated = !$seenformat;
5370 $seenformat = true;
5371 break;
5372 default:
5373 # Most other things appear to be empty or numeric...
5374 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5375 }
5376 }
5377
5378 if ( $validated ) {
5379 $params[$type][$paramName] = $value;
5380 }
5381 }
5382 }
5383 if ( !$validated ) {
5384 $caption = $part;
5385 }
5386 }
5387
5388 # Process alignment parameters
5389 if ( $params['horizAlign'] ) {
5390 $params['frame']['align'] = key( $params['horizAlign'] );
5391 }
5392 if ( $params['vertAlign'] ) {
5393 $params['frame']['valign'] = key( $params['vertAlign'] );
5394 }
5395
5396 $params['frame']['caption'] = $caption;
5397
5398 # Will the image be presented in a frame, with the caption below?
5399 $imageIsFramed = isset( $params['frame']['frame'] )
5400 || isset( $params['frame']['framed'] )
5401 || isset( $params['frame']['thumbnail'] )
5402 || isset( $params['frame']['manualthumb'] );
5403
5404 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5405 # came to also set the caption, ordinary text after the image -- which
5406 # makes no sense, because that just repeats the text multiple times in
5407 # screen readers. It *also* came to set the title attribute.
5408 # Now that we have an alt attribute, we should not set the alt text to
5409 # equal the caption: that's worse than useless, it just repeats the
5410 # text. This is the framed/thumbnail case. If there's no caption, we
5411 # use the unnamed parameter for alt text as well, just for the time be-
5412 # ing, if the unnamed param is set and the alt param is not.
5413 # For the future, we need to figure out if we want to tweak this more,
5414 # e.g., introducing a title= parameter for the title; ignoring the un-
5415 # named parameter entirely for images without a caption; adding an ex-
5416 # plicit caption= parameter and preserving the old magic unnamed para-
5417 # meter for BC; ...
5418 if ( $imageIsFramed ) { # Framed image
5419 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5420 # No caption or alt text, add the filename as the alt text so
5421 # that screen readers at least get some description of the image
5422 $params['frame']['alt'] = $title->getText();
5423 }
5424 # Do not set $params['frame']['title'] because tooltips don't make sense
5425 # for framed images
5426 } else { # Inline image
5427 if ( !isset( $params['frame']['alt'] ) ) {
5428 # No alt text, use the "caption" for the alt text
5429 if ( $caption !== '' ) {
5430 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5431 } else {
5432 # No caption, fall back to using the filename for the
5433 # alt text
5434 $params['frame']['alt'] = $title->getText();
5435 }
5436 }
5437 # Use the "caption" for the tooltip text
5438 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5439 }
5440 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5441
5442 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5443
5444 # Linker does the rest
5445 $time = $options['time'] ?? false;
5446 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5447 $time, $descQuery, $this->mOptions->getThumbSize() );
5448
5449 # Give the handler a chance to modify the parser object
5450 if ( $handler ) {
5451 $handler->parserTransformHook( $this, $file );
5452 }
5453
5454 return $ret;
5455 }
5456
5457 /**
5458 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5459 *
5460 * Adds an entry to appropriate link tables.
5461 *
5462 * @since 1.32
5463 * @param string $value
5464 * @return array of `[ type, target ]`, where:
5465 * - `type` is one of:
5466 * - `null`: Given value is not a valid link target, use default
5467 * - `'no-link'`: Given value is empty, do not generate a link
5468 * - `'link-url'`: Given value is a valid external link
5469 * - `'link-title'`: Given value is a valid internal link
5470 * - `target` is:
5471 * - When `type` is `null` or `'no-link'`: `false`
5472 * - When `type` is `'link-url'`: URL string corresponding to given value
5473 * - When `type` is `'link-title'`: Title object corresponding to given value
5474 */
5475 public function parseLinkParameter( $value ) {
5476 $chars = self::EXT_LINK_URL_CLASS;
5477 $addr = self::EXT_LINK_ADDR;
5478 $prots = $this->mUrlProtocols;
5479 $type = null;
5480 $target = false;
5481 if ( $value === '' ) {
5482 $type = 'no-link';
5483 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5484 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5485 $this->mOutput->addExternalLink( $value );
5486 $type = 'link-url';
5487 $target = $value;
5488 }
5489 } else {
5490 $linkTitle = Title::newFromText( $value );
5491 if ( $linkTitle ) {
5492 $this->mOutput->addLink( $linkTitle );
5493 $type = 'link-title';
5494 $target = $linkTitle;
5495 }
5496 }
5497 return [ $type, $target ];
5498 }
5499
5500 /**
5501 * @param string $caption
5502 * @param LinkHolderArray|bool $holders
5503 * @return mixed|string
5504 */
5505 protected function stripAltText( $caption, $holders ) {
5506 # Strip bad stuff out of the title (tooltip). We can't just use
5507 # replaceLinkHoldersText() here, because if this function is called
5508 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5509 if ( $holders ) {
5510 $tooltip = $holders->replaceText( $caption );
5511 } else {
5512 $tooltip = $this->replaceLinkHoldersText( $caption );
5513 }
5514
5515 # make sure there are no placeholders in thumbnail attributes
5516 # that are later expanded to html- so expand them now and
5517 # remove the tags
5518 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5519 # Compatibility hack! In HTML certain entity references not terminated
5520 # by a semicolon are decoded (but not if we're in an attribute; that's
5521 # how link URLs get away without properly escaping & in queries).
5522 # But wikitext has always required semicolon-termination of entities,
5523 # so encode & where needed to avoid decode of semicolon-less entities.
5524 # See T209236 and
5525 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5526 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5527 $tooltip = preg_replace( "/
5528 & # 1. entity prefix
5529 (?= # 2. followed by:
5530 (?: # a. one of the legacy semicolon-less named entities
5531 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5532 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5533 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5534 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5535 U(?:acute|circ|grave|uml)|Yacute|
5536 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5537 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5538 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5539 frac(?:1(?:2|4)|34)|
5540 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5541 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5542 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5543 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5544 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5545 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5546 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5547 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5548 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5549 )
5550 (?:[^;]|$)) # b. and not followed by a semicolon
5551 # S = study, for efficiency
5552 /Sx", '&amp;', $tooltip );
5553 $tooltip = Sanitizer::stripAllTags( $tooltip );
5554
5555 return $tooltip;
5556 }
5557
5558 /**
5559 * Set a flag in the output object indicating that the content is dynamic and
5560 * shouldn't be cached.
5561 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5562 */
5563 public function disableCache() {
5564 wfDebug( "Parser output marked as uncacheable.\n" );
5565 if ( !$this->mOutput ) {
5566 throw new MWException( __METHOD__ .
5567 " can only be called when actually parsing something" );
5568 }
5569 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5570 }
5571
5572 /**
5573 * Callback from the Sanitizer for expanding items found in HTML attribute
5574 * values, so they can be safely tested and escaped.
5575 *
5576 * @param string &$text
5577 * @param bool|PPFrame $frame
5578 * @return string
5579 */
5580 public function attributeStripCallback( &$text, $frame = false ) {
5581 $text = $this->replaceVariables( $text, $frame );
5582 $text = $this->mStripState->unstripBoth( $text );
5583 return $text;
5584 }
5585
5586 /**
5587 * Accessor
5588 *
5589 * @return array
5590 */
5591 public function getTags() {
5592 $this->firstCallInit();
5593 return array_merge(
5594 array_keys( $this->mTransparentTagHooks ),
5595 array_keys( $this->mTagHooks ),
5596 array_keys( $this->mFunctionTagHooks )
5597 );
5598 }
5599
5600 /**
5601 * @since 1.32
5602 * @return array
5603 */
5604 public function getFunctionSynonyms() {
5605 $this->firstCallInit();
5606 return $this->mFunctionSynonyms;
5607 }
5608
5609 /**
5610 * @since 1.32
5611 * @return string
5612 */
5613 public function getUrlProtocols() {
5614 return $this->mUrlProtocols;
5615 }
5616
5617 /**
5618 * Replace transparent tags in $text with the values given by the callbacks.
5619 *
5620 * Transparent tag hooks are like regular XML-style tag hooks, except they
5621 * operate late in the transformation sequence, on HTML instead of wikitext.
5622 *
5623 * @param string $text
5624 *
5625 * @return string
5626 */
5627 public function replaceTransparentTags( $text ) {
5628 $matches = [];
5629 $elements = array_keys( $this->mTransparentTagHooks );
5630 $text = self::extractTagsAndParams( $elements, $text, $matches );
5631 $replacements = [];
5632
5633 foreach ( $matches as $marker => $data ) {
5634 list( $element, $content, $params, $tag ) = $data;
5635 $tagName = strtolower( $element );
5636 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5637 $output = call_user_func_array(
5638 $this->mTransparentTagHooks[$tagName],
5639 [ $content, $params, $this ]
5640 );
5641 } else {
5642 $output = $tag;
5643 }
5644 $replacements[$marker] = $output;
5645 }
5646 return strtr( $text, $replacements );
5647 }
5648
5649 /**
5650 * Break wikitext input into sections, and either pull or replace
5651 * some particular section's text.
5652 *
5653 * External callers should use the getSection and replaceSection methods.
5654 *
5655 * @param string $text Page wikitext
5656 * @param string|int $sectionId A section identifier string of the form:
5657 * "<flag1> - <flag2> - ... - <section number>"
5658 *
5659 * Currently the only recognised flag is "T", which means the target section number
5660 * was derived during a template inclusion parse, in other words this is a template
5661 * section edit link. If no flags are given, it was an ordinary section edit link.
5662 * This flag is required to avoid a section numbering mismatch when a section is
5663 * enclosed by "<includeonly>" (T8563).
5664 *
5665 * The section number 0 pulls the text before the first heading; other numbers will
5666 * pull the given section along with its lower-level subsections. If the section is
5667 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5668 *
5669 * Section 0 is always considered to exist, even if it only contains the empty
5670 * string. If $text is the empty string and section 0 is replaced, $newText is
5671 * returned.
5672 *
5673 * @param string $mode One of "get" or "replace"
5674 * @param string $newText Replacement text for section data.
5675 * @return string For "get", the extracted section text.
5676 * for "replace", the whole page with the section replaced.
5677 */
5678 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5679 global $wgTitle; # not generally used but removes an ugly failure mode
5680
5681 $magicScopeVariable = $this->lock();
5682 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5683 $outText = '';
5684 $frame = $this->getPreprocessor()->newFrame();
5685
5686 # Process section extraction flags
5687 $flags = 0;
5688 $sectionParts = explode( '-', $sectionId );
5689 $sectionIndex = array_pop( $sectionParts );
5690 foreach ( $sectionParts as $part ) {
5691 if ( $part === 'T' ) {
5692 $flags |= self::PTD_FOR_INCLUSION;
5693 }
5694 }
5695
5696 # Check for empty input
5697 if ( strval( $text ) === '' ) {
5698 # Only sections 0 and T-0 exist in an empty document
5699 if ( $sectionIndex == 0 ) {
5700 if ( $mode === 'get' ) {
5701 return '';
5702 } else {
5703 return $newText;
5704 }
5705 } else {
5706 if ( $mode === 'get' ) {
5707 return $newText;
5708 } else {
5709 return $text;
5710 }
5711 }
5712 }
5713
5714 # Preprocess the text
5715 $root = $this->preprocessToDom( $text, $flags );
5716
5717 # <h> nodes indicate section breaks
5718 # They can only occur at the top level, so we can find them by iterating the root's children
5719 $node = $root->getFirstChild();
5720
5721 # Find the target section
5722 if ( $sectionIndex == 0 ) {
5723 # Section zero doesn't nest, level=big
5724 $targetLevel = 1000;
5725 } else {
5726 while ( $node ) {
5727 if ( $node->getName() === 'h' ) {
5728 $bits = $node->splitHeading();
5729 if ( $bits['i'] == $sectionIndex ) {
5730 $targetLevel = $bits['level'];
5731 break;
5732 }
5733 }
5734 if ( $mode === 'replace' ) {
5735 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5736 }
5737 $node = $node->getNextSibling();
5738 }
5739 }
5740
5741 if ( !$node ) {
5742 # Not found
5743 if ( $mode === 'get' ) {
5744 return $newText;
5745 } else {
5746 return $text;
5747 }
5748 }
5749
5750 # Find the end of the section, including nested sections
5751 do {
5752 if ( $node->getName() === 'h' ) {
5753 $bits = $node->splitHeading();
5754 $curLevel = $bits['level'];
5755 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5756 break;
5757 }
5758 }
5759 if ( $mode === 'get' ) {
5760 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5761 }
5762 $node = $node->getNextSibling();
5763 } while ( $node );
5764
5765 # Write out the remainder (in replace mode only)
5766 if ( $mode === 'replace' ) {
5767 # Output the replacement text
5768 # Add two newlines on -- trailing whitespace in $newText is conventionally
5769 # stripped by the editor, so we need both newlines to restore the paragraph gap
5770 # Only add trailing whitespace if there is newText
5771 if ( $newText != "" ) {
5772 $outText .= $newText . "\n\n";
5773 }
5774
5775 while ( $node ) {
5776 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5777 $node = $node->getNextSibling();
5778 }
5779 }
5780
5781 if ( is_string( $outText ) ) {
5782 # Re-insert stripped tags
5783 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5784 }
5785
5786 return $outText;
5787 }
5788
5789 /**
5790 * This function returns the text of a section, specified by a number ($section).
5791 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5792 * the first section before any such heading (section 0).
5793 *
5794 * If a section contains subsections, these are also returned.
5795 *
5796 * @param string $text Text to look in
5797 * @param string|int $sectionId Section identifier as a number or string
5798 * (e.g. 0, 1 or 'T-1').
5799 * @param string $defaultText Default to return if section is not found
5800 *
5801 * @return string Text of the requested section
5802 */
5803 public function getSection( $text, $sectionId, $defaultText = '' ) {
5804 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5805 }
5806
5807 /**
5808 * This function returns $oldtext after the content of the section
5809 * specified by $section has been replaced with $text. If the target
5810 * section does not exist, $oldtext is returned unchanged.
5811 *
5812 * @param string $oldText Former text of the article
5813 * @param string|int $sectionId Section identifier as a number or string
5814 * (e.g. 0, 1 or 'T-1').
5815 * @param string $newText Replacing text
5816 *
5817 * @return string Modified text
5818 */
5819 public function replaceSection( $oldText, $sectionId, $newText ) {
5820 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5821 }
5822
5823 /**
5824 * Get the ID of the revision we are parsing
5825 *
5826 * @return int|null
5827 */
5828 public function getRevisionId() {
5829 return $this->mRevisionId;
5830 }
5831
5832 /**
5833 * Get the revision object for $this->mRevisionId
5834 *
5835 * @return Revision|null Either a Revision object or null
5836 * @since 1.23 (public since 1.23)
5837 */
5838 public function getRevisionObject() {
5839 if ( !is_null( $this->mRevisionObject ) ) {
5840 return $this->mRevisionObject;
5841 }
5842
5843 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5844 // This is useful when parsing revision that has not yet been saved.
5845 // However, if we get back a saved revision even though we are in
5846 // preview mode, we'll have to ignore it, see below.
5847 // NOTE: This callback may be used to inject an OLD revision that was
5848 // already loaded, so "current" is a bit of a misnomer. We can't just
5849 // skip it if mRevisionId is set.
5850 $rev = call_user_func(
5851 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5852 );
5853
5854 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
5855 // We are in preview mode (mRevisionId is null), and the current revision callback
5856 // returned an existing revision. Ignore it and return null, it's probably the page's
5857 // current revision, which is not what we want here. Note that we do want to call the
5858 // callback to allow the unsaved revision to be injected here, e.g. for
5859 // self-transclusion previews.
5860 return null;
5861 }
5862
5863 // If the parse is for a new revision, then the callback should have
5864 // already been set to force the object and should match mRevisionId.
5865 // If not, try to fetch by mRevisionId for sanity.
5866 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
5867 $rev = Revision::newFromId( $this->mRevisionId );
5868 }
5869
5870 $this->mRevisionObject = $rev;
5871
5872 return $this->mRevisionObject;
5873 }
5874
5875 /**
5876 * Get the timestamp associated with the current revision, adjusted for
5877 * the default server-local timestamp
5878 * @return string
5879 */
5880 public function getRevisionTimestamp() {
5881 if ( is_null( $this->mRevisionTimestamp ) ) {
5882 $revObject = $this->getRevisionObject();
5883 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5884
5885 # The cryptic '' timezone parameter tells to use the site-default
5886 # timezone offset instead of the user settings.
5887 # Since this value will be saved into the parser cache, served
5888 # to other users, and potentially even used inside links and such,
5889 # it needs to be consistent for all visitors.
5890 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
5891 }
5892 return $this->mRevisionTimestamp;
5893 }
5894
5895 /**
5896 * Get the name of the user that edited the last revision
5897 *
5898 * @return string User name
5899 */
5900 public function getRevisionUser() {
5901 if ( is_null( $this->mRevisionUser ) ) {
5902 $revObject = $this->getRevisionObject();
5903
5904 # if this template is subst: the revision id will be blank,
5905 # so just use the current user's name
5906 if ( $revObject ) {
5907 $this->mRevisionUser = $revObject->getUserText();
5908 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5909 $this->mRevisionUser = $this->getUser()->getName();
5910 }
5911 }
5912 return $this->mRevisionUser;
5913 }
5914
5915 /**
5916 * Get the size of the revision
5917 *
5918 * @return int|null Revision size
5919 */
5920 public function getRevisionSize() {
5921 if ( is_null( $this->mRevisionSize ) ) {
5922 $revObject = $this->getRevisionObject();
5923
5924 # if this variable is subst: the revision id will be blank,
5925 # so just use the parser input size, because the own substituation
5926 # will change the size.
5927 if ( $revObject ) {
5928 $this->mRevisionSize = $revObject->getSize();
5929 } else {
5930 $this->mRevisionSize = $this->mInputSize;
5931 }
5932 }
5933 return $this->mRevisionSize;
5934 }
5935
5936 /**
5937 * Mutator for $mDefaultSort
5938 *
5939 * @param string $sort New value
5940 */
5941 public function setDefaultSort( $sort ) {
5942 $this->mDefaultSort = $sort;
5943 $this->mOutput->setProperty( 'defaultsort', $sort );
5944 }
5945
5946 /**
5947 * Accessor for $mDefaultSort
5948 * Will use the empty string if none is set.
5949 *
5950 * This value is treated as a prefix, so the
5951 * empty string is equivalent to sorting by
5952 * page name.
5953 *
5954 * @return string
5955 */
5956 public function getDefaultSort() {
5957 if ( $this->mDefaultSort !== false ) {
5958 return $this->mDefaultSort;
5959 } else {
5960 return '';
5961 }
5962 }
5963
5964 /**
5965 * Accessor for $mDefaultSort
5966 * Unlike getDefaultSort(), will return false if none is set
5967 *
5968 * @return string|bool
5969 */
5970 public function getCustomDefaultSort() {
5971 return $this->mDefaultSort;
5972 }
5973
5974 private static function getSectionNameFromStrippedText( $text ) {
5975 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5976 $text = Sanitizer::decodeCharReferences( $text );
5977 $text = self::normalizeSectionName( $text );
5978 return $text;
5979 }
5980
5981 private static function makeAnchor( $sectionName ) {
5982 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5983 }
5984
5985 private function makeLegacyAnchor( $sectionName ) {
5986 $fragmentMode = $this->siteConfig->get( 'FragmentMode' );
5987 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
5988 // ForAttribute() and ForLink() are the same for legacy encoding
5989 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5990 } else {
5991 $id = Sanitizer::escapeIdForLink( $sectionName );
5992 }
5993
5994 return "#$id";
5995 }
5996
5997 /**
5998 * Try to guess the section anchor name based on a wikitext fragment
5999 * presumably extracted from a heading, for example "Header" from
6000 * "== Header ==".
6001 *
6002 * @param string $text
6003 * @return string Anchor (starting with '#')
6004 */
6005 public function guessSectionNameFromWikiText( $text ) {
6006 # Strip out wikitext links(they break the anchor)
6007 $text = $this->stripSectionName( $text );
6008 $sectionName = self::getSectionNameFromStrippedText( $text );
6009 return self::makeAnchor( $sectionName );
6010 }
6011
6012 /**
6013 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6014 * instead, if possible. For use in redirects, since various versions
6015 * of Microsoft browsers interpret Location: headers as something other
6016 * than UTF-8, resulting in breakage.
6017 *
6018 * @param string $text The section name
6019 * @return string Anchor (starting with '#')
6020 */
6021 public function guessLegacySectionNameFromWikiText( $text ) {
6022 # Strip out wikitext links(they break the anchor)
6023 $text = $this->stripSectionName( $text );
6024 $sectionName = self::getSectionNameFromStrippedText( $text );
6025 return $this->makeLegacyAnchor( $sectionName );
6026 }
6027
6028 /**
6029 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6030 * @param string $text Section name (plain text)
6031 * @return string Anchor (starting with '#')
6032 */
6033 public static function guessSectionNameFromStrippedText( $text ) {
6034 $sectionName = self::getSectionNameFromStrippedText( $text );
6035 return self::makeAnchor( $sectionName );
6036 }
6037
6038 /**
6039 * Apply the same normalization as code making links to this section would
6040 *
6041 * @param string $text
6042 * @return string
6043 */
6044 private static function normalizeSectionName( $text ) {
6045 # T90902: ensure the same normalization is applied for IDs as to links
6046 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
6047 try {
6048
6049 $parts = $titleParser->splitTitleString( "#$text" );
6050 } catch ( MalformedTitleException $ex ) {
6051 return $text;
6052 }
6053 return $parts['fragment'];
6054 }
6055
6056 /**
6057 * Strips a text string of wikitext for use in a section anchor
6058 *
6059 * Accepts a text string and then removes all wikitext from the
6060 * string and leaves only the resultant text (i.e. the result of
6061 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6062 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6063 * to create valid section anchors by mimicing the output of the
6064 * parser when headings are parsed.
6065 *
6066 * @param string $text Text string to be stripped of wikitext
6067 * for use in a Section anchor
6068 * @return string Filtered text string
6069 */
6070 public function stripSectionName( $text ) {
6071 # Strip internal link markup
6072 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6073 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6074
6075 # Strip external link markup
6076 # @todo FIXME: Not tolerant to blank link text
6077 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6078 # on how many empty links there are on the page - need to figure that out.
6079 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6080
6081 # Parse wikitext quotes (italics & bold)
6082 $text = $this->doQuotes( $text );
6083
6084 # Strip HTML tags
6085 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6086 return $text;
6087 }
6088
6089 /**
6090 * strip/replaceVariables/unstrip for preprocessor regression testing
6091 *
6092 * @param string $text
6093 * @param Title $title
6094 * @param ParserOptions $options
6095 * @param int $outputType
6096 *
6097 * @return string
6098 */
6099 public function testSrvus( $text, Title $title, ParserOptions $options,
6100 $outputType = self::OT_HTML
6101 ) {
6102 $magicScopeVariable = $this->lock();
6103 $this->startParse( $title, $options, $outputType, true );
6104
6105 $text = $this->replaceVariables( $text );
6106 $text = $this->mStripState->unstripBoth( $text );
6107 $text = Sanitizer::removeHTMLtags( $text );
6108 return $text;
6109 }
6110
6111 /**
6112 * @param string $text
6113 * @param Title $title
6114 * @param ParserOptions $options
6115 * @return string
6116 */
6117 public function testPst( $text, Title $title, ParserOptions $options ) {
6118 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6119 }
6120
6121 /**
6122 * @param string $text
6123 * @param Title $title
6124 * @param ParserOptions $options
6125 * @return string
6126 */
6127 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6128 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6129 }
6130
6131 /**
6132 * Call a callback function on all regions of the given text that are not
6133 * inside strip markers, and replace those regions with the return value
6134 * of the callback. For example, with input:
6135 *
6136 * aaa<MARKER>bbb
6137 *
6138 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6139 * two strings will be replaced with the value returned by the callback in
6140 * each case.
6141 *
6142 * @param string $s
6143 * @param callable $callback
6144 *
6145 * @return string
6146 */
6147 public function markerSkipCallback( $s, $callback ) {
6148 $i = 0;
6149 $out = '';
6150 while ( $i < strlen( $s ) ) {
6151 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6152 if ( $markerStart === false ) {
6153 $out .= call_user_func( $callback, substr( $s, $i ) );
6154 break;
6155 } else {
6156 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6157 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6158 if ( $markerEnd === false ) {
6159 $out .= substr( $s, $markerStart );
6160 break;
6161 } else {
6162 $markerEnd += strlen( self::MARKER_SUFFIX );
6163 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6164 $i = $markerEnd;
6165 }
6166 }
6167 }
6168 return $out;
6169 }
6170
6171 /**
6172 * Remove any strip markers found in the given text.
6173 *
6174 * @param string $text
6175 * @return string
6176 */
6177 public function killMarkers( $text ) {
6178 return $this->mStripState->killMarkers( $text );
6179 }
6180
6181 /**
6182 * Save the parser state required to convert the given half-parsed text to
6183 * HTML. "Half-parsed" in this context means the output of
6184 * recursiveTagParse() or internalParse(). This output has strip markers
6185 * from replaceVariables (extensionSubstitution() etc.), and link
6186 * placeholders from replaceLinkHolders().
6187 *
6188 * Returns an array which can be serialized and stored persistently. This
6189 * array can later be loaded into another parser instance with
6190 * unserializeHalfParsedText(). The text can then be safely incorporated into
6191 * the return value of a parser hook.
6192 *
6193 * @deprecated since 1.31
6194 * @param string $text
6195 *
6196 * @return array
6197 */
6198 public function serializeHalfParsedText( $text ) {
6199 wfDeprecated( __METHOD__, '1.31' );
6200 $data = [
6201 'text' => $text,
6202 'version' => self::HALF_PARSED_VERSION,
6203 'stripState' => $this->mStripState->getSubState( $text ),
6204 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6205 ];
6206 return $data;
6207 }
6208
6209 /**
6210 * Load the parser state given in the $data array, which is assumed to
6211 * have been generated by serializeHalfParsedText(). The text contents is
6212 * extracted from the array, and its markers are transformed into markers
6213 * appropriate for the current Parser instance. This transformed text is
6214 * returned, and can be safely included in the return value of a parser
6215 * hook.
6216 *
6217 * If the $data array has been stored persistently, the caller should first
6218 * check whether it is still valid, by calling isValidHalfParsedText().
6219 *
6220 * @deprecated since 1.31
6221 * @param array $data Serialized data
6222 * @throws MWException
6223 * @return string
6224 */
6225 public function unserializeHalfParsedText( $data ) {
6226 wfDeprecated( __METHOD__, '1.31' );
6227 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6228 throw new MWException( __METHOD__ . ': invalid version' );
6229 }
6230
6231 # First, extract the strip state.
6232 $texts = [ $data['text'] ];
6233 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6234
6235 # Now renumber links
6236 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6237
6238 # Should be good to go.
6239 return $texts[0];
6240 }
6241
6242 /**
6243 * Returns true if the given array, presumed to be generated by
6244 * serializeHalfParsedText(), is compatible with the current version of the
6245 * parser.
6246 *
6247 * @deprecated since 1.31
6248 * @param array $data
6249 *
6250 * @return bool
6251 */
6252 public function isValidHalfParsedText( $data ) {
6253 wfDeprecated( __METHOD__, '1.31' );
6254 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6255 }
6256
6257 /**
6258 * Parsed a width param of imagelink like 300px or 200x300px
6259 *
6260 * @param string $value
6261 * @param bool $parseHeight
6262 *
6263 * @return array
6264 * @since 1.20
6265 */
6266 public static function parseWidthParam( $value, $parseHeight = true ) {
6267 $parsedWidthParam = [];
6268 if ( $value === '' ) {
6269 return $parsedWidthParam;
6270 }
6271 $m = [];
6272 # (T15500) In both cases (width/height and width only),
6273 # permit trailing "px" for backward compatibility.
6274 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6275 $width = intval( $m[1] );
6276 $height = intval( $m[2] );
6277 $parsedWidthParam['width'] = $width;
6278 $parsedWidthParam['height'] = $height;
6279 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6280 $width = intval( $value );
6281 $parsedWidthParam['width'] = $width;
6282 }
6283 return $parsedWidthParam;
6284 }
6285
6286 /**
6287 * Lock the current instance of the parser.
6288 *
6289 * This is meant to stop someone from calling the parser
6290 * recursively and messing up all the strip state.
6291 *
6292 * @throws MWException If parser is in a parse
6293 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6294 */
6295 protected function lock() {
6296 if ( $this->mInParse ) {
6297 throw new MWException( "Parser state cleared while parsing. "
6298 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6299 }
6300
6301 // Save the backtrace when locking, so that if some code tries locking again,
6302 // we can print the lock owner's backtrace for easier debugging
6303 $e = new Exception;
6304 $this->mInParse = $e->getTraceAsString();
6305
6306 $recursiveCheck = new ScopedCallback( function () {
6307 $this->mInParse = false;
6308 } );
6309
6310 return $recursiveCheck;
6311 }
6312
6313 /**
6314 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6315 *
6316 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6317 * or if there is more than one <p/> tag in the input HTML.
6318 *
6319 * @param string $html
6320 * @return string
6321 * @since 1.24
6322 */
6323 public static function stripOuterParagraph( $html ) {
6324 $m = [];
6325 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6326 if ( strpos( $m[1], '</p>' ) === false ) {
6327 $html = $m[1];
6328 }
6329 }
6330
6331 return $html;
6332 }
6333
6334 /**
6335 * Return this parser if it is not doing anything, otherwise
6336 * get a fresh parser. You can use this method by doing
6337 * $myParser = $wgParser->getFreshParser(), or more simply
6338 * $wgParser->getFreshParser()->parse( ... );
6339 * if you're unsure if $wgParser is safe to use.
6340 *
6341 * @since 1.24
6342 * @return Parser A parser object that is not parsing anything
6343 */
6344 public function getFreshParser() {
6345 if ( $this->mInParse ) {
6346 return $this->factory->create();
6347 } else {
6348 return $this;
6349 }
6350 }
6351
6352 /**
6353 * Set's up the PHP implementation of OOUI for use in this request
6354 * and instructs OutputPage to enable OOUI for itself.
6355 *
6356 * @since 1.26
6357 */
6358 public function enableOOUI() {
6359 OutputPage::setupOOUI();
6360 $this->mOutput->setEnableOOUI( true );
6361 }
6362 }
Wiklou, le Wiki du Wiklou