dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Improve docs for Title::getInternalURL/getCanonicalURL"
[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( $this->mDoubleUnderscores['nocontentconvert'] ) )
1424 && !$this->mOptions->getInterfaceMessage()
1425 ) {
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 $text = $this->mStripState->unstripNoWiki( $text );
1433
1434 if ( $isMain ) {
1435 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1436 }
1437
1438 $text = $this->replaceTransparentTags( $text );
1439 $text = $this->mStripState->unstripGeneral( $text );
1440
1441 $text = Sanitizer::normalizeCharReferences( $text );
1442
1443 if ( MWTidy::isEnabled() ) {
1444 if ( $this->mOptions->getTidy() ) {
1445 $text = MWTidy::tidy( $text );
1446 }
1447 } else {
1448 # attempt to sanitize at least some nesting problems
1449 # (T4702 and quite a few others)
1450 # This code path is buggy and deprecated!
1451 wfDeprecated( 'disabling tidy', '1.33' );
1452 $tidyregs = [
1453 # ''Something [http://www.cool.com cool''] -->
1454 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1455 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1456 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1457 # fix up an anchor inside another anchor, only
1458 # at least for a single single nested link (T5695)
1459 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1460 '\\1\\2</a>\\3</a>\\1\\4</a>',
1461 # fix div inside inline elements- doBlockLevels won't wrap a line which
1462 # contains a div, so fix it up here; replace
1463 # div with escaped text
1464 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1465 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1466 # remove empty italic or bold tag pairs, some
1467 # introduced by rules above
1468 '/<([bi])><\/\\1>/' => '',
1469 ];
1470
1471 $text = preg_replace(
1472 array_keys( $tidyregs ),
1473 array_values( $tidyregs ),
1474 $text );
1475 }
1476
1477 if ( $isMain ) {
1478 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1479 }
1480
1481 return $text;
1482 }
1483
1484 /**
1485 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1486 * magic external links.
1487 *
1488 * DML
1489 * @private
1490 *
1491 * @param string $text
1492 *
1493 * @return string
1494 */
1495 public function doMagicLinks( $text ) {
1496 $prots = wfUrlProtocolsWithoutProtRel();
1497 $urlChar = self::EXT_LINK_URL_CLASS;
1498 $addr = self::EXT_LINK_ADDR;
1499 $space = self::SPACE_NOT_NL; # non-newline space
1500 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1501 $spaces = "$space++"; # possessive match of 1 or more spaces
1502 $text = preg_replace_callback(
1503 '!(?: # Start cases
1504 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1505 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1506 (\b # m[3]: Free external links
1507 (?i:$prots)
1508 ($addr$urlChar*) # m[4]: Post-protocol path
1509 ) |
1510 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1511 ([0-9]+)\b |
1512 \bISBN $spaces ( # m[6]: ISBN, capture number
1513 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1514 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1515 [0-9Xx] # check digit
1516 )\b
1517 )!xu", [ $this, 'magicLinkCallback' ], $text );
1518 return $text;
1519 }
1520
1521 /**
1522 * @throws MWException
1523 * @param array $m
1524 * @return string HTML
1525 */
1526 public function magicLinkCallback( $m ) {
1527 if ( isset( $m[1] ) && $m[1] !== '' ) {
1528 # Skip anchor
1529 return $m[0];
1530 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1531 # Skip HTML element
1532 return $m[0];
1533 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1534 # Free external link
1535 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1536 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1537 # RFC or PMID
1538 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1539 if ( !$this->mOptions->getMagicRFCLinks() ) {
1540 return $m[0];
1541 }
1542 $keyword = 'RFC';
1543 $urlmsg = 'rfcurl';
1544 $cssClass = 'mw-magiclink-rfc';
1545 $trackingCat = 'magiclink-tracking-rfc';
1546 $id = $m[5];
1547 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1548 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1549 return $m[0];
1550 }
1551 $keyword = 'PMID';
1552 $urlmsg = 'pubmedurl';
1553 $cssClass = 'mw-magiclink-pmid';
1554 $trackingCat = 'magiclink-tracking-pmid';
1555 $id = $m[5];
1556 } else {
1557 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1558 substr( $m[0], 0, 20 ) . '"' );
1559 }
1560 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1561 $this->addTrackingCategory( $trackingCat );
1562 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1563 } elseif ( isset( $m[6] ) && $m[6] !== ''
1564 && $this->mOptions->getMagicISBNLinks()
1565 ) {
1566 # ISBN
1567 $isbn = $m[6];
1568 $space = self::SPACE_NOT_NL; # non-newline space
1569 $isbn = preg_replace( "/$space/", ' ', $isbn );
1570 $num = strtr( $isbn, [
1571 '-' => '',
1572 ' ' => '',
1573 'x' => 'X',
1574 ] );
1575 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1576 return $this->getLinkRenderer()->makeKnownLink(
1577 SpecialPage::getTitleFor( 'Booksources', $num ),
1578 "ISBN $isbn",
1579 [
1580 'class' => 'internal mw-magiclink-isbn',
1581 'title' => false // suppress title attribute
1582 ]
1583 );
1584 } else {
1585 return $m[0];
1586 }
1587 }
1588
1589 /**
1590 * Make a free external link, given a user-supplied URL
1591 *
1592 * @param string $url
1593 * @param int $numPostProto
1594 * The number of characters after the protocol.
1595 * @return string HTML
1596 * @private
1597 */
1598 public function makeFreeExternalLink( $url, $numPostProto ) {
1599 $trail = '';
1600
1601 # The characters '<' and '>' (which were escaped by
1602 # removeHTMLtags()) should not be included in
1603 # URLs, per RFC 2396.
1604 # Make &nbsp; terminate a URL as well (bug T84937)
1605 $m2 = [];
1606 if ( preg_match(
1607 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1608 $url,
1609 $m2,
1610 PREG_OFFSET_CAPTURE
1611 ) ) {
1612 $trail = substr( $url, $m2[0][1] ) . $trail;
1613 $url = substr( $url, 0, $m2[0][1] );
1614 }
1615
1616 # Move trailing punctuation to $trail
1617 $sep = ',;\.:!?';
1618 # If there is no left bracket, then consider right brackets fair game too
1619 if ( strpos( $url, '(' ) === false ) {
1620 $sep .= ')';
1621 }
1622
1623 $urlRev = strrev( $url );
1624 $numSepChars = strspn( $urlRev, $sep );
1625 # Don't break a trailing HTML entity by moving the ; into $trail
1626 # This is in hot code, so use substr_compare to avoid having to
1627 # create a new string object for the comparison
1628 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1629 # more optimization: instead of running preg_match with a $
1630 # anchor, which can be slow, do the match on the reversed
1631 # string starting at the desired offset.
1632 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1633 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1634 $numSepChars--;
1635 }
1636 }
1637 if ( $numSepChars ) {
1638 $trail = substr( $url, -$numSepChars ) . $trail;
1639 $url = substr( $url, 0, -$numSepChars );
1640 }
1641
1642 # Verify that we still have a real URL after trail removal, and
1643 # not just lone protocol
1644 if ( strlen( $trail ) >= $numPostProto ) {
1645 return $url . $trail;
1646 }
1647
1648 $url = Sanitizer::cleanUrl( $url );
1649
1650 # Is this an external image?
1651 $text = $this->maybeMakeExternalImage( $url );
1652 if ( $text === false ) {
1653 # Not an image, make a link
1654 $text = Linker::makeExternalLink( $url,
1655 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1656 true, 'free',
1657 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1658 # Register it in the output object...
1659 $this->mOutput->addExternalLink( $url );
1660 }
1661 return $text . $trail;
1662 }
1663
1664 /**
1665 * Parse headers and return html
1666 *
1667 * @private
1668 *
1669 * @param string $text
1670 *
1671 * @return string
1672 */
1673 public function doHeadings( $text ) {
1674 for ( $i = 6; $i >= 1; --$i ) {
1675 $h = str_repeat( '=', $i );
1676 // Trim non-newline whitespace from headings
1677 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1678 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1679 }
1680 return $text;
1681 }
1682
1683 /**
1684 * Replace single quotes with HTML markup
1685 * @private
1686 *
1687 * @param string $text
1688 *
1689 * @return string The altered text
1690 */
1691 public function doAllQuotes( $text ) {
1692 $outtext = '';
1693 $lines = StringUtils::explode( "\n", $text );
1694 foreach ( $lines as $line ) {
1695 $outtext .= $this->doQuotes( $line ) . "\n";
1696 }
1697 $outtext = substr( $outtext, 0, -1 );
1698 return $outtext;
1699 }
1700
1701 /**
1702 * Helper function for doAllQuotes()
1703 *
1704 * @param string $text
1705 *
1706 * @return string
1707 */
1708 public function doQuotes( $text ) {
1709 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1710 $countarr = count( $arr );
1711 if ( $countarr == 1 ) {
1712 return $text;
1713 }
1714
1715 // First, do some preliminary work. This may shift some apostrophes from
1716 // being mark-up to being text. It also counts the number of occurrences
1717 // of bold and italics mark-ups.
1718 $numbold = 0;
1719 $numitalics = 0;
1720 for ( $i = 1; $i < $countarr; $i += 2 ) {
1721 $thislen = strlen( $arr[$i] );
1722 // If there are ever four apostrophes, assume the first is supposed to
1723 // be text, and the remaining three constitute mark-up for bold text.
1724 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1725 if ( $thislen == 4 ) {
1726 $arr[$i - 1] .= "'";
1727 $arr[$i] = "'''";
1728 $thislen = 3;
1729 } elseif ( $thislen > 5 ) {
1730 // If there are more than 5 apostrophes in a row, assume they're all
1731 // text except for the last 5.
1732 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1733 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1734 $arr[$i] = "'''''";
1735 $thislen = 5;
1736 }
1737 // Count the number of occurrences of bold and italics mark-ups.
1738 if ( $thislen == 2 ) {
1739 $numitalics++;
1740 } elseif ( $thislen == 3 ) {
1741 $numbold++;
1742 } elseif ( $thislen == 5 ) {
1743 $numitalics++;
1744 $numbold++;
1745 }
1746 }
1747
1748 // If there is an odd number of both bold and italics, it is likely
1749 // that one of the bold ones was meant to be an apostrophe followed
1750 // by italics. Which one we cannot know for certain, but it is more
1751 // likely to be one that has a single-letter word before it.
1752 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1753 $firstsingleletterword = -1;
1754 $firstmultiletterword = -1;
1755 $firstspace = -1;
1756 for ( $i = 1; $i < $countarr; $i += 2 ) {
1757 if ( strlen( $arr[$i] ) == 3 ) {
1758 $x1 = substr( $arr[$i - 1], -1 );
1759 $x2 = substr( $arr[$i - 1], -2, 1 );
1760 if ( $x1 === ' ' ) {
1761 if ( $firstspace == -1 ) {
1762 $firstspace = $i;
1763 }
1764 } elseif ( $x2 === ' ' ) {
1765 $firstsingleletterword = $i;
1766 // if $firstsingleletterword is set, we don't
1767 // look at the other options, so we can bail early.
1768 break;
1769 } elseif ( $firstmultiletterword == -1 ) {
1770 $firstmultiletterword = $i;
1771 }
1772 }
1773 }
1774
1775 // If there is a single-letter word, use it!
1776 if ( $firstsingleletterword > -1 ) {
1777 $arr[$firstsingleletterword] = "''";
1778 $arr[$firstsingleletterword - 1] .= "'";
1779 } elseif ( $firstmultiletterword > -1 ) {
1780 // If not, but there's a multi-letter word, use that one.
1781 $arr[$firstmultiletterword] = "''";
1782 $arr[$firstmultiletterword - 1] .= "'";
1783 } elseif ( $firstspace > -1 ) {
1784 // ... otherwise use the first one that has neither.
1785 // (notice that it is possible for all three to be -1 if, for example,
1786 // there is only one pentuple-apostrophe in the line)
1787 $arr[$firstspace] = "''";
1788 $arr[$firstspace - 1] .= "'";
1789 }
1790 }
1791
1792 // Now let's actually convert our apostrophic mush to HTML!
1793 $output = '';
1794 $buffer = '';
1795 $state = '';
1796 $i = 0;
1797 foreach ( $arr as $r ) {
1798 if ( ( $i % 2 ) == 0 ) {
1799 if ( $state === 'both' ) {
1800 $buffer .= $r;
1801 } else {
1802 $output .= $r;
1803 }
1804 } else {
1805 $thislen = strlen( $r );
1806 if ( $thislen == 2 ) {
1807 if ( $state === 'i' ) {
1808 $output .= '</i>';
1809 $state = '';
1810 } elseif ( $state === 'bi' ) {
1811 $output .= '</i>';
1812 $state = 'b';
1813 } elseif ( $state === 'ib' ) {
1814 $output .= '</b></i><b>';
1815 $state = 'b';
1816 } elseif ( $state === 'both' ) {
1817 $output .= '<b><i>' . $buffer . '</i>';
1818 $state = 'b';
1819 } else { // $state can be 'b' or ''
1820 $output .= '<i>';
1821 $state .= 'i';
1822 }
1823 } elseif ( $thislen == 3 ) {
1824 if ( $state === 'b' ) {
1825 $output .= '</b>';
1826 $state = '';
1827 } elseif ( $state === 'bi' ) {
1828 $output .= '</i></b><i>';
1829 $state = 'i';
1830 } elseif ( $state === 'ib' ) {
1831 $output .= '</b>';
1832 $state = 'i';
1833 } elseif ( $state === 'both' ) {
1834 $output .= '<i><b>' . $buffer . '</b>';
1835 $state = 'i';
1836 } else { // $state can be 'i' or ''
1837 $output .= '<b>';
1838 $state .= 'b';
1839 }
1840 } elseif ( $thislen == 5 ) {
1841 if ( $state === 'b' ) {
1842 $output .= '</b><i>';
1843 $state = 'i';
1844 } elseif ( $state === 'i' ) {
1845 $output .= '</i><b>';
1846 $state = 'b';
1847 } elseif ( $state === 'bi' ) {
1848 $output .= '</i></b>';
1849 $state = '';
1850 } elseif ( $state === 'ib' ) {
1851 $output .= '</b></i>';
1852 $state = '';
1853 } elseif ( $state === 'both' ) {
1854 $output .= '<i><b>' . $buffer . '</b></i>';
1855 $state = '';
1856 } else { // ($state == '')
1857 $buffer = '';
1858 $state = 'both';
1859 }
1860 }
1861 }
1862 $i++;
1863 }
1864 // Now close all remaining tags. Notice that the order is important.
1865 if ( $state === 'b' || $state === 'ib' ) {
1866 $output .= '</b>';
1867 }
1868 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1869 $output .= '</i>';
1870 }
1871 if ( $state === 'bi' ) {
1872 $output .= '</b>';
1873 }
1874 // There might be lonely ''''', so make sure we have a buffer
1875 if ( $state === 'both' && $buffer ) {
1876 $output .= '<b><i>' . $buffer . '</i></b>';
1877 }
1878 return $output;
1879 }
1880
1881 /**
1882 * Replace external links (REL)
1883 *
1884 * Note: this is all very hackish and the order of execution matters a lot.
1885 * Make sure to run tests/parser/parserTests.php if you change this code.
1886 *
1887 * @private
1888 *
1889 * @param string $text
1890 *
1891 * @throws MWException
1892 * @return string
1893 */
1894 public function replaceExternalLinks( $text ) {
1895 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1896 if ( $bits === false ) {
1897 throw new MWException( "PCRE needs to be compiled with "
1898 . "--enable-unicode-properties in order for MediaWiki to function" );
1899 }
1900 $s = array_shift( $bits );
1901
1902 $i = 0;
1903 while ( $i < count( $bits ) ) {
1904 $url = $bits[$i++];
1905 $i++; // protocol
1906 $text = $bits[$i++];
1907 $trail = $bits[$i++];
1908
1909 # The characters '<' and '>' (which were escaped by
1910 # removeHTMLtags()) should not be included in
1911 # URLs, per RFC 2396.
1912 $m2 = [];
1913 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1914 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1915 $url = substr( $url, 0, $m2[0][1] );
1916 }
1917
1918 # If the link text is an image URL, replace it with an <img> tag
1919 # This happened by accident in the original parser, but some people used it extensively
1920 $img = $this->maybeMakeExternalImage( $text );
1921 if ( $img !== false ) {
1922 $text = $img;
1923 }
1924
1925 $dtrail = '';
1926
1927 # Set linktype for CSS
1928 $linktype = 'text';
1929
1930 # No link text, e.g. [http://domain.tld/some.link]
1931 if ( $text == '' ) {
1932 # Autonumber
1933 $langObj = $this->getTargetLanguage();
1934 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1935 $linktype = 'autonumber';
1936 } else {
1937 # Have link text, e.g. [http://domain.tld/some.link text]s
1938 # Check for trail
1939 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1940 }
1941
1942 // Excluding protocol-relative URLs may avoid many false positives.
1943 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1944 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1945 }
1946
1947 $url = Sanitizer::cleanUrl( $url );
1948
1949 # Use the encoded URL
1950 # This means that users can paste URLs directly into the text
1951 # Funny characters like ö aren't valid in URLs anyway
1952 # This was changed in August 2004
1953 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1954 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1955
1956 # Register link in the output object.
1957 $this->mOutput->addExternalLink( $url );
1958 }
1959
1960 return $s;
1961 }
1962
1963 /**
1964 * Get the rel attribute for a particular external link.
1965 *
1966 * @since 1.21
1967 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1968 * nofollow if appropriate
1969 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1970 * @return string|null Rel attribute for $url
1971 */
1972 public static function getExternalLinkRel( $url = false, $title = null ) {
1973 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1974 $ns = $title ? $title->getNamespace() : false;
1975 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1976 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1977 ) {
1978 return 'nofollow';
1979 }
1980 return null;
1981 }
1982
1983 /**
1984 * Get an associative array of additional HTML attributes appropriate for a
1985 * particular external link. This currently may include rel => nofollow
1986 * (depending on configuration, namespace, and the URL's domain) and/or a
1987 * target attribute (depending on configuration).
1988 *
1989 * @param string $url URL to extract the domain from for rel =>
1990 * nofollow if appropriate
1991 * @return array Associative array of HTML attributes
1992 */
1993 public function getExternalLinkAttribs( $url ) {
1994 $attribs = [];
1995 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1996
1997 $target = $this->mOptions->getExternalLinkTarget();
1998 if ( $target ) {
1999 $attribs['target'] = $target;
2000 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2001 // T133507. New windows can navigate parent cross-origin.
2002 // Including noreferrer due to lacking browser
2003 // support of noopener. Eventually noreferrer should be removed.
2004 if ( $rel !== '' ) {
2005 $rel .= ' ';
2006 }
2007 $rel .= 'noreferrer noopener';
2008 }
2009 }
2010 $attribs['rel'] = $rel;
2011 return $attribs;
2012 }
2013
2014 /**
2015 * Replace unusual escape codes in a URL with their equivalent characters
2016 *
2017 * This generally follows the syntax defined in RFC 3986, with special
2018 * consideration for HTTP query strings.
2019 *
2020 * @param string $url
2021 * @return string
2022 */
2023 public static function normalizeLinkUrl( $url ) {
2024 # Test for RFC 3986 IPv6 syntax
2025 $scheme = '[a-z][a-z0-9+.-]*:';
2026 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2027 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2028 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2029 IP::isValid( rawurldecode( $m[1] ) )
2030 ) {
2031 $isIPv6 = rawurldecode( $m[1] );
2032 } else {
2033 $isIPv6 = false;
2034 }
2035
2036 # Make sure unsafe characters are encoded
2037 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2038 function ( $m ) {
2039 return rawurlencode( $m[0] );
2040 },
2041 $url
2042 );
2043
2044 $ret = '';
2045 $end = strlen( $url );
2046
2047 # Fragment part - 'fragment'
2048 $start = strpos( $url, '#' );
2049 if ( $start !== false && $start < $end ) {
2050 $ret = self::normalizeUrlComponent(
2051 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2052 $end = $start;
2053 }
2054
2055 # Query part - 'query' minus &=+;
2056 $start = strpos( $url, '?' );
2057 if ( $start !== false && $start < $end ) {
2058 $ret = self::normalizeUrlComponent(
2059 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2060 $end = $start;
2061 }
2062
2063 # Scheme and path part - 'pchar'
2064 # (we assume no userinfo or encoded colons in the host)
2065 $ret = self::normalizeUrlComponent(
2066 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2067
2068 # Fix IPv6 syntax
2069 if ( $isIPv6 !== false ) {
2070 $ipv6Host = "%5B({$isIPv6})%5D";
2071 $ret = preg_replace(
2072 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2073 "$1[$2]",
2074 $ret
2075 );
2076 }
2077
2078 return $ret;
2079 }
2080
2081 private static function normalizeUrlComponent( $component, $unsafe ) {
2082 $callback = function ( $matches ) use ( $unsafe ) {
2083 $char = urldecode( $matches[0] );
2084 $ord = ord( $char );
2085 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2086 # Unescape it
2087 return $char;
2088 } else {
2089 # Leave it escaped, but use uppercase for a-f
2090 return strtoupper( $matches[0] );
2091 }
2092 };
2093 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2094 }
2095
2096 /**
2097 * make an image if it's allowed, either through the global
2098 * option, through the exception, or through the on-wiki whitelist
2099 *
2100 * @param string $url
2101 *
2102 * @return string
2103 */
2104 private function maybeMakeExternalImage( $url ) {
2105 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2106 $imagesexception = !empty( $imagesfrom );
2107 $text = false;
2108 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2109 if ( $imagesexception && is_array( $imagesfrom ) ) {
2110 $imagematch = false;
2111 foreach ( $imagesfrom as $match ) {
2112 if ( strpos( $url, $match ) === 0 ) {
2113 $imagematch = true;
2114 break;
2115 }
2116 }
2117 } elseif ( $imagesexception ) {
2118 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2119 } else {
2120 $imagematch = false;
2121 }
2122
2123 if ( $this->mOptions->getAllowExternalImages()
2124 || ( $imagesexception && $imagematch )
2125 ) {
2126 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2127 # Image found
2128 $text = Linker::makeExternalImage( $url );
2129 }
2130 }
2131 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2132 && preg_match( self::EXT_IMAGE_REGEX, $url )
2133 ) {
2134 $whitelist = explode(
2135 "\n",
2136 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2137 );
2138
2139 foreach ( $whitelist as $entry ) {
2140 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2141 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2142 continue;
2143 }
2144 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2145 # Image matches a whitelist entry
2146 $text = Linker::makeExternalImage( $url );
2147 break;
2148 }
2149 }
2150 }
2151 return $text;
2152 }
2153
2154 /**
2155 * Process [[ ]] wikilinks
2156 *
2157 * @param string $s
2158 *
2159 * @return string Processed text
2160 *
2161 * @private
2162 */
2163 public function replaceInternalLinks( $s ) {
2164 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2165 return $s;
2166 }
2167
2168 /**
2169 * Process [[ ]] wikilinks (RIL)
2170 * @param string &$s
2171 * @throws MWException
2172 * @return LinkHolderArray
2173 *
2174 * @private
2175 */
2176 public function replaceInternalLinks2( &$s ) {
2177 static $tc = false, $e1, $e1_img;
2178 # the % is needed to support urlencoded titles as well
2179 if ( !$tc ) {
2180 $tc = Title::legalChars() . '#%';
2181 # Match a link having the form [[namespace:link|alternate]]trail
2182 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2183 # Match cases where there is no "]]", which might still be images
2184 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2185 }
2186
2187 $holders = new LinkHolderArray( $this );
2188
2189 # split the entire text string on occurrences of [[
2190 $a = StringUtils::explode( '[[', ' ' . $s );
2191 # get the first element (all text up to first [[), and remove the space we added
2192 $s = $a->current();
2193 $a->next();
2194 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2195 $s = substr( $s, 1 );
2196
2197 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2198 $e2 = null;
2199 if ( $useLinkPrefixExtension ) {
2200 # Match the end of a line for a word that's not followed by whitespace,
2201 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2202 $charset = $this->contLang->linkPrefixCharset();
2203 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2204 }
2205
2206 if ( is_null( $this->mTitle ) ) {
2207 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2208 }
2209 $nottalk = !$this->mTitle->isTalkPage();
2210
2211 if ( $useLinkPrefixExtension ) {
2212 $m = [];
2213 if ( preg_match( $e2, $s, $m ) ) {
2214 $first_prefix = $m[2];
2215 } else {
2216 $first_prefix = false;
2217 }
2218 } else {
2219 $prefix = '';
2220 }
2221
2222 $useSubpages = $this->areSubpagesAllowed();
2223
2224 # Loop for each link
2225 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2226 # Check for excessive memory usage
2227 if ( $holders->isBig() ) {
2228 # Too big
2229 # Do the existence check, replace the link holders and clear the array
2230 $holders->replace( $s );
2231 $holders->clear();
2232 }
2233
2234 if ( $useLinkPrefixExtension ) {
2235 if ( preg_match( $e2, $s, $m ) ) {
2236 list( , $s, $prefix ) = $m;
2237 } else {
2238 $prefix = '';
2239 }
2240 # first link
2241 if ( $first_prefix ) {
2242 $prefix = $first_prefix;
2243 $first_prefix = false;
2244 }
2245 }
2246
2247 $might_be_img = false;
2248
2249 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2250 $text = $m[2];
2251 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2252 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2253 # the real problem is with the $e1 regex
2254 # See T1500.
2255 # Still some problems for cases where the ] is meant to be outside punctuation,
2256 # and no image is in sight. See T4095.
2257 if ( $text !== ''
2258 && substr( $m[3], 0, 1 ) === ']'
2259 && strpos( $text, '[' ) !== false
2260 ) {
2261 $text .= ']'; # so that replaceExternalLinks($text) works later
2262 $m[3] = substr( $m[3], 1 );
2263 }
2264 # fix up urlencoded title texts
2265 if ( strpos( $m[1], '%' ) !== false ) {
2266 # Should anchors '#' also be rejected?
2267 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2268 }
2269 $trail = $m[3];
2270 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2271 # Invalid, but might be an image with a link in its caption
2272 $might_be_img = true;
2273 $text = $m[2];
2274 if ( strpos( $m[1], '%' ) !== false ) {
2275 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2276 }
2277 $trail = "";
2278 } else { # Invalid form; output directly
2279 $s .= $prefix . '[[' . $line;
2280 continue;
2281 }
2282
2283 $origLink = ltrim( $m[1], ' ' );
2284
2285 # Don't allow internal links to pages containing
2286 # PROTO: where PROTO is a valid URL protocol; these
2287 # should be external links.
2288 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2289 $s .= $prefix . '[[' . $line;
2290 continue;
2291 }
2292
2293 # Make subpage if necessary
2294 if ( $useSubpages ) {
2295 $link = $this->maybeDoSubpageLink( $origLink, $text );
2296 } else {
2297 $link = $origLink;
2298 }
2299
2300 // \x7f isn't a default legal title char, so most likely strip
2301 // markers will force us into the "invalid form" path above. But,
2302 // just in case, let's assert that xmlish tags aren't valid in
2303 // the title position.
2304 $unstrip = $this->mStripState->killMarkers( $link );
2305 $noMarkers = ( $unstrip === $link );
2306
2307 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2308 if ( $nt === null ) {
2309 $s .= $prefix . '[[' . $line;
2310 continue;
2311 }
2312
2313 $ns = $nt->getNamespace();
2314 $iw = $nt->getInterwiki();
2315
2316 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2317
2318 if ( $might_be_img ) { # if this is actually an invalid link
2319 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2320 $found = false;
2321 while ( true ) {
2322 # look at the next 'line' to see if we can close it there
2323 $a->next();
2324 $next_line = $a->current();
2325 if ( $next_line === false || $next_line === null ) {
2326 break;
2327 }
2328 $m = explode( ']]', $next_line, 3 );
2329 if ( count( $m ) == 3 ) {
2330 # the first ]] closes the inner link, the second the image
2331 $found = true;
2332 $text .= "[[{$m[0]}]]{$m[1]}";
2333 $trail = $m[2];
2334 break;
2335 } elseif ( count( $m ) == 2 ) {
2336 # if there's exactly one ]] that's fine, we'll keep looking
2337 $text .= "[[{$m[0]}]]{$m[1]}";
2338 } else {
2339 # if $next_line is invalid too, we need look no further
2340 $text .= '[[' . $next_line;
2341 break;
2342 }
2343 }
2344 if ( !$found ) {
2345 # we couldn't find the end of this imageLink, so output it raw
2346 # but don't ignore what might be perfectly normal links in the text we've examined
2347 $holders->merge( $this->replaceInternalLinks2( $text ) );
2348 $s .= "{$prefix}[[$link|$text";
2349 # note: no $trail, because without an end, there *is* no trail
2350 continue;
2351 }
2352 } else { # it's not an image, so output it raw
2353 $s .= "{$prefix}[[$link|$text";
2354 # note: no $trail, because without an end, there *is* no trail
2355 continue;
2356 }
2357 }
2358
2359 $wasblank = ( $text == '' );
2360 if ( $wasblank ) {
2361 $text = $link;
2362 if ( !$noforce ) {
2363 # Strip off leading ':'
2364 $text = substr( $text, 1 );
2365 }
2366 } else {
2367 # T6598 madness. Handle the quotes only if they come from the alternate part
2368 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2369 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2370 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2371 $text = $this->doQuotes( $text );
2372 }
2373
2374 # Link not escaped by : , create the various objects
2375 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2376 # Interwikis
2377 if (
2378 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2379 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2380 in_array( $iw, $this->siteConfig->get( 'ExtraInterlanguageLinkPrefixes' ) )
2381 )
2382 ) {
2383 # T26502: filter duplicates
2384 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2385 $this->mLangLinkLanguages[$iw] = true;
2386 $this->mOutput->addLanguageLink( $nt->getFullText() );
2387 }
2388
2389 /**
2390 * Strip the whitespace interwiki links produce, see T10897
2391 */
2392 $s = rtrim( $s . $prefix ) . $trail; # T175416
2393 continue;
2394 }
2395
2396 if ( $ns == NS_FILE ) {
2397 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2398 if ( $wasblank ) {
2399 # if no parameters were passed, $text
2400 # becomes something like "File:Foo.png",
2401 # which we don't want to pass on to the
2402 # image generator
2403 $text = '';
2404 } else {
2405 # recursively parse links inside the image caption
2406 # actually, this will parse them in any other parameters, too,
2407 # but it might be hard to fix that, and it doesn't matter ATM
2408 $text = $this->replaceExternalLinks( $text );
2409 $holders->merge( $this->replaceInternalLinks2( $text ) );
2410 }
2411 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2412 $s .= $prefix . $this->armorLinks(
2413 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2414 continue;
2415 }
2416 } elseif ( $ns == NS_CATEGORY ) {
2417 /**
2418 * Strip the whitespace Category links produce, see T2087
2419 */
2420 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2421
2422 if ( $wasblank ) {
2423 $sortkey = $this->getDefaultSort();
2424 } else {
2425 $sortkey = $text;
2426 }
2427 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2428 $sortkey = str_replace( "\n", '', $sortkey );
2429 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2430 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2431
2432 continue;
2433 }
2434 }
2435
2436 # Self-link checking. For some languages, variants of the title are checked in
2437 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2438 # for linking to a different variant.
2439 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2440 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2441 continue;
2442 }
2443
2444 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2445 # @todo FIXME: Should do batch file existence checks, see comment below
2446 if ( $ns == NS_MEDIA ) {
2447 # Give extensions a chance to select the file revision for us
2448 $options = [];
2449 $descQuery = false;
2450 Hooks::run( 'BeforeParserFetchFileAndTitle',
2451 [ $this, $nt, &$options, &$descQuery ] );
2452 # Fetch and register the file (file title may be different via hooks)
2453 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2454 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2455 $s .= $prefix . $this->armorLinks(
2456 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2457 continue;
2458 }
2459
2460 # Some titles, such as valid special pages or files in foreign repos, should
2461 # be shown as bluelinks even though they're not included in the page table
2462 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2463 # batch file existence checks for NS_FILE and NS_MEDIA
2464 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2465 $this->mOutput->addLink( $nt );
2466 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2467 } else {
2468 # Links will be added to the output link list after checking
2469 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2470 }
2471 }
2472 return $holders;
2473 }
2474
2475 /**
2476 * Render a forced-blue link inline; protect against double expansion of
2477 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2478 * Since this little disaster has to split off the trail text to avoid
2479 * breaking URLs in the following text without breaking trails on the
2480 * wiki links, it's been made into a horrible function.
2481 *
2482 * @param Title $nt
2483 * @param string $text
2484 * @param string $trail
2485 * @param string $prefix
2486 * @return string HTML-wikitext mix oh yuck
2487 */
2488 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2489 list( $inside, $trail ) = Linker::splitTrail( $trail );
2490
2491 if ( $text == '' ) {
2492 $text = htmlspecialchars( $nt->getPrefixedText() );
2493 }
2494
2495 $link = $this->getLinkRenderer()->makeKnownLink(
2496 $nt, new HtmlArmor( "$prefix$text$inside" )
2497 );
2498
2499 return $this->armorLinks( $link ) . $trail;
2500 }
2501
2502 /**
2503 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2504 * going to go through further parsing steps before inline URL expansion.
2505 *
2506 * Not needed quite as much as it used to be since free links are a bit
2507 * more sensible these days. But bracketed links are still an issue.
2508 *
2509 * @param string $text More-or-less HTML
2510 * @return string Less-or-more HTML with NOPARSE bits
2511 */
2512 public function armorLinks( $text ) {
2513 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2514 self::MARKER_PREFIX . "NOPARSE$1", $text );
2515 }
2516
2517 /**
2518 * Return true if subpage links should be expanded on this page.
2519 * @return bool
2520 */
2521 public function areSubpagesAllowed() {
2522 # Some namespaces don't allow subpages
2523 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2524 }
2525
2526 /**
2527 * Handle link to subpage if necessary
2528 *
2529 * @param string $target The source of the link
2530 * @param string &$text The link text, modified as necessary
2531 * @return string The full name of the link
2532 * @private
2533 */
2534 public function maybeDoSubpageLink( $target, &$text ) {
2535 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2536 }
2537
2538 /**
2539 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2540 *
2541 * @param string $text
2542 * @param bool $linestart Whether or not this is at the start of a line.
2543 * @private
2544 * @return string The lists rendered as HTML
2545 */
2546 public function doBlockLevels( $text, $linestart ) {
2547 return BlockLevelPass::doBlockLevels( $text, $linestart );
2548 }
2549
2550 /**
2551 * Return value of a magic variable (like PAGENAME)
2552 *
2553 * @private
2554 *
2555 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2556 * @param bool|PPFrame $frame
2557 *
2558 * @throws MWException
2559 * @return string
2560 */
2561 public function getVariableValue( $index, $frame = false ) {
2562 if ( is_null( $this->mTitle ) ) {
2563 // If no title set, bad things are going to happen
2564 // later. Title should always be set since this
2565 // should only be called in the middle of a parse
2566 // operation (but the unit-tests do funky stuff)
2567 throw new MWException( __METHOD__ . ' Should only be '
2568 . ' called while parsing (no title set)' );
2569 }
2570
2571 // Avoid PHP 7.1 warning from passing $this by reference
2572 $parser = $this;
2573
2574 /**
2575 * Some of these require message or data lookups and can be
2576 * expensive to check many times.
2577 */
2578 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] )
2579 && isset( $this->mVarCache[$index] )
2580 ) {
2581 return $this->mVarCache[$index];
2582 }
2583
2584 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2585 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2586
2587 // In miser mode, disable words that always cause double-parses on page save (T137900)
2588 static $slowRevWords = [ 'revisionid' => true ]; // @TODO: 'revisiontimestamp'
2589 if (
2590 isset( $slowRevWords[$index] ) &&
2591 $this->siteConfig->get( 'MiserMode' ) &&
2592 !$this->mOptions->getInterfaceMessage() &&
2593 // @TODO: disallow this word on all namespaces
2594 MWNamespace::isContent( $this->mTitle->getNamespace() )
2595 ) {
2596 return $this->mRevisionId ? '-' : '';
2597 };
2598
2599 $pageLang = $this->getFunctionLang();
2600
2601 switch ( $index ) {
2602 case '!':
2603 $value = '|';
2604 break;
2605 case 'currentmonth':
2606 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2607 break;
2608 case 'currentmonth1':
2609 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2610 break;
2611 case 'currentmonthname':
2612 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2613 break;
2614 case 'currentmonthnamegen':
2615 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2616 break;
2617 case 'currentmonthabbrev':
2618 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2619 break;
2620 case 'currentday':
2621 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2622 break;
2623 case 'currentday2':
2624 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2625 break;
2626 case 'localmonth':
2627 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2628 break;
2629 case 'localmonth1':
2630 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2631 break;
2632 case 'localmonthname':
2633 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2634 break;
2635 case 'localmonthnamegen':
2636 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2637 break;
2638 case 'localmonthabbrev':
2639 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2640 break;
2641 case 'localday':
2642 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2643 break;
2644 case 'localday2':
2645 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2646 break;
2647 case 'pagename':
2648 $value = wfEscapeWikiText( $this->mTitle->getText() );
2649 break;
2650 case 'pagenamee':
2651 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2652 break;
2653 case 'fullpagename':
2654 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2655 break;
2656 case 'fullpagenamee':
2657 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2658 break;
2659 case 'subpagename':
2660 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2661 break;
2662 case 'subpagenamee':
2663 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2664 break;
2665 case 'rootpagename':
2666 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2667 break;
2668 case 'rootpagenamee':
2669 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2670 ' ',
2671 '_',
2672 $this->mTitle->getRootText()
2673 ) ) );
2674 break;
2675 case 'basepagename':
2676 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2677 break;
2678 case 'basepagenamee':
2679 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2680 ' ',
2681 '_',
2682 $this->mTitle->getBaseText()
2683 ) ) );
2684 break;
2685 case 'talkpagename':
2686 if ( $this->mTitle->canHaveTalkPage() ) {
2687 $talkPage = $this->mTitle->getTalkPage();
2688 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2689 } else {
2690 $value = '';
2691 }
2692 break;
2693 case 'talkpagenamee':
2694 if ( $this->mTitle->canHaveTalkPage() ) {
2695 $talkPage = $this->mTitle->getTalkPage();
2696 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2697 } else {
2698 $value = '';
2699 }
2700 break;
2701 case 'subjectpagename':
2702 $subjPage = $this->mTitle->getSubjectPage();
2703 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2704 break;
2705 case 'subjectpagenamee':
2706 $subjPage = $this->mTitle->getSubjectPage();
2707 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2708 break;
2709 case 'pageid': // requested in T25427
2710 $pageid = $this->getTitle()->getArticleID();
2711 if ( $pageid == 0 ) {
2712 # 0 means the page doesn't exist in the database,
2713 # which means the user is previewing a new page.
2714 # The vary-revision flag must be set, because the magic word
2715 # will have a different value once the page is saved.
2716 $this->mOutput->setFlag( 'vary-revision' );
2717 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2718 }
2719 $value = $pageid ?: null;
2720 break;
2721 case 'revisionid':
2722 # Let the edit saving system know we should parse the page
2723 # *after* a revision ID has been assigned.
2724 $this->mOutput->setFlag( 'vary-revision-id' );
2725 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2726 $value = $this->mRevisionId;
2727
2728 if ( !$value ) {
2729 $rev = $this->getRevisionObject();
2730 if ( $rev ) {
2731 $value = $rev->getId();
2732 }
2733 }
2734
2735 if ( !$value ) {
2736 $value = $this->mOptions->getSpeculativeRevId();
2737 if ( $value ) {
2738 $this->mOutput->setSpeculativeRevIdUsed( $value );
2739 }
2740 }
2741 break;
2742 case 'revisionday':
2743 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2744 break;
2745 case 'revisionday2':
2746 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2747 break;
2748 case 'revisionmonth':
2749 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2750 break;
2751 case 'revisionmonth1':
2752 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2753 break;
2754 case 'revisionyear':
2755 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2756 break;
2757 case 'revisiontimestamp':
2758 # Let the edit saving system know we should parse the page
2759 # *after* a revision ID has been assigned. This is for null edits.
2760 $this->mOutput->setFlag( 'vary-revision' );
2761 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2762 $value = $this->getRevisionTimestamp();
2763 break;
2764 case 'revisionuser':
2765 # Let the edit saving system know we should parse the page
2766 # *after* a revision ID has been assigned for null edits.
2767 $this->mOutput->setFlag( 'vary-user' );
2768 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2769 $value = $this->getRevisionUser();
2770 break;
2771 case 'revisionsize':
2772 $value = $this->getRevisionSize();
2773 break;
2774 case 'namespace':
2775 $value = str_replace( '_', ' ',
2776 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2777 break;
2778 case 'namespacee':
2779 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2780 break;
2781 case 'namespacenumber':
2782 $value = $this->mTitle->getNamespace();
2783 break;
2784 case 'talkspace':
2785 $value = $this->mTitle->canHaveTalkPage()
2786 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2787 : '';
2788 break;
2789 case 'talkspacee':
2790 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2791 break;
2792 case 'subjectspace':
2793 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2794 break;
2795 case 'subjectspacee':
2796 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2797 break;
2798 case 'currentdayname':
2799 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2800 break;
2801 case 'currentyear':
2802 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2803 break;
2804 case 'currenttime':
2805 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2806 break;
2807 case 'currenthour':
2808 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2809 break;
2810 case 'currentweek':
2811 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2812 # int to remove the padding
2813 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2814 break;
2815 case 'currentdow':
2816 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2817 break;
2818 case 'localdayname':
2819 $value = $pageLang->getWeekdayName(
2820 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2821 );
2822 break;
2823 case 'localyear':
2824 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2825 break;
2826 case 'localtime':
2827 $value = $pageLang->time(
2828 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2829 false,
2830 false
2831 );
2832 break;
2833 case 'localhour':
2834 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2835 break;
2836 case 'localweek':
2837 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2838 # int to remove the padding
2839 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2840 break;
2841 case 'localdow':
2842 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2843 break;
2844 case 'numberofarticles':
2845 $value = $pageLang->formatNum( SiteStats::articles() );
2846 break;
2847 case 'numberoffiles':
2848 $value = $pageLang->formatNum( SiteStats::images() );
2849 break;
2850 case 'numberofusers':
2851 $value = $pageLang->formatNum( SiteStats::users() );
2852 break;
2853 case 'numberofactiveusers':
2854 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2855 break;
2856 case 'numberofpages':
2857 $value = $pageLang->formatNum( SiteStats::pages() );
2858 break;
2859 case 'numberofadmins':
2860 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2861 break;
2862 case 'numberofedits':
2863 $value = $pageLang->formatNum( SiteStats::edits() );
2864 break;
2865 case 'currenttimestamp':
2866 $value = wfTimestamp( TS_MW, $ts );
2867 break;
2868 case 'localtimestamp':
2869 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2870 break;
2871 case 'currentversion':
2872 $value = SpecialVersion::getVersion();
2873 break;
2874 case 'articlepath':
2875 return $this->siteConfig->get( 'ArticlePath' );
2876 case 'sitename':
2877 return $this->siteConfig->get( 'Sitename' );
2878 case 'server':
2879 return $this->siteConfig->get( 'Server' );
2880 case 'servername':
2881 return $this->siteConfig->get( 'ServerName' );
2882 case 'scriptpath':
2883 return $this->siteConfig->get( 'ScriptPath' );
2884 case 'stylepath':
2885 return $this->siteConfig->get( 'StylePath' );
2886 case 'directionmark':
2887 return $pageLang->getDirMark();
2888 case 'contentlanguage':
2889 return $this->siteConfig->get( 'LanguageCode' );
2890 case 'pagelanguage':
2891 $value = $pageLang->getCode();
2892 break;
2893 case 'cascadingsources':
2894 $value = CoreParserFunctions::cascadingsources( $this );
2895 break;
2896 default:
2897 $ret = null;
2898 Hooks::run(
2899 'ParserGetVariableValueSwitch',
2900 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2901 );
2902
2903 return $ret;
2904 }
2905
2906 if ( $index ) {
2907 $this->mVarCache[$index] = $value;
2908 }
2909
2910 return $value;
2911 }
2912
2913 /**
2914 * @param int $start
2915 * @param int $len
2916 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2917 * @param string $variable Parser variable name
2918 * @return string
2919 */
2920 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2921 # Get the timezone-adjusted timestamp to be used for this revision
2922 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2923 # Possibly set vary-revision if there is not yet an associated revision
2924 if ( !$this->getRevisionObject() ) {
2925 # Get the timezone-adjusted timestamp $mtts seconds in the future
2926 $resThen = substr(
2927 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2928 $start,
2929 $len
2930 );
2931
2932 if ( $resNow !== $resThen ) {
2933 # Let the edit saving system know we should parse the page
2934 # *after* a revision ID has been assigned. This is for null edits.
2935 $this->mOutput->setFlag( 'vary-revision' );
2936 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2937 }
2938 }
2939
2940 return $resNow;
2941 }
2942
2943 /**
2944 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2945 *
2946 * @private
2947 */
2948 public function initialiseVariables() {
2949 $variableIDs = $this->magicWordFactory->getVariableIDs();
2950 $substIDs = $this->magicWordFactory->getSubstIDs();
2951
2952 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
2953 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
2954 }
2955
2956 /**
2957 * Preprocess some wikitext and return the document tree.
2958 * This is the ghost of replace_variables().
2959 *
2960 * @param string $text The text to parse
2961 * @param int $flags Bitwise combination of:
2962 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2963 * included. Default is to assume a direct page view.
2964 *
2965 * The generated DOM tree must depend only on the input text and the flags.
2966 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2967 *
2968 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2969 * change in the DOM tree for a given text, must be passed through the section identifier
2970 * in the section edit link and thus back to extractSections().
2971 *
2972 * The output of this function is currently only cached in process memory, but a persistent
2973 * cache may be implemented at a later date which takes further advantage of these strict
2974 * dependency requirements.
2975 *
2976 * @return PPNode
2977 */
2978 public function preprocessToDom( $text, $flags = 0 ) {
2979 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2980 return $dom;
2981 }
2982
2983 /**
2984 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2985 *
2986 * @param string $s
2987 *
2988 * @return array
2989 */
2990 public static function splitWhitespace( $s ) {
2991 $ltrimmed = ltrim( $s );
2992 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2993 $trimmed = rtrim( $ltrimmed );
2994 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2995 if ( $diff > 0 ) {
2996 $w2 = substr( $ltrimmed, -$diff );
2997 } else {
2998 $w2 = '';
2999 }
3000 return [ $w1, $trimmed, $w2 ];
3001 }
3002
3003 /**
3004 * Replace magic variables, templates, and template arguments
3005 * with the appropriate text. Templates are substituted recursively,
3006 * taking care to avoid infinite loops.
3007 *
3008 * Note that the substitution depends on value of $mOutputType:
3009 * self::OT_WIKI: only {{subst:}} templates
3010 * self::OT_PREPROCESS: templates but not extension tags
3011 * self::OT_HTML: all templates and extension tags
3012 *
3013 * @param string $text The text to transform
3014 * @param bool|PPFrame $frame Object describing the arguments passed to the
3015 * template. Arguments may also be provided as an associative array, as
3016 * was the usual case before MW1.12. Providing arguments this way may be
3017 * useful for extensions wishing to perform variable replacement
3018 * explicitly.
3019 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3020 * double-brace expansion.
3021 * @return string
3022 */
3023 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3024 # Is there any text? Also, Prevent too big inclusions!
3025 $textSize = strlen( $text );
3026 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3027 return $text;
3028 }
3029
3030 if ( $frame === false ) {
3031 $frame = $this->getPreprocessor()->newFrame();
3032 } elseif ( !( $frame instanceof PPFrame ) ) {
3033 wfDebug( __METHOD__ . " called using plain parameters instead of "
3034 . "a PPFrame instance. Creating custom frame.\n" );
3035 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3036 }
3037
3038 $dom = $this->preprocessToDom( $text );
3039 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3040 $text = $frame->expand( $dom, $flags );
3041
3042 return $text;
3043 }
3044
3045 /**
3046 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3047 *
3048 * @param array $args
3049 *
3050 * @return array
3051 */
3052 public static function createAssocArgs( $args ) {
3053 $assocArgs = [];
3054 $index = 1;
3055 foreach ( $args as $arg ) {
3056 $eqpos = strpos( $arg, '=' );
3057 if ( $eqpos === false ) {
3058 $assocArgs[$index++] = $arg;
3059 } else {
3060 $name = trim( substr( $arg, 0, $eqpos ) );
3061 $value = trim( substr( $arg, $eqpos + 1 ) );
3062 if ( $value === false ) {
3063 $value = '';
3064 }
3065 if ( $name !== false ) {
3066 $assocArgs[$name] = $value;
3067 }
3068 }
3069 }
3070
3071 return $assocArgs;
3072 }
3073
3074 /**
3075 * Warn the user when a parser limitation is reached
3076 * Will warn at most once the user per limitation type
3077 *
3078 * The results are shown during preview and run through the Parser (See EditPage.php)
3079 *
3080 * @param string $limitationType Should be one of:
3081 * 'expensive-parserfunction' (corresponding messages:
3082 * 'expensive-parserfunction-warning',
3083 * 'expensive-parserfunction-category')
3084 * 'post-expand-template-argument' (corresponding messages:
3085 * 'post-expand-template-argument-warning',
3086 * 'post-expand-template-argument-category')
3087 * 'post-expand-template-inclusion' (corresponding messages:
3088 * 'post-expand-template-inclusion-warning',
3089 * 'post-expand-template-inclusion-category')
3090 * 'node-count-exceeded' (corresponding messages:
3091 * 'node-count-exceeded-warning',
3092 * 'node-count-exceeded-category')
3093 * 'expansion-depth-exceeded' (corresponding messages:
3094 * 'expansion-depth-exceeded-warning',
3095 * 'expansion-depth-exceeded-category')
3096 * @param string|int|null $current Current value
3097 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3098 * exceeded, provide the values (optional)
3099 */
3100 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3101 # does no harm if $current and $max are present but are unnecessary for the message
3102 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3103 # only during preview, and that would split the parser cache unnecessarily.
3104 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3105 ->text();
3106 $this->mOutput->addWarning( $warning );
3107 $this->addTrackingCategory( "$limitationType-category" );
3108 }
3109
3110 /**
3111 * Return the text of a template, after recursively
3112 * replacing any variables or templates within the template.
3113 *
3114 * @param array $piece The parts of the template
3115 * $piece['title']: the title, i.e. the part before the |
3116 * $piece['parts']: the parameter array
3117 * $piece['lineStart']: whether the brace was at the start of a line
3118 * @param PPFrame $frame The current frame, contains template arguments
3119 * @throws Exception
3120 * @return string The text of the template
3121 */
3122 public function braceSubstitution( $piece, $frame ) {
3123 // Flags
3124
3125 // $text has been filled
3126 $found = false;
3127 // wiki markup in $text should be escaped
3128 $nowiki = false;
3129 // $text is HTML, armour it against wikitext transformation
3130 $isHTML = false;
3131 // Force interwiki transclusion to be done in raw mode not rendered
3132 $forceRawInterwiki = false;
3133 // $text is a DOM node needing expansion in a child frame
3134 $isChildObj = false;
3135 // $text is a DOM node needing expansion in the current frame
3136 $isLocalObj = false;
3137
3138 # Title object, where $text came from
3139 $title = false;
3140
3141 # $part1 is the bit before the first |, and must contain only title characters.
3142 # Various prefixes will be stripped from it later.
3143 $titleWithSpaces = $frame->expand( $piece['title'] );
3144 $part1 = trim( $titleWithSpaces );
3145 $titleText = false;
3146
3147 # Original title text preserved for various purposes
3148 $originalTitle = $part1;
3149
3150 # $args is a list of argument nodes, starting from index 0, not including $part1
3151 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3152 # below won't work b/c this $args isn't an object
3153 $args = ( $piece['parts'] == null ) ? [] : $piece['parts'];
3154
3155 $profileSection = null; // profile templates
3156
3157 # SUBST
3158 if ( !$found ) {
3159 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3160
3161 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3162 # Decide whether to expand template or keep wikitext as-is.
3163 if ( $this->ot['wiki'] ) {
3164 if ( $substMatch === false ) {
3165 $literal = true; # literal when in PST with no prefix
3166 } else {
3167 $literal = false; # expand when in PST with subst: or safesubst:
3168 }
3169 } else {
3170 if ( $substMatch == 'subst' ) {
3171 $literal = true; # literal when not in PST with plain subst:
3172 } else {
3173 $literal = false; # expand when not in PST with safesubst: or no prefix
3174 }
3175 }
3176 if ( $literal ) {
3177 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3178 $isLocalObj = true;
3179 $found = true;
3180 }
3181 }
3182
3183 # Variables
3184 if ( !$found && $args->getLength() == 0 ) {
3185 $id = $this->mVariables->matchStartToEnd( $part1 );
3186 if ( $id !== false ) {
3187 $text = $this->getVariableValue( $id, $frame );
3188 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3189 $this->mOutput->updateCacheExpiry(
3190 $this->magicWordFactory->getCacheTTL( $id ) );
3191 }
3192 $found = true;
3193 }
3194 }
3195
3196 # MSG, MSGNW and RAW
3197 if ( !$found ) {
3198 # Check for MSGNW:
3199 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3200 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3201 $nowiki = true;
3202 } else {
3203 # Remove obsolete MSG:
3204 $mwMsg = $this->magicWordFactory->get( 'msg' );
3205 $mwMsg->matchStartAndRemove( $part1 );
3206 }
3207
3208 # Check for RAW:
3209 $mwRaw = $this->magicWordFactory->get( 'raw' );
3210 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3211 $forceRawInterwiki = true;
3212 }
3213 }
3214
3215 # Parser functions
3216 if ( !$found ) {
3217 $colonPos = strpos( $part1, ':' );
3218 if ( $colonPos !== false ) {
3219 $func = substr( $part1, 0, $colonPos );
3220 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3221 $argsLength = $args->getLength();
3222 for ( $i = 0; $i < $argsLength; $i++ ) {
3223 $funcArgs[] = $args->item( $i );
3224 }
3225
3226 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3227
3228 // Extract any forwarded flags
3229 if ( isset( $result['title'] ) ) {
3230 $title = $result['title'];
3231 }
3232 if ( isset( $result['found'] ) ) {
3233 $found = $result['found'];
3234 }
3235 if ( array_key_exists( 'text', $result ) ) {
3236 // a string or null
3237 $text = $result['text'];
3238 }
3239 if ( isset( $result['nowiki'] ) ) {
3240 $nowiki = $result['nowiki'];
3241 }
3242 if ( isset( $result['isHTML'] ) ) {
3243 $isHTML = $result['isHTML'];
3244 }
3245 if ( isset( $result['forceRawInterwiki'] ) ) {
3246 $forceRawInterwiki = $result['forceRawInterwiki'];
3247 }
3248 if ( isset( $result['isChildObj'] ) ) {
3249 $isChildObj = $result['isChildObj'];
3250 }
3251 if ( isset( $result['isLocalObj'] ) ) {
3252 $isLocalObj = $result['isLocalObj'];
3253 }
3254 }
3255 }
3256
3257 # Finish mangling title and then check for loops.
3258 # Set $title to a Title object and $titleText to the PDBK
3259 if ( !$found ) {
3260 $ns = NS_TEMPLATE;
3261 # Split the title into page and subpage
3262 $subpage = '';
3263 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3264 if ( $part1 !== $relative ) {
3265 $part1 = $relative;
3266 $ns = $this->mTitle->getNamespace();
3267 }
3268 $title = Title::newFromText( $part1, $ns );
3269 if ( $title ) {
3270 $titleText = $title->getPrefixedText();
3271 # Check for language variants if the template is not found
3272 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3273 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3274 }
3275 # Do recursion depth check
3276 $limit = $this->mOptions->getMaxTemplateDepth();
3277 if ( $frame->depth >= $limit ) {
3278 $found = true;
3279 $text = '<span class="error">'
3280 . wfMessage( 'parser-template-recursion-depth-warning' )
3281 ->numParams( $limit )->inContentLanguage()->text()
3282 . '</span>';
3283 }
3284 }
3285 }
3286
3287 # Load from database
3288 if ( !$found && $title ) {
3289 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3290 if ( !$title->isExternal() ) {
3291 if ( $title->isSpecialPage()
3292 && $this->mOptions->getAllowSpecialInclusion()
3293 && $this->ot['html']
3294 ) {
3295 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3296 // Pass the template arguments as URL parameters.
3297 // "uselang" will have no effect since the Language object
3298 // is forced to the one defined in ParserOptions.
3299 $pageArgs = [];
3300 $argsLength = $args->getLength();
3301 for ( $i = 0; $i < $argsLength; $i++ ) {
3302 $bits = $args->item( $i )->splitArg();
3303 if ( strval( $bits['index'] ) === '' ) {
3304 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3305 $value = trim( $frame->expand( $bits['value'] ) );
3306 $pageArgs[$name] = $value;
3307 }
3308 }
3309
3310 // Create a new context to execute the special page
3311 $context = new RequestContext;
3312 $context->setTitle( $title );
3313 $context->setRequest( new FauxRequest( $pageArgs ) );
3314 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3315 $context->setUser( $this->getUser() );
3316 } else {
3317 // If this page is cached, then we better not be per user.
3318 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3319 }
3320 $context->setLanguage( $this->mOptions->getUserLangObj() );
3321 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3322 if ( $ret ) {
3323 $text = $context->getOutput()->getHTML();
3324 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3325 $found = true;
3326 $isHTML = true;
3327 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3328 $this->mOutput->updateRuntimeAdaptiveExpiry(
3329 $specialPage->maxIncludeCacheTime()
3330 );
3331 }
3332 }
3333 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3334 $found = false; # access denied
3335 wfDebug( __METHOD__ . ": template inclusion denied for " .
3336 $title->getPrefixedDBkey() . "\n" );
3337 } else {
3338 list( $text, $title ) = $this->getTemplateDom( $title );
3339 if ( $text !== false ) {
3340 $found = true;
3341 $isChildObj = true;
3342 }
3343 }
3344
3345 # If the title is valid but undisplayable, make a link to it
3346 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3347 $text = "[[:$titleText]]";
3348 $found = true;
3349 }
3350 } elseif ( $title->isTrans() ) {
3351 # Interwiki transclusion
3352 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3353 $text = $this->interwikiTransclude( $title, 'render' );
3354 $isHTML = true;
3355 } else {
3356 $text = $this->interwikiTransclude( $title, 'raw' );
3357 # Preprocess it like a template
3358 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3359 $isChildObj = true;
3360 }
3361 $found = true;
3362 }
3363
3364 # Do infinite loop check
3365 # This has to be done after redirect resolution to avoid infinite loops via redirects
3366 if ( !$frame->loopCheck( $title ) ) {
3367 $found = true;
3368 $text = '<span class="error">'
3369 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3370 . '</span>';
3371 $this->addTrackingCategory( 'template-loop-category' );
3372 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3373 wfEscapeWikiText( $titleText ) )->text() );
3374 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3375 }
3376 }
3377
3378 # If we haven't found text to substitute by now, we're done
3379 # Recover the source wikitext and return it
3380 if ( !$found ) {
3381 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3382 if ( $profileSection ) {
3383 $this->mProfiler->scopedProfileOut( $profileSection );
3384 }
3385 return [ 'object' => $text ];
3386 }
3387
3388 # Expand DOM-style return values in a child frame
3389 if ( $isChildObj ) {
3390 # Clean up argument array
3391 $newFrame = $frame->newChild( $args, $title );
3392
3393 if ( $nowiki ) {
3394 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3395 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3396 # Expansion is eligible for the empty-frame cache
3397 $text = $newFrame->cachedExpand( $titleText, $text );
3398 } else {
3399 # Uncached expansion
3400 $text = $newFrame->expand( $text );
3401 }
3402 }
3403 if ( $isLocalObj && $nowiki ) {
3404 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3405 $isLocalObj = false;
3406 }
3407
3408 if ( $profileSection ) {
3409 $this->mProfiler->scopedProfileOut( $profileSection );
3410 }
3411
3412 # Replace raw HTML by a placeholder
3413 if ( $isHTML ) {
3414 $text = $this->insertStripItem( $text );
3415 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3416 # Escape nowiki-style return values
3417 $text = wfEscapeWikiText( $text );
3418 } elseif ( is_string( $text )
3419 && !$piece['lineStart']
3420 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3421 ) {
3422 # T2529: if the template begins with a table or block-level
3423 # element, it should be treated as beginning a new line.
3424 # This behavior is somewhat controversial.
3425 $text = "\n" . $text;
3426 }
3427
3428 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3429 # Error, oversize inclusion
3430 if ( $titleText !== false ) {
3431 # Make a working, properly escaped link if possible (T25588)
3432 $text = "[[:$titleText]]";
3433 } else {
3434 # This will probably not be a working link, but at least it may
3435 # provide some hint of where the problem is
3436 preg_replace( '/^:/', '', $originalTitle );
3437 $text = "[[:$originalTitle]]";
3438 }
3439 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3440 . 'post-expand include size too large -->' );
3441 $this->limitationWarn( 'post-expand-template-inclusion' );
3442 }
3443
3444 if ( $isLocalObj ) {
3445 $ret = [ 'object' => $text ];
3446 } else {
3447 $ret = [ 'text' => $text ];
3448 }
3449
3450 return $ret;
3451 }
3452
3453 /**
3454 * Call a parser function and return an array with text and flags.
3455 *
3456 * The returned array will always contain a boolean 'found', indicating
3457 * whether the parser function was found or not. It may also contain the
3458 * following:
3459 * text: string|object, resulting wikitext or PP DOM object
3460 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3461 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3462 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3463 * nowiki: bool, wiki markup in $text should be escaped
3464 *
3465 * @since 1.21
3466 * @param PPFrame $frame The current frame, contains template arguments
3467 * @param string $function Function name
3468 * @param array $args Arguments to the function
3469 * @throws MWException
3470 * @return array
3471 */
3472 public function callParserFunction( $frame, $function, array $args = [] ) {
3473 # Case sensitive functions
3474 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3475 $function = $this->mFunctionSynonyms[1][$function];
3476 } else {
3477 # Case insensitive functions
3478 $function = $this->contLang->lc( $function );
3479 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3480 $function = $this->mFunctionSynonyms[0][$function];
3481 } else {
3482 return [ 'found' => false ];
3483 }
3484 }
3485
3486 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3487
3488 // Avoid PHP 7.1 warning from passing $this by reference
3489 $parser = $this;
3490
3491 $allArgs = [ &$parser ];
3492 if ( $flags & self::SFH_OBJECT_ARGS ) {
3493 # Convert arguments to PPNodes and collect for appending to $allArgs
3494 $funcArgs = [];
3495 foreach ( $args as $k => $v ) {
3496 if ( $v instanceof PPNode || $k === 0 ) {
3497 $funcArgs[] = $v;
3498 } else {
3499 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3500 }
3501 }
3502
3503 # Add a frame parameter, and pass the arguments as an array
3504 $allArgs[] = $frame;
3505 $allArgs[] = $funcArgs;
3506 } else {
3507 # Convert arguments to plain text and append to $allArgs
3508 foreach ( $args as $k => $v ) {
3509 if ( $v instanceof PPNode ) {
3510 $allArgs[] = trim( $frame->expand( $v ) );
3511 } elseif ( is_int( $k ) && $k >= 0 ) {
3512 $allArgs[] = trim( $v );
3513 } else {
3514 $allArgs[] = trim( "$k=$v" );
3515 }
3516 }
3517 }
3518
3519 $result = $callback( ...$allArgs );
3520
3521 # The interface for function hooks allows them to return a wikitext
3522 # string or an array containing the string and any flags. This mungs
3523 # things around to match what this method should return.
3524 if ( !is_array( $result ) ) {
3525 $result = [
3526 'found' => true,
3527 'text' => $result,
3528 ];
3529 } else {
3530 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3531 $result['text'] = $result[0];
3532 }
3533 unset( $result[0] );
3534 $result += [
3535 'found' => true,
3536 ];
3537 }
3538
3539 $noparse = true;
3540 $preprocessFlags = 0;
3541 if ( isset( $result['noparse'] ) ) {
3542 $noparse = $result['noparse'];
3543 }
3544 if ( isset( $result['preprocessFlags'] ) ) {
3545 $preprocessFlags = $result['preprocessFlags'];
3546 }
3547
3548 if ( !$noparse ) {
3549 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3550 $result['isChildObj'] = true;
3551 }
3552
3553 return $result;
3554 }
3555
3556 /**
3557 * Get the semi-parsed DOM representation of a template with a given title,
3558 * and its redirect destination title. Cached.
3559 *
3560 * @param Title $title
3561 *
3562 * @return array
3563 */
3564 public function getTemplateDom( $title ) {
3565 $cacheTitle = $title;
3566 $titleText = $title->getPrefixedDBkey();
3567
3568 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3569 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3570 $title = Title::makeTitle( $ns, $dbk );
3571 $titleText = $title->getPrefixedDBkey();
3572 }
3573 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3574 return [ $this->mTplDomCache[$titleText], $title ];
3575 }
3576
3577 # Cache miss, go to the database
3578 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3579
3580 if ( $text === false ) {
3581 $this->mTplDomCache[$titleText] = false;
3582 return [ false, $title ];
3583 }
3584
3585 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3586 $this->mTplDomCache[$titleText] = $dom;
3587
3588 if ( !$title->equals( $cacheTitle ) ) {
3589 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3590 [ $title->getNamespace(), $title->getDBkey() ];
3591 }
3592
3593 return [ $dom, $title ];
3594 }
3595
3596 /**
3597 * Fetch the current revision of a given title. Note that the revision
3598 * (and even the title) may not exist in the database, so everything
3599 * contributing to the output of the parser should use this method
3600 * where possible, rather than getting the revisions themselves. This
3601 * method also caches its results, so using it benefits performance.
3602 *
3603 * @since 1.24
3604 * @param Title $title
3605 * @return Revision
3606 */
3607 public function fetchCurrentRevisionOfTitle( $title ) {
3608 $cacheKey = $title->getPrefixedDBkey();
3609 if ( !$this->currentRevisionCache ) {
3610 $this->currentRevisionCache = new MapCacheLRU( 100 );
3611 }
3612 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3613 $this->currentRevisionCache->set( $cacheKey,
3614 // Defaults to Parser::statelessFetchRevision()
3615 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3616 );
3617 }
3618 return $this->currentRevisionCache->get( $cacheKey );
3619 }
3620
3621 /**
3622 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3623 * without passing them on to it.
3624 *
3625 * @since 1.24
3626 * @param Title $title
3627 * @param Parser|bool $parser
3628 * @return Revision|bool False if missing
3629 */
3630 public static function statelessFetchRevision( Title $title, $parser = false ) {
3631 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3632
3633 return $rev;
3634 }
3635
3636 /**
3637 * Fetch the unparsed text of a template and register a reference to it.
3638 * @param Title $title
3639 * @return array ( string or false, Title )
3640 */
3641 public function fetchTemplateAndTitle( $title ) {
3642 // Defaults to Parser::statelessFetchTemplate()
3643 $templateCb = $this->mOptions->getTemplateCallback();
3644 $stuff = call_user_func( $templateCb, $title, $this );
3645 // We use U+007F DELETE to distinguish strip markers from regular text.
3646 $text = $stuff['text'];
3647 if ( is_string( $stuff['text'] ) ) {
3648 $text = strtr( $text, "\x7f", "?" );
3649 }
3650 $finalTitle = $stuff['finalTitle'] ?? $title;
3651 if ( isset( $stuff['deps'] ) ) {
3652 foreach ( $stuff['deps'] as $dep ) {
3653 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3654 if ( $dep['title']->equals( $this->getTitle() ) ) {
3655 // If we transclude ourselves, the final result
3656 // will change based on the new version of the page
3657 $this->mOutput->setFlag( 'vary-revision' );
3658 }
3659 }
3660 }
3661 return [ $text, $finalTitle ];
3662 }
3663
3664 /**
3665 * Fetch the unparsed text of a template and register a reference to it.
3666 * @param Title $title
3667 * @return string|bool
3668 */
3669 public function fetchTemplate( $title ) {
3670 return $this->fetchTemplateAndTitle( $title )[0];
3671 }
3672
3673 /**
3674 * Static function to get a template
3675 * Can be overridden via ParserOptions::setTemplateCallback().
3676 *
3677 * @param Title $title
3678 * @param bool|Parser $parser
3679 *
3680 * @return array
3681 */
3682 public static function statelessFetchTemplate( $title, $parser = false ) {
3683 $text = $skip = false;
3684 $finalTitle = $title;
3685 $deps = [];
3686
3687 # Loop to fetch the article, with up to 1 redirect
3688 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3689 # Give extensions a chance to select the revision instead
3690 $id = false; # Assume current
3691 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3692 [ $parser, $title, &$skip, &$id ] );
3693
3694 if ( $skip ) {
3695 $text = false;
3696 $deps[] = [
3697 'title' => $title,
3698 'page_id' => $title->getArticleID(),
3699 'rev_id' => null
3700 ];
3701 break;
3702 }
3703 # Get the revision
3704 if ( $id ) {
3705 $rev = Revision::newFromId( $id );
3706 } elseif ( $parser ) {
3707 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3708 } else {
3709 $rev = Revision::newFromTitle( $title );
3710 }
3711 $rev_id = $rev ? $rev->getId() : 0;
3712 # If there is no current revision, there is no page
3713 if ( $id === false && !$rev ) {
3714 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3715 $linkCache->addBadLinkObj( $title );
3716 }
3717
3718 $deps[] = [
3719 'title' => $title,
3720 'page_id' => $title->getArticleID(),
3721 'rev_id' => $rev_id ];
3722 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3723 # We fetched a rev from a different title; register it too...
3724 $deps[] = [
3725 'title' => $rev->getTitle(),
3726 'page_id' => $rev->getPage(),
3727 'rev_id' => $rev_id ];
3728 }
3729
3730 if ( $rev ) {
3731 $content = $rev->getContent();
3732 $text = $content ? $content->getWikitextForTransclusion() : null;
3733
3734 Hooks::run( 'ParserFetchTemplate',
3735 [ $parser, $title, $rev, &$text, &$deps ] );
3736
3737 if ( $text === false || $text === null ) {
3738 $text = false;
3739 break;
3740 }
3741 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3742 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
3743 lcfirst( $title->getText() ) )->inContentLanguage();
3744 if ( !$message->exists() ) {
3745 $text = false;
3746 break;
3747 }
3748 $content = $message->content();
3749 $text = $message->plain();
3750 } else {
3751 break;
3752 }
3753 if ( !$content ) {
3754 break;
3755 }
3756 # Redirect?
3757 $finalTitle = $title;
3758 $title = $content->getRedirectTarget();
3759 }
3760 return [
3761 'text' => $text,
3762 'finalTitle' => $finalTitle,
3763 'deps' => $deps ];
3764 }
3765
3766 /**
3767 * Fetch a file and its title and register a reference to it.
3768 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3769 * @param Title $title
3770 * @param array $options Array of options to RepoGroup::findFile
3771 * @return File|bool
3772 * @deprecated since 1.32, use fetchFileAndTitle instead
3773 */
3774 public function fetchFile( $title, $options = [] ) {
3775 wfDeprecated( __METHOD__, '1.32' );
3776 return $this->fetchFileAndTitle( $title, $options )[0];
3777 }
3778
3779 /**
3780 * Fetch a file and its title and register a reference to it.
3781 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3782 * @param Title $title
3783 * @param array $options Array of options to RepoGroup::findFile
3784 * @return array ( File or false, Title of file )
3785 */
3786 public function fetchFileAndTitle( $title, $options = [] ) {
3787 $file = $this->fetchFileNoRegister( $title, $options );
3788
3789 $time = $file ? $file->getTimestamp() : false;
3790 $sha1 = $file ? $file->getSha1() : false;
3791 # Register the file as a dependency...
3792 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3793 if ( $file && !$title->equals( $file->getTitle() ) ) {
3794 # Update fetched file title
3795 $title = $file->getTitle();
3796 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3797 }
3798 return [ $file, $title ];
3799 }
3800
3801 /**
3802 * Helper function for fetchFileAndTitle.
3803 *
3804 * Also useful if you need to fetch a file but not use it yet,
3805 * for example to get the file's handler.
3806 *
3807 * @param Title $title
3808 * @param array $options Array of options to RepoGroup::findFile
3809 * @return File|bool
3810 */
3811 protected function fetchFileNoRegister( $title, $options = [] ) {
3812 if ( isset( $options['broken'] ) ) {
3813 $file = false; // broken thumbnail forced by hook
3814 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3815 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3816 } else { // get by (name,timestamp)
3817 $file = wfFindFile( $title, $options );
3818 }
3819 return $file;
3820 }
3821
3822 /**
3823 * Transclude an interwiki link.
3824 *
3825 * @param Title $title
3826 * @param string $action Usually one of (raw, render)
3827 *
3828 * @return string
3829 */
3830 public function interwikiTransclude( $title, $action ) {
3831 if ( !$this->siteConfig->get( 'EnableScaryTranscluding' ) ) {
3832 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3833 }
3834
3835 $url = $title->getFullURL( [ 'action' => $action ] );
3836 if ( strlen( $url ) > 1024 ) {
3837 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3838 }
3839
3840 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3841
3842 $fname = __METHOD__;
3843 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3844
3845 $data = $cache->getWithSetCallback(
3846 $cache->makeGlobalKey(
3847 'interwiki-transclude',
3848 ( $wikiId !== false ) ? $wikiId : 'external',
3849 sha1( $url )
3850 ),
3851 $this->siteConfig->get( 'TranscludeCacheExpiry' ),
3852 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3853 $req = MWHttpRequest::factory( $url, [], $fname );
3854
3855 $status = $req->execute(); // Status object
3856 if ( !$status->isOK() ) {
3857 $ttl = $cache::TTL_UNCACHEABLE;
3858 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3859 $ttl = min( $cache::TTL_LAGGED, $ttl );
3860 }
3861
3862 return [
3863 'text' => $status->isOK() ? $req->getContent() : null,
3864 'code' => $req->getStatus()
3865 ];
3866 },
3867 [
3868 'checkKeys' => ( $wikiId !== false )
3869 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3870 : [],
3871 'pcGroup' => 'interwiki-transclude:5',
3872 'pcTTL' => $cache::TTL_PROC_LONG
3873 ]
3874 );
3875
3876 if ( is_string( $data['text'] ) ) {
3877 $text = $data['text'];
3878 } elseif ( $data['code'] != 200 ) {
3879 // Though we failed to fetch the content, this status is useless.
3880 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3881 ->params( $url, $data['code'] )->inContentLanguage()->text();
3882 } else {
3883 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3884 }
3885
3886 return $text;
3887 }
3888
3889 /**
3890 * Triple brace replacement -- used for template arguments
3891 * @private
3892 *
3893 * @param array $piece
3894 * @param PPFrame $frame
3895 *
3896 * @return array
3897 */
3898 public function argSubstitution( $piece, $frame ) {
3899 $error = false;
3900 $parts = $piece['parts'];
3901 $nameWithSpaces = $frame->expand( $piece['title'] );
3902 $argName = trim( $nameWithSpaces );
3903 $object = false;
3904 $text = $frame->getArgument( $argName );
3905 if ( $text === false && $parts->getLength() > 0
3906 && ( $this->ot['html']
3907 || $this->ot['pre']
3908 || ( $this->ot['wiki'] && $frame->isTemplate() )
3909 )
3910 ) {
3911 # No match in frame, use the supplied default
3912 $object = $parts->item( 0 )->getChildren();
3913 }
3914 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3915 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3916 $this->limitationWarn( 'post-expand-template-argument' );
3917 }
3918
3919 if ( $text === false && $object === false ) {
3920 # No match anywhere
3921 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3922 }
3923 if ( $error !== false ) {
3924 $text .= $error;
3925 }
3926 if ( $object !== false ) {
3927 $ret = [ 'object' => $object ];
3928 } else {
3929 $ret = [ 'text' => $text ];
3930 }
3931
3932 return $ret;
3933 }
3934
3935 /**
3936 * Return the text to be used for a given extension tag.
3937 * This is the ghost of strip().
3938 *
3939 * @param array $params Associative array of parameters:
3940 * name PPNode for the tag name
3941 * attr PPNode for unparsed text where tag attributes are thought to be
3942 * attributes Optional associative array of parsed attributes
3943 * inner Contents of extension element
3944 * noClose Original text did not have a close tag
3945 * @param PPFrame $frame
3946 *
3947 * @throws MWException
3948 * @return string
3949 */
3950 public function extensionSubstitution( $params, $frame ) {
3951 static $errorStr = '<span class="error">';
3952 static $errorLen = 20;
3953
3954 $name = $frame->expand( $params['name'] );
3955 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3956 // Probably expansion depth or node count exceeded. Just punt the
3957 // error up.
3958 return $name;
3959 }
3960
3961 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3962 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3963 // See above
3964 return $attrText;
3965 }
3966
3967 // We can't safely check if the expansion for $content resulted in an
3968 // error, because the content could happen to be the error string
3969 // (T149622).
3970 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3971
3972 $marker = self::MARKER_PREFIX . "-$name-"
3973 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3974
3975 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3976 ( $this->ot['html'] || $this->ot['pre'] );
3977 if ( $isFunctionTag ) {
3978 $markerType = 'none';
3979 } else {
3980 $markerType = 'general';
3981 }
3982 if ( $this->ot['html'] || $isFunctionTag ) {
3983 $name = strtolower( $name );
3984 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3985 if ( isset( $params['attributes'] ) ) {
3986 $attributes += $params['attributes'];
3987 }
3988
3989 if ( isset( $this->mTagHooks[$name] ) ) {
3990 $output = call_user_func_array( $this->mTagHooks[$name],
3991 [ $content, $attributes, $this, $frame ] );
3992 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3993 list( $callback, ) = $this->mFunctionTagHooks[$name];
3994
3995 // Avoid PHP 7.1 warning from passing $this by reference
3996 $parser = $this;
3997 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3998 } else {
3999 $output = '<span class="error">Invalid tag extension name: ' .
4000 htmlspecialchars( $name ) . '</span>';
4001 }
4002
4003 if ( is_array( $output ) ) {
4004 // Extract flags
4005 $flags = $output;
4006 $output = $flags[0];
4007 if ( isset( $flags['markerType'] ) ) {
4008 $markerType = $flags['markerType'];
4009 }
4010 }
4011 } else {
4012 if ( is_null( $attrText ) ) {
4013 $attrText = '';
4014 }
4015 if ( isset( $params['attributes'] ) ) {
4016 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4017 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4018 htmlspecialchars( $attrValue ) . '"';
4019 }
4020 }
4021 if ( $content === null ) {
4022 $output = "<$name$attrText/>";
4023 } else {
4024 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4025 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4026 // See above
4027 return $close;
4028 }
4029 $output = "<$name$attrText>$content$close";
4030 }
4031 }
4032
4033 if ( $markerType === 'none' ) {
4034 return $output;
4035 } elseif ( $markerType === 'nowiki' ) {
4036 $this->mStripState->addNoWiki( $marker, $output );
4037 } elseif ( $markerType === 'general' ) {
4038 $this->mStripState->addGeneral( $marker, $output );
4039 } else {
4040 throw new MWException( __METHOD__ . ': invalid marker type' );
4041 }
4042 return $marker;
4043 }
4044
4045 /**
4046 * Increment an include size counter
4047 *
4048 * @param string $type The type of expansion
4049 * @param int $size The size of the text
4050 * @return bool False if this inclusion would take it over the maximum, true otherwise
4051 */
4052 public function incrementIncludeSize( $type, $size ) {
4053 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4054 return false;
4055 } else {
4056 $this->mIncludeSizes[$type] += $size;
4057 return true;
4058 }
4059 }
4060
4061 /**
4062 * Increment the expensive function count
4063 *
4064 * @return bool False if the limit has been exceeded
4065 */
4066 public function incrementExpensiveFunctionCount() {
4067 $this->mExpensiveFunctionCount++;
4068 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4069 }
4070
4071 /**
4072 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4073 * Fills $this->mDoubleUnderscores, returns the modified text
4074 *
4075 * @param string $text
4076 *
4077 * @return string
4078 */
4079 public function doDoubleUnderscore( $text ) {
4080 # The position of __TOC__ needs to be recorded
4081 $mw = $this->magicWordFactory->get( 'toc' );
4082 if ( $mw->match( $text ) ) {
4083 $this->mShowToc = true;
4084 $this->mForceTocPosition = true;
4085
4086 # Set a placeholder. At the end we'll fill it in with the TOC.
4087 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4088
4089 # Only keep the first one.
4090 $text = $mw->replace( '', $text );
4091 }
4092
4093 # Now match and remove the rest of them
4094 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4095 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4096
4097 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4098 $this->mOutput->mNoGallery = true;
4099 }
4100 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4101 $this->mShowToc = false;
4102 }
4103 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4104 && $this->mTitle->getNamespace() == NS_CATEGORY
4105 ) {
4106 $this->addTrackingCategory( 'hidden-category-category' );
4107 }
4108 # (T10068) Allow control over whether robots index a page.
4109 # __INDEX__ always overrides __NOINDEX__, see T16899
4110 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4111 $this->mOutput->setIndexPolicy( 'noindex' );
4112 $this->addTrackingCategory( 'noindex-category' );
4113 }
4114 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4115 $this->mOutput->setIndexPolicy( 'index' );
4116 $this->addTrackingCategory( 'index-category' );
4117 }
4118
4119 # Cache all double underscores in the database
4120 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4121 $this->mOutput->setProperty( $key, '' );
4122 }
4123
4124 return $text;
4125 }
4126
4127 /**
4128 * @see ParserOutput::addTrackingCategory()
4129 * @param string $msg Message key
4130 * @return bool Whether the addition was successful
4131 */
4132 public function addTrackingCategory( $msg ) {
4133 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4134 }
4135
4136 /**
4137 * This function accomplishes several tasks:
4138 * 1) Auto-number headings if that option is enabled
4139 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4140 * 3) Add a Table of contents on the top for users who have enabled the option
4141 * 4) Auto-anchor headings
4142 *
4143 * It loops through all headlines, collects the necessary data, then splits up the
4144 * string and re-inserts the newly formatted headlines.
4145 *
4146 * @param string $text
4147 * @param string $origText Original, untouched wikitext
4148 * @param bool $isMain
4149 * @return mixed|string
4150 * @private
4151 */
4152 public function formatHeadings( $text, $origText, $isMain = true ) {
4153 # Inhibit editsection links if requested in the page
4154 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4155 $maybeShowEditLink = false;
4156 } else {
4157 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4158 }
4159
4160 # Get all headlines for numbering them and adding funky stuff like [edit]
4161 # links - this is for later, but we need the number of headlines right now
4162 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4163 # be trimmed here since whitespace in HTML headings is significant.
4164 $matches = [];
4165 $numMatches = preg_match_all(
4166 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4167 $text,
4168 $matches
4169 );
4170
4171 # if there are fewer than 4 headlines in the article, do not show TOC
4172 # unless it's been explicitly enabled.
4173 $enoughToc = $this->mShowToc &&
4174 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4175
4176 # Allow user to stipulate that a page should have a "new section"
4177 # link added via __NEWSECTIONLINK__
4178 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4179 $this->mOutput->setNewSection( true );
4180 }
4181
4182 # Allow user to remove the "new section"
4183 # link via __NONEWSECTIONLINK__
4184 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4185 $this->mOutput->hideNewSection( true );
4186 }
4187
4188 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4189 # override above conditions and always show TOC above first header
4190 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4191 $this->mShowToc = true;
4192 $enoughToc = true;
4193 }
4194
4195 # headline counter
4196 $headlineCount = 0;
4197 $numVisible = 0;
4198
4199 # Ugh .. the TOC should have neat indentation levels which can be
4200 # passed to the skin functions. These are determined here
4201 $toc = '';
4202 $full = '';
4203 $head = [];
4204 $sublevelCount = [];
4205 $levelCount = [];
4206 $level = 0;
4207 $prevlevel = 0;
4208 $toclevel = 0;
4209 $prevtoclevel = 0;
4210 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4211 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4212 $oldType = $this->mOutputType;
4213 $this->setOutputType( self::OT_WIKI );
4214 $frame = $this->getPreprocessor()->newFrame();
4215 $root = $this->preprocessToDom( $origText );
4216 $node = $root->getFirstChild();
4217 $byteOffset = 0;
4218 $tocraw = [];
4219 $refers = [];
4220
4221 $headlines = $numMatches !== false ? $matches[3] : [];
4222
4223 $maxTocLevel = $this->siteConfig->get( 'MaxTocLevel' );
4224 foreach ( $headlines as $headline ) {
4225 $isTemplate = false;
4226 $titleText = false;
4227 $sectionIndex = false;
4228 $numbering = '';
4229 $markerMatches = [];
4230 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4231 $serial = $markerMatches[1];
4232 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4233 $isTemplate = ( $titleText != $baseTitleText );
4234 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4235 }
4236
4237 if ( $toclevel ) {
4238 $prevlevel = $level;
4239 }
4240 $level = $matches[1][$headlineCount];
4241
4242 if ( $level > $prevlevel ) {
4243 # Increase TOC level
4244 $toclevel++;
4245 $sublevelCount[$toclevel] = 0;
4246 if ( $toclevel < $maxTocLevel ) {
4247 $prevtoclevel = $toclevel;
4248 $toc .= Linker::tocIndent();
4249 $numVisible++;
4250 }
4251 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4252 # Decrease TOC level, find level to jump to
4253
4254 for ( $i = $toclevel; $i > 0; $i-- ) {
4255 if ( $levelCount[$i] == $level ) {
4256 # Found last matching level
4257 $toclevel = $i;
4258 break;
4259 } elseif ( $levelCount[$i] < $level ) {
4260 # Found first matching level below current level
4261 $toclevel = $i + 1;
4262 break;
4263 }
4264 }
4265 if ( $i == 0 ) {
4266 $toclevel = 1;
4267 }
4268 if ( $toclevel < $maxTocLevel ) {
4269 if ( $prevtoclevel < $maxTocLevel ) {
4270 # Unindent only if the previous toc level was shown :p
4271 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4272 $prevtoclevel = $toclevel;
4273 } else {
4274 $toc .= Linker::tocLineEnd();
4275 }
4276 }
4277 } else {
4278 # No change in level, end TOC line
4279 if ( $toclevel < $maxTocLevel ) {
4280 $toc .= Linker::tocLineEnd();
4281 }
4282 }
4283
4284 $levelCount[$toclevel] = $level;
4285
4286 # count number of headlines for each level
4287 $sublevelCount[$toclevel]++;
4288 $dot = 0;
4289 for ( $i = 1; $i <= $toclevel; $i++ ) {
4290 if ( !empty( $sublevelCount[$i] ) ) {
4291 if ( $dot ) {
4292 $numbering .= '.';
4293 }
4294 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4295 $dot = 1;
4296 }
4297 }
4298
4299 # The safe header is a version of the header text safe to use for links
4300
4301 # Remove link placeholders by the link text.
4302 # <!--LINK number-->
4303 # turns into
4304 # link text with suffix
4305 # Do this before unstrip since link text can contain strip markers
4306 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4307
4308 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4309 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4310
4311 # Remove any <style> or <script> tags (T198618)
4312 $safeHeadline = preg_replace(
4313 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4314 '',
4315 $safeHeadline
4316 );
4317
4318 # Strip out HTML (first regex removes any tag not allowed)
4319 # Allowed tags are:
4320 # * <sup> and <sub> (T10393)
4321 # * <i> (T28375)
4322 # * <b> (r105284)
4323 # * <bdi> (T74884)
4324 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4325 # * <s> and <strike> (T35715)
4326 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4327 # to allow setting directionality in toc items.
4328 $tocline = preg_replace(
4329 [
4330 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4331 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4332 ],
4333 [ '', '<$1>' ],
4334 $safeHeadline
4335 );
4336
4337 # Strip '<span></span>', which is the result from the above if
4338 # <span id="foo"></span> is used to produce an additional anchor
4339 # for a section.
4340 $tocline = str_replace( '<span></span>', '', $tocline );
4341
4342 $tocline = trim( $tocline );
4343
4344 # For the anchor, strip out HTML-y stuff period
4345 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4346 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4347
4348 # Save headline for section edit hint before it's escaped
4349 $headlineHint = $safeHeadline;
4350
4351 # Decode HTML entities
4352 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4353
4354 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4355
4356 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4357 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4358 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4359 if ( $fallbackHeadline === $safeHeadline ) {
4360 # No reason to have both (in fact, we can't)
4361 $fallbackHeadline = false;
4362 }
4363
4364 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4365 # @todo FIXME: We may be changing them depending on the current locale.
4366 $arrayKey = strtolower( $safeHeadline );
4367 if ( $fallbackHeadline === false ) {
4368 $fallbackArrayKey = false;
4369 } else {
4370 $fallbackArrayKey = strtolower( $fallbackHeadline );
4371 }
4372
4373 # Create the anchor for linking from the TOC to the section
4374 $anchor = $safeHeadline;
4375 $fallbackAnchor = $fallbackHeadline;
4376 if ( isset( $refers[$arrayKey] ) ) {
4377 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4378 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4379 $anchor .= "_$i";
4380 $linkAnchor .= "_$i";
4381 $refers["${arrayKey}_$i"] = true;
4382 } else {
4383 $refers[$arrayKey] = true;
4384 }
4385 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4386 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4387 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4388 $fallbackAnchor .= "_$i";
4389 $refers["${fallbackArrayKey}_$i"] = true;
4390 } else {
4391 $refers[$fallbackArrayKey] = true;
4392 }
4393
4394 # Don't number the heading if it is the only one (looks silly)
4395 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4396 # the two are different if the line contains a link
4397 $headline = Html::element(
4398 'span',
4399 [ 'class' => 'mw-headline-number' ],
4400 $numbering
4401 ) . ' ' . $headline;
4402 }
4403
4404 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4405 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4406 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4407 }
4408
4409 # Add the section to the section tree
4410 # Find the DOM node for this header
4411 $noOffset = ( $isTemplate || $sectionIndex === false );
4412 while ( $node && !$noOffset ) {
4413 if ( $node->getName() === 'h' ) {
4414 $bits = $node->splitHeading();
4415 if ( $bits['i'] == $sectionIndex ) {
4416 break;
4417 }
4418 }
4419 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4420 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4421 $node = $node->getNextSibling();
4422 }
4423 $tocraw[] = [
4424 'toclevel' => $toclevel,
4425 'level' => $level,
4426 'line' => $tocline,
4427 'number' => $numbering,
4428 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4429 'fromtitle' => $titleText,
4430 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4431 'anchor' => $anchor,
4432 ];
4433
4434 # give headline the correct <h#> tag
4435 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4436 // Output edit section links as markers with styles that can be customized by skins
4437 if ( $isTemplate ) {
4438 # Put a T flag in the section identifier, to indicate to extractSections()
4439 # that sections inside <includeonly> should be counted.
4440 $editsectionPage = $titleText;
4441 $editsectionSection = "T-$sectionIndex";
4442 $editsectionContent = null;
4443 } else {
4444 $editsectionPage = $this->mTitle->getPrefixedText();
4445 $editsectionSection = $sectionIndex;
4446 $editsectionContent = $headlineHint;
4447 }
4448 // We use a bit of pesudo-xml for editsection markers. The
4449 // language converter is run later on. Using a UNIQ style marker
4450 // leads to the converter screwing up the tokens when it
4451 // converts stuff. And trying to insert strip tags fails too. At
4452 // this point all real inputted tags have already been escaped,
4453 // so we don't have to worry about a user trying to input one of
4454 // these markers directly. We use a page and section attribute
4455 // to stop the language converter from converting these
4456 // important bits of data, but put the headline hint inside a
4457 // content block because the language converter is supposed to
4458 // be able to convert that piece of data.
4459 // Gets replaced with html in ParserOutput::getText
4460 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4461 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4462 if ( $editsectionContent !== null ) {
4463 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4464 } else {
4465 $editlink .= '/>';
4466 }
4467 } else {
4468 $editlink = '';
4469 }
4470 $head[$headlineCount] = Linker::makeHeadline( $level,
4471 $matches['attrib'][$headlineCount], $anchor, $headline,
4472 $editlink, $fallbackAnchor );
4473
4474 $headlineCount++;
4475 }
4476
4477 $this->setOutputType( $oldType );
4478
4479 # Never ever show TOC if no headers
4480 if ( $numVisible < 1 ) {
4481 $enoughToc = false;
4482 }
4483
4484 if ( $enoughToc ) {
4485 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4486 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4487 }
4488 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4489 $this->mOutput->setTOCHTML( $toc );
4490 $toc = self::TOC_START . $toc . self::TOC_END;
4491 }
4492
4493 if ( $isMain ) {
4494 $this->mOutput->setSections( $tocraw );
4495 }
4496
4497 # split up and insert constructed headlines
4498 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4499 $i = 0;
4500
4501 // build an array of document sections
4502 $sections = [];
4503 foreach ( $blocks as $block ) {
4504 // $head is zero-based, sections aren't.
4505 if ( empty( $head[$i - 1] ) ) {
4506 $sections[$i] = $block;
4507 } else {
4508 $sections[$i] = $head[$i - 1] . $block;
4509 }
4510
4511 /**
4512 * Send a hook, one per section.
4513 * The idea here is to be able to make section-level DIVs, but to do so in a
4514 * lower-impact, more correct way than r50769
4515 *
4516 * $this : caller
4517 * $section : the section number
4518 * &$sectionContent : ref to the content of the section
4519 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4520 */
4521 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4522
4523 $i++;
4524 }
4525
4526 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4527 // append the TOC at the beginning
4528 // Top anchor now in skin
4529 $sections[0] .= $toc . "\n";
4530 }
4531
4532 $full .= implode( '', $sections );
4533
4534 if ( $this->mForceTocPosition ) {
4535 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4536 } else {
4537 return $full;
4538 }
4539 }
4540
4541 /**
4542 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4543 * conversion, substituting signatures, {{subst:}} templates, etc.
4544 *
4545 * @param string $text The text to transform
4546 * @param Title $title The Title object for the current article
4547 * @param User $user The User object describing the current user
4548 * @param ParserOptions $options Parsing options
4549 * @param bool $clearState Whether to clear the parser state first
4550 * @return string The altered wiki markup
4551 */
4552 public function preSaveTransform( $text, Title $title, User $user,
4553 ParserOptions $options, $clearState = true
4554 ) {
4555 if ( $clearState ) {
4556 $magicScopeVariable = $this->lock();
4557 }
4558 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4559 $this->setUser( $user );
4560
4561 // Strip U+0000 NULL (T159174)
4562 $text = str_replace( "\000", '', $text );
4563
4564 // We still normalize line endings for backwards-compatibility
4565 // with other code that just calls PST, but this should already
4566 // be handled in TextContent subclasses
4567 $text = TextContent::normalizeLineEndings( $text );
4568
4569 if ( $options->getPreSaveTransform() ) {
4570 $text = $this->pstPass2( $text, $user );
4571 }
4572 $text = $this->mStripState->unstripBoth( $text );
4573
4574 $this->setUser( null ); # Reset
4575
4576 return $text;
4577 }
4578
4579 /**
4580 * Pre-save transform helper function
4581 *
4582 * @param string $text
4583 * @param User $user
4584 *
4585 * @return string
4586 */
4587 private function pstPass2( $text, $user ) {
4588 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4589 # $this->contLang here in order to give everyone the same signature and use the default one
4590 # rather than the one selected in each user's preferences. (see also T14815)
4591 $ts = $this->mOptions->getTimestamp();
4592 $timestamp = MWTimestamp::getLocalInstance( $ts );
4593 $ts = $timestamp->format( 'YmdHis' );
4594 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4595
4596 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4597
4598 # Variable replacement
4599 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4600 $text = $this->replaceVariables( $text );
4601
4602 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4603 # which may corrupt this parser instance via its wfMessage()->text() call-
4604
4605 # Signatures
4606 if ( strpos( $text, '~~~' ) !== false ) {
4607 $sigText = $this->getUserSig( $user );
4608 $text = strtr( $text, [
4609 '~~~~~' => $d,
4610 '~~~~' => "$sigText $d",
4611 '~~~' => $sigText
4612 ] );
4613 # The main two signature forms used above are time-sensitive
4614 $this->mOutput->setFlag( 'user-signature' );
4615 }
4616
4617 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4618 $tc = '[' . Title::legalChars() . ']';
4619 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4620
4621 // [[ns:page (context)|]]
4622 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4623 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4624 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4625 // [[ns:page (context), context|]] (using either single or double-width comma)
4626 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4627 // [[|page]] (reverse pipe trick: add context from page title)
4628 $p2 = "/\[\[\\|($tc+)]]/";
4629
4630 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4631 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4632 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4633 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4634
4635 $t = $this->mTitle->getText();
4636 $m = [];
4637 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4638 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4639 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4640 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4641 } else {
4642 # if there's no context, don't bother duplicating the title
4643 $text = preg_replace( $p2, '[[\\1]]', $text );
4644 }
4645
4646 return $text;
4647 }
4648
4649 /**
4650 * Fetch the user's signature text, if any, and normalize to
4651 * validated, ready-to-insert wikitext.
4652 * If you have pre-fetched the nickname or the fancySig option, you can
4653 * specify them here to save a database query.
4654 * Do not reuse this parser instance after calling getUserSig(),
4655 * as it may have changed if it's the $wgParser.
4656 *
4657 * @param User &$user
4658 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4659 * @param bool|null $fancySig whether the nicknname is the complete signature
4660 * or null to use default value
4661 * @return string
4662 */
4663 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4664 $username = $user->getName();
4665
4666 # If not given, retrieve from the user object.
4667 if ( $nickname === false ) {
4668 $nickname = $user->getOption( 'nickname' );
4669 }
4670
4671 if ( is_null( $fancySig ) ) {
4672 $fancySig = $user->getBoolOption( 'fancysig' );
4673 }
4674
4675 $nickname = $nickname == null ? $username : $nickname;
4676
4677 if ( mb_strlen( $nickname ) > $this->siteConfig->get( 'MaxSigChars' ) ) {
4678 $nickname = $username;
4679 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4680 } elseif ( $fancySig !== false ) {
4681 # Sig. might contain markup; validate this
4682 if ( $this->validateSig( $nickname ) !== false ) {
4683 # Validated; clean up (if needed) and return it
4684 return $this->cleanSig( $nickname, true );
4685 } else {
4686 # Failed to validate; fall back to the default
4687 $nickname = $username;
4688 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4689 }
4690 }
4691
4692 # Make sure nickname doesnt get a sig in a sig
4693 $nickname = self::cleanSigInSig( $nickname );
4694
4695 # If we're still here, make it a link to the user page
4696 $userText = wfEscapeWikiText( $username );
4697 $nickText = wfEscapeWikiText( $nickname );
4698 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4699
4700 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4701 ->title( $this->getTitle() )->text();
4702 }
4703
4704 /**
4705 * Check that the user's signature contains no bad XML
4706 *
4707 * @param string $text
4708 * @return string|bool An expanded string, or false if invalid.
4709 */
4710 public function validateSig( $text ) {
4711 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4712 }
4713
4714 /**
4715 * Clean up signature text
4716 *
4717 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4718 * 2) Substitute all transclusions
4719 *
4720 * @param string $text
4721 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4722 * @return string Signature text
4723 */
4724 public function cleanSig( $text, $parsing = false ) {
4725 if ( !$parsing ) {
4726 global $wgTitle;
4727 $magicScopeVariable = $this->lock();
4728 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4729 }
4730
4731 # Option to disable this feature
4732 if ( !$this->mOptions->getCleanSignatures() ) {
4733 return $text;
4734 }
4735
4736 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4737 # => Move this logic to braceSubstitution()
4738 $substWord = $this->magicWordFactory->get( 'subst' );
4739 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4740 $substText = '{{' . $substWord->getSynonym( 0 );
4741
4742 $text = preg_replace( $substRegex, $substText, $text );
4743 $text = self::cleanSigInSig( $text );
4744 $dom = $this->preprocessToDom( $text );
4745 $frame = $this->getPreprocessor()->newFrame();
4746 $text = $frame->expand( $dom );
4747
4748 if ( !$parsing ) {
4749 $text = $this->mStripState->unstripBoth( $text );
4750 }
4751
4752 return $text;
4753 }
4754
4755 /**
4756 * Strip 3, 4 or 5 tildes out of signatures.
4757 *
4758 * @param string $text
4759 * @return string Signature text with /~{3,5}/ removed
4760 */
4761 public static function cleanSigInSig( $text ) {
4762 $text = preg_replace( '/~{3,5}/', '', $text );
4763 return $text;
4764 }
4765
4766 /**
4767 * Set up some variables which are usually set up in parse()
4768 * so that an external function can call some class members with confidence
4769 *
4770 * @param Title|null $title
4771 * @param ParserOptions $options
4772 * @param int $outputType
4773 * @param bool $clearState
4774 */
4775 public function startExternalParse( Title $title = null, ParserOptions $options,
4776 $outputType, $clearState = true
4777 ) {
4778 $this->startParse( $title, $options, $outputType, $clearState );
4779 }
4780
4781 /**
4782 * @param Title|null $title
4783 * @param ParserOptions $options
4784 * @param int $outputType
4785 * @param bool $clearState
4786 */
4787 private function startParse( Title $title = null, ParserOptions $options,
4788 $outputType, $clearState = true
4789 ) {
4790 $this->setTitle( $title );
4791 $this->mOptions = $options;
4792 $this->setOutputType( $outputType );
4793 if ( $clearState ) {
4794 $this->clearState();
4795 }
4796 }
4797
4798 /**
4799 * Wrapper for preprocess()
4800 *
4801 * @param string $text The text to preprocess
4802 * @param ParserOptions $options
4803 * @param Title|null $title Title object or null to use $wgTitle
4804 * @return string
4805 */
4806 public function transformMsg( $text, $options, $title = null ) {
4807 static $executing = false;
4808
4809 # Guard against infinite recursion
4810 if ( $executing ) {
4811 return $text;
4812 }
4813 $executing = true;
4814
4815 if ( !$title ) {
4816 global $wgTitle;
4817 $title = $wgTitle;
4818 }
4819
4820 $text = $this->preprocess( $text, $title, $options );
4821
4822 $executing = false;
4823 return $text;
4824 }
4825
4826 /**
4827 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4828 * The callback should have the following form:
4829 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4830 *
4831 * Transform and return $text. Use $parser for any required context, e.g. use
4832 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4833 *
4834 * Hooks may return extended information by returning an array, of which the
4835 * first numbered element (index 0) must be the return string, and all other
4836 * entries are extracted into local variables within an internal function
4837 * in the Parser class.
4838 *
4839 * This interface (introduced r61913) appears to be undocumented, but
4840 * 'markerType' is used by some core tag hooks to override which strip
4841 * array their results are placed in. **Use great caution if attempting
4842 * this interface, as it is not documented and injudicious use could smash
4843 * private variables.**
4844 *
4845 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4846 * @param callable $callback The callback function (and object) to use for the tag
4847 * @throws MWException
4848 * @return callable|null The old value of the mTagHooks array associated with the hook
4849 */
4850 public function setHook( $tag, callable $callback ) {
4851 $tag = strtolower( $tag );
4852 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4853 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4854 }
4855 $oldVal = $this->mTagHooks[$tag] ?? null;
4856 $this->mTagHooks[$tag] = $callback;
4857 if ( !in_array( $tag, $this->mStripList ) ) {
4858 $this->mStripList[] = $tag;
4859 }
4860
4861 return $oldVal;
4862 }
4863
4864 /**
4865 * As setHook(), but letting the contents be parsed.
4866 *
4867 * Transparent tag hooks are like regular XML-style tag hooks, except they
4868 * operate late in the transformation sequence, on HTML instead of wikitext.
4869 *
4870 * This is probably obsoleted by things dealing with parser frames?
4871 * The only extension currently using it is geoserver.
4872 *
4873 * @since 1.10
4874 * @todo better document or deprecate this
4875 *
4876 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4877 * @param callable $callback The callback function (and object) to use for the tag
4878 * @throws MWException
4879 * @return callable|null The old value of the mTagHooks array associated with the hook
4880 */
4881 public function setTransparentTagHook( $tag, callable $callback ) {
4882 $tag = strtolower( $tag );
4883 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4884 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4885 }
4886 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4887 $this->mTransparentTagHooks[$tag] = $callback;
4888
4889 return $oldVal;
4890 }
4891
4892 /**
4893 * Remove all tag hooks
4894 */
4895 public function clearTagHooks() {
4896 $this->mTagHooks = [];
4897 $this->mFunctionTagHooks = [];
4898 $this->mStripList = $this->mDefaultStripList;
4899 }
4900
4901 /**
4902 * Create a function, e.g. {{sum:1|2|3}}
4903 * The callback function should have the form:
4904 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4905 *
4906 * Or with Parser::SFH_OBJECT_ARGS:
4907 * function myParserFunction( $parser, $frame, $args ) { ... }
4908 *
4909 * The callback may either return the text result of the function, or an array with the text
4910 * in element 0, and a number of flags in the other elements. The names of the flags are
4911 * specified in the keys. Valid flags are:
4912 * found The text returned is valid, stop processing the template. This
4913 * is on by default.
4914 * nowiki Wiki markup in the return value should be escaped
4915 * isHTML The returned text is HTML, armour it against wikitext transformation
4916 *
4917 * @param string $id The magic word ID
4918 * @param callable $callback The callback function (and object) to use
4919 * @param int $flags A combination of the following flags:
4920 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4921 *
4922 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4923 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4924 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4925 * the arguments, and to control the way they are expanded.
4926 *
4927 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4928 * arguments, for instance:
4929 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4930 *
4931 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4932 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4933 * working if/when this is changed.
4934 *
4935 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4936 * expansion.
4937 *
4938 * Please read the documentation in includes/parser/Preprocessor.php for more information
4939 * about the methods available in PPFrame and PPNode.
4940 *
4941 * @throws MWException
4942 * @return string|callable The old callback function for this name, if any
4943 */
4944 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4945 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4946 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4947
4948 # Add to function cache
4949 $mw = $this->magicWordFactory->get( $id );
4950 if ( !$mw ) {
4951 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4952 }
4953
4954 $synonyms = $mw->getSynonyms();
4955 $sensitive = intval( $mw->isCaseSensitive() );
4956
4957 foreach ( $synonyms as $syn ) {
4958 # Case
4959 if ( !$sensitive ) {
4960 $syn = $this->contLang->lc( $syn );
4961 }
4962 # Add leading hash
4963 if ( !( $flags & self::SFH_NO_HASH ) ) {
4964 $syn = '#' . $syn;
4965 }
4966 # Remove trailing colon
4967 if ( substr( $syn, -1, 1 ) === ':' ) {
4968 $syn = substr( $syn, 0, -1 );
4969 }
4970 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4971 }
4972 return $oldVal;
4973 }
4974
4975 /**
4976 * Get all registered function hook identifiers
4977 *
4978 * @return array
4979 */
4980 public function getFunctionHooks() {
4981 $this->firstCallInit();
4982 return array_keys( $this->mFunctionHooks );
4983 }
4984
4985 /**
4986 * Create a tag function, e.g. "<test>some stuff</test>".
4987 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4988 * Unlike parser functions, their content is not preprocessed.
4989 * @param string $tag
4990 * @param callable $callback
4991 * @param int $flags
4992 * @throws MWException
4993 * @return null
4994 */
4995 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4996 $tag = strtolower( $tag );
4997 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4998 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4999 }
5000 $old = $this->mFunctionTagHooks[$tag] ?? null;
5001 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
5002
5003 if ( !in_array( $tag, $this->mStripList ) ) {
5004 $this->mStripList[] = $tag;
5005 }
5006
5007 return $old;
5008 }
5009
5010 /**
5011 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5012 * Placeholders created in Linker::link()
5013 *
5014 * @param string &$text
5015 * @param int $options
5016 */
5017 public function replaceLinkHolders( &$text, $options = 0 ) {
5018 $this->mLinkHolders->replace( $text );
5019 }
5020
5021 /**
5022 * Replace "<!--LINK-->" link placeholders with plain text of links
5023 * (not HTML-formatted).
5024 *
5025 * @param string $text
5026 * @return string
5027 */
5028 public function replaceLinkHoldersText( $text ) {
5029 return $this->mLinkHolders->replaceText( $text );
5030 }
5031
5032 /**
5033 * Renders an image gallery from a text with one line per image.
5034 * text labels may be given by using |-style alternative text. E.g.
5035 * Image:one.jpg|The number "1"
5036 * Image:tree.jpg|A tree
5037 * given as text will return the HTML of a gallery with two images,
5038 * labeled 'The number "1"' and
5039 * 'A tree'.
5040 *
5041 * @param string $text
5042 * @param array $params
5043 * @return string HTML
5044 */
5045 public function renderImageGallery( $text, $params ) {
5046 $mode = false;
5047 if ( isset( $params['mode'] ) ) {
5048 $mode = $params['mode'];
5049 }
5050
5051 try {
5052 $ig = ImageGalleryBase::factory( $mode );
5053 } catch ( Exception $e ) {
5054 // If invalid type set, fallback to default.
5055 $ig = ImageGalleryBase::factory( false );
5056 }
5057
5058 $ig->setContextTitle( $this->mTitle );
5059 $ig->setShowBytes( false );
5060 $ig->setShowDimensions( false );
5061 $ig->setShowFilename( false );
5062 $ig->setParser( $this );
5063 $ig->setHideBadImages();
5064 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5065
5066 if ( isset( $params['showfilename'] ) ) {
5067 $ig->setShowFilename( true );
5068 } else {
5069 $ig->setShowFilename( false );
5070 }
5071 if ( isset( $params['caption'] ) ) {
5072 // NOTE: We aren't passing a frame here or below. Frame info
5073 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5074 // See T107332#4030581
5075 $caption = $this->recursiveTagParse( $params['caption'] );
5076 $ig->setCaptionHtml( $caption );
5077 }
5078 if ( isset( $params['perrow'] ) ) {
5079 $ig->setPerRow( $params['perrow'] );
5080 }
5081 if ( isset( $params['widths'] ) ) {
5082 $ig->setWidths( $params['widths'] );
5083 }
5084 if ( isset( $params['heights'] ) ) {
5085 $ig->setHeights( $params['heights'] );
5086 }
5087 $ig->setAdditionalOptions( $params );
5088
5089 // Avoid PHP 7.1 warning from passing $this by reference
5090 $parser = $this;
5091 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5092
5093 $lines = StringUtils::explode( "\n", $text );
5094 foreach ( $lines as $line ) {
5095 # match lines like these:
5096 # Image:someimage.jpg|This is some image
5097 $matches = [];
5098 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5099 # Skip empty lines
5100 if ( count( $matches ) == 0 ) {
5101 continue;
5102 }
5103
5104 if ( strpos( $matches[0], '%' ) !== false ) {
5105 $matches[1] = rawurldecode( $matches[1] );
5106 }
5107 $title = Title::newFromText( $matches[1], NS_FILE );
5108 if ( is_null( $title ) ) {
5109 # Bogus title. Ignore these so we don't bomb out later.
5110 continue;
5111 }
5112
5113 # We need to get what handler the file uses, to figure out parameters.
5114 # Note, a hook can overide the file name, and chose an entirely different
5115 # file (which potentially could be of a different type and have different handler).
5116 $options = [];
5117 $descQuery = false;
5118 Hooks::run( 'BeforeParserFetchFileAndTitle',
5119 [ $this, $title, &$options, &$descQuery ] );
5120 # Don't register it now, as TraditionalImageGallery does that later.
5121 $file = $this->fetchFileNoRegister( $title, $options );
5122 $handler = $file ? $file->getHandler() : false;
5123
5124 $paramMap = [
5125 'img_alt' => 'gallery-internal-alt',
5126 'img_link' => 'gallery-internal-link',
5127 ];
5128 if ( $handler ) {
5129 $paramMap += $handler->getParamMap();
5130 // We don't want people to specify per-image widths.
5131 // Additionally the width parameter would need special casing anyhow.
5132 unset( $paramMap['img_width'] );
5133 }
5134
5135 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5136
5137 $label = '';
5138 $alt = '';
5139 $link = '';
5140 $handlerOptions = [];
5141 if ( isset( $matches[3] ) ) {
5142 // look for an |alt= definition while trying not to break existing
5143 // captions with multiple pipes (|) in it, until a more sensible grammar
5144 // is defined for images in galleries
5145
5146 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5147 // splitting on '|' is a bit odd, and different from makeImage.
5148 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5149 // Protect LanguageConverter markup
5150 $parameterMatches = StringUtils::delimiterExplode(
5151 '-{', '}-', '|', $matches[3], true /* nested */
5152 );
5153
5154 foreach ( $parameterMatches as $parameterMatch ) {
5155 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5156 if ( $magicName ) {
5157 $paramName = $paramMap[$magicName];
5158
5159 switch ( $paramName ) {
5160 case 'gallery-internal-alt':
5161 $alt = $this->stripAltText( $match, false );
5162 break;
5163 case 'gallery-internal-link':
5164 $linkValue = $this->stripAltText( $match, false );
5165 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5166 // Result of LanguageConverter::markNoConversion
5167 // invoked on an external link.
5168 $linkValue = substr( $linkValue, 4, -2 );
5169 }
5170 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5171 if ( $type === 'link-url' ) {
5172 $link = $target;
5173 $this->mOutput->addExternalLink( $target );
5174 } elseif ( $type === 'link-title' ) {
5175 $link = $target->getLinkURL();
5176 $this->mOutput->addLink( $target );
5177 }
5178 break;
5179 default:
5180 // Must be a handler specific parameter.
5181 if ( $handler->validateParam( $paramName, $match ) ) {
5182 $handlerOptions[$paramName] = $match;
5183 } else {
5184 // Guess not, consider it as caption.
5185 wfDebug( "$parameterMatch failed parameter validation\n" );
5186 $label = $parameterMatch;
5187 }
5188 }
5189
5190 } else {
5191 // Last pipe wins.
5192 $label = $parameterMatch;
5193 }
5194 }
5195 }
5196
5197 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5198 }
5199 $html = $ig->toHTML();
5200 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5201 return $html;
5202 }
5203
5204 /**
5205 * @param MediaHandler $handler
5206 * @return array
5207 */
5208 public function getImageParams( $handler ) {
5209 if ( $handler ) {
5210 $handlerClass = get_class( $handler );
5211 } else {
5212 $handlerClass = '';
5213 }
5214 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5215 # Initialise static lists
5216 static $internalParamNames = [
5217 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5218 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5219 'bottom', 'text-bottom' ],
5220 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5221 'upright', 'border', 'link', 'alt', 'class' ],
5222 ];
5223 static $internalParamMap;
5224 if ( !$internalParamMap ) {
5225 $internalParamMap = [];
5226 foreach ( $internalParamNames as $type => $names ) {
5227 foreach ( $names as $name ) {
5228 // For grep: img_left, img_right, img_center, img_none,
5229 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5230 // img_bottom, img_text_bottom,
5231 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5232 // img_border, img_link, img_alt, img_class
5233 $magicName = str_replace( '-', '_', "img_$name" );
5234 $internalParamMap[$magicName] = [ $type, $name ];
5235 }
5236 }
5237 }
5238
5239 # Add handler params
5240 $paramMap = $internalParamMap;
5241 if ( $handler ) {
5242 $handlerParamMap = $handler->getParamMap();
5243 foreach ( $handlerParamMap as $magic => $paramName ) {
5244 $paramMap[$magic] = [ 'handler', $paramName ];
5245 }
5246 }
5247 $this->mImageParams[$handlerClass] = $paramMap;
5248 $this->mImageParamsMagicArray[$handlerClass] =
5249 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5250 }
5251 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5252 }
5253
5254 /**
5255 * Parse image options text and use it to make an image
5256 *
5257 * @param Title $title
5258 * @param string $options
5259 * @param LinkHolderArray|bool $holders
5260 * @return string HTML
5261 */
5262 public function makeImage( $title, $options, $holders = false ) {
5263 # Check if the options text is of the form "options|alt text"
5264 # Options are:
5265 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5266 # * left no resizing, just left align. label is used for alt= only
5267 # * right same, but right aligned
5268 # * none same, but not aligned
5269 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5270 # * center center the image
5271 # * frame Keep original image size, no magnify-button.
5272 # * framed Same as "frame"
5273 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5274 # * upright reduce width for upright images, rounded to full __0 px
5275 # * border draw a 1px border around the image
5276 # * alt Text for HTML alt attribute (defaults to empty)
5277 # * class Set a class for img node
5278 # * link Set the target of the image link. Can be external, interwiki, or local
5279 # vertical-align values (no % or length right now):
5280 # * baseline
5281 # * sub
5282 # * super
5283 # * top
5284 # * text-top
5285 # * middle
5286 # * bottom
5287 # * text-bottom
5288
5289 # Protect LanguageConverter markup when splitting into parts
5290 $parts = StringUtils::delimiterExplode(
5291 '-{', '}-', '|', $options, true /* allow nesting */
5292 );
5293
5294 # Give extensions a chance to select the file revision for us
5295 $options = [];
5296 $descQuery = false;
5297 Hooks::run( 'BeforeParserFetchFileAndTitle',
5298 [ $this, $title, &$options, &$descQuery ] );
5299 # Fetch and register the file (file title may be different via hooks)
5300 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5301
5302 # Get parameter map
5303 $handler = $file ? $file->getHandler() : false;
5304
5305 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5306
5307 if ( !$file ) {
5308 $this->addTrackingCategory( 'broken-file-category' );
5309 }
5310
5311 # Process the input parameters
5312 $caption = '';
5313 $params = [ 'frame' => [], 'handler' => [],
5314 'horizAlign' => [], 'vertAlign' => [] ];
5315 $seenformat = false;
5316 foreach ( $parts as $part ) {
5317 $part = trim( $part );
5318 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5319 $validated = false;
5320 if ( isset( $paramMap[$magicName] ) ) {
5321 list( $type, $paramName ) = $paramMap[$magicName];
5322
5323 # Special case; width and height come in one variable together
5324 if ( $type === 'handler' && $paramName === 'width' ) {
5325 $parsedWidthParam = self::parseWidthParam( $value );
5326 if ( isset( $parsedWidthParam['width'] ) ) {
5327 $width = $parsedWidthParam['width'];
5328 if ( $handler->validateParam( 'width', $width ) ) {
5329 $params[$type]['width'] = $width;
5330 $validated = true;
5331 }
5332 }
5333 if ( isset( $parsedWidthParam['height'] ) ) {
5334 $height = $parsedWidthParam['height'];
5335 if ( $handler->validateParam( 'height', $height ) ) {
5336 $params[$type]['height'] = $height;
5337 $validated = true;
5338 }
5339 }
5340 # else no validation -- T15436
5341 } else {
5342 if ( $type === 'handler' ) {
5343 # Validate handler parameter
5344 $validated = $handler->validateParam( $paramName, $value );
5345 } else {
5346 # Validate internal parameters
5347 switch ( $paramName ) {
5348 case 'manualthumb':
5349 case 'alt':
5350 case 'class':
5351 # @todo FIXME: Possibly check validity here for
5352 # manualthumb? downstream behavior seems odd with
5353 # missing manual thumbs.
5354 $validated = true;
5355 $value = $this->stripAltText( $value, $holders );
5356 break;
5357 case 'link':
5358 list( $paramName, $value ) =
5359 $this->parseLinkParameter(
5360 $this->stripAltText( $value, $holders )
5361 );
5362 if ( $paramName ) {
5363 $validated = true;
5364 if ( $paramName === 'no-link' ) {
5365 $value = true;
5366 }
5367 if ( ( $paramName === 'link-url' ) && $this->mOptions->getExternalLinkTarget() ) {
5368 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5369 }
5370 }
5371 break;
5372 case 'frameless':
5373 case 'framed':
5374 case 'thumbnail':
5375 // use first appearing option, discard others.
5376 $validated = !$seenformat;
5377 $seenformat = true;
5378 break;
5379 default:
5380 # Most other things appear to be empty or numeric...
5381 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5382 }
5383 }
5384
5385 if ( $validated ) {
5386 $params[$type][$paramName] = $value;
5387 }
5388 }
5389 }
5390 if ( !$validated ) {
5391 $caption = $part;
5392 }
5393 }
5394
5395 # Process alignment parameters
5396 if ( $params['horizAlign'] ) {
5397 $params['frame']['align'] = key( $params['horizAlign'] );
5398 }
5399 if ( $params['vertAlign'] ) {
5400 $params['frame']['valign'] = key( $params['vertAlign'] );
5401 }
5402
5403 $params['frame']['caption'] = $caption;
5404
5405 # Will the image be presented in a frame, with the caption below?
5406 $imageIsFramed = isset( $params['frame']['frame'] )
5407 || isset( $params['frame']['framed'] )
5408 || isset( $params['frame']['thumbnail'] )
5409 || isset( $params['frame']['manualthumb'] );
5410
5411 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5412 # came to also set the caption, ordinary text after the image -- which
5413 # makes no sense, because that just repeats the text multiple times in
5414 # screen readers. It *also* came to set the title attribute.
5415 # Now that we have an alt attribute, we should not set the alt text to
5416 # equal the caption: that's worse than useless, it just repeats the
5417 # text. This is the framed/thumbnail case. If there's no caption, we
5418 # use the unnamed parameter for alt text as well, just for the time be-
5419 # ing, if the unnamed param is set and the alt param is not.
5420 # For the future, we need to figure out if we want to tweak this more,
5421 # e.g., introducing a title= parameter for the title; ignoring the un-
5422 # named parameter entirely for images without a caption; adding an ex-
5423 # plicit caption= parameter and preserving the old magic unnamed para-
5424 # meter for BC; ...
5425 if ( $imageIsFramed ) { # Framed image
5426 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5427 # No caption or alt text, add the filename as the alt text so
5428 # that screen readers at least get some description of the image
5429 $params['frame']['alt'] = $title->getText();
5430 }
5431 # Do not set $params['frame']['title'] because tooltips don't make sense
5432 # for framed images
5433 } else { # Inline image
5434 if ( !isset( $params['frame']['alt'] ) ) {
5435 # No alt text, use the "caption" for the alt text
5436 if ( $caption !== '' ) {
5437 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5438 } else {
5439 # No caption, fall back to using the filename for the
5440 # alt text
5441 $params['frame']['alt'] = $title->getText();
5442 }
5443 }
5444 # Use the "caption" for the tooltip text
5445 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5446 }
5447 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5448
5449 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5450
5451 # Linker does the rest
5452 $time = $options['time'] ?? false;
5453 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5454 $time, $descQuery, $this->mOptions->getThumbSize() );
5455
5456 # Give the handler a chance to modify the parser object
5457 if ( $handler ) {
5458 $handler->parserTransformHook( $this, $file );
5459 }
5460
5461 return $ret;
5462 }
5463
5464 /**
5465 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5466 *
5467 * Adds an entry to appropriate link tables.
5468 *
5469 * @since 1.32
5470 * @param string $value
5471 * @return array of `[ type, target ]`, where:
5472 * - `type` is one of:
5473 * - `null`: Given value is not a valid link target, use default
5474 * - `'no-link'`: Given value is empty, do not generate a link
5475 * - `'link-url'`: Given value is a valid external link
5476 * - `'link-title'`: Given value is a valid internal link
5477 * - `target` is:
5478 * - When `type` is `null` or `'no-link'`: `false`
5479 * - When `type` is `'link-url'`: URL string corresponding to given value
5480 * - When `type` is `'link-title'`: Title object corresponding to given value
5481 */
5482 public function parseLinkParameter( $value ) {
5483 $chars = self::EXT_LINK_URL_CLASS;
5484 $addr = self::EXT_LINK_ADDR;
5485 $prots = $this->mUrlProtocols;
5486 $type = null;
5487 $target = false;
5488 if ( $value === '' ) {
5489 $type = 'no-link';
5490 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5491 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5492 $this->mOutput->addExternalLink( $value );
5493 $type = 'link-url';
5494 $target = $value;
5495 }
5496 } else {
5497 $linkTitle = Title::newFromText( $value );
5498 if ( $linkTitle ) {
5499 $this->mOutput->addLink( $linkTitle );
5500 $type = 'link-title';
5501 $target = $linkTitle;
5502 }
5503 }
5504 return [ $type, $target ];
5505 }
5506
5507 /**
5508 * @param string $caption
5509 * @param LinkHolderArray|bool $holders
5510 * @return mixed|string
5511 */
5512 protected function stripAltText( $caption, $holders ) {
5513 # Strip bad stuff out of the title (tooltip). We can't just use
5514 # replaceLinkHoldersText() here, because if this function is called
5515 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5516 if ( $holders ) {
5517 $tooltip = $holders->replaceText( $caption );
5518 } else {
5519 $tooltip = $this->replaceLinkHoldersText( $caption );
5520 }
5521
5522 # make sure there are no placeholders in thumbnail attributes
5523 # that are later expanded to html- so expand them now and
5524 # remove the tags
5525 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5526 # Compatibility hack! In HTML certain entity references not terminated
5527 # by a semicolon are decoded (but not if we're in an attribute; that's
5528 # how link URLs get away without properly escaping & in queries).
5529 # But wikitext has always required semicolon-termination of entities,
5530 # so encode & where needed to avoid decode of semicolon-less entities.
5531 # See T209236 and
5532 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5533 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5534 $tooltip = preg_replace( "/
5535 & # 1. entity prefix
5536 (?= # 2. followed by:
5537 (?: # a. one of the legacy semicolon-less named entities
5538 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5539 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5540 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5541 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5542 U(?:acute|circ|grave|uml)|Yacute|
5543 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5544 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5545 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5546 frac(?:1(?:2|4)|34)|
5547 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5548 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5549 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5550 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5551 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5552 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5553 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5554 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5555 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5556 )
5557 (?:[^;]|$)) # b. and not followed by a semicolon
5558 # S = study, for efficiency
5559 /Sx", '&amp;', $tooltip );
5560 $tooltip = Sanitizer::stripAllTags( $tooltip );
5561
5562 return $tooltip;
5563 }
5564
5565 /**
5566 * Set a flag in the output object indicating that the content is dynamic and
5567 * shouldn't be cached.
5568 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5569 */
5570 public function disableCache() {
5571 wfDebug( "Parser output marked as uncacheable.\n" );
5572 if ( !$this->mOutput ) {
5573 throw new MWException( __METHOD__ .
5574 " can only be called when actually parsing something" );
5575 }
5576 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5577 }
5578
5579 /**
5580 * Callback from the Sanitizer for expanding items found in HTML attribute
5581 * values, so they can be safely tested and escaped.
5582 *
5583 * @param string &$text
5584 * @param bool|PPFrame $frame
5585 * @return string
5586 */
5587 public function attributeStripCallback( &$text, $frame = false ) {
5588 $text = $this->replaceVariables( $text, $frame );
5589 $text = $this->mStripState->unstripBoth( $text );
5590 return $text;
5591 }
5592
5593 /**
5594 * Accessor
5595 *
5596 * @return array
5597 */
5598 public function getTags() {
5599 $this->firstCallInit();
5600 return array_merge(
5601 array_keys( $this->mTransparentTagHooks ),
5602 array_keys( $this->mTagHooks ),
5603 array_keys( $this->mFunctionTagHooks )
5604 );
5605 }
5606
5607 /**
5608 * @since 1.32
5609 * @return array
5610 */
5611 public function getFunctionSynonyms() {
5612 $this->firstCallInit();
5613 return $this->mFunctionSynonyms;
5614 }
5615
5616 /**
5617 * @since 1.32
5618 * @return string
5619 */
5620 public function getUrlProtocols() {
5621 return $this->mUrlProtocols;
5622 }
5623
5624 /**
5625 * Replace transparent tags in $text with the values given by the callbacks.
5626 *
5627 * Transparent tag hooks are like regular XML-style tag hooks, except they
5628 * operate late in the transformation sequence, on HTML instead of wikitext.
5629 *
5630 * @param string $text
5631 *
5632 * @return string
5633 */
5634 public function replaceTransparentTags( $text ) {
5635 $matches = [];
5636 $elements = array_keys( $this->mTransparentTagHooks );
5637 $text = self::extractTagsAndParams( $elements, $text, $matches );
5638 $replacements = [];
5639
5640 foreach ( $matches as $marker => $data ) {
5641 list( $element, $content, $params, $tag ) = $data;
5642 $tagName = strtolower( $element );
5643 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5644 $output = call_user_func_array(
5645 $this->mTransparentTagHooks[$tagName],
5646 [ $content, $params, $this ]
5647 );
5648 } else {
5649 $output = $tag;
5650 }
5651 $replacements[$marker] = $output;
5652 }
5653 return strtr( $text, $replacements );
5654 }
5655
5656 /**
5657 * Break wikitext input into sections, and either pull or replace
5658 * some particular section's text.
5659 *
5660 * External callers should use the getSection and replaceSection methods.
5661 *
5662 * @param string $text Page wikitext
5663 * @param string|int $sectionId A section identifier string of the form:
5664 * "<flag1> - <flag2> - ... - <section number>"
5665 *
5666 * Currently the only recognised flag is "T", which means the target section number
5667 * was derived during a template inclusion parse, in other words this is a template
5668 * section edit link. If no flags are given, it was an ordinary section edit link.
5669 * This flag is required to avoid a section numbering mismatch when a section is
5670 * enclosed by "<includeonly>" (T8563).
5671 *
5672 * The section number 0 pulls the text before the first heading; other numbers will
5673 * pull the given section along with its lower-level subsections. If the section is
5674 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5675 *
5676 * Section 0 is always considered to exist, even if it only contains the empty
5677 * string. If $text is the empty string and section 0 is replaced, $newText is
5678 * returned.
5679 *
5680 * @param string $mode One of "get" or "replace"
5681 * @param string $newText Replacement text for section data.
5682 * @return string For "get", the extracted section text.
5683 * for "replace", the whole page with the section replaced.
5684 */
5685 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5686 global $wgTitle; # not generally used but removes an ugly failure mode
5687
5688 $magicScopeVariable = $this->lock();
5689 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5690 $outText = '';
5691 $frame = $this->getPreprocessor()->newFrame();
5692
5693 # Process section extraction flags
5694 $flags = 0;
5695 $sectionParts = explode( '-', $sectionId );
5696 $sectionIndex = array_pop( $sectionParts );
5697 foreach ( $sectionParts as $part ) {
5698 if ( $part === 'T' ) {
5699 $flags |= self::PTD_FOR_INCLUSION;
5700 }
5701 }
5702
5703 # Check for empty input
5704 if ( strval( $text ) === '' ) {
5705 # Only sections 0 and T-0 exist in an empty document
5706 if ( $sectionIndex == 0 ) {
5707 if ( $mode === 'get' ) {
5708 return '';
5709 }
5710
5711 return $newText;
5712 } else {
5713 if ( $mode === 'get' ) {
5714 return $newText;
5715 }
5716
5717 return $text;
5718 }
5719 }
5720
5721 # Preprocess the text
5722 $root = $this->preprocessToDom( $text, $flags );
5723
5724 # <h> nodes indicate section breaks
5725 # They can only occur at the top level, so we can find them by iterating the root's children
5726 $node = $root->getFirstChild();
5727
5728 # Find the target section
5729 if ( $sectionIndex == 0 ) {
5730 # Section zero doesn't nest, level=big
5731 $targetLevel = 1000;
5732 } else {
5733 while ( $node ) {
5734 if ( $node->getName() === 'h' ) {
5735 $bits = $node->splitHeading();
5736 if ( $bits['i'] == $sectionIndex ) {
5737 $targetLevel = $bits['level'];
5738 break;
5739 }
5740 }
5741 if ( $mode === 'replace' ) {
5742 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5743 }
5744 $node = $node->getNextSibling();
5745 }
5746 }
5747
5748 if ( !$node ) {
5749 # Not found
5750 if ( $mode === 'get' ) {
5751 return $newText;
5752 } else {
5753 return $text;
5754 }
5755 }
5756
5757 # Find the end of the section, including nested sections
5758 do {
5759 if ( $node->getName() === 'h' ) {
5760 $bits = $node->splitHeading();
5761 $curLevel = $bits['level'];
5762 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5763 break;
5764 }
5765 }
5766 if ( $mode === 'get' ) {
5767 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5768 }
5769 $node = $node->getNextSibling();
5770 } while ( $node );
5771
5772 # Write out the remainder (in replace mode only)
5773 if ( $mode === 'replace' ) {
5774 # Output the replacement text
5775 # Add two newlines on -- trailing whitespace in $newText is conventionally
5776 # stripped by the editor, so we need both newlines to restore the paragraph gap
5777 # Only add trailing whitespace if there is newText
5778 if ( $newText != "" ) {
5779 $outText .= $newText . "\n\n";
5780 }
5781
5782 while ( $node ) {
5783 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5784 $node = $node->getNextSibling();
5785 }
5786 }
5787
5788 if ( is_string( $outText ) ) {
5789 # Re-insert stripped tags
5790 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5791 }
5792
5793 return $outText;
5794 }
5795
5796 /**
5797 * This function returns the text of a section, specified by a number ($section).
5798 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5799 * the first section before any such heading (section 0).
5800 *
5801 * If a section contains subsections, these are also returned.
5802 *
5803 * @param string $text Text to look in
5804 * @param string|int $sectionId Section identifier as a number or string
5805 * (e.g. 0, 1 or 'T-1').
5806 * @param string $defaultText Default to return if section is not found
5807 *
5808 * @return string Text of the requested section
5809 */
5810 public function getSection( $text, $sectionId, $defaultText = '' ) {
5811 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5812 }
5813
5814 /**
5815 * This function returns $oldtext after the content of the section
5816 * specified by $section has been replaced with $text. If the target
5817 * section does not exist, $oldtext is returned unchanged.
5818 *
5819 * @param string $oldText Former text of the article
5820 * @param string|int $sectionId Section identifier as a number or string
5821 * (e.g. 0, 1 or 'T-1').
5822 * @param string $newText Replacing text
5823 *
5824 * @return string Modified text
5825 */
5826 public function replaceSection( $oldText, $sectionId, $newText ) {
5827 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5828 }
5829
5830 /**
5831 * Get the ID of the revision we are parsing
5832 *
5833 * @return int|null
5834 */
5835 public function getRevisionId() {
5836 return $this->mRevisionId;
5837 }
5838
5839 /**
5840 * Get the revision object for $this->mRevisionId
5841 *
5842 * @return Revision|null Either a Revision object or null
5843 * @since 1.23 (public since 1.23)
5844 */
5845 public function getRevisionObject() {
5846 if ( !is_null( $this->mRevisionObject ) ) {
5847 return $this->mRevisionObject;
5848 }
5849
5850 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5851 // This is useful when parsing revision that has not yet been saved.
5852 // However, if we get back a saved revision even though we are in
5853 // preview mode, we'll have to ignore it, see below.
5854 // NOTE: This callback may be used to inject an OLD revision that was
5855 // already loaded, so "current" is a bit of a misnomer. We can't just
5856 // skip it if mRevisionId is set.
5857 $rev = call_user_func(
5858 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5859 );
5860
5861 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
5862 // We are in preview mode (mRevisionId is null), and the current revision callback
5863 // returned an existing revision. Ignore it and return null, it's probably the page's
5864 // current revision, which is not what we want here. Note that we do want to call the
5865 // callback to allow the unsaved revision to be injected here, e.g. for
5866 // self-transclusion previews.
5867 return null;
5868 }
5869
5870 // If the parse is for a new revision, then the callback should have
5871 // already been set to force the object and should match mRevisionId.
5872 // If not, try to fetch by mRevisionId for sanity.
5873 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
5874 $rev = Revision::newFromId( $this->mRevisionId );
5875 }
5876
5877 $this->mRevisionObject = $rev;
5878
5879 return $this->mRevisionObject;
5880 }
5881
5882 /**
5883 * Get the timestamp associated with the current revision, adjusted for
5884 * the default server-local timestamp
5885 * @return string
5886 */
5887 public function getRevisionTimestamp() {
5888 if ( is_null( $this->mRevisionTimestamp ) ) {
5889 $revObject = $this->getRevisionObject();
5890 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5891
5892 # The cryptic '' timezone parameter tells to use the site-default
5893 # timezone offset instead of the user settings.
5894 # Since this value will be saved into the parser cache, served
5895 # to other users, and potentially even used inside links and such,
5896 # it needs to be consistent for all visitors.
5897 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
5898 }
5899 return $this->mRevisionTimestamp;
5900 }
5901
5902 /**
5903 * Get the name of the user that edited the last revision
5904 *
5905 * @return string User name
5906 */
5907 public function getRevisionUser() {
5908 if ( is_null( $this->mRevisionUser ) ) {
5909 $revObject = $this->getRevisionObject();
5910
5911 # if this template is subst: the revision id will be blank,
5912 # so just use the current user's name
5913 if ( $revObject ) {
5914 $this->mRevisionUser = $revObject->getUserText();
5915 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5916 $this->mRevisionUser = $this->getUser()->getName();
5917 }
5918 }
5919 return $this->mRevisionUser;
5920 }
5921
5922 /**
5923 * Get the size of the revision
5924 *
5925 * @return int|null Revision size
5926 */
5927 public function getRevisionSize() {
5928 if ( is_null( $this->mRevisionSize ) ) {
5929 $revObject = $this->getRevisionObject();
5930
5931 # if this variable is subst: the revision id will be blank,
5932 # so just use the parser input size, because the own substituation
5933 # will change the size.
5934 if ( $revObject ) {
5935 $this->mRevisionSize = $revObject->getSize();
5936 } else {
5937 $this->mRevisionSize = $this->mInputSize;
5938 }
5939 }
5940 return $this->mRevisionSize;
5941 }
5942
5943 /**
5944 * Mutator for $mDefaultSort
5945 *
5946 * @param string $sort New value
5947 */
5948 public function setDefaultSort( $sort ) {
5949 $this->mDefaultSort = $sort;
5950 $this->mOutput->setProperty( 'defaultsort', $sort );
5951 }
5952
5953 /**
5954 * Accessor for $mDefaultSort
5955 * Will use the empty string if none is set.
5956 *
5957 * This value is treated as a prefix, so the
5958 * empty string is equivalent to sorting by
5959 * page name.
5960 *
5961 * @return string
5962 */
5963 public function getDefaultSort() {
5964 if ( $this->mDefaultSort !== false ) {
5965 return $this->mDefaultSort;
5966 } else {
5967 return '';
5968 }
5969 }
5970
5971 /**
5972 * Accessor for $mDefaultSort
5973 * Unlike getDefaultSort(), will return false if none is set
5974 *
5975 * @return string|bool
5976 */
5977 public function getCustomDefaultSort() {
5978 return $this->mDefaultSort;
5979 }
5980
5981 private static function getSectionNameFromStrippedText( $text ) {
5982 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5983 $text = Sanitizer::decodeCharReferences( $text );
5984 $text = self::normalizeSectionName( $text );
5985 return $text;
5986 }
5987
5988 private static function makeAnchor( $sectionName ) {
5989 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5990 }
5991
5992 private function makeLegacyAnchor( $sectionName ) {
5993 $fragmentMode = $this->siteConfig->get( 'FragmentMode' );
5994 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
5995 // ForAttribute() and ForLink() are the same for legacy encoding
5996 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5997 } else {
5998 $id = Sanitizer::escapeIdForLink( $sectionName );
5999 }
6000
6001 return "#$id";
6002 }
6003
6004 /**
6005 * Try to guess the section anchor name based on a wikitext fragment
6006 * presumably extracted from a heading, for example "Header" from
6007 * "== Header ==".
6008 *
6009 * @param string $text
6010 * @return string Anchor (starting with '#')
6011 */
6012 public function guessSectionNameFromWikiText( $text ) {
6013 # Strip out wikitext links(they break the anchor)
6014 $text = $this->stripSectionName( $text );
6015 $sectionName = self::getSectionNameFromStrippedText( $text );
6016 return self::makeAnchor( $sectionName );
6017 }
6018
6019 /**
6020 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6021 * instead, if possible. For use in redirects, since various versions
6022 * of Microsoft browsers interpret Location: headers as something other
6023 * than UTF-8, resulting in breakage.
6024 *
6025 * @param string $text The section name
6026 * @return string Anchor (starting with '#')
6027 */
6028 public function guessLegacySectionNameFromWikiText( $text ) {
6029 # Strip out wikitext links(they break the anchor)
6030 $text = $this->stripSectionName( $text );
6031 $sectionName = self::getSectionNameFromStrippedText( $text );
6032 return $this->makeLegacyAnchor( $sectionName );
6033 }
6034
6035 /**
6036 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6037 * @param string $text Section name (plain text)
6038 * @return string Anchor (starting with '#')
6039 */
6040 public static function guessSectionNameFromStrippedText( $text ) {
6041 $sectionName = self::getSectionNameFromStrippedText( $text );
6042 return self::makeAnchor( $sectionName );
6043 }
6044
6045 /**
6046 * Apply the same normalization as code making links to this section would
6047 *
6048 * @param string $text
6049 * @return string
6050 */
6051 private static function normalizeSectionName( $text ) {
6052 # T90902: ensure the same normalization is applied for IDs as to links
6053 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
6054 try {
6055
6056 $parts = $titleParser->splitTitleString( "#$text" );
6057 } catch ( MalformedTitleException $ex ) {
6058 return $text;
6059 }
6060 return $parts['fragment'];
6061 }
6062
6063 /**
6064 * Strips a text string of wikitext for use in a section anchor
6065 *
6066 * Accepts a text string and then removes all wikitext from the
6067 * string and leaves only the resultant text (i.e. the result of
6068 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6069 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6070 * to create valid section anchors by mimicing the output of the
6071 * parser when headings are parsed.
6072 *
6073 * @param string $text Text string to be stripped of wikitext
6074 * for use in a Section anchor
6075 * @return string Filtered text string
6076 */
6077 public function stripSectionName( $text ) {
6078 # Strip internal link markup
6079 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6080 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6081
6082 # Strip external link markup
6083 # @todo FIXME: Not tolerant to blank link text
6084 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6085 # on how many empty links there are on the page - need to figure that out.
6086 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6087
6088 # Parse wikitext quotes (italics & bold)
6089 $text = $this->doQuotes( $text );
6090
6091 # Strip HTML tags
6092 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6093 return $text;
6094 }
6095
6096 /**
6097 * strip/replaceVariables/unstrip for preprocessor regression testing
6098 *
6099 * @param string $text
6100 * @param Title $title
6101 * @param ParserOptions $options
6102 * @param int $outputType
6103 *
6104 * @return string
6105 */
6106 public function testSrvus( $text, Title $title, ParserOptions $options,
6107 $outputType = self::OT_HTML
6108 ) {
6109 $magicScopeVariable = $this->lock();
6110 $this->startParse( $title, $options, $outputType, true );
6111
6112 $text = $this->replaceVariables( $text );
6113 $text = $this->mStripState->unstripBoth( $text );
6114 $text = Sanitizer::removeHTMLtags( $text );
6115 return $text;
6116 }
6117
6118 /**
6119 * @param string $text
6120 * @param Title $title
6121 * @param ParserOptions $options
6122 * @return string
6123 */
6124 public function testPst( $text, Title $title, ParserOptions $options ) {
6125 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6126 }
6127
6128 /**
6129 * @param string $text
6130 * @param Title $title
6131 * @param ParserOptions $options
6132 * @return string
6133 */
6134 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6135 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6136 }
6137
6138 /**
6139 * Call a callback function on all regions of the given text that are not
6140 * inside strip markers, and replace those regions with the return value
6141 * of the callback. For example, with input:
6142 *
6143 * aaa<MARKER>bbb
6144 *
6145 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6146 * two strings will be replaced with the value returned by the callback in
6147 * each case.
6148 *
6149 * @param string $s
6150 * @param callable $callback
6151 *
6152 * @return string
6153 */
6154 public function markerSkipCallback( $s, $callback ) {
6155 $i = 0;
6156 $out = '';
6157 while ( $i < strlen( $s ) ) {
6158 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6159 if ( $markerStart === false ) {
6160 $out .= call_user_func( $callback, substr( $s, $i ) );
6161 break;
6162 } else {
6163 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6164 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6165 if ( $markerEnd === false ) {
6166 $out .= substr( $s, $markerStart );
6167 break;
6168 } else {
6169 $markerEnd += strlen( self::MARKER_SUFFIX );
6170 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6171 $i = $markerEnd;
6172 }
6173 }
6174 }
6175 return $out;
6176 }
6177
6178 /**
6179 * Remove any strip markers found in the given text.
6180 *
6181 * @param string $text
6182 * @return string
6183 */
6184 public function killMarkers( $text ) {
6185 return $this->mStripState->killMarkers( $text );
6186 }
6187
6188 /**
6189 * Save the parser state required to convert the given half-parsed text to
6190 * HTML. "Half-parsed" in this context means the output of
6191 * recursiveTagParse() or internalParse(). This output has strip markers
6192 * from replaceVariables (extensionSubstitution() etc.), and link
6193 * placeholders from replaceLinkHolders().
6194 *
6195 * Returns an array which can be serialized and stored persistently. This
6196 * array can later be loaded into another parser instance with
6197 * unserializeHalfParsedText(). The text can then be safely incorporated into
6198 * the return value of a parser hook.
6199 *
6200 * @deprecated since 1.31
6201 * @param string $text
6202 *
6203 * @return array
6204 */
6205 public function serializeHalfParsedText( $text ) {
6206 wfDeprecated( __METHOD__, '1.31' );
6207 $data = [
6208 'text' => $text,
6209 'version' => self::HALF_PARSED_VERSION,
6210 'stripState' => $this->mStripState->getSubState( $text ),
6211 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6212 ];
6213 return $data;
6214 }
6215
6216 /**
6217 * Load the parser state given in the $data array, which is assumed to
6218 * have been generated by serializeHalfParsedText(). The text contents is
6219 * extracted from the array, and its markers are transformed into markers
6220 * appropriate for the current Parser instance. This transformed text is
6221 * returned, and can be safely included in the return value of a parser
6222 * hook.
6223 *
6224 * If the $data array has been stored persistently, the caller should first
6225 * check whether it is still valid, by calling isValidHalfParsedText().
6226 *
6227 * @deprecated since 1.31
6228 * @param array $data Serialized data
6229 * @throws MWException
6230 * @return string
6231 */
6232 public function unserializeHalfParsedText( $data ) {
6233 wfDeprecated( __METHOD__, '1.31' );
6234 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6235 throw new MWException( __METHOD__ . ': invalid version' );
6236 }
6237
6238 # First, extract the strip state.
6239 $texts = [ $data['text'] ];
6240 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6241
6242 # Now renumber links
6243 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6244
6245 # Should be good to go.
6246 return $texts[0];
6247 }
6248
6249 /**
6250 * Returns true if the given array, presumed to be generated by
6251 * serializeHalfParsedText(), is compatible with the current version of the
6252 * parser.
6253 *
6254 * @deprecated since 1.31
6255 * @param array $data
6256 *
6257 * @return bool
6258 */
6259 public function isValidHalfParsedText( $data ) {
6260 wfDeprecated( __METHOD__, '1.31' );
6261 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6262 }
6263
6264 /**
6265 * Parsed a width param of imagelink like 300px or 200x300px
6266 *
6267 * @param string $value
6268 * @param bool $parseHeight
6269 *
6270 * @return array
6271 * @since 1.20
6272 */
6273 public static function parseWidthParam( $value, $parseHeight = true ) {
6274 $parsedWidthParam = [];
6275 if ( $value === '' ) {
6276 return $parsedWidthParam;
6277 }
6278 $m = [];
6279 # (T15500) In both cases (width/height and width only),
6280 # permit trailing "px" for backward compatibility.
6281 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6282 $width = intval( $m[1] );
6283 $height = intval( $m[2] );
6284 $parsedWidthParam['width'] = $width;
6285 $parsedWidthParam['height'] = $height;
6286 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6287 $width = intval( $value );
6288 $parsedWidthParam['width'] = $width;
6289 }
6290 return $parsedWidthParam;
6291 }
6292
6293 /**
6294 * Lock the current instance of the parser.
6295 *
6296 * This is meant to stop someone from calling the parser
6297 * recursively and messing up all the strip state.
6298 *
6299 * @throws MWException If parser is in a parse
6300 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6301 */
6302 protected function lock() {
6303 if ( $this->mInParse ) {
6304 throw new MWException( "Parser state cleared while parsing. "
6305 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6306 }
6307
6308 // Save the backtrace when locking, so that if some code tries locking again,
6309 // we can print the lock owner's backtrace for easier debugging
6310 $e = new Exception;
6311 $this->mInParse = $e->getTraceAsString();
6312
6313 $recursiveCheck = new ScopedCallback( function () {
6314 $this->mInParse = false;
6315 } );
6316
6317 return $recursiveCheck;
6318 }
6319
6320 /**
6321 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6322 *
6323 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6324 * or if there is more than one <p/> tag in the input HTML.
6325 *
6326 * @param string $html
6327 * @return string
6328 * @since 1.24
6329 */
6330 public static function stripOuterParagraph( $html ) {
6331 $m = [];
6332 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6333 $html = $m[1];
6334 }
6335
6336 return $html;
6337 }
6338
6339 /**
6340 * Return this parser if it is not doing anything, otherwise
6341 * get a fresh parser. You can use this method by doing
6342 * $myParser = $wgParser->getFreshParser(), or more simply
6343 * $wgParser->getFreshParser()->parse( ... );
6344 * if you're unsure if $wgParser is safe to use.
6345 *
6346 * @since 1.24
6347 * @return Parser A parser object that is not parsing anything
6348 */
6349 public function getFreshParser() {
6350 if ( $this->mInParse ) {
6351 return $this->factory->create();
6352 } else {
6353 return $this;
6354 }
6355 }
6356
6357 /**
6358 * Set's up the PHP implementation of OOUI for use in this request
6359 * and instructs OutputPage to enable OOUI for itself.
6360 *
6361 * @since 1.26
6362 */
6363 public function enableOOUI() {
6364 OutputPage::setupOOUI();
6365 $this->mOutput->setEnableOOUI( true );
6366 }
6367 }
Wiklou, le Wiki du Wiklou