dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "resourceloader: Omit non-existent messages in MessageBlobStore"
[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\Linker\LinkRendererFactory;
25 use MediaWiki\Linker\LinkTarget;
26 use MediaWiki\MediaWikiServices;
27 use MediaWiki\Special\SpecialPageFactory;
28 use Wikimedia\ScopedCallback;
29
30 /**
31 * @defgroup Parser Parser
32 */
33
34 /**
35 * PHP Parser - Processes wiki markup (which uses a more user-friendly
36 * syntax, such as "[[link]]" for making links), and provides a one-way
37 * transformation of that wiki markup it into (X)HTML output / markup
38 * (which in turn the browser understands, and can display).
39 *
40 * There are seven main entry points into the Parser class:
41 *
42 * - Parser::parse()
43 * produces HTML output
44 * - Parser::preSaveTransform()
45 * produces altered wiki markup
46 * - Parser::preprocess()
47 * removes HTML comments and expands templates
48 * - Parser::cleanSig() and Parser::cleanSigInSig()
49 * cleans a signature before saving it to preferences
50 * - Parser::getSection()
51 * return the content of a section from an article for section editing
52 * - Parser::replaceSection()
53 * replaces a section by number inside an article
54 * - Parser::getPreloadText()
55 * removes <noinclude> sections and <includeonly> tags
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // phpcs:ignore Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102
103 # Regular expression for a non-newline space
104 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 /**
118 * @var string Prefix and suffix for temporary replacement strings
119 * for the multipass parser.
120 *
121 * \x7f should never appear in input as it's disallowed in XML.
122 * Using it at the front also gives us a little extra robustness
123 * since it shouldn't match when butted up against identifier-like
124 * string constructs.
125 *
126 * Must not consist of all title characters, or else it will change
127 * the behavior of <nowiki> in a link.
128 *
129 * Must have a character that needs escaping in attributes, otherwise
130 * someone could put a strip marker in an attribute, to get around
131 * escaping quote marks, and break out of the attribute. Thus we add
132 * `'".
133 */
134 const MARKER_SUFFIX = "-QINU`\"'\x7f";
135 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
136
137 # Markers used for wrapping the table of contents
138 const TOC_START = '<mw:toc>';
139 const TOC_END = '</mw:toc>';
140
141 /** @var int Assume that no output will later be saved this many seconds after parsing */
142 const MAX_TTS = 900;
143
144 # Persistent:
145 public $mTagHooks = [];
146 public $mTransparentTagHooks = [];
147 public $mFunctionHooks = [];
148 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
149 public $mFunctionTagHooks = [];
150 public $mStripList = [];
151 public $mDefaultStripList = [];
152 public $mVarCache = [];
153 public $mImageParams = [];
154 public $mImageParamsMagicArray = [];
155 public $mMarkerIndex = 0;
156 /**
157 * @var bool Whether firstCallInit still needs to be called
158 */
159 public $mFirstCall = true;
160
161 # Initialised by initialiseVariables()
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mVariables;
167
168 /**
169 * @var MagicWordArray
170 */
171 public $mSubstWords;
172 # Initialised in constructor
173 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
174
175 # Initialized in getPreprocessor()
176 /** @var Preprocessor */
177 public $mPreprocessor;
178
179 # Cleared with clearState():
180 /**
181 * @var ParserOutput
182 */
183 public $mOutput;
184 public $mAutonumber;
185
186 /**
187 * @var StripState
188 */
189 public $mStripState;
190
191 public $mIncludeCount;
192 /**
193 * @var LinkHolderArray
194 */
195 public $mLinkHolders;
196
197 public $mLinkID;
198 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
199 public $mDefaultSort;
200 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
201 public $mExpensiveFunctionCount; # number of expensive parser function calls
202 public $mShowToc, $mForceTocPosition;
203
204 /**
205 * @var User
206 */
207 public $mUser; # User object; only used when doing pre-save transform
208
209 # Temporary
210 # These are variables reset at least once per parse regardless of $clearState
211
212 /**
213 * @var ParserOptions
214 */
215 public $mOptions;
216
217 /**
218 * @var Title
219 */
220 public $mTitle; # Title context, used for self-link rendering and similar things
221 public $mOutputType; # Output type, one of the OT_xxx constants
222 public $ot; # Shortcut alias, see setOutputType()
223 public $mRevisionObject; # The revision object of the specified revision ID
224 public $mRevisionId; # ID to display in {{REVISIONID}} tags
225 public $mRevisionTimestamp; # The timestamp of the specified revision ID
226 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
227 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
228 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
229 public $mInputSize = false; # For {{PAGESIZE}} on current page.
230
231 /**
232 * @var string Deprecated accessor for the strip marker prefix.
233 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
234 */
235 public $mUniqPrefix = self::MARKER_PREFIX;
236
237 /**
238 * @var array Array with the language name of each language link (i.e. the
239 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
240 * duplicate language links to the ParserOutput.
241 */
242 public $mLangLinkLanguages;
243
244 /**
245 * @var MapCacheLRU|null
246 * @since 1.24
247 *
248 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
249 */
250 public $currentRevisionCache;
251
252 /**
253 * @var bool|string Recursive call protection.
254 * This variable should be treated as if it were private.
255 */
256 public $mInParse = false;
257
258 /** @var SectionProfiler */
259 protected $mProfiler;
260
261 /**
262 * @var LinkRenderer
263 */
264 protected $mLinkRenderer;
265
266 /** @var MagicWordFactory */
267 private $magicWordFactory;
268
269 /** @var Language */
270 private $contLang;
271
272 /** @var ParserFactory */
273 private $factory;
274
275 /** @var SpecialPageFactory */
276 private $specialPageFactory;
277
278 /** @var Config */
279 private $siteConfig;
280
281 /** @var LinkRendererFactory */
282 private $linkRendererFactory;
283
284 /** @var NamespaceInfo */
285 private $nsInfo;
286
287 /**
288 * @param array $parserConf See $wgParserConf documentation
289 * @param MagicWordFactory|null $magicWordFactory
290 * @param Language|null $contLang Content language
291 * @param ParserFactory|null $factory
292 * @param string|null $urlProtocols As returned from wfUrlProtocols()
293 * @param SpecialPageFactory|null $spFactory
294 * @param Config|null $siteConfig
295 * @param LinkRendererFactory|null $linkRendererFactory
296 * @param NamespaceInfo|null $nsInfo
297 */
298 public function __construct(
299 array $parserConf = [], MagicWordFactory $magicWordFactory = null,
300 Language $contLang = null, ParserFactory $factory = null, $urlProtocols = null,
301 SpecialPageFactory $spFactory = null, Config $siteConfig = null,
302 LinkRendererFactory $linkRendererFactory = null,
303 NamespaceInfo $nsInfo = null
304 ) {
305 $this->mConf = $parserConf;
306 $this->mUrlProtocols = $urlProtocols ?? wfUrlProtocols();
307 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
308 self::EXT_LINK_ADDR .
309 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
310 if ( isset( $parserConf['preprocessorClass'] ) ) {
311 $this->mPreprocessorClass = $parserConf['preprocessorClass'];
312 } elseif ( wfIsHHVM() ) {
313 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
314 $this->mPreprocessorClass = Preprocessor_Hash::class;
315 } elseif ( extension_loaded( 'domxml' ) ) {
316 # PECL extension that conflicts with the core DOM extension (T15770)
317 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
318 $this->mPreprocessorClass = Preprocessor_Hash::class;
319 } elseif ( extension_loaded( 'dom' ) ) {
320 $this->mPreprocessorClass = Preprocessor_DOM::class;
321 } else {
322 $this->mPreprocessorClass = Preprocessor_Hash::class;
323 }
324 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
325
326 $services = MediaWikiServices::getInstance();
327 $this->magicWordFactory = $magicWordFactory ??
328 $services->getMagicWordFactory();
329
330 $this->contLang = $contLang ?? $services->getContentLanguage();
331
332 $this->factory = $factory ?? $services->getParserFactory();
333 $this->specialPageFactory = $spFactory ?? $services->getSpecialPageFactory();
334 $this->siteConfig = $siteConfig ?? $services->getMainConfig();
335 $this->linkRendererFactory =
336 $linkRendererFactory ?? $services->getLinkRendererFactory();
337 $this->nsInfo = $nsInfo ?? $services->getNamespaceInfo();
338 }
339
340 /**
341 * Reduce memory usage to reduce the impact of circular references
342 */
343 public function __destruct() {
344 if ( isset( $this->mLinkHolders ) ) {
345 unset( $this->mLinkHolders );
346 }
347 foreach ( $this as $name => $value ) {
348 unset( $this->$name );
349 }
350 }
351
352 /**
353 * Allow extensions to clean up when the parser is cloned
354 */
355 public function __clone() {
356 $this->mInParse = false;
357
358 // T58226: When you create a reference "to" an object field, that
359 // makes the object field itself be a reference too (until the other
360 // reference goes out of scope). When cloning, any field that's a
361 // reference is copied as a reference in the new object. Both of these
362 // are defined PHP5 behaviors, as inconvenient as it is for us when old
363 // hooks from PHP4 days are passing fields by reference.
364 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
365 // Make a non-reference copy of the field, then rebind the field to
366 // reference the new copy.
367 $tmp = $this->$k;
368 $this->$k =& $tmp;
369 unset( $tmp );
370 }
371
372 Hooks::run( 'ParserCloned', [ $this ] );
373 }
374
375 /**
376 * Do various kinds of initialisation on the first call of the parser
377 */
378 public function firstCallInit() {
379 if ( !$this->mFirstCall ) {
380 return;
381 }
382 $this->mFirstCall = false;
383
384 CoreParserFunctions::register( $this );
385 CoreTagHooks::register( $this );
386 $this->initialiseVariables();
387
388 // Avoid PHP 7.1 warning from passing $this by reference
389 $parser = $this;
390 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
391 }
392
393 /**
394 * Clear Parser state
395 *
396 * @private
397 */
398 public function clearState() {
399 $this->firstCallInit();
400 $this->mOutput = new ParserOutput;
401 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
402 $this->mAutonumber = 0;
403 $this->mIncludeCount = [];
404 $this->mLinkHolders = new LinkHolderArray( $this );
405 $this->mLinkID = 0;
406 $this->mRevisionObject = $this->mRevisionTimestamp =
407 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
408 $this->mVarCache = [];
409 $this->mUser = null;
410 $this->mLangLinkLanguages = [];
411 $this->currentRevisionCache = null;
412
413 $this->mStripState = new StripState( $this );
414
415 # Clear these on every parse, T6549
416 $this->mTplRedirCache = $this->mTplDomCache = [];
417
418 $this->mShowToc = true;
419 $this->mForceTocPosition = false;
420 $this->mIncludeSizes = [
421 'post-expand' => 0,
422 'arg' => 0,
423 ];
424 $this->mPPNodeCount = 0;
425 $this->mGeneratedPPNodeCount = 0;
426 $this->mHighestExpansionDepth = 0;
427 $this->mDefaultSort = false;
428 $this->mHeadings = [];
429 $this->mDoubleUnderscores = [];
430 $this->mExpensiveFunctionCount = 0;
431
432 # Fix cloning
433 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
434 $this->mPreprocessor = null;
435 }
436
437 $this->mProfiler = new SectionProfiler();
438
439 // Avoid PHP 7.1 warning from passing $this by reference
440 $parser = $this;
441 Hooks::run( 'ParserClearState', [ &$parser ] );
442 }
443
444 /**
445 * Convert wikitext to HTML
446 * Do not call this function recursively.
447 *
448 * @param string $text Text we want to parse
449 * @param-taint $text escapes_htmlnoent
450 * @param Title $title
451 * @param ParserOptions $options
452 * @param bool $linestart
453 * @param bool $clearState
454 * @param int|null $revid Number to pass in {{REVISIONID}}
455 * @return ParserOutput A ParserOutput
456 * @return-taint escaped
457 */
458 public function parse(
459 $text, Title $title, ParserOptions $options,
460 $linestart = true, $clearState = true, $revid = null
461 ) {
462 if ( $clearState ) {
463 // We use U+007F DELETE to construct strip markers, so we have to make
464 // sure that this character does not occur in the input text.
465 $text = strtr( $text, "\x7f", "?" );
466 $magicScopeVariable = $this->lock();
467 }
468 // Strip U+0000 NULL (T159174)
469 $text = str_replace( "\000", '', $text );
470
471 $this->startParse( $title, $options, self::OT_HTML, $clearState );
472
473 $this->currentRevisionCache = null;
474 $this->mInputSize = strlen( $text );
475 if ( $this->mOptions->getEnableLimitReport() ) {
476 $this->mOutput->resetParseStartTime();
477 }
478
479 $oldRevisionId = $this->mRevisionId;
480 $oldRevisionObject = $this->mRevisionObject;
481 $oldRevisionTimestamp = $this->mRevisionTimestamp;
482 $oldRevisionUser = $this->mRevisionUser;
483 $oldRevisionSize = $this->mRevisionSize;
484 if ( $revid !== null ) {
485 $this->mRevisionId = $revid;
486 $this->mRevisionObject = null;
487 $this->mRevisionTimestamp = null;
488 $this->mRevisionUser = null;
489 $this->mRevisionSize = null;
490 }
491
492 // Avoid PHP 7.1 warning from passing $this by reference
493 $parser = $this;
494 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
495 # No more strip!
496 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
497 $text = $this->internalParse( $text );
498 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
499
500 $text = $this->internalParseHalfParsed( $text, true, $linestart );
501
502 /**
503 * A converted title will be provided in the output object if title and
504 * content conversion are enabled, the article text does not contain
505 * a conversion-suppressing double-underscore tag, and no
506 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
507 * automatic link conversion.
508 */
509 if ( !( $options->getDisableTitleConversion()
510 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
511 || isset( $this->mDoubleUnderscores['notitleconvert'] )
512 || $this->mOutput->getDisplayTitle() !== false )
513 ) {
514 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
515 if ( $convruletitle ) {
516 $this->mOutput->setTitleText( $convruletitle );
517 } else {
518 $titleText = $this->getTargetLanguage()->convertTitle( $title );
519 $this->mOutput->setTitleText( $titleText );
520 }
521 }
522
523 # Compute runtime adaptive expiry if set
524 $this->mOutput->finalizeAdaptiveCacheExpiry();
525
526 # Warn if too many heavyweight parser functions were used
527 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
528 $this->limitationWarn( 'expensive-parserfunction',
529 $this->mExpensiveFunctionCount,
530 $this->mOptions->getExpensiveParserFunctionLimit()
531 );
532 }
533
534 # Information on limits, for the benefit of users who try to skirt them
535 if ( $this->mOptions->getEnableLimitReport() ) {
536 $text .= $this->makeLimitReport();
537 }
538
539 # Wrap non-interface parser output in a <div> so it can be targeted
540 # with CSS (T37247)
541 $class = $this->mOptions->getWrapOutputClass();
542 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
543 $this->mOutput->addWrapperDivClass( $class );
544 }
545
546 $this->mOutput->setText( $text );
547
548 $this->mRevisionId = $oldRevisionId;
549 $this->mRevisionObject = $oldRevisionObject;
550 $this->mRevisionTimestamp = $oldRevisionTimestamp;
551 $this->mRevisionUser = $oldRevisionUser;
552 $this->mRevisionSize = $oldRevisionSize;
553 $this->mInputSize = false;
554 $this->currentRevisionCache = null;
555
556 return $this->mOutput;
557 }
558
559 /**
560 * Set the limit report data in the current ParserOutput, and return the
561 * limit report HTML comment.
562 *
563 * @return string
564 */
565 protected function makeLimitReport() {
566 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
567
568 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
569 if ( $cpuTime !== null ) {
570 $this->mOutput->setLimitReportData( 'limitreport-cputime',
571 sprintf( "%.3f", $cpuTime )
572 );
573 }
574
575 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
576 $this->mOutput->setLimitReportData( 'limitreport-walltime',
577 sprintf( "%.3f", $wallTime )
578 );
579
580 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
581 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
582 );
583 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
584 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
585 );
586 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
587 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
588 );
589 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
590 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
591 );
592 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
593 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
594 );
595 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
596 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
597 );
598
599 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
600 $this->mOutput->setLimitReportData( $key, $value );
601 }
602
603 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
604
605 $limitReport = "NewPP limit report\n";
606 if ( $this->siteConfig->get( 'ShowHostnames' ) ) {
607 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
608 }
609 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
610 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
611 $limitReport .= 'Dynamic content: ' .
612 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
613 "\n";
614
615 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
616 if ( Hooks::run( 'ParserLimitReportFormat',
617 [ $key, &$value, &$limitReport, false, false ]
618 ) ) {
619 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
620 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
621 ->inLanguage( 'en' )->useDatabase( false );
622 if ( !$valueMsg->exists() ) {
623 $valueMsg = new RawMessage( '$1' );
624 }
625 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
626 $valueMsg->params( $value );
627 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
628 }
629 }
630 }
631 // Since we're not really outputting HTML, decode the entities and
632 // then re-encode the things that need hiding inside HTML comments.
633 $limitReport = htmlspecialchars_decode( $limitReport );
634
635 // Sanitize for comment. Note '‐' in the replacement is U+2010,
636 // which looks much like the problematic '-'.
637 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
638 $text = "\n<!-- \n$limitReport-->\n";
639
640 // Add on template profiling data in human/machine readable way
641 $dataByFunc = $this->mProfiler->getFunctionStats();
642 uasort( $dataByFunc, function ( $a, $b ) {
643 return $b['real'] <=> $a['real']; // descending order
644 } );
645 $profileReport = [];
646 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
647 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
648 $item['%real'], $item['real'], $item['calls'],
649 htmlspecialchars( $item['name'] ) );
650 }
651 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
652 $text .= implode( "\n", $profileReport ) . "\n-->\n";
653
654 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
655
656 // Add other cache related metadata
657 if ( $this->siteConfig->get( 'ShowHostnames' ) ) {
658 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
659 }
660 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
661 $this->mOutput->getCacheTime() );
662 $this->mOutput->setLimitReportData( 'cachereport-ttl',
663 $this->mOutput->getCacheExpiry() );
664 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
665 $this->mOutput->hasDynamicContent() );
666
667 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
668 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
669 $this->mTitle->getPrefixedDBkey() );
670 }
671 return $text;
672 }
673
674 /**
675 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
676 * can be called from an extension tag hook.
677 *
678 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
679 * instead, which means that lists and links have not been fully parsed yet,
680 * and strip markers are still present.
681 *
682 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
683 *
684 * Use this function if you're a parser tag hook and you want to parse
685 * wikitext before or after applying additional transformations, and you
686 * intend to *return the result as hook output*, which will cause it to go
687 * through the rest of parsing process automatically.
688 *
689 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
690 * $text are not expanded
691 *
692 * @param string $text Text extension wants to have parsed
693 * @param-taint $text escapes_htmlnoent
694 * @param bool|PPFrame $frame The frame to use for expanding any template variables
695 * @return string UNSAFE half-parsed HTML
696 * @return-taint escaped
697 */
698 public function recursiveTagParse( $text, $frame = false ) {
699 // Avoid PHP 7.1 warning from passing $this by reference
700 $parser = $this;
701 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
702 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
703 $text = $this->internalParse( $text, false, $frame );
704 return $text;
705 }
706
707 /**
708 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
709 * point can be called from an extension tag hook.
710 *
711 * The output of this function is fully-parsed HTML that is safe for output.
712 * If you're a parser tag hook, you might want to use recursiveTagParse()
713 * instead.
714 *
715 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
716 * $text are not expanded
717 *
718 * @since 1.25
719 *
720 * @param string $text Text extension wants to have parsed
721 * @param-taint $text escapes_htmlnoent
722 * @param bool|PPFrame $frame The frame to use for expanding any template variables
723 * @return string Fully parsed HTML
724 * @return-taint escaped
725 */
726 public function recursiveTagParseFully( $text, $frame = false ) {
727 $text = $this->recursiveTagParse( $text, $frame );
728 $text = $this->internalParseHalfParsed( $text, false );
729 return $text;
730 }
731
732 /**
733 * Expand templates and variables in the text, producing valid, static wikitext.
734 * Also removes comments.
735 * Do not call this function recursively.
736 * @param string $text
737 * @param Title|null $title
738 * @param ParserOptions $options
739 * @param int|null $revid
740 * @param bool|PPFrame $frame
741 * @return mixed|string
742 */
743 public function preprocess( $text, Title $title = null,
744 ParserOptions $options, $revid = null, $frame = false
745 ) {
746 $magicScopeVariable = $this->lock();
747 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
748 if ( $revid !== null ) {
749 $this->mRevisionId = $revid;
750 }
751 // Avoid PHP 7.1 warning from passing $this by reference
752 $parser = $this;
753 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
754 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
755 $text = $this->replaceVariables( $text, $frame );
756 $text = $this->mStripState->unstripBoth( $text );
757 return $text;
758 }
759
760 /**
761 * Recursive parser entry point that can be called from an extension tag
762 * hook.
763 *
764 * @param string $text Text to be expanded
765 * @param bool|PPFrame $frame The frame to use for expanding any template variables
766 * @return string
767 * @since 1.19
768 */
769 public function recursivePreprocess( $text, $frame = false ) {
770 $text = $this->replaceVariables( $text, $frame );
771 $text = $this->mStripState->unstripBoth( $text );
772 return $text;
773 }
774
775 /**
776 * Process the wikitext for the "?preload=" feature. (T7210)
777 *
778 * "<noinclude>", "<includeonly>" etc. are parsed as for template
779 * transclusion, comments, templates, arguments, tags hooks and parser
780 * functions are untouched.
781 *
782 * @param string $text
783 * @param Title $title
784 * @param ParserOptions $options
785 * @param array $params
786 * @return string
787 */
788 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
789 $msg = new RawMessage( $text );
790 $text = $msg->params( $params )->plain();
791
792 # Parser (re)initialisation
793 $magicScopeVariable = $this->lock();
794 $this->startParse( $title, $options, self::OT_PLAIN, true );
795
796 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
797 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
798 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
799 $text = $this->mStripState->unstripBoth( $text );
800 return $text;
801 }
802
803 /**
804 * Set the current user.
805 * Should only be used when doing pre-save transform.
806 *
807 * @param User|null $user User object or null (to reset)
808 */
809 public function setUser( $user ) {
810 $this->mUser = $user;
811 }
812
813 /**
814 * Set the context title
815 *
816 * @param Title $t
817 */
818 public function setTitle( $t ) {
819 if ( !$t ) {
820 $t = Title::newFromText( 'NO TITLE' );
821 }
822
823 if ( $t->hasFragment() ) {
824 # Strip the fragment to avoid various odd effects
825 $this->mTitle = $t->createFragmentTarget( '' );
826 } else {
827 $this->mTitle = $t;
828 }
829 }
830
831 /**
832 * Accessor for the Title object
833 *
834 * @return Title
835 */
836 public function getTitle() {
837 return $this->mTitle;
838 }
839
840 /**
841 * Accessor/mutator for the Title object
842 *
843 * @param Title|null $x Title object or null to just get the current one
844 * @return Title
845 */
846 public function Title( $x = null ) {
847 return wfSetVar( $this->mTitle, $x );
848 }
849
850 /**
851 * Set the output type
852 *
853 * @param int $ot New value
854 */
855 public function setOutputType( $ot ) {
856 $this->mOutputType = $ot;
857 # Shortcut alias
858 $this->ot = [
859 'html' => $ot == self::OT_HTML,
860 'wiki' => $ot == self::OT_WIKI,
861 'pre' => $ot == self::OT_PREPROCESS,
862 'plain' => $ot == self::OT_PLAIN,
863 ];
864 }
865
866 /**
867 * Accessor/mutator for the output type
868 *
869 * @param int|null $x New value or null to just get the current one
870 * @return int
871 */
872 public function OutputType( $x = null ) {
873 return wfSetVar( $this->mOutputType, $x );
874 }
875
876 /**
877 * Get the ParserOutput object
878 *
879 * @return ParserOutput
880 */
881 public function getOutput() {
882 return $this->mOutput;
883 }
884
885 /**
886 * Get the ParserOptions object
887 *
888 * @return ParserOptions
889 */
890 public function getOptions() {
891 return $this->mOptions;
892 }
893
894 /**
895 * Accessor/mutator for the ParserOptions object
896 *
897 * @param ParserOptions|null $x New value or null to just get the current one
898 * @return ParserOptions Current ParserOptions object
899 */
900 public function Options( $x = null ) {
901 return wfSetVar( $this->mOptions, $x );
902 }
903
904 /**
905 * @return int
906 */
907 public function nextLinkID() {
908 return $this->mLinkID++;
909 }
910
911 /**
912 * @param int $id
913 */
914 public function setLinkID( $id ) {
915 $this->mLinkID = $id;
916 }
917
918 /**
919 * Get a language object for use in parser functions such as {{FORMATNUM:}}
920 * @return Language
921 */
922 public function getFunctionLang() {
923 return $this->getTargetLanguage();
924 }
925
926 /**
927 * Get the target language for the content being parsed. This is usually the
928 * language that the content is in.
929 *
930 * @since 1.19
931 *
932 * @throws MWException
933 * @return Language
934 */
935 public function getTargetLanguage() {
936 $target = $this->mOptions->getTargetLanguage();
937
938 if ( $target !== null ) {
939 return $target;
940 } elseif ( $this->mOptions->getInterfaceMessage() ) {
941 return $this->mOptions->getUserLangObj();
942 } elseif ( is_null( $this->mTitle ) ) {
943 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
944 }
945
946 return $this->mTitle->getPageLanguage();
947 }
948
949 /**
950 * Get the language object for language conversion
951 * @deprecated since 1.32, just use getTargetLanguage()
952 * @return Language|null
953 */
954 public function getConverterLanguage() {
955 return $this->getTargetLanguage();
956 }
957
958 /**
959 * Get a User object either from $this->mUser, if set, or from the
960 * ParserOptions object otherwise
961 *
962 * @return User
963 */
964 public function getUser() {
965 if ( !is_null( $this->mUser ) ) {
966 return $this->mUser;
967 }
968 return $this->mOptions->getUser();
969 }
970
971 /**
972 * Get a preprocessor object
973 *
974 * @return Preprocessor
975 */
976 public function getPreprocessor() {
977 if ( !isset( $this->mPreprocessor ) ) {
978 $class = $this->mPreprocessorClass;
979 $this->mPreprocessor = new $class( $this );
980 }
981 return $this->mPreprocessor;
982 }
983
984 /**
985 * Get a LinkRenderer instance to make links with
986 *
987 * @since 1.28
988 * @return LinkRenderer
989 */
990 public function getLinkRenderer() {
991 // XXX We make the LinkRenderer with current options and then cache it forever
992 if ( !$this->mLinkRenderer ) {
993 $this->mLinkRenderer = $this->linkRendererFactory->create();
994 $this->mLinkRenderer->setStubThreshold(
995 $this->getOptions()->getStubThreshold()
996 );
997 }
998
999 return $this->mLinkRenderer;
1000 }
1001
1002 /**
1003 * Get the MagicWordFactory that this Parser is using
1004 *
1005 * @since 1.32
1006 * @return MagicWordFactory
1007 */
1008 public function getMagicWordFactory() {
1009 return $this->magicWordFactory;
1010 }
1011
1012 /**
1013 * Get the content language that this Parser is using
1014 *
1015 * @since 1.32
1016 * @return Language
1017 */
1018 public function getContentLanguage() {
1019 return $this->contLang;
1020 }
1021
1022 /**
1023 * Replaces all occurrences of HTML-style comments and the given tags
1024 * in the text with a random marker and returns the next text. The output
1025 * parameter $matches will be an associative array filled with data in
1026 * the form:
1027 *
1028 * @code
1029 * 'UNIQ-xxxxx' => [
1030 * 'element',
1031 * 'tag content',
1032 * [ 'param' => 'x' ],
1033 * '<element param="x">tag content</element>' ]
1034 * @endcode
1035 *
1036 * @param array $elements List of element names. Comments are always extracted.
1037 * @param string $text Source text string.
1038 * @param array &$matches Out parameter, Array: extracted tags
1039 * @return string Stripped text
1040 */
1041 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1042 static $n = 1;
1043 $stripped = '';
1044 $matches = [];
1045
1046 $taglist = implode( '|', $elements );
1047 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1048
1049 while ( $text != '' ) {
1050 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
1051 $stripped .= $p[0];
1052 if ( count( $p ) < 5 ) {
1053 break;
1054 }
1055 if ( count( $p ) > 5 ) {
1056 # comment
1057 $element = $p[4];
1058 $attributes = '';
1059 $close = '';
1060 $inside = $p[5];
1061 } else {
1062 # tag
1063 list( , $element, $attributes, $close, $inside ) = $p;
1064 }
1065
1066 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1067 $stripped .= $marker;
1068
1069 if ( $close === '/>' ) {
1070 # Empty element tag, <tag />
1071 $content = null;
1072 $text = $inside;
1073 $tail = null;
1074 } else {
1075 if ( $element === '!--' ) {
1076 $end = '/(-->)/';
1077 } else {
1078 $end = "/(<\\/$element\\s*>)/i";
1079 }
1080 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1081 $content = $q[0];
1082 if ( count( $q ) < 3 ) {
1083 # No end tag -- let it run out to the end of the text.
1084 $tail = '';
1085 $text = '';
1086 } else {
1087 list( , $tail, $text ) = $q;
1088 }
1089 }
1090
1091 $matches[$marker] = [ $element,
1092 $content,
1093 Sanitizer::decodeTagAttributes( $attributes ),
1094 "<$element$attributes$close$content$tail" ];
1095 }
1096 return $stripped;
1097 }
1098
1099 /**
1100 * Get a list of strippable XML-like elements
1101 *
1102 * @return array
1103 */
1104 public function getStripList() {
1105 return $this->mStripList;
1106 }
1107
1108 /**
1109 * Add an item to the strip state
1110 * Returns the unique tag which must be inserted into the stripped text
1111 * The tag will be replaced with the original text in unstrip()
1112 *
1113 * @param string $text
1114 *
1115 * @return string
1116 */
1117 public function insertStripItem( $text ) {
1118 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1119 $this->mMarkerIndex++;
1120 $this->mStripState->addGeneral( $marker, $text );
1121 return $marker;
1122 }
1123
1124 /**
1125 * parse the wiki syntax used to render tables
1126 *
1127 * @private
1128 * @param string $text
1129 * @return string
1130 */
1131 public function doTableStuff( $text ) {
1132 $lines = StringUtils::explode( "\n", $text );
1133 $out = '';
1134 $td_history = []; # Is currently a td tag open?
1135 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1136 $tr_history = []; # Is currently a tr tag open?
1137 $tr_attributes = []; # history of tr attributes
1138 $has_opened_tr = []; # Did this table open a <tr> element?
1139 $indent_level = 0; # indent level of the table
1140
1141 foreach ( $lines as $outLine ) {
1142 $line = trim( $outLine );
1143
1144 if ( $line === '' ) { # empty line, go to next line
1145 $out .= $outLine . "\n";
1146 continue;
1147 }
1148
1149 $first_character = $line[0];
1150 $first_two = substr( $line, 0, 2 );
1151 $matches = [];
1152
1153 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1154 # First check if we are starting a new table
1155 $indent_level = strlen( $matches[1] );
1156
1157 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1158 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1159
1160 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1161 array_push( $td_history, false );
1162 array_push( $last_tag_history, '' );
1163 array_push( $tr_history, false );
1164 array_push( $tr_attributes, '' );
1165 array_push( $has_opened_tr, false );
1166 } elseif ( count( $td_history ) == 0 ) {
1167 # Don't do any of the following
1168 $out .= $outLine . "\n";
1169 continue;
1170 } elseif ( $first_two === '|}' ) {
1171 # We are ending a table
1172 $line = '</table>' . substr( $line, 2 );
1173 $last_tag = array_pop( $last_tag_history );
1174
1175 if ( !array_pop( $has_opened_tr ) ) {
1176 $line = "<tr><td></td></tr>{$line}";
1177 }
1178
1179 if ( array_pop( $tr_history ) ) {
1180 $line = "</tr>{$line}";
1181 }
1182
1183 if ( array_pop( $td_history ) ) {
1184 $line = "</{$last_tag}>{$line}";
1185 }
1186 array_pop( $tr_attributes );
1187 if ( $indent_level > 0 ) {
1188 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1189 } else {
1190 $outLine = $line;
1191 }
1192 } elseif ( $first_two === '|-' ) {
1193 # Now we have a table row
1194 $line = preg_replace( '#^\|-+#', '', $line );
1195
1196 # Whats after the tag is now only attributes
1197 $attributes = $this->mStripState->unstripBoth( $line );
1198 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1199 array_pop( $tr_attributes );
1200 array_push( $tr_attributes, $attributes );
1201
1202 $line = '';
1203 $last_tag = array_pop( $last_tag_history );
1204 array_pop( $has_opened_tr );
1205 array_push( $has_opened_tr, true );
1206
1207 if ( array_pop( $tr_history ) ) {
1208 $line = '</tr>';
1209 }
1210
1211 if ( array_pop( $td_history ) ) {
1212 $line = "</{$last_tag}>{$line}";
1213 }
1214
1215 $outLine = $line;
1216 array_push( $tr_history, false );
1217 array_push( $td_history, false );
1218 array_push( $last_tag_history, '' );
1219 } elseif ( $first_character === '|'
1220 || $first_character === '!'
1221 || $first_two === '|+'
1222 ) {
1223 # This might be cell elements, td, th or captions
1224 if ( $first_two === '|+' ) {
1225 $first_character = '+';
1226 $line = substr( $line, 2 );
1227 } else {
1228 $line = substr( $line, 1 );
1229 }
1230
1231 // Implies both are valid for table headings.
1232 if ( $first_character === '!' ) {
1233 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1234 }
1235
1236 # Split up multiple cells on the same line.
1237 # FIXME : This can result in improper nesting of tags processed
1238 # by earlier parser steps.
1239 $cells = explode( '||', $line );
1240
1241 $outLine = '';
1242
1243 # Loop through each table cell
1244 foreach ( $cells as $cell ) {
1245 $previous = '';
1246 if ( $first_character !== '+' ) {
1247 $tr_after = array_pop( $tr_attributes );
1248 if ( !array_pop( $tr_history ) ) {
1249 $previous = "<tr{$tr_after}>\n";
1250 }
1251 array_push( $tr_history, true );
1252 array_push( $tr_attributes, '' );
1253 array_pop( $has_opened_tr );
1254 array_push( $has_opened_tr, true );
1255 }
1256
1257 $last_tag = array_pop( $last_tag_history );
1258
1259 if ( array_pop( $td_history ) ) {
1260 $previous = "</{$last_tag}>\n{$previous}";
1261 }
1262
1263 if ( $first_character === '|' ) {
1264 $last_tag = 'td';
1265 } elseif ( $first_character === '!' ) {
1266 $last_tag = 'th';
1267 } elseif ( $first_character === '+' ) {
1268 $last_tag = 'caption';
1269 } else {
1270 $last_tag = '';
1271 }
1272
1273 array_push( $last_tag_history, $last_tag );
1274
1275 # A cell could contain both parameters and data
1276 $cell_data = explode( '|', $cell, 2 );
1277
1278 # T2553: Note that a '|' inside an invalid link should not
1279 # be mistaken as delimiting cell parameters
1280 # Bug T153140: Neither should language converter markup.
1281 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1282 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1283 } elseif ( count( $cell_data ) == 1 ) {
1284 // Whitespace in cells is trimmed
1285 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1286 } else {
1287 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1288 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1289 // Whitespace in cells is trimmed
1290 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1291 }
1292
1293 $outLine .= $cell;
1294 array_push( $td_history, true );
1295 }
1296 }
1297 $out .= $outLine . "\n";
1298 }
1299
1300 # Closing open td, tr && table
1301 while ( count( $td_history ) > 0 ) {
1302 if ( array_pop( $td_history ) ) {
1303 $out .= "</td>\n";
1304 }
1305 if ( array_pop( $tr_history ) ) {
1306 $out .= "</tr>\n";
1307 }
1308 if ( !array_pop( $has_opened_tr ) ) {
1309 $out .= "<tr><td></td></tr>\n";
1310 }
1311
1312 $out .= "</table>\n";
1313 }
1314
1315 # Remove trailing line-ending (b/c)
1316 if ( substr( $out, -1 ) === "\n" ) {
1317 $out = substr( $out, 0, -1 );
1318 }
1319
1320 # special case: don't return empty table
1321 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1322 $out = '';
1323 }
1324
1325 return $out;
1326 }
1327
1328 /**
1329 * Helper function for parse() that transforms wiki markup into half-parsed
1330 * HTML. Only called for $mOutputType == self::OT_HTML.
1331 *
1332 * @private
1333 *
1334 * @param string $text The text to parse
1335 * @param-taint $text escapes_html
1336 * @param bool $isMain Whether this is being called from the main parse() function
1337 * @param PPFrame|bool $frame A pre-processor frame
1338 *
1339 * @return string
1340 */
1341 public function internalParse( $text, $isMain = true, $frame = false ) {
1342 $origText = $text;
1343
1344 // Avoid PHP 7.1 warning from passing $this by reference
1345 $parser = $this;
1346
1347 # Hook to suspend the parser in this state
1348 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1349 return $text;
1350 }
1351
1352 # if $frame is provided, then use $frame for replacing any variables
1353 if ( $frame ) {
1354 # use frame depth to infer how include/noinclude tags should be handled
1355 # depth=0 means this is the top-level document; otherwise it's an included document
1356 if ( !$frame->depth ) {
1357 $flag = 0;
1358 } else {
1359 $flag = self::PTD_FOR_INCLUSION;
1360 }
1361 $dom = $this->preprocessToDom( $text, $flag );
1362 $text = $frame->expand( $dom );
1363 } else {
1364 # if $frame is not provided, then use old-style replaceVariables
1365 $text = $this->replaceVariables( $text );
1366 }
1367
1368 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1369 $text = Sanitizer::removeHTMLtags(
1370 $text,
1371 [ $this, 'attributeStripCallback' ],
1372 false,
1373 array_keys( $this->mTransparentTagHooks ),
1374 [],
1375 [ $this, 'addTrackingCategory' ]
1376 );
1377 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1378
1379 # Tables need to come after variable replacement for things to work
1380 # properly; putting them before other transformations should keep
1381 # exciting things like link expansions from showing up in surprising
1382 # places.
1383 $text = $this->doTableStuff( $text );
1384
1385 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1386
1387 $text = $this->doDoubleUnderscore( $text );
1388
1389 $text = $this->doHeadings( $text );
1390 $text = $this->replaceInternalLinks( $text );
1391 $text = $this->doAllQuotes( $text );
1392 $text = $this->replaceExternalLinks( $text );
1393
1394 # replaceInternalLinks may sometimes leave behind
1395 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1396 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1397
1398 $text = $this->doMagicLinks( $text );
1399 $text = $this->formatHeadings( $text, $origText, $isMain );
1400
1401 return $text;
1402 }
1403
1404 /**
1405 * Helper function for parse() that transforms half-parsed HTML into fully
1406 * parsed HTML.
1407 *
1408 * @param string $text
1409 * @param bool $isMain
1410 * @param bool $linestart
1411 * @return string
1412 */
1413 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1414 $text = $this->mStripState->unstripGeneral( $text );
1415
1416 // Avoid PHP 7.1 warning from passing $this by reference
1417 $parser = $this;
1418
1419 if ( $isMain ) {
1420 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1421 }
1422
1423 # Clean up special characters, only run once, next-to-last before doBlockLevels
1424 $text = Sanitizer::armorFrenchSpaces( $text );
1425
1426 $text = $this->doBlockLevels( $text, $linestart );
1427
1428 $this->replaceLinkHolders( $text );
1429
1430 /**
1431 * The input doesn't get language converted if
1432 * a) It's disabled
1433 * b) Content isn't converted
1434 * c) It's a conversion table
1435 * d) it is an interface message (which is in the user language)
1436 */
1437 if ( !( $this->mOptions->getDisableContentConversion()
1438 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1439 && !$this->mOptions->getInterfaceMessage()
1440 ) {
1441 # The position of the convert() call should not be changed. it
1442 # assumes that the links are all replaced and the only thing left
1443 # is the <nowiki> mark.
1444 $text = $this->getTargetLanguage()->convert( $text );
1445 }
1446
1447 $text = $this->mStripState->unstripNoWiki( $text );
1448
1449 if ( $isMain ) {
1450 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1451 }
1452
1453 $text = $this->replaceTransparentTags( $text );
1454 $text = $this->mStripState->unstripGeneral( $text );
1455
1456 $text = Sanitizer::normalizeCharReferences( $text );
1457
1458 if ( MWTidy::isEnabled() ) {
1459 if ( $this->mOptions->getTidy() ) {
1460 $text = MWTidy::tidy( $text );
1461 }
1462 } else {
1463 # attempt to sanitize at least some nesting problems
1464 # (T4702 and quite a few others)
1465 # This code path is buggy and deprecated!
1466 wfDeprecated( 'disabling tidy', '1.33' );
1467 $tidyregs = [
1468 # ''Something [http://www.cool.com cool''] -->
1469 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1470 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1471 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1472 # fix up an anchor inside another anchor, only
1473 # at least for a single single nested link (T5695)
1474 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1475 '\\1\\2</a>\\3</a>\\1\\4</a>',
1476 # fix div inside inline elements- doBlockLevels won't wrap a line which
1477 # contains a div, so fix it up here; replace
1478 # div with escaped text
1479 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1480 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1481 # remove empty italic or bold tag pairs, some
1482 # introduced by rules above
1483 '/<([bi])><\/\\1>/' => '',
1484 ];
1485
1486 $text = preg_replace(
1487 array_keys( $tidyregs ),
1488 array_values( $tidyregs ),
1489 $text );
1490 }
1491
1492 if ( $isMain ) {
1493 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1494 }
1495
1496 return $text;
1497 }
1498
1499 /**
1500 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1501 * magic external links.
1502 *
1503 * DML
1504 * @private
1505 *
1506 * @param string $text
1507 *
1508 * @return string
1509 */
1510 public function doMagicLinks( $text ) {
1511 $prots = wfUrlProtocolsWithoutProtRel();
1512 $urlChar = self::EXT_LINK_URL_CLASS;
1513 $addr = self::EXT_LINK_ADDR;
1514 $space = self::SPACE_NOT_NL; # non-newline space
1515 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1516 $spaces = "$space++"; # possessive match of 1 or more spaces
1517 $text = preg_replace_callback(
1518 '!(?: # Start cases
1519 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1520 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1521 (\b # m[3]: Free external links
1522 (?i:$prots)
1523 ($addr$urlChar*) # m[4]: Post-protocol path
1524 ) |
1525 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1526 ([0-9]+)\b |
1527 \bISBN $spaces ( # m[6]: ISBN, capture number
1528 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1529 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1530 [0-9Xx] # check digit
1531 )\b
1532 )!xu", [ $this, 'magicLinkCallback' ], $text );
1533 return $text;
1534 }
1535
1536 /**
1537 * @throws MWException
1538 * @param array $m
1539 * @return string HTML
1540 */
1541 public function magicLinkCallback( $m ) {
1542 if ( isset( $m[1] ) && $m[1] !== '' ) {
1543 # Skip anchor
1544 return $m[0];
1545 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1546 # Skip HTML element
1547 return $m[0];
1548 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1549 # Free external link
1550 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1551 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1552 # RFC or PMID
1553 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1554 if ( !$this->mOptions->getMagicRFCLinks() ) {
1555 return $m[0];
1556 }
1557 $keyword = 'RFC';
1558 $urlmsg = 'rfcurl';
1559 $cssClass = 'mw-magiclink-rfc';
1560 $trackingCat = 'magiclink-tracking-rfc';
1561 $id = $m[5];
1562 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1563 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1564 return $m[0];
1565 }
1566 $keyword = 'PMID';
1567 $urlmsg = 'pubmedurl';
1568 $cssClass = 'mw-magiclink-pmid';
1569 $trackingCat = 'magiclink-tracking-pmid';
1570 $id = $m[5];
1571 } else {
1572 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1573 substr( $m[0], 0, 20 ) . '"' );
1574 }
1575 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1576 $this->addTrackingCategory( $trackingCat );
1577 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1578 } elseif ( isset( $m[6] ) && $m[6] !== ''
1579 && $this->mOptions->getMagicISBNLinks()
1580 ) {
1581 # ISBN
1582 $isbn = $m[6];
1583 $space = self::SPACE_NOT_NL; # non-newline space
1584 $isbn = preg_replace( "/$space/", ' ', $isbn );
1585 $num = strtr( $isbn, [
1586 '-' => '',
1587 ' ' => '',
1588 'x' => 'X',
1589 ] );
1590 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1591 return $this->getLinkRenderer()->makeKnownLink(
1592 SpecialPage::getTitleFor( 'Booksources', $num ),
1593 "ISBN $isbn",
1594 [
1595 'class' => 'internal mw-magiclink-isbn',
1596 'title' => false // suppress title attribute
1597 ]
1598 );
1599 } else {
1600 return $m[0];
1601 }
1602 }
1603
1604 /**
1605 * Make a free external link, given a user-supplied URL
1606 *
1607 * @param string $url
1608 * @param int $numPostProto
1609 * The number of characters after the protocol.
1610 * @return string HTML
1611 * @private
1612 */
1613 public function makeFreeExternalLink( $url, $numPostProto ) {
1614 $trail = '';
1615
1616 # The characters '<' and '>' (which were escaped by
1617 # removeHTMLtags()) should not be included in
1618 # URLs, per RFC 2396.
1619 # Make &nbsp; terminate a URL as well (bug T84937)
1620 $m2 = [];
1621 if ( preg_match(
1622 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1623 $url,
1624 $m2,
1625 PREG_OFFSET_CAPTURE
1626 ) ) {
1627 $trail = substr( $url, $m2[0][1] ) . $trail;
1628 $url = substr( $url, 0, $m2[0][1] );
1629 }
1630
1631 # Move trailing punctuation to $trail
1632 $sep = ',;\.:!?';
1633 # If there is no left bracket, then consider right brackets fair game too
1634 if ( strpos( $url, '(' ) === false ) {
1635 $sep .= ')';
1636 }
1637
1638 $urlRev = strrev( $url );
1639 $numSepChars = strspn( $urlRev, $sep );
1640 # Don't break a trailing HTML entity by moving the ; into $trail
1641 # This is in hot code, so use substr_compare to avoid having to
1642 # create a new string object for the comparison
1643 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1644 # more optimization: instead of running preg_match with a $
1645 # anchor, which can be slow, do the match on the reversed
1646 # string starting at the desired offset.
1647 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1648 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1649 $numSepChars--;
1650 }
1651 }
1652 if ( $numSepChars ) {
1653 $trail = substr( $url, -$numSepChars ) . $trail;
1654 $url = substr( $url, 0, -$numSepChars );
1655 }
1656
1657 # Verify that we still have a real URL after trail removal, and
1658 # not just lone protocol
1659 if ( strlen( $trail ) >= $numPostProto ) {
1660 return $url . $trail;
1661 }
1662
1663 $url = Sanitizer::cleanUrl( $url );
1664
1665 # Is this an external image?
1666 $text = $this->maybeMakeExternalImage( $url );
1667 if ( $text === false ) {
1668 # Not an image, make a link
1669 $text = Linker::makeExternalLink( $url,
1670 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1671 true, 'free',
1672 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1673 # Register it in the output object...
1674 $this->mOutput->addExternalLink( $url );
1675 }
1676 return $text . $trail;
1677 }
1678
1679 /**
1680 * Parse headers and return html
1681 *
1682 * @private
1683 *
1684 * @param string $text
1685 *
1686 * @return string
1687 */
1688 public function doHeadings( $text ) {
1689 for ( $i = 6; $i >= 1; --$i ) {
1690 $h = str_repeat( '=', $i );
1691 // Trim non-newline whitespace from headings
1692 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1693 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1694 }
1695 return $text;
1696 }
1697
1698 /**
1699 * Replace single quotes with HTML markup
1700 * @private
1701 *
1702 * @param string $text
1703 *
1704 * @return string The altered text
1705 */
1706 public function doAllQuotes( $text ) {
1707 $outtext = '';
1708 $lines = StringUtils::explode( "\n", $text );
1709 foreach ( $lines as $line ) {
1710 $outtext .= $this->doQuotes( $line ) . "\n";
1711 }
1712 $outtext = substr( $outtext, 0, -1 );
1713 return $outtext;
1714 }
1715
1716 /**
1717 * Helper function for doAllQuotes()
1718 *
1719 * @param string $text
1720 *
1721 * @return string
1722 */
1723 public function doQuotes( $text ) {
1724 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1725 $countarr = count( $arr );
1726 if ( $countarr == 1 ) {
1727 return $text;
1728 }
1729
1730 // First, do some preliminary work. This may shift some apostrophes from
1731 // being mark-up to being text. It also counts the number of occurrences
1732 // of bold and italics mark-ups.
1733 $numbold = 0;
1734 $numitalics = 0;
1735 for ( $i = 1; $i < $countarr; $i += 2 ) {
1736 $thislen = strlen( $arr[$i] );
1737 // If there are ever four apostrophes, assume the first is supposed to
1738 // be text, and the remaining three constitute mark-up for bold text.
1739 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1740 if ( $thislen == 4 ) {
1741 $arr[$i - 1] .= "'";
1742 $arr[$i] = "'''";
1743 $thislen = 3;
1744 } elseif ( $thislen > 5 ) {
1745 // If there are more than 5 apostrophes in a row, assume they're all
1746 // text except for the last 5.
1747 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1748 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1749 $arr[$i] = "'''''";
1750 $thislen = 5;
1751 }
1752 // Count the number of occurrences of bold and italics mark-ups.
1753 if ( $thislen == 2 ) {
1754 $numitalics++;
1755 } elseif ( $thislen == 3 ) {
1756 $numbold++;
1757 } elseif ( $thislen == 5 ) {
1758 $numitalics++;
1759 $numbold++;
1760 }
1761 }
1762
1763 // If there is an odd number of both bold and italics, it is likely
1764 // that one of the bold ones was meant to be an apostrophe followed
1765 // by italics. Which one we cannot know for certain, but it is more
1766 // likely to be one that has a single-letter word before it.
1767 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1768 $firstsingleletterword = -1;
1769 $firstmultiletterword = -1;
1770 $firstspace = -1;
1771 for ( $i = 1; $i < $countarr; $i += 2 ) {
1772 if ( strlen( $arr[$i] ) == 3 ) {
1773 $x1 = substr( $arr[$i - 1], -1 );
1774 $x2 = substr( $arr[$i - 1], -2, 1 );
1775 if ( $x1 === ' ' ) {
1776 if ( $firstspace == -1 ) {
1777 $firstspace = $i;
1778 }
1779 } elseif ( $x2 === ' ' ) {
1780 $firstsingleletterword = $i;
1781 // if $firstsingleletterword is set, we don't
1782 // look at the other options, so we can bail early.
1783 break;
1784 } elseif ( $firstmultiletterword == -1 ) {
1785 $firstmultiletterword = $i;
1786 }
1787 }
1788 }
1789
1790 // If there is a single-letter word, use it!
1791 if ( $firstsingleletterword > -1 ) {
1792 $arr[$firstsingleletterword] = "''";
1793 $arr[$firstsingleletterword - 1] .= "'";
1794 } elseif ( $firstmultiletterword > -1 ) {
1795 // If not, but there's a multi-letter word, use that one.
1796 $arr[$firstmultiletterword] = "''";
1797 $arr[$firstmultiletterword - 1] .= "'";
1798 } elseif ( $firstspace > -1 ) {
1799 // ... otherwise use the first one that has neither.
1800 // (notice that it is possible for all three to be -1 if, for example,
1801 // there is only one pentuple-apostrophe in the line)
1802 $arr[$firstspace] = "''";
1803 $arr[$firstspace - 1] .= "'";
1804 }
1805 }
1806
1807 // Now let's actually convert our apostrophic mush to HTML!
1808 $output = '';
1809 $buffer = '';
1810 $state = '';
1811 $i = 0;
1812 foreach ( $arr as $r ) {
1813 if ( ( $i % 2 ) == 0 ) {
1814 if ( $state === 'both' ) {
1815 $buffer .= $r;
1816 } else {
1817 $output .= $r;
1818 }
1819 } else {
1820 $thislen = strlen( $r );
1821 if ( $thislen == 2 ) {
1822 if ( $state === 'i' ) {
1823 $output .= '</i>';
1824 $state = '';
1825 } elseif ( $state === 'bi' ) {
1826 $output .= '</i>';
1827 $state = 'b';
1828 } elseif ( $state === 'ib' ) {
1829 $output .= '</b></i><b>';
1830 $state = 'b';
1831 } elseif ( $state === 'both' ) {
1832 $output .= '<b><i>' . $buffer . '</i>';
1833 $state = 'b';
1834 } else { // $state can be 'b' or ''
1835 $output .= '<i>';
1836 $state .= 'i';
1837 }
1838 } elseif ( $thislen == 3 ) {
1839 if ( $state === 'b' ) {
1840 $output .= '</b>';
1841 $state = '';
1842 } elseif ( $state === 'bi' ) {
1843 $output .= '</i></b><i>';
1844 $state = 'i';
1845 } elseif ( $state === 'ib' ) {
1846 $output .= '</b>';
1847 $state = 'i';
1848 } elseif ( $state === 'both' ) {
1849 $output .= '<i><b>' . $buffer . '</b>';
1850 $state = 'i';
1851 } else { // $state can be 'i' or ''
1852 $output .= '<b>';
1853 $state .= 'b';
1854 }
1855 } elseif ( $thislen == 5 ) {
1856 if ( $state === 'b' ) {
1857 $output .= '</b><i>';
1858 $state = 'i';
1859 } elseif ( $state === 'i' ) {
1860 $output .= '</i><b>';
1861 $state = 'b';
1862 } elseif ( $state === 'bi' ) {
1863 $output .= '</i></b>';
1864 $state = '';
1865 } elseif ( $state === 'ib' ) {
1866 $output .= '</b></i>';
1867 $state = '';
1868 } elseif ( $state === 'both' ) {
1869 $output .= '<i><b>' . $buffer . '</b></i>';
1870 $state = '';
1871 } else { // ($state == '')
1872 $buffer = '';
1873 $state = 'both';
1874 }
1875 }
1876 }
1877 $i++;
1878 }
1879 // Now close all remaining tags. Notice that the order is important.
1880 if ( $state === 'b' || $state === 'ib' ) {
1881 $output .= '</b>';
1882 }
1883 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1884 $output .= '</i>';
1885 }
1886 if ( $state === 'bi' ) {
1887 $output .= '</b>';
1888 }
1889 // There might be lonely ''''', so make sure we have a buffer
1890 if ( $state === 'both' && $buffer ) {
1891 $output .= '<b><i>' . $buffer . '</i></b>';
1892 }
1893 return $output;
1894 }
1895
1896 /**
1897 * Replace external links (REL)
1898 *
1899 * Note: this is all very hackish and the order of execution matters a lot.
1900 * Make sure to run tests/parser/parserTests.php if you change this code.
1901 *
1902 * @private
1903 *
1904 * @param string $text
1905 *
1906 * @throws MWException
1907 * @return string
1908 */
1909 public function replaceExternalLinks( $text ) {
1910 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1911 if ( $bits === false ) {
1912 throw new MWException( "PCRE needs to be compiled with "
1913 . "--enable-unicode-properties in order for MediaWiki to function" );
1914 }
1915 $s = array_shift( $bits );
1916
1917 $i = 0;
1918 while ( $i < count( $bits ) ) {
1919 $url = $bits[$i++];
1920 $i++; // protocol
1921 $text = $bits[$i++];
1922 $trail = $bits[$i++];
1923
1924 # The characters '<' and '>' (which were escaped by
1925 # removeHTMLtags()) should not be included in
1926 # URLs, per RFC 2396.
1927 $m2 = [];
1928 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1929 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1930 $url = substr( $url, 0, $m2[0][1] );
1931 }
1932
1933 # If the link text is an image URL, replace it with an <img> tag
1934 # This happened by accident in the original parser, but some people used it extensively
1935 $img = $this->maybeMakeExternalImage( $text );
1936 if ( $img !== false ) {
1937 $text = $img;
1938 }
1939
1940 $dtrail = '';
1941
1942 # Set linktype for CSS
1943 $linktype = 'text';
1944
1945 # No link text, e.g. [http://domain.tld/some.link]
1946 if ( $text == '' ) {
1947 # Autonumber
1948 $langObj = $this->getTargetLanguage();
1949 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1950 $linktype = 'autonumber';
1951 } else {
1952 # Have link text, e.g. [http://domain.tld/some.link text]s
1953 # Check for trail
1954 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1955 }
1956
1957 // Excluding protocol-relative URLs may avoid many false positives.
1958 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1959 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1960 }
1961
1962 $url = Sanitizer::cleanUrl( $url );
1963
1964 # Use the encoded URL
1965 # This means that users can paste URLs directly into the text
1966 # Funny characters like ö aren't valid in URLs anyway
1967 # This was changed in August 2004
1968 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1969 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1970
1971 # Register link in the output object.
1972 $this->mOutput->addExternalLink( $url );
1973 }
1974
1975 return $s;
1976 }
1977
1978 /**
1979 * Get the rel attribute for a particular external link.
1980 *
1981 * @since 1.21
1982 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1983 * nofollow if appropriate
1984 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
1985 * @return string|null Rel attribute for $url
1986 */
1987 public static function getExternalLinkRel( $url = false, $title = null ) {
1988 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1989 $ns = $title ? $title->getNamespace() : false;
1990 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1991 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1992 ) {
1993 return 'nofollow';
1994 }
1995 return null;
1996 }
1997
1998 /**
1999 * Get an associative array of additional HTML attributes appropriate for a
2000 * particular external link. This currently may include rel => nofollow
2001 * (depending on configuration, namespace, and the URL's domain) and/or a
2002 * target attribute (depending on configuration).
2003 *
2004 * @param string $url URL to extract the domain from for rel =>
2005 * nofollow if appropriate
2006 * @return array Associative array of HTML attributes
2007 */
2008 public function getExternalLinkAttribs( $url ) {
2009 $attribs = [];
2010 $rel = self::getExternalLinkRel( $url, $this->mTitle );
2011
2012 $target = $this->mOptions->getExternalLinkTarget();
2013 if ( $target ) {
2014 $attribs['target'] = $target;
2015 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2016 // T133507. New windows can navigate parent cross-origin.
2017 // Including noreferrer due to lacking browser
2018 // support of noopener. Eventually noreferrer should be removed.
2019 if ( $rel !== '' ) {
2020 $rel .= ' ';
2021 }
2022 $rel .= 'noreferrer noopener';
2023 }
2024 }
2025 $attribs['rel'] = $rel;
2026 return $attribs;
2027 }
2028
2029 /**
2030 * Replace unusual escape codes in a URL with their equivalent characters
2031 *
2032 * This generally follows the syntax defined in RFC 3986, with special
2033 * consideration for HTTP query strings.
2034 *
2035 * @param string $url
2036 * @return string
2037 */
2038 public static function normalizeLinkUrl( $url ) {
2039 # Test for RFC 3986 IPv6 syntax
2040 $scheme = '[a-z][a-z0-9+.-]*:';
2041 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2042 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2043 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2044 IP::isValid( rawurldecode( $m[1] ) )
2045 ) {
2046 $isIPv6 = rawurldecode( $m[1] );
2047 } else {
2048 $isIPv6 = false;
2049 }
2050
2051 # Make sure unsafe characters are encoded
2052 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2053 function ( $m ) {
2054 return rawurlencode( $m[0] );
2055 },
2056 $url
2057 );
2058
2059 $ret = '';
2060 $end = strlen( $url );
2061
2062 # Fragment part - 'fragment'
2063 $start = strpos( $url, '#' );
2064 if ( $start !== false && $start < $end ) {
2065 $ret = self::normalizeUrlComponent(
2066 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2067 $end = $start;
2068 }
2069
2070 # Query part - 'query' minus &=+;
2071 $start = strpos( $url, '?' );
2072 if ( $start !== false && $start < $end ) {
2073 $ret = self::normalizeUrlComponent(
2074 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2075 $end = $start;
2076 }
2077
2078 # Scheme and path part - 'pchar'
2079 # (we assume no userinfo or encoded colons in the host)
2080 $ret = self::normalizeUrlComponent(
2081 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2082
2083 # Fix IPv6 syntax
2084 if ( $isIPv6 !== false ) {
2085 $ipv6Host = "%5B({$isIPv6})%5D";
2086 $ret = preg_replace(
2087 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2088 "$1[$2]",
2089 $ret
2090 );
2091 }
2092
2093 return $ret;
2094 }
2095
2096 private static function normalizeUrlComponent( $component, $unsafe ) {
2097 $callback = function ( $matches ) use ( $unsafe ) {
2098 $char = urldecode( $matches[0] );
2099 $ord = ord( $char );
2100 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2101 # Unescape it
2102 return $char;
2103 } else {
2104 # Leave it escaped, but use uppercase for a-f
2105 return strtoupper( $matches[0] );
2106 }
2107 };
2108 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2109 }
2110
2111 /**
2112 * make an image if it's allowed, either through the global
2113 * option, through the exception, or through the on-wiki whitelist
2114 *
2115 * @param string $url
2116 *
2117 * @return string
2118 */
2119 private function maybeMakeExternalImage( $url ) {
2120 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2121 $imagesexception = !empty( $imagesfrom );
2122 $text = false;
2123 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2124 if ( $imagesexception && is_array( $imagesfrom ) ) {
2125 $imagematch = false;
2126 foreach ( $imagesfrom as $match ) {
2127 if ( strpos( $url, $match ) === 0 ) {
2128 $imagematch = true;
2129 break;
2130 }
2131 }
2132 } elseif ( $imagesexception ) {
2133 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2134 } else {
2135 $imagematch = false;
2136 }
2137
2138 if ( $this->mOptions->getAllowExternalImages()
2139 || ( $imagesexception && $imagematch )
2140 ) {
2141 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2142 # Image found
2143 $text = Linker::makeExternalImage( $url );
2144 }
2145 }
2146 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2147 && preg_match( self::EXT_IMAGE_REGEX, $url )
2148 ) {
2149 $whitelist = explode(
2150 "\n",
2151 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2152 );
2153
2154 foreach ( $whitelist as $entry ) {
2155 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2156 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2157 continue;
2158 }
2159 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2160 # Image matches a whitelist entry
2161 $text = Linker::makeExternalImage( $url );
2162 break;
2163 }
2164 }
2165 }
2166 return $text;
2167 }
2168
2169 /**
2170 * Process [[ ]] wikilinks
2171 *
2172 * @param string $s
2173 *
2174 * @return string Processed text
2175 *
2176 * @private
2177 */
2178 public function replaceInternalLinks( $s ) {
2179 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2180 return $s;
2181 }
2182
2183 /**
2184 * Process [[ ]] wikilinks (RIL)
2185 * @param string &$s
2186 * @throws MWException
2187 * @return LinkHolderArray
2188 *
2189 * @private
2190 */
2191 public function replaceInternalLinks2( &$s ) {
2192 static $tc = false, $e1, $e1_img;
2193 # the % is needed to support urlencoded titles as well
2194 if ( !$tc ) {
2195 $tc = Title::legalChars() . '#%';
2196 # Match a link having the form [[namespace:link|alternate]]trail
2197 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2198 # Match cases where there is no "]]", which might still be images
2199 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2200 }
2201
2202 $holders = new LinkHolderArray( $this );
2203
2204 # split the entire text string on occurrences of [[
2205 $a = StringUtils::explode( '[[', ' ' . $s );
2206 # get the first element (all text up to first [[), and remove the space we added
2207 $s = $a->current();
2208 $a->next();
2209 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2210 $s = substr( $s, 1 );
2211
2212 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2213 $e2 = null;
2214 if ( $useLinkPrefixExtension ) {
2215 # Match the end of a line for a word that's not followed by whitespace,
2216 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2217 $charset = $this->contLang->linkPrefixCharset();
2218 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2219 }
2220
2221 if ( is_null( $this->mTitle ) ) {
2222 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2223 }
2224 $nottalk = !$this->mTitle->isTalkPage();
2225
2226 if ( $useLinkPrefixExtension ) {
2227 $m = [];
2228 if ( preg_match( $e2, $s, $m ) ) {
2229 $first_prefix = $m[2];
2230 } else {
2231 $first_prefix = false;
2232 }
2233 } else {
2234 $prefix = '';
2235 }
2236
2237 $useSubpages = $this->areSubpagesAllowed();
2238
2239 # Loop for each link
2240 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2241 # Check for excessive memory usage
2242 if ( $holders->isBig() ) {
2243 # Too big
2244 # Do the existence check, replace the link holders and clear the array
2245 $holders->replace( $s );
2246 $holders->clear();
2247 }
2248
2249 if ( $useLinkPrefixExtension ) {
2250 if ( preg_match( $e2, $s, $m ) ) {
2251 list( , $s, $prefix ) = $m;
2252 } else {
2253 $prefix = '';
2254 }
2255 # first link
2256 if ( $first_prefix ) {
2257 $prefix = $first_prefix;
2258 $first_prefix = false;
2259 }
2260 }
2261
2262 $might_be_img = false;
2263
2264 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2265 $text = $m[2];
2266 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2267 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2268 # the real problem is with the $e1 regex
2269 # See T1500.
2270 # Still some problems for cases where the ] is meant to be outside punctuation,
2271 # and no image is in sight. See T4095.
2272 if ( $text !== ''
2273 && substr( $m[3], 0, 1 ) === ']'
2274 && strpos( $text, '[' ) !== false
2275 ) {
2276 $text .= ']'; # so that replaceExternalLinks($text) works later
2277 $m[3] = substr( $m[3], 1 );
2278 }
2279 # fix up urlencoded title texts
2280 if ( strpos( $m[1], '%' ) !== false ) {
2281 # Should anchors '#' also be rejected?
2282 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2283 }
2284 $trail = $m[3];
2285 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2286 # Invalid, but might be an image with a link in its caption
2287 $might_be_img = true;
2288 $text = $m[2];
2289 if ( strpos( $m[1], '%' ) !== false ) {
2290 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2291 }
2292 $trail = "";
2293 } else { # Invalid form; output directly
2294 $s .= $prefix . '[[' . $line;
2295 continue;
2296 }
2297
2298 $origLink = ltrim( $m[1], ' ' );
2299
2300 # Don't allow internal links to pages containing
2301 # PROTO: where PROTO is a valid URL protocol; these
2302 # should be external links.
2303 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2304 $s .= $prefix . '[[' . $line;
2305 continue;
2306 }
2307
2308 # Make subpage if necessary
2309 if ( $useSubpages ) {
2310 $link = $this->maybeDoSubpageLink( $origLink, $text );
2311 } else {
2312 $link = $origLink;
2313 }
2314
2315 // \x7f isn't a default legal title char, so most likely strip
2316 // markers will force us into the "invalid form" path above. But,
2317 // just in case, let's assert that xmlish tags aren't valid in
2318 // the title position.
2319 $unstrip = $this->mStripState->killMarkers( $link );
2320 $noMarkers = ( $unstrip === $link );
2321
2322 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2323 if ( $nt === null ) {
2324 $s .= $prefix . '[[' . $line;
2325 continue;
2326 }
2327
2328 $ns = $nt->getNamespace();
2329 $iw = $nt->getInterwiki();
2330
2331 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2332
2333 if ( $might_be_img ) { # if this is actually an invalid link
2334 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2335 $found = false;
2336 while ( true ) {
2337 # look at the next 'line' to see if we can close it there
2338 $a->next();
2339 $next_line = $a->current();
2340 if ( $next_line === false || $next_line === null ) {
2341 break;
2342 }
2343 $m = explode( ']]', $next_line, 3 );
2344 if ( count( $m ) == 3 ) {
2345 # the first ]] closes the inner link, the second the image
2346 $found = true;
2347 $text .= "[[{$m[0]}]]{$m[1]}";
2348 $trail = $m[2];
2349 break;
2350 } elseif ( count( $m ) == 2 ) {
2351 # if there's exactly one ]] that's fine, we'll keep looking
2352 $text .= "[[{$m[0]}]]{$m[1]}";
2353 } else {
2354 # if $next_line is invalid too, we need look no further
2355 $text .= '[[' . $next_line;
2356 break;
2357 }
2358 }
2359 if ( !$found ) {
2360 # we couldn't find the end of this imageLink, so output it raw
2361 # but don't ignore what might be perfectly normal links in the text we've examined
2362 $holders->merge( $this->replaceInternalLinks2( $text ) );
2363 $s .= "{$prefix}[[$link|$text";
2364 # note: no $trail, because without an end, there *is* no trail
2365 continue;
2366 }
2367 } else { # it's not an image, so output it raw
2368 $s .= "{$prefix}[[$link|$text";
2369 # note: no $trail, because without an end, there *is* no trail
2370 continue;
2371 }
2372 }
2373
2374 $wasblank = ( $text == '' );
2375 if ( $wasblank ) {
2376 $text = $link;
2377 if ( !$noforce ) {
2378 # Strip off leading ':'
2379 $text = substr( $text, 1 );
2380 }
2381 } else {
2382 # T6598 madness. Handle the quotes only if they come from the alternate part
2383 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2384 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2385 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2386 $text = $this->doQuotes( $text );
2387 }
2388
2389 # Link not escaped by : , create the various objects
2390 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2391 # Interwikis
2392 if (
2393 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2394 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2395 in_array( $iw, $this->siteConfig->get( 'ExtraInterlanguageLinkPrefixes' ) )
2396 )
2397 ) {
2398 # T26502: filter duplicates
2399 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2400 $this->mLangLinkLanguages[$iw] = true;
2401 $this->mOutput->addLanguageLink( $nt->getFullText() );
2402 }
2403
2404 /**
2405 * Strip the whitespace interwiki links produce, see T10897
2406 */
2407 $s = rtrim( $s . $prefix ) . $trail; # T175416
2408 continue;
2409 }
2410
2411 if ( $ns == NS_FILE ) {
2412 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2413 if ( $wasblank ) {
2414 # if no parameters were passed, $text
2415 # becomes something like "File:Foo.png",
2416 # which we don't want to pass on to the
2417 # image generator
2418 $text = '';
2419 } else {
2420 # recursively parse links inside the image caption
2421 # actually, this will parse them in any other parameters, too,
2422 # but it might be hard to fix that, and it doesn't matter ATM
2423 $text = $this->replaceExternalLinks( $text );
2424 $holders->merge( $this->replaceInternalLinks2( $text ) );
2425 }
2426 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2427 $s .= $prefix . $this->armorLinks(
2428 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2429 continue;
2430 }
2431 } elseif ( $ns == NS_CATEGORY ) {
2432 /**
2433 * Strip the whitespace Category links produce, see T2087
2434 */
2435 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2436
2437 if ( $wasblank ) {
2438 $sortkey = $this->getDefaultSort();
2439 } else {
2440 $sortkey = $text;
2441 }
2442 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2443 $sortkey = str_replace( "\n", '', $sortkey );
2444 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2445 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2446
2447 continue;
2448 }
2449 }
2450
2451 # Self-link checking. For some languages, variants of the title are checked in
2452 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2453 # for linking to a different variant.
2454 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2455 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2456 continue;
2457 }
2458
2459 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2460 # @todo FIXME: Should do batch file existence checks, see comment below
2461 if ( $ns == NS_MEDIA ) {
2462 # Give extensions a chance to select the file revision for us
2463 $options = [];
2464 $descQuery = false;
2465 Hooks::run( 'BeforeParserFetchFileAndTitle',
2466 [ $this, $nt, &$options, &$descQuery ] );
2467 # Fetch and register the file (file title may be different via hooks)
2468 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2469 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2470 $s .= $prefix . $this->armorLinks(
2471 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2472 continue;
2473 }
2474
2475 # Some titles, such as valid special pages or files in foreign repos, should
2476 # be shown as bluelinks even though they're not included in the page table
2477 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2478 # batch file existence checks for NS_FILE and NS_MEDIA
2479 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2480 $this->mOutput->addLink( $nt );
2481 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2482 } else {
2483 # Links will be added to the output link list after checking
2484 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2485 }
2486 }
2487 return $holders;
2488 }
2489
2490 /**
2491 * Render a forced-blue link inline; protect against double expansion of
2492 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2493 * Since this little disaster has to split off the trail text to avoid
2494 * breaking URLs in the following text without breaking trails on the
2495 * wiki links, it's been made into a horrible function.
2496 *
2497 * @param Title $nt
2498 * @param string $text
2499 * @param string $trail
2500 * @param string $prefix
2501 * @return string HTML-wikitext mix oh yuck
2502 */
2503 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2504 list( $inside, $trail ) = Linker::splitTrail( $trail );
2505
2506 if ( $text == '' ) {
2507 $text = htmlspecialchars( $nt->getPrefixedText() );
2508 }
2509
2510 $link = $this->getLinkRenderer()->makeKnownLink(
2511 $nt, new HtmlArmor( "$prefix$text$inside" )
2512 );
2513
2514 return $this->armorLinks( $link ) . $trail;
2515 }
2516
2517 /**
2518 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2519 * going to go through further parsing steps before inline URL expansion.
2520 *
2521 * Not needed quite as much as it used to be since free links are a bit
2522 * more sensible these days. But bracketed links are still an issue.
2523 *
2524 * @param string $text More-or-less HTML
2525 * @return string Less-or-more HTML with NOPARSE bits
2526 */
2527 public function armorLinks( $text ) {
2528 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2529 self::MARKER_PREFIX . "NOPARSE$1", $text );
2530 }
2531
2532 /**
2533 * Return true if subpage links should be expanded on this page.
2534 * @return bool
2535 */
2536 public function areSubpagesAllowed() {
2537 # Some namespaces don't allow subpages
2538 return $this->nsInfo->hasSubpages( $this->mTitle->getNamespace() );
2539 }
2540
2541 /**
2542 * Handle link to subpage if necessary
2543 *
2544 * @param string $target The source of the link
2545 * @param string &$text The link text, modified as necessary
2546 * @return string The full name of the link
2547 * @private
2548 */
2549 public function maybeDoSubpageLink( $target, &$text ) {
2550 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2551 }
2552
2553 /**
2554 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2555 *
2556 * @param string $text
2557 * @param bool $linestart Whether or not this is at the start of a line.
2558 * @private
2559 * @return string The lists rendered as HTML
2560 */
2561 public function doBlockLevels( $text, $linestart ) {
2562 return BlockLevelPass::doBlockLevels( $text, $linestart );
2563 }
2564
2565 /**
2566 * Return value of a magic variable (like PAGENAME)
2567 *
2568 * @private
2569 *
2570 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2571 * @param bool|PPFrame $frame
2572 *
2573 * @throws MWException
2574 * @return string
2575 */
2576 public function getVariableValue( $index, $frame = false ) {
2577 if ( is_null( $this->mTitle ) ) {
2578 // If no title set, bad things are going to happen
2579 // later. Title should always be set since this
2580 // should only be called in the middle of a parse
2581 // operation (but the unit-tests do funky stuff)
2582 throw new MWException( __METHOD__ . ' Should only be '
2583 . ' called while parsing (no title set)' );
2584 }
2585
2586 // Avoid PHP 7.1 warning from passing $this by reference
2587 $parser = $this;
2588
2589 /**
2590 * Some of these require message or data lookups and can be
2591 * expensive to check many times.
2592 */
2593 if (
2594 Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) &&
2595 isset( $this->mVarCache[$index] )
2596 ) {
2597 return $this->mVarCache[$index];
2598 }
2599
2600 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2601 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2602
2603 $pageLang = $this->getFunctionLang();
2604
2605 switch ( $index ) {
2606 case '!':
2607 $value = '|';
2608 break;
2609 case 'currentmonth':
2610 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2611 break;
2612 case 'currentmonth1':
2613 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2614 break;
2615 case 'currentmonthname':
2616 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2617 break;
2618 case 'currentmonthnamegen':
2619 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2620 break;
2621 case 'currentmonthabbrev':
2622 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2623 break;
2624 case 'currentday':
2625 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2626 break;
2627 case 'currentday2':
2628 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2629 break;
2630 case 'localmonth':
2631 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2632 break;
2633 case 'localmonth1':
2634 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2635 break;
2636 case 'localmonthname':
2637 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2638 break;
2639 case 'localmonthnamegen':
2640 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2641 break;
2642 case 'localmonthabbrev':
2643 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2644 break;
2645 case 'localday':
2646 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2647 break;
2648 case 'localday2':
2649 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2650 break;
2651 case 'pagename':
2652 $value = wfEscapeWikiText( $this->mTitle->getText() );
2653 break;
2654 case 'pagenamee':
2655 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2656 break;
2657 case 'fullpagename':
2658 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2659 break;
2660 case 'fullpagenamee':
2661 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2662 break;
2663 case 'subpagename':
2664 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2665 break;
2666 case 'subpagenamee':
2667 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2668 break;
2669 case 'rootpagename':
2670 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2671 break;
2672 case 'rootpagenamee':
2673 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2674 ' ',
2675 '_',
2676 $this->mTitle->getRootText()
2677 ) ) );
2678 break;
2679 case 'basepagename':
2680 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2681 break;
2682 case 'basepagenamee':
2683 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2684 ' ',
2685 '_',
2686 $this->mTitle->getBaseText()
2687 ) ) );
2688 break;
2689 case 'talkpagename':
2690 if ( $this->mTitle->canHaveTalkPage() ) {
2691 $talkPage = $this->mTitle->getTalkPage();
2692 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2693 } else {
2694 $value = '';
2695 }
2696 break;
2697 case 'talkpagenamee':
2698 if ( $this->mTitle->canHaveTalkPage() ) {
2699 $talkPage = $this->mTitle->getTalkPage();
2700 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2701 } else {
2702 $value = '';
2703 }
2704 break;
2705 case 'subjectpagename':
2706 $subjPage = $this->mTitle->getSubjectPage();
2707 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2708 break;
2709 case 'subjectpagenamee':
2710 $subjPage = $this->mTitle->getSubjectPage();
2711 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2712 break;
2713 case 'pageid': // requested in T25427
2714 $pageid = $this->getTitle()->getArticleID();
2715 if ( $pageid == 0 ) {
2716 # 0 means the page doesn't exist in the database,
2717 # which means the user is previewing a new page.
2718 # The vary-revision flag must be set, because the magic word
2719 # will have a different value once the page is saved.
2720 $this->mOutput->setFlag( 'vary-revision' );
2721 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2722 }
2723 $value = $pageid ?: null;
2724 break;
2725 case 'revisionid':
2726 if (
2727 $this->siteConfig->get( 'MiserMode' ) &&
2728 !$this->mOptions->getInterfaceMessage() &&
2729 // @TODO: disallow this word on all namespaces
2730 $this->nsInfo->isContent( $this->mTitle->getNamespace() )
2731 ) {
2732 // Use a stub result instead of the actual revision ID in order to avoid
2733 // double parses on page save but still allow preview detection (T137900)
2734 if ( $this->getRevisionId() || $this->mOptions->getSpeculativeRevId() ) {
2735 $value = '-';
2736 } else {
2737 $this->mOutput->setFlag( 'vary-revision-exists' );
2738 $value = '';
2739 }
2740 } else {
2741 # Inform the edit saving system that getting the canonical output after
2742 # revision insertion requires another parse using the actual revision ID
2743 $this->mOutput->setFlag( 'vary-revision-id' );
2744 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2745 $value = $this->getRevisionId();
2746 if ( $value === 0 ) {
2747 $rev = $this->getRevisionObject();
2748 $value = $rev ? $rev->getId() : $value;
2749 }
2750 if ( !$value ) {
2751 $value = $this->mOptions->getSpeculativeRevId();
2752 if ( $value ) {
2753 $this->mOutput->setSpeculativeRevIdUsed( $value );
2754 }
2755 }
2756 }
2757 break;
2758 case 'revisionday':
2759 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2760 break;
2761 case 'revisionday2':
2762 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2763 break;
2764 case 'revisionmonth':
2765 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2766 break;
2767 case 'revisionmonth1':
2768 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2769 break;
2770 case 'revisionyear':
2771 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2772 break;
2773 case 'revisiontimestamp':
2774 # Let the edit saving system know we should parse the page
2775 # *after* a revision ID has been assigned. This is for null edits.
2776 $this->mOutput->setFlag( 'vary-revision' );
2777 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2778 $value = $this->getRevisionTimestamp();
2779 break;
2780 case 'revisionuser':
2781 # Let the edit saving system know we should parse the page
2782 # *after* a revision ID has been assigned for null edits.
2783 $this->mOutput->setFlag( 'vary-user' );
2784 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2785 $value = $this->getRevisionUser();
2786 break;
2787 case 'revisionsize':
2788 $value = $this->getRevisionSize();
2789 break;
2790 case 'namespace':
2791 $value = str_replace( '_', ' ',
2792 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2793 break;
2794 case 'namespacee':
2795 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2796 break;
2797 case 'namespacenumber':
2798 $value = $this->mTitle->getNamespace();
2799 break;
2800 case 'talkspace':
2801 $value = $this->mTitle->canHaveTalkPage()
2802 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2803 : '';
2804 break;
2805 case 'talkspacee':
2806 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2807 break;
2808 case 'subjectspace':
2809 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2810 break;
2811 case 'subjectspacee':
2812 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2813 break;
2814 case 'currentdayname':
2815 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2816 break;
2817 case 'currentyear':
2818 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2819 break;
2820 case 'currenttime':
2821 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2822 break;
2823 case 'currenthour':
2824 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2825 break;
2826 case 'currentweek':
2827 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2828 # int to remove the padding
2829 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2830 break;
2831 case 'currentdow':
2832 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2833 break;
2834 case 'localdayname':
2835 $value = $pageLang->getWeekdayName(
2836 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2837 );
2838 break;
2839 case 'localyear':
2840 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2841 break;
2842 case 'localtime':
2843 $value = $pageLang->time(
2844 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2845 false,
2846 false
2847 );
2848 break;
2849 case 'localhour':
2850 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2851 break;
2852 case 'localweek':
2853 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2854 # int to remove the padding
2855 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2856 break;
2857 case 'localdow':
2858 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2859 break;
2860 case 'numberofarticles':
2861 $value = $pageLang->formatNum( SiteStats::articles() );
2862 break;
2863 case 'numberoffiles':
2864 $value = $pageLang->formatNum( SiteStats::images() );
2865 break;
2866 case 'numberofusers':
2867 $value = $pageLang->formatNum( SiteStats::users() );
2868 break;
2869 case 'numberofactiveusers':
2870 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2871 break;
2872 case 'numberofpages':
2873 $value = $pageLang->formatNum( SiteStats::pages() );
2874 break;
2875 case 'numberofadmins':
2876 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2877 break;
2878 case 'numberofedits':
2879 $value = $pageLang->formatNum( SiteStats::edits() );
2880 break;
2881 case 'currenttimestamp':
2882 $value = wfTimestamp( TS_MW, $ts );
2883 break;
2884 case 'localtimestamp':
2885 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2886 break;
2887 case 'currentversion':
2888 $value = SpecialVersion::getVersion();
2889 break;
2890 case 'articlepath':
2891 return $this->siteConfig->get( 'ArticlePath' );
2892 case 'sitename':
2893 return $this->siteConfig->get( 'Sitename' );
2894 case 'server':
2895 return $this->siteConfig->get( 'Server' );
2896 case 'servername':
2897 return $this->siteConfig->get( 'ServerName' );
2898 case 'scriptpath':
2899 return $this->siteConfig->get( 'ScriptPath' );
2900 case 'stylepath':
2901 return $this->siteConfig->get( 'StylePath' );
2902 case 'directionmark':
2903 return $pageLang->getDirMark();
2904 case 'contentlanguage':
2905 return $this->siteConfig->get( 'LanguageCode' );
2906 case 'pagelanguage':
2907 $value = $pageLang->getCode();
2908 break;
2909 case 'cascadingsources':
2910 $value = CoreParserFunctions::cascadingsources( $this );
2911 break;
2912 default:
2913 $ret = null;
2914 Hooks::run(
2915 'ParserGetVariableValueSwitch',
2916 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2917 );
2918
2919 return $ret;
2920 }
2921
2922 if ( $index ) {
2923 $this->mVarCache[$index] = $value;
2924 }
2925
2926 return $value;
2927 }
2928
2929 /**
2930 * @param int $start
2931 * @param int $len
2932 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2933 * @param string $variable Parser variable name
2934 * @return string
2935 */
2936 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2937 # Get the timezone-adjusted timestamp to be used for this revision
2938 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2939 # Possibly set vary-revision if there is not yet an associated revision
2940 if ( !$this->getRevisionObject() ) {
2941 # Get the timezone-adjusted timestamp $mtts seconds in the future
2942 $resThen = substr(
2943 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2944 $start,
2945 $len
2946 );
2947
2948 if ( $resNow !== $resThen ) {
2949 # Let the edit saving system know we should parse the page
2950 # *after* a revision ID has been assigned. This is for null edits.
2951 $this->mOutput->setFlag( 'vary-revision' );
2952 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2953 }
2954 }
2955
2956 return $resNow;
2957 }
2958
2959 /**
2960 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2961 *
2962 * @private
2963 */
2964 public function initialiseVariables() {
2965 $variableIDs = $this->magicWordFactory->getVariableIDs();
2966 $substIDs = $this->magicWordFactory->getSubstIDs();
2967
2968 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
2969 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
2970 }
2971
2972 /**
2973 * Preprocess some wikitext and return the document tree.
2974 * This is the ghost of replace_variables().
2975 *
2976 * @param string $text The text to parse
2977 * @param int $flags Bitwise combination of:
2978 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2979 * included. Default is to assume a direct page view.
2980 *
2981 * The generated DOM tree must depend only on the input text and the flags.
2982 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2983 *
2984 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2985 * change in the DOM tree for a given text, must be passed through the section identifier
2986 * in the section edit link and thus back to extractSections().
2987 *
2988 * The output of this function is currently only cached in process memory, but a persistent
2989 * cache may be implemented at a later date which takes further advantage of these strict
2990 * dependency requirements.
2991 *
2992 * @return PPNode
2993 */
2994 public function preprocessToDom( $text, $flags = 0 ) {
2995 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2996 return $dom;
2997 }
2998
2999 /**
3000 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3001 *
3002 * @param string $s
3003 *
3004 * @return array
3005 */
3006 public static function splitWhitespace( $s ) {
3007 $ltrimmed = ltrim( $s );
3008 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3009 $trimmed = rtrim( $ltrimmed );
3010 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3011 if ( $diff > 0 ) {
3012 $w2 = substr( $ltrimmed, -$diff );
3013 } else {
3014 $w2 = '';
3015 }
3016 return [ $w1, $trimmed, $w2 ];
3017 }
3018
3019 /**
3020 * Replace magic variables, templates, and template arguments
3021 * with the appropriate text. Templates are substituted recursively,
3022 * taking care to avoid infinite loops.
3023 *
3024 * Note that the substitution depends on value of $mOutputType:
3025 * self::OT_WIKI: only {{subst:}} templates
3026 * self::OT_PREPROCESS: templates but not extension tags
3027 * self::OT_HTML: all templates and extension tags
3028 *
3029 * @param string $text The text to transform
3030 * @param bool|PPFrame $frame Object describing the arguments passed to the
3031 * template. Arguments may also be provided as an associative array, as
3032 * was the usual case before MW1.12. Providing arguments this way may be
3033 * useful for extensions wishing to perform variable replacement
3034 * explicitly.
3035 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3036 * double-brace expansion.
3037 * @return string
3038 */
3039 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3040 # Is there any text? Also, Prevent too big inclusions!
3041 $textSize = strlen( $text );
3042 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3043 return $text;
3044 }
3045
3046 if ( $frame === false ) {
3047 $frame = $this->getPreprocessor()->newFrame();
3048 } elseif ( !( $frame instanceof PPFrame ) ) {
3049 wfDebug( __METHOD__ . " called using plain parameters instead of "
3050 . "a PPFrame instance. Creating custom frame.\n" );
3051 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3052 }
3053
3054 $dom = $this->preprocessToDom( $text );
3055 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3056 $text = $frame->expand( $dom, $flags );
3057
3058 return $text;
3059 }
3060
3061 /**
3062 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3063 *
3064 * @param array $args
3065 *
3066 * @return array
3067 */
3068 public static function createAssocArgs( $args ) {
3069 $assocArgs = [];
3070 $index = 1;
3071 foreach ( $args as $arg ) {
3072 $eqpos = strpos( $arg, '=' );
3073 if ( $eqpos === false ) {
3074 $assocArgs[$index++] = $arg;
3075 } else {
3076 $name = trim( substr( $arg, 0, $eqpos ) );
3077 $value = trim( substr( $arg, $eqpos + 1 ) );
3078 if ( $value === false ) {
3079 $value = '';
3080 }
3081 if ( $name !== false ) {
3082 $assocArgs[$name] = $value;
3083 }
3084 }
3085 }
3086
3087 return $assocArgs;
3088 }
3089
3090 /**
3091 * Warn the user when a parser limitation is reached
3092 * Will warn at most once the user per limitation type
3093 *
3094 * The results are shown during preview and run through the Parser (See EditPage.php)
3095 *
3096 * @param string $limitationType Should be one of:
3097 * 'expensive-parserfunction' (corresponding messages:
3098 * 'expensive-parserfunction-warning',
3099 * 'expensive-parserfunction-category')
3100 * 'post-expand-template-argument' (corresponding messages:
3101 * 'post-expand-template-argument-warning',
3102 * 'post-expand-template-argument-category')
3103 * 'post-expand-template-inclusion' (corresponding messages:
3104 * 'post-expand-template-inclusion-warning',
3105 * 'post-expand-template-inclusion-category')
3106 * 'node-count-exceeded' (corresponding messages:
3107 * 'node-count-exceeded-warning',
3108 * 'node-count-exceeded-category')
3109 * 'expansion-depth-exceeded' (corresponding messages:
3110 * 'expansion-depth-exceeded-warning',
3111 * 'expansion-depth-exceeded-category')
3112 * @param string|int|null $current Current value
3113 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3114 * exceeded, provide the values (optional)
3115 */
3116 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3117 # does no harm if $current and $max are present but are unnecessary for the message
3118 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3119 # only during preview, and that would split the parser cache unnecessarily.
3120 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3121 ->text();
3122 $this->mOutput->addWarning( $warning );
3123 $this->addTrackingCategory( "$limitationType-category" );
3124 }
3125
3126 /**
3127 * Return the text of a template, after recursively
3128 * replacing any variables or templates within the template.
3129 *
3130 * @param array $piece The parts of the template
3131 * $piece['title']: the title, i.e. the part before the |
3132 * $piece['parts']: the parameter array
3133 * $piece['lineStart']: whether the brace was at the start of a line
3134 * @param PPFrame $frame The current frame, contains template arguments
3135 * @throws Exception
3136 * @return string The text of the template
3137 */
3138 public function braceSubstitution( $piece, $frame ) {
3139 // Flags
3140
3141 // $text has been filled
3142 $found = false;
3143 // wiki markup in $text should be escaped
3144 $nowiki = false;
3145 // $text is HTML, armour it against wikitext transformation
3146 $isHTML = false;
3147 // Force interwiki transclusion to be done in raw mode not rendered
3148 $forceRawInterwiki = false;
3149 // $text is a DOM node needing expansion in a child frame
3150 $isChildObj = false;
3151 // $text is a DOM node needing expansion in the current frame
3152 $isLocalObj = false;
3153
3154 # Title object, where $text came from
3155 $title = false;
3156
3157 # $part1 is the bit before the first |, and must contain only title characters.
3158 # Various prefixes will be stripped from it later.
3159 $titleWithSpaces = $frame->expand( $piece['title'] );
3160 $part1 = trim( $titleWithSpaces );
3161 $titleText = false;
3162
3163 # Original title text preserved for various purposes
3164 $originalTitle = $part1;
3165
3166 # $args is a list of argument nodes, starting from index 0, not including $part1
3167 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3168 # below won't work b/c this $args isn't an object
3169 $args = ( $piece['parts'] == null ) ? [] : $piece['parts'];
3170
3171 $profileSection = null; // profile templates
3172
3173 # SUBST
3174 if ( !$found ) {
3175 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3176
3177 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3178 # Decide whether to expand template or keep wikitext as-is.
3179 if ( $this->ot['wiki'] ) {
3180 if ( $substMatch === false ) {
3181 $literal = true; # literal when in PST with no prefix
3182 } else {
3183 $literal = false; # expand when in PST with subst: or safesubst:
3184 }
3185 } else {
3186 if ( $substMatch == 'subst' ) {
3187 $literal = true; # literal when not in PST with plain subst:
3188 } else {
3189 $literal = false; # expand when not in PST with safesubst: or no prefix
3190 }
3191 }
3192 if ( $literal ) {
3193 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3194 $isLocalObj = true;
3195 $found = true;
3196 }
3197 }
3198
3199 # Variables
3200 if ( !$found && $args->getLength() == 0 ) {
3201 $id = $this->mVariables->matchStartToEnd( $part1 );
3202 if ( $id !== false ) {
3203 $text = $this->getVariableValue( $id, $frame );
3204 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3205 $this->mOutput->updateCacheExpiry(
3206 $this->magicWordFactory->getCacheTTL( $id ) );
3207 }
3208 $found = true;
3209 }
3210 }
3211
3212 # MSG, MSGNW and RAW
3213 if ( !$found ) {
3214 # Check for MSGNW:
3215 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3216 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3217 $nowiki = true;
3218 } else {
3219 # Remove obsolete MSG:
3220 $mwMsg = $this->magicWordFactory->get( 'msg' );
3221 $mwMsg->matchStartAndRemove( $part1 );
3222 }
3223
3224 # Check for RAW:
3225 $mwRaw = $this->magicWordFactory->get( 'raw' );
3226 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3227 $forceRawInterwiki = true;
3228 }
3229 }
3230
3231 # Parser functions
3232 if ( !$found ) {
3233 $colonPos = strpos( $part1, ':' );
3234 if ( $colonPos !== false ) {
3235 $func = substr( $part1, 0, $colonPos );
3236 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3237 $argsLength = $args->getLength();
3238 for ( $i = 0; $i < $argsLength; $i++ ) {
3239 $funcArgs[] = $args->item( $i );
3240 }
3241
3242 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3243
3244 // Extract any forwarded flags
3245 if ( isset( $result['title'] ) ) {
3246 $title = $result['title'];
3247 }
3248 if ( isset( $result['found'] ) ) {
3249 $found = $result['found'];
3250 }
3251 if ( array_key_exists( 'text', $result ) ) {
3252 // a string or null
3253 $text = $result['text'];
3254 }
3255 if ( isset( $result['nowiki'] ) ) {
3256 $nowiki = $result['nowiki'];
3257 }
3258 if ( isset( $result['isHTML'] ) ) {
3259 $isHTML = $result['isHTML'];
3260 }
3261 if ( isset( $result['forceRawInterwiki'] ) ) {
3262 $forceRawInterwiki = $result['forceRawInterwiki'];
3263 }
3264 if ( isset( $result['isChildObj'] ) ) {
3265 $isChildObj = $result['isChildObj'];
3266 }
3267 if ( isset( $result['isLocalObj'] ) ) {
3268 $isLocalObj = $result['isLocalObj'];
3269 }
3270 }
3271 }
3272
3273 # Finish mangling title and then check for loops.
3274 # Set $title to a Title object and $titleText to the PDBK
3275 if ( !$found ) {
3276 $ns = NS_TEMPLATE;
3277 # Split the title into page and subpage
3278 $subpage = '';
3279 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3280 if ( $part1 !== $relative ) {
3281 $part1 = $relative;
3282 $ns = $this->mTitle->getNamespace();
3283 }
3284 $title = Title::newFromText( $part1, $ns );
3285 if ( $title ) {
3286 $titleText = $title->getPrefixedText();
3287 # Check for language variants if the template is not found
3288 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3289 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3290 }
3291 # Do recursion depth check
3292 $limit = $this->mOptions->getMaxTemplateDepth();
3293 if ( $frame->depth >= $limit ) {
3294 $found = true;
3295 $text = '<span class="error">'
3296 . wfMessage( 'parser-template-recursion-depth-warning' )
3297 ->numParams( $limit )->inContentLanguage()->text()
3298 . '</span>';
3299 }
3300 }
3301 }
3302
3303 # Load from database
3304 if ( !$found && $title ) {
3305 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3306 if ( !$title->isExternal() ) {
3307 if ( $title->isSpecialPage()
3308 && $this->mOptions->getAllowSpecialInclusion()
3309 && $this->ot['html']
3310 ) {
3311 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3312 // Pass the template arguments as URL parameters.
3313 // "uselang" will have no effect since the Language object
3314 // is forced to the one defined in ParserOptions.
3315 $pageArgs = [];
3316 $argsLength = $args->getLength();
3317 for ( $i = 0; $i < $argsLength; $i++ ) {
3318 $bits = $args->item( $i )->splitArg();
3319 if ( strval( $bits['index'] ) === '' ) {
3320 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3321 $value = trim( $frame->expand( $bits['value'] ) );
3322 $pageArgs[$name] = $value;
3323 }
3324 }
3325
3326 // Create a new context to execute the special page
3327 $context = new RequestContext;
3328 $context->setTitle( $title );
3329 $context->setRequest( new FauxRequest( $pageArgs ) );
3330 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3331 $context->setUser( $this->getUser() );
3332 } else {
3333 // If this page is cached, then we better not be per user.
3334 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3335 }
3336 $context->setLanguage( $this->mOptions->getUserLangObj() );
3337 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3338 if ( $ret ) {
3339 $text = $context->getOutput()->getHTML();
3340 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3341 $found = true;
3342 $isHTML = true;
3343 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3344 $this->mOutput->updateRuntimeAdaptiveExpiry(
3345 $specialPage->maxIncludeCacheTime()
3346 );
3347 }
3348 }
3349 } elseif ( $this->nsInfo->isNonincludable( $title->getNamespace() ) ) {
3350 $found = false; # access denied
3351 wfDebug( __METHOD__ . ": template inclusion denied for " .
3352 $title->getPrefixedDBkey() . "\n" );
3353 } else {
3354 list( $text, $title ) = $this->getTemplateDom( $title );
3355 if ( $text !== false ) {
3356 $found = true;
3357 $isChildObj = true;
3358 }
3359 }
3360
3361 # If the title is valid but undisplayable, make a link to it
3362 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3363 $text = "[[:$titleText]]";
3364 $found = true;
3365 }
3366 } elseif ( $title->isTrans() ) {
3367 # Interwiki transclusion
3368 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3369 $text = $this->interwikiTransclude( $title, 'render' );
3370 $isHTML = true;
3371 } else {
3372 $text = $this->interwikiTransclude( $title, 'raw' );
3373 # Preprocess it like a template
3374 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3375 $isChildObj = true;
3376 }
3377 $found = true;
3378 }
3379
3380 # Do infinite loop check
3381 # This has to be done after redirect resolution to avoid infinite loops via redirects
3382 if ( !$frame->loopCheck( $title ) ) {
3383 $found = true;
3384 $text = '<span class="error">'
3385 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3386 . '</span>';
3387 $this->addTrackingCategory( 'template-loop-category' );
3388 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3389 wfEscapeWikiText( $titleText ) )->text() );
3390 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3391 }
3392 }
3393
3394 # If we haven't found text to substitute by now, we're done
3395 # Recover the source wikitext and return it
3396 if ( !$found ) {
3397 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3398 if ( $profileSection ) {
3399 $this->mProfiler->scopedProfileOut( $profileSection );
3400 }
3401 return [ 'object' => $text ];
3402 }
3403
3404 # Expand DOM-style return values in a child frame
3405 if ( $isChildObj ) {
3406 # Clean up argument array
3407 $newFrame = $frame->newChild( $args, $title );
3408
3409 if ( $nowiki ) {
3410 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3411 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3412 # Expansion is eligible for the empty-frame cache
3413 $text = $newFrame->cachedExpand( $titleText, $text );
3414 } else {
3415 # Uncached expansion
3416 $text = $newFrame->expand( $text );
3417 }
3418 }
3419 if ( $isLocalObj && $nowiki ) {
3420 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3421 $isLocalObj = false;
3422 }
3423
3424 if ( $profileSection ) {
3425 $this->mProfiler->scopedProfileOut( $profileSection );
3426 }
3427
3428 # Replace raw HTML by a placeholder
3429 if ( $isHTML ) {
3430 $text = $this->insertStripItem( $text );
3431 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3432 # Escape nowiki-style return values
3433 $text = wfEscapeWikiText( $text );
3434 } elseif ( is_string( $text )
3435 && !$piece['lineStart']
3436 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3437 ) {
3438 # T2529: if the template begins with a table or block-level
3439 # element, it should be treated as beginning a new line.
3440 # This behavior is somewhat controversial.
3441 $text = "\n" . $text;
3442 }
3443
3444 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3445 # Error, oversize inclusion
3446 if ( $titleText !== false ) {
3447 # Make a working, properly escaped link if possible (T25588)
3448 $text = "[[:$titleText]]";
3449 } else {
3450 # This will probably not be a working link, but at least it may
3451 # provide some hint of where the problem is
3452 preg_replace( '/^:/', '', $originalTitle );
3453 $text = "[[:$originalTitle]]";
3454 }
3455 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3456 . 'post-expand include size too large -->' );
3457 $this->limitationWarn( 'post-expand-template-inclusion' );
3458 }
3459
3460 if ( $isLocalObj ) {
3461 $ret = [ 'object' => $text ];
3462 } else {
3463 $ret = [ 'text' => $text ];
3464 }
3465
3466 return $ret;
3467 }
3468
3469 /**
3470 * Call a parser function and return an array with text and flags.
3471 *
3472 * The returned array will always contain a boolean 'found', indicating
3473 * whether the parser function was found or not. It may also contain the
3474 * following:
3475 * text: string|object, resulting wikitext or PP DOM object
3476 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3477 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3478 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3479 * nowiki: bool, wiki markup in $text should be escaped
3480 *
3481 * @since 1.21
3482 * @param PPFrame $frame The current frame, contains template arguments
3483 * @param string $function Function name
3484 * @param array $args Arguments to the function
3485 * @throws MWException
3486 * @return array
3487 */
3488 public function callParserFunction( $frame, $function, array $args = [] ) {
3489 # Case sensitive functions
3490 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3491 $function = $this->mFunctionSynonyms[1][$function];
3492 } else {
3493 # Case insensitive functions
3494 $function = $this->contLang->lc( $function );
3495 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3496 $function = $this->mFunctionSynonyms[0][$function];
3497 } else {
3498 return [ 'found' => false ];
3499 }
3500 }
3501
3502 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3503
3504 // Avoid PHP 7.1 warning from passing $this by reference
3505 $parser = $this;
3506
3507 $allArgs = [ &$parser ];
3508 if ( $flags & self::SFH_OBJECT_ARGS ) {
3509 # Convert arguments to PPNodes and collect for appending to $allArgs
3510 $funcArgs = [];
3511 foreach ( $args as $k => $v ) {
3512 if ( $v instanceof PPNode || $k === 0 ) {
3513 $funcArgs[] = $v;
3514 } else {
3515 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3516 }
3517 }
3518
3519 # Add a frame parameter, and pass the arguments as an array
3520 $allArgs[] = $frame;
3521 $allArgs[] = $funcArgs;
3522 } else {
3523 # Convert arguments to plain text and append to $allArgs
3524 foreach ( $args as $k => $v ) {
3525 if ( $v instanceof PPNode ) {
3526 $allArgs[] = trim( $frame->expand( $v ) );
3527 } elseif ( is_int( $k ) && $k >= 0 ) {
3528 $allArgs[] = trim( $v );
3529 } else {
3530 $allArgs[] = trim( "$k=$v" );
3531 }
3532 }
3533 }
3534
3535 $result = $callback( ...$allArgs );
3536
3537 # The interface for function hooks allows them to return a wikitext
3538 # string or an array containing the string and any flags. This mungs
3539 # things around to match what this method should return.
3540 if ( !is_array( $result ) ) {
3541 $result = [
3542 'found' => true,
3543 'text' => $result,
3544 ];
3545 } else {
3546 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3547 $result['text'] = $result[0];
3548 }
3549 unset( $result[0] );
3550 $result += [
3551 'found' => true,
3552 ];
3553 }
3554
3555 $noparse = true;
3556 $preprocessFlags = 0;
3557 if ( isset( $result['noparse'] ) ) {
3558 $noparse = $result['noparse'];
3559 }
3560 if ( isset( $result['preprocessFlags'] ) ) {
3561 $preprocessFlags = $result['preprocessFlags'];
3562 }
3563
3564 if ( !$noparse ) {
3565 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3566 $result['isChildObj'] = true;
3567 }
3568
3569 return $result;
3570 }
3571
3572 /**
3573 * Get the semi-parsed DOM representation of a template with a given title,
3574 * and its redirect destination title. Cached.
3575 *
3576 * @param Title $title
3577 *
3578 * @return array
3579 */
3580 public function getTemplateDom( $title ) {
3581 $cacheTitle = $title;
3582 $titleText = $title->getPrefixedDBkey();
3583
3584 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3585 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3586 $title = Title::makeTitle( $ns, $dbk );
3587 $titleText = $title->getPrefixedDBkey();
3588 }
3589 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3590 return [ $this->mTplDomCache[$titleText], $title ];
3591 }
3592
3593 # Cache miss, go to the database
3594 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3595
3596 if ( $text === false ) {
3597 $this->mTplDomCache[$titleText] = false;
3598 return [ false, $title ];
3599 }
3600
3601 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3602 $this->mTplDomCache[$titleText] = $dom;
3603
3604 if ( !$title->equals( $cacheTitle ) ) {
3605 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3606 [ $title->getNamespace(), $title->getDBkey() ];
3607 }
3608
3609 return [ $dom, $title ];
3610 }
3611
3612 /**
3613 * Fetch the current revision of a given title. Note that the revision
3614 * (and even the title) may not exist in the database, so everything
3615 * contributing to the output of the parser should use this method
3616 * where possible, rather than getting the revisions themselves. This
3617 * method also caches its results, so using it benefits performance.
3618 *
3619 * @since 1.24
3620 * @param Title $title
3621 * @return Revision
3622 */
3623 public function fetchCurrentRevisionOfTitle( $title ) {
3624 $cacheKey = $title->getPrefixedDBkey();
3625 if ( !$this->currentRevisionCache ) {
3626 $this->currentRevisionCache = new MapCacheLRU( 100 );
3627 }
3628 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3629 $this->currentRevisionCache->set( $cacheKey,
3630 // Defaults to Parser::statelessFetchRevision()
3631 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3632 );
3633 }
3634 return $this->currentRevisionCache->get( $cacheKey );
3635 }
3636
3637 /**
3638 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3639 * without passing them on to it.
3640 *
3641 * @since 1.24
3642 * @param Title $title
3643 * @param Parser|bool $parser
3644 * @return Revision|bool False if missing
3645 */
3646 public static function statelessFetchRevision( Title $title, $parser = false ) {
3647 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3648
3649 return $rev;
3650 }
3651
3652 /**
3653 * Fetch the unparsed text of a template and register a reference to it.
3654 * @param Title $title
3655 * @return array ( string or false, Title )
3656 */
3657 public function fetchTemplateAndTitle( $title ) {
3658 // Defaults to Parser::statelessFetchTemplate()
3659 $templateCb = $this->mOptions->getTemplateCallback();
3660 $stuff = call_user_func( $templateCb, $title, $this );
3661 // We use U+007F DELETE to distinguish strip markers from regular text.
3662 $text = $stuff['text'];
3663 if ( is_string( $stuff['text'] ) ) {
3664 $text = strtr( $text, "\x7f", "?" );
3665 }
3666 $finalTitle = $stuff['finalTitle'] ?? $title;
3667 if ( isset( $stuff['deps'] ) ) {
3668 foreach ( $stuff['deps'] as $dep ) {
3669 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3670 if ( $dep['title']->equals( $this->getTitle() ) ) {
3671 // If we transclude ourselves, the final result
3672 // will change based on the new version of the page
3673 $this->mOutput->setFlag( 'vary-revision' );
3674 }
3675 }
3676 }
3677 return [ $text, $finalTitle ];
3678 }
3679
3680 /**
3681 * Fetch the unparsed text of a template and register a reference to it.
3682 * @param Title $title
3683 * @return string|bool
3684 */
3685 public function fetchTemplate( $title ) {
3686 return $this->fetchTemplateAndTitle( $title )[0];
3687 }
3688
3689 /**
3690 * Static function to get a template
3691 * Can be overridden via ParserOptions::setTemplateCallback().
3692 *
3693 * @param Title $title
3694 * @param bool|Parser $parser
3695 *
3696 * @return array
3697 */
3698 public static function statelessFetchTemplate( $title, $parser = false ) {
3699 $text = $skip = false;
3700 $finalTitle = $title;
3701 $deps = [];
3702
3703 # Loop to fetch the article, with up to 1 redirect
3704 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3705 # Give extensions a chance to select the revision instead
3706 $id = false; # Assume current
3707 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3708 [ $parser, $title, &$skip, &$id ] );
3709
3710 if ( $skip ) {
3711 $text = false;
3712 $deps[] = [
3713 'title' => $title,
3714 'page_id' => $title->getArticleID(),
3715 'rev_id' => null
3716 ];
3717 break;
3718 }
3719 # Get the revision
3720 if ( $id ) {
3721 $rev = Revision::newFromId( $id );
3722 } elseif ( $parser ) {
3723 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3724 } else {
3725 $rev = Revision::newFromTitle( $title );
3726 }
3727 $rev_id = $rev ? $rev->getId() : 0;
3728 # If there is no current revision, there is no page
3729 if ( $id === false && !$rev ) {
3730 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3731 $linkCache->addBadLinkObj( $title );
3732 }
3733
3734 $deps[] = [
3735 'title' => $title,
3736 'page_id' => $title->getArticleID(),
3737 'rev_id' => $rev_id ];
3738 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3739 # We fetched a rev from a different title; register it too...
3740 $deps[] = [
3741 'title' => $rev->getTitle(),
3742 'page_id' => $rev->getPage(),
3743 'rev_id' => $rev_id ];
3744 }
3745
3746 if ( $rev ) {
3747 $content = $rev->getContent();
3748 $text = $content ? $content->getWikitextForTransclusion() : null;
3749
3750 Hooks::run( 'ParserFetchTemplate',
3751 [ $parser, $title, $rev, &$text, &$deps ] );
3752
3753 if ( $text === false || $text === null ) {
3754 $text = false;
3755 break;
3756 }
3757 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3758 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
3759 lcfirst( $title->getText() ) )->inContentLanguage();
3760 if ( !$message->exists() ) {
3761 $text = false;
3762 break;
3763 }
3764 $content = $message->content();
3765 $text = $message->plain();
3766 } else {
3767 break;
3768 }
3769 if ( !$content ) {
3770 break;
3771 }
3772 # Redirect?
3773 $finalTitle = $title;
3774 $title = $content->getRedirectTarget();
3775 }
3776 return [
3777 'text' => $text,
3778 'finalTitle' => $finalTitle,
3779 'deps' => $deps ];
3780 }
3781
3782 /**
3783 * Fetch a file and its title and register a reference to it.
3784 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3785 * @param Title $title
3786 * @param array $options Array of options to RepoGroup::findFile
3787 * @return File|bool
3788 * @deprecated since 1.32, use fetchFileAndTitle instead
3789 */
3790 public function fetchFile( $title, $options = [] ) {
3791 wfDeprecated( __METHOD__, '1.32' );
3792 return $this->fetchFileAndTitle( $title, $options )[0];
3793 }
3794
3795 /**
3796 * Fetch a file and its title and register a reference to it.
3797 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3798 * @param Title $title
3799 * @param array $options Array of options to RepoGroup::findFile
3800 * @return array ( File or false, Title of file )
3801 */
3802 public function fetchFileAndTitle( $title, $options = [] ) {
3803 $file = $this->fetchFileNoRegister( $title, $options );
3804
3805 $time = $file ? $file->getTimestamp() : false;
3806 $sha1 = $file ? $file->getSha1() : false;
3807 # Register the file as a dependency...
3808 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3809 if ( $file && !$title->equals( $file->getTitle() ) ) {
3810 # Update fetched file title
3811 $title = $file->getTitle();
3812 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3813 }
3814 return [ $file, $title ];
3815 }
3816
3817 /**
3818 * Helper function for fetchFileAndTitle.
3819 *
3820 * Also useful if you need to fetch a file but not use it yet,
3821 * for example to get the file's handler.
3822 *
3823 * @param Title $title
3824 * @param array $options Array of options to RepoGroup::findFile
3825 * @return File|bool
3826 */
3827 protected function fetchFileNoRegister( $title, $options = [] ) {
3828 if ( isset( $options['broken'] ) ) {
3829 $file = false; // broken thumbnail forced by hook
3830 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3831 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3832 } else { // get by (name,timestamp)
3833 $file = wfFindFile( $title, $options );
3834 }
3835 return $file;
3836 }
3837
3838 /**
3839 * Transclude an interwiki link.
3840 *
3841 * @param Title $title
3842 * @param string $action Usually one of (raw, render)
3843 *
3844 * @return string
3845 */
3846 public function interwikiTransclude( $title, $action ) {
3847 if ( !$this->siteConfig->get( 'EnableScaryTranscluding' ) ) {
3848 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3849 }
3850
3851 $url = $title->getFullURL( [ 'action' => $action ] );
3852 if ( strlen( $url ) > 1024 ) {
3853 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3854 }
3855
3856 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3857
3858 $fname = __METHOD__;
3859 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3860
3861 $data = $cache->getWithSetCallback(
3862 $cache->makeGlobalKey(
3863 'interwiki-transclude',
3864 ( $wikiId !== false ) ? $wikiId : 'external',
3865 sha1( $url )
3866 ),
3867 $this->siteConfig->get( 'TranscludeCacheExpiry' ),
3868 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3869 $req = MWHttpRequest::factory( $url, [], $fname );
3870
3871 $status = $req->execute(); // Status object
3872 if ( !$status->isOK() ) {
3873 $ttl = $cache::TTL_UNCACHEABLE;
3874 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3875 $ttl = min( $cache::TTL_LAGGED, $ttl );
3876 }
3877
3878 return [
3879 'text' => $status->isOK() ? $req->getContent() : null,
3880 'code' => $req->getStatus()
3881 ];
3882 },
3883 [
3884 'checkKeys' => ( $wikiId !== false )
3885 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3886 : [],
3887 'pcGroup' => 'interwiki-transclude:5',
3888 'pcTTL' => $cache::TTL_PROC_LONG
3889 ]
3890 );
3891
3892 if ( is_string( $data['text'] ) ) {
3893 $text = $data['text'];
3894 } elseif ( $data['code'] != 200 ) {
3895 // Though we failed to fetch the content, this status is useless.
3896 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3897 ->params( $url, $data['code'] )->inContentLanguage()->text();
3898 } else {
3899 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3900 }
3901
3902 return $text;
3903 }
3904
3905 /**
3906 * Triple brace replacement -- used for template arguments
3907 * @private
3908 *
3909 * @param array $piece
3910 * @param PPFrame $frame
3911 *
3912 * @return array
3913 */
3914 public function argSubstitution( $piece, $frame ) {
3915 $error = false;
3916 $parts = $piece['parts'];
3917 $nameWithSpaces = $frame->expand( $piece['title'] );
3918 $argName = trim( $nameWithSpaces );
3919 $object = false;
3920 $text = $frame->getArgument( $argName );
3921 if ( $text === false && $parts->getLength() > 0
3922 && ( $this->ot['html']
3923 || $this->ot['pre']
3924 || ( $this->ot['wiki'] && $frame->isTemplate() )
3925 )
3926 ) {
3927 # No match in frame, use the supplied default
3928 $object = $parts->item( 0 )->getChildren();
3929 }
3930 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3931 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3932 $this->limitationWarn( 'post-expand-template-argument' );
3933 }
3934
3935 if ( $text === false && $object === false ) {
3936 # No match anywhere
3937 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3938 }
3939 if ( $error !== false ) {
3940 $text .= $error;
3941 }
3942 if ( $object !== false ) {
3943 $ret = [ 'object' => $object ];
3944 } else {
3945 $ret = [ 'text' => $text ];
3946 }
3947
3948 return $ret;
3949 }
3950
3951 /**
3952 * Return the text to be used for a given extension tag.
3953 * This is the ghost of strip().
3954 *
3955 * @param array $params Associative array of parameters:
3956 * name PPNode for the tag name
3957 * attr PPNode for unparsed text where tag attributes are thought to be
3958 * attributes Optional associative array of parsed attributes
3959 * inner Contents of extension element
3960 * noClose Original text did not have a close tag
3961 * @param PPFrame $frame
3962 *
3963 * @throws MWException
3964 * @return string
3965 */
3966 public function extensionSubstitution( $params, $frame ) {
3967 static $errorStr = '<span class="error">';
3968 static $errorLen = 20;
3969
3970 $name = $frame->expand( $params['name'] );
3971 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3972 // Probably expansion depth or node count exceeded. Just punt the
3973 // error up.
3974 return $name;
3975 }
3976
3977 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3978 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3979 // See above
3980 return $attrText;
3981 }
3982
3983 // We can't safely check if the expansion for $content resulted in an
3984 // error, because the content could happen to be the error string
3985 // (T149622).
3986 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3987
3988 $marker = self::MARKER_PREFIX . "-$name-"
3989 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3990
3991 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3992 ( $this->ot['html'] || $this->ot['pre'] );
3993 if ( $isFunctionTag ) {
3994 $markerType = 'none';
3995 } else {
3996 $markerType = 'general';
3997 }
3998 if ( $this->ot['html'] || $isFunctionTag ) {
3999 $name = strtolower( $name );
4000 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4001 if ( isset( $params['attributes'] ) ) {
4002 $attributes += $params['attributes'];
4003 }
4004
4005 if ( isset( $this->mTagHooks[$name] ) ) {
4006 $output = call_user_func_array( $this->mTagHooks[$name],
4007 [ $content, $attributes, $this, $frame ] );
4008 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4009 list( $callback, ) = $this->mFunctionTagHooks[$name];
4010
4011 // Avoid PHP 7.1 warning from passing $this by reference
4012 $parser = $this;
4013 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4014 } else {
4015 $output = '<span class="error">Invalid tag extension name: ' .
4016 htmlspecialchars( $name ) . '</span>';
4017 }
4018
4019 if ( is_array( $output ) ) {
4020 // Extract flags
4021 $flags = $output;
4022 $output = $flags[0];
4023 if ( isset( $flags['markerType'] ) ) {
4024 $markerType = $flags['markerType'];
4025 }
4026 }
4027 } else {
4028 if ( is_null( $attrText ) ) {
4029 $attrText = '';
4030 }
4031 if ( isset( $params['attributes'] ) ) {
4032 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4033 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4034 htmlspecialchars( $attrValue ) . '"';
4035 }
4036 }
4037 if ( $content === null ) {
4038 $output = "<$name$attrText/>";
4039 } else {
4040 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4041 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4042 // See above
4043 return $close;
4044 }
4045 $output = "<$name$attrText>$content$close";
4046 }
4047 }
4048
4049 if ( $markerType === 'none' ) {
4050 return $output;
4051 } elseif ( $markerType === 'nowiki' ) {
4052 $this->mStripState->addNoWiki( $marker, $output );
4053 } elseif ( $markerType === 'general' ) {
4054 $this->mStripState->addGeneral( $marker, $output );
4055 } else {
4056 throw new MWException( __METHOD__ . ': invalid marker type' );
4057 }
4058 return $marker;
4059 }
4060
4061 /**
4062 * Increment an include size counter
4063 *
4064 * @param string $type The type of expansion
4065 * @param int $size The size of the text
4066 * @return bool False if this inclusion would take it over the maximum, true otherwise
4067 */
4068 public function incrementIncludeSize( $type, $size ) {
4069 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4070 return false;
4071 } else {
4072 $this->mIncludeSizes[$type] += $size;
4073 return true;
4074 }
4075 }
4076
4077 /**
4078 * Increment the expensive function count
4079 *
4080 * @return bool False if the limit has been exceeded
4081 */
4082 public function incrementExpensiveFunctionCount() {
4083 $this->mExpensiveFunctionCount++;
4084 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4085 }
4086
4087 /**
4088 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4089 * Fills $this->mDoubleUnderscores, returns the modified text
4090 *
4091 * @param string $text
4092 *
4093 * @return string
4094 */
4095 public function doDoubleUnderscore( $text ) {
4096 # The position of __TOC__ needs to be recorded
4097 $mw = $this->magicWordFactory->get( 'toc' );
4098 if ( $mw->match( $text ) ) {
4099 $this->mShowToc = true;
4100 $this->mForceTocPosition = true;
4101
4102 # Set a placeholder. At the end we'll fill it in with the TOC.
4103 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4104
4105 # Only keep the first one.
4106 $text = $mw->replace( '', $text );
4107 }
4108
4109 # Now match and remove the rest of them
4110 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4111 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4112
4113 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4114 $this->mOutput->mNoGallery = true;
4115 }
4116 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4117 $this->mShowToc = false;
4118 }
4119 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4120 && $this->mTitle->getNamespace() == NS_CATEGORY
4121 ) {
4122 $this->addTrackingCategory( 'hidden-category-category' );
4123 }
4124 # (T10068) Allow control over whether robots index a page.
4125 # __INDEX__ always overrides __NOINDEX__, see T16899
4126 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4127 $this->mOutput->setIndexPolicy( 'noindex' );
4128 $this->addTrackingCategory( 'noindex-category' );
4129 }
4130 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4131 $this->mOutput->setIndexPolicy( 'index' );
4132 $this->addTrackingCategory( 'index-category' );
4133 }
4134
4135 # Cache all double underscores in the database
4136 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4137 $this->mOutput->setProperty( $key, '' );
4138 }
4139
4140 return $text;
4141 }
4142
4143 /**
4144 * @see ParserOutput::addTrackingCategory()
4145 * @param string $msg Message key
4146 * @return bool Whether the addition was successful
4147 */
4148 public function addTrackingCategory( $msg ) {
4149 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4150 }
4151
4152 /**
4153 * This function accomplishes several tasks:
4154 * 1) Auto-number headings if that option is enabled
4155 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4156 * 3) Add a Table of contents on the top for users who have enabled the option
4157 * 4) Auto-anchor headings
4158 *
4159 * It loops through all headlines, collects the necessary data, then splits up the
4160 * string and re-inserts the newly formatted headlines.
4161 *
4162 * @param string $text
4163 * @param string $origText Original, untouched wikitext
4164 * @param bool $isMain
4165 * @return mixed|string
4166 * @private
4167 */
4168 public function formatHeadings( $text, $origText, $isMain = true ) {
4169 # Inhibit editsection links if requested in the page
4170 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4171 $maybeShowEditLink = false;
4172 } else {
4173 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4174 }
4175
4176 # Get all headlines for numbering them and adding funky stuff like [edit]
4177 # links - this is for later, but we need the number of headlines right now
4178 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4179 # be trimmed here since whitespace in HTML headings is significant.
4180 $matches = [];
4181 $numMatches = preg_match_all(
4182 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4183 $text,
4184 $matches
4185 );
4186
4187 # if there are fewer than 4 headlines in the article, do not show TOC
4188 # unless it's been explicitly enabled.
4189 $enoughToc = $this->mShowToc &&
4190 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4191
4192 # Allow user to stipulate that a page should have a "new section"
4193 # link added via __NEWSECTIONLINK__
4194 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4195 $this->mOutput->setNewSection( true );
4196 }
4197
4198 # Allow user to remove the "new section"
4199 # link via __NONEWSECTIONLINK__
4200 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4201 $this->mOutput->hideNewSection( true );
4202 }
4203
4204 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4205 # override above conditions and always show TOC above first header
4206 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4207 $this->mShowToc = true;
4208 $enoughToc = true;
4209 }
4210
4211 # headline counter
4212 $headlineCount = 0;
4213 $numVisible = 0;
4214
4215 # Ugh .. the TOC should have neat indentation levels which can be
4216 # passed to the skin functions. These are determined here
4217 $toc = '';
4218 $full = '';
4219 $head = [];
4220 $sublevelCount = [];
4221 $levelCount = [];
4222 $level = 0;
4223 $prevlevel = 0;
4224 $toclevel = 0;
4225 $prevtoclevel = 0;
4226 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4227 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4228 $oldType = $this->mOutputType;
4229 $this->setOutputType( self::OT_WIKI );
4230 $frame = $this->getPreprocessor()->newFrame();
4231 $root = $this->preprocessToDom( $origText );
4232 $node = $root->getFirstChild();
4233 $byteOffset = 0;
4234 $tocraw = [];
4235 $refers = [];
4236
4237 $headlines = $numMatches !== false ? $matches[3] : [];
4238
4239 $maxTocLevel = $this->siteConfig->get( 'MaxTocLevel' );
4240 foreach ( $headlines as $headline ) {
4241 $isTemplate = false;
4242 $titleText = false;
4243 $sectionIndex = false;
4244 $numbering = '';
4245 $markerMatches = [];
4246 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4247 $serial = $markerMatches[1];
4248 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4249 $isTemplate = ( $titleText != $baseTitleText );
4250 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4251 }
4252
4253 if ( $toclevel ) {
4254 $prevlevel = $level;
4255 }
4256 $level = $matches[1][$headlineCount];
4257
4258 if ( $level > $prevlevel ) {
4259 # Increase TOC level
4260 $toclevel++;
4261 $sublevelCount[$toclevel] = 0;
4262 if ( $toclevel < $maxTocLevel ) {
4263 $prevtoclevel = $toclevel;
4264 $toc .= Linker::tocIndent();
4265 $numVisible++;
4266 }
4267 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4268 # Decrease TOC level, find level to jump to
4269
4270 for ( $i = $toclevel; $i > 0; $i-- ) {
4271 if ( $levelCount[$i] == $level ) {
4272 # Found last matching level
4273 $toclevel = $i;
4274 break;
4275 } elseif ( $levelCount[$i] < $level ) {
4276 # Found first matching level below current level
4277 $toclevel = $i + 1;
4278 break;
4279 }
4280 }
4281 if ( $i == 0 ) {
4282 $toclevel = 1;
4283 }
4284 if ( $toclevel < $maxTocLevel ) {
4285 if ( $prevtoclevel < $maxTocLevel ) {
4286 # Unindent only if the previous toc level was shown :p
4287 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4288 $prevtoclevel = $toclevel;
4289 } else {
4290 $toc .= Linker::tocLineEnd();
4291 }
4292 }
4293 } else {
4294 # No change in level, end TOC line
4295 if ( $toclevel < $maxTocLevel ) {
4296 $toc .= Linker::tocLineEnd();
4297 }
4298 }
4299
4300 $levelCount[$toclevel] = $level;
4301
4302 # count number of headlines for each level
4303 $sublevelCount[$toclevel]++;
4304 $dot = 0;
4305 for ( $i = 1; $i <= $toclevel; $i++ ) {
4306 if ( !empty( $sublevelCount[$i] ) ) {
4307 if ( $dot ) {
4308 $numbering .= '.';
4309 }
4310 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4311 $dot = 1;
4312 }
4313 }
4314
4315 # The safe header is a version of the header text safe to use for links
4316
4317 # Remove link placeholders by the link text.
4318 # <!--LINK number-->
4319 # turns into
4320 # link text with suffix
4321 # Do this before unstrip since link text can contain strip markers
4322 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4323
4324 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4325 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4326
4327 # Remove any <style> or <script> tags (T198618)
4328 $safeHeadline = preg_replace(
4329 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4330 '',
4331 $safeHeadline
4332 );
4333
4334 # Strip out HTML (first regex removes any tag not allowed)
4335 # Allowed tags are:
4336 # * <sup> and <sub> (T10393)
4337 # * <i> (T28375)
4338 # * <b> (r105284)
4339 # * <bdi> (T74884)
4340 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4341 # * <s> and <strike> (T35715)
4342 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4343 # to allow setting directionality in toc items.
4344 $tocline = preg_replace(
4345 [
4346 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4347 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4348 ],
4349 [ '', '<$1>' ],
4350 $safeHeadline
4351 );
4352
4353 # Strip '<span></span>', which is the result from the above if
4354 # <span id="foo"></span> is used to produce an additional anchor
4355 # for a section.
4356 $tocline = str_replace( '<span></span>', '', $tocline );
4357
4358 $tocline = trim( $tocline );
4359
4360 # For the anchor, strip out HTML-y stuff period
4361 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4362 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4363
4364 # Save headline for section edit hint before it's escaped
4365 $headlineHint = $safeHeadline;
4366
4367 # Decode HTML entities
4368 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4369
4370 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4371
4372 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4373 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4374 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4375 if ( $fallbackHeadline === $safeHeadline ) {
4376 # No reason to have both (in fact, we can't)
4377 $fallbackHeadline = false;
4378 }
4379
4380 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4381 # @todo FIXME: We may be changing them depending on the current locale.
4382 $arrayKey = strtolower( $safeHeadline );
4383 if ( $fallbackHeadline === false ) {
4384 $fallbackArrayKey = false;
4385 } else {
4386 $fallbackArrayKey = strtolower( $fallbackHeadline );
4387 }
4388
4389 # Create the anchor for linking from the TOC to the section
4390 $anchor = $safeHeadline;
4391 $fallbackAnchor = $fallbackHeadline;
4392 if ( isset( $refers[$arrayKey] ) ) {
4393 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4394 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4395 $anchor .= "_$i";
4396 $linkAnchor .= "_$i";
4397 $refers["${arrayKey}_$i"] = true;
4398 } else {
4399 $refers[$arrayKey] = true;
4400 }
4401 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4402 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4403 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4404 $fallbackAnchor .= "_$i";
4405 $refers["${fallbackArrayKey}_$i"] = true;
4406 } else {
4407 $refers[$fallbackArrayKey] = true;
4408 }
4409
4410 # Don't number the heading if it is the only one (looks silly)
4411 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4412 # the two are different if the line contains a link
4413 $headline = Html::element(
4414 'span',
4415 [ 'class' => 'mw-headline-number' ],
4416 $numbering
4417 ) . ' ' . $headline;
4418 }
4419
4420 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4421 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4422 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4423 }
4424
4425 # Add the section to the section tree
4426 # Find the DOM node for this header
4427 $noOffset = ( $isTemplate || $sectionIndex === false );
4428 while ( $node && !$noOffset ) {
4429 if ( $node->getName() === 'h' ) {
4430 $bits = $node->splitHeading();
4431 if ( $bits['i'] == $sectionIndex ) {
4432 break;
4433 }
4434 }
4435 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4436 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4437 $node = $node->getNextSibling();
4438 }
4439 $tocraw[] = [
4440 'toclevel' => $toclevel,
4441 'level' => $level,
4442 'line' => $tocline,
4443 'number' => $numbering,
4444 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4445 'fromtitle' => $titleText,
4446 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4447 'anchor' => $anchor,
4448 ];
4449
4450 # give headline the correct <h#> tag
4451 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4452 // Output edit section links as markers with styles that can be customized by skins
4453 if ( $isTemplate ) {
4454 # Put a T flag in the section identifier, to indicate to extractSections()
4455 # that sections inside <includeonly> should be counted.
4456 $editsectionPage = $titleText;
4457 $editsectionSection = "T-$sectionIndex";
4458 $editsectionContent = null;
4459 } else {
4460 $editsectionPage = $this->mTitle->getPrefixedText();
4461 $editsectionSection = $sectionIndex;
4462 $editsectionContent = $headlineHint;
4463 }
4464 // We use a bit of pesudo-xml for editsection markers. The
4465 // language converter is run later on. Using a UNIQ style marker
4466 // leads to the converter screwing up the tokens when it
4467 // converts stuff. And trying to insert strip tags fails too. At
4468 // this point all real inputted tags have already been escaped,
4469 // so we don't have to worry about a user trying to input one of
4470 // these markers directly. We use a page and section attribute
4471 // to stop the language converter from converting these
4472 // important bits of data, but put the headline hint inside a
4473 // content block because the language converter is supposed to
4474 // be able to convert that piece of data.
4475 // Gets replaced with html in ParserOutput::getText
4476 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4477 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4478 if ( $editsectionContent !== null ) {
4479 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4480 } else {
4481 $editlink .= '/>';
4482 }
4483 } else {
4484 $editlink = '';
4485 }
4486 $head[$headlineCount] = Linker::makeHeadline( $level,
4487 $matches['attrib'][$headlineCount], $anchor, $headline,
4488 $editlink, $fallbackAnchor );
4489
4490 $headlineCount++;
4491 }
4492
4493 $this->setOutputType( $oldType );
4494
4495 # Never ever show TOC if no headers
4496 if ( $numVisible < 1 ) {
4497 $enoughToc = false;
4498 }
4499
4500 if ( $enoughToc ) {
4501 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4502 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4503 }
4504 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4505 $this->mOutput->setTOCHTML( $toc );
4506 $toc = self::TOC_START . $toc . self::TOC_END;
4507 }
4508
4509 if ( $isMain ) {
4510 $this->mOutput->setSections( $tocraw );
4511 }
4512
4513 # split up and insert constructed headlines
4514 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4515 $i = 0;
4516
4517 // build an array of document sections
4518 $sections = [];
4519 foreach ( $blocks as $block ) {
4520 // $head is zero-based, sections aren't.
4521 if ( empty( $head[$i - 1] ) ) {
4522 $sections[$i] = $block;
4523 } else {
4524 $sections[$i] = $head[$i - 1] . $block;
4525 }
4526
4527 /**
4528 * Send a hook, one per section.
4529 * The idea here is to be able to make section-level DIVs, but to do so in a
4530 * lower-impact, more correct way than r50769
4531 *
4532 * $this : caller
4533 * $section : the section number
4534 * &$sectionContent : ref to the content of the section
4535 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4536 */
4537 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4538
4539 $i++;
4540 }
4541
4542 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4543 // append the TOC at the beginning
4544 // Top anchor now in skin
4545 $sections[0] .= $toc . "\n";
4546 }
4547
4548 $full .= implode( '', $sections );
4549
4550 if ( $this->mForceTocPosition ) {
4551 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4552 } else {
4553 return $full;
4554 }
4555 }
4556
4557 /**
4558 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4559 * conversion, substituting signatures, {{subst:}} templates, etc.
4560 *
4561 * @param string $text The text to transform
4562 * @param Title $title The Title object for the current article
4563 * @param User $user The User object describing the current user
4564 * @param ParserOptions $options Parsing options
4565 * @param bool $clearState Whether to clear the parser state first
4566 * @return string The altered wiki markup
4567 */
4568 public function preSaveTransform( $text, Title $title, User $user,
4569 ParserOptions $options, $clearState = true
4570 ) {
4571 if ( $clearState ) {
4572 $magicScopeVariable = $this->lock();
4573 }
4574 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4575 $this->setUser( $user );
4576
4577 // Strip U+0000 NULL (T159174)
4578 $text = str_replace( "\000", '', $text );
4579
4580 // We still normalize line endings for backwards-compatibility
4581 // with other code that just calls PST, but this should already
4582 // be handled in TextContent subclasses
4583 $text = TextContent::normalizeLineEndings( $text );
4584
4585 if ( $options->getPreSaveTransform() ) {
4586 $text = $this->pstPass2( $text, $user );
4587 }
4588 $text = $this->mStripState->unstripBoth( $text );
4589
4590 $this->setUser( null ); # Reset
4591
4592 return $text;
4593 }
4594
4595 /**
4596 * Pre-save transform helper function
4597 *
4598 * @param string $text
4599 * @param User $user
4600 *
4601 * @return string
4602 */
4603 private function pstPass2( $text, $user ) {
4604 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4605 # $this->contLang here in order to give everyone the same signature and use the default one
4606 # rather than the one selected in each user's preferences. (see also T14815)
4607 $ts = $this->mOptions->getTimestamp();
4608 $timestamp = MWTimestamp::getLocalInstance( $ts );
4609 $ts = $timestamp->format( 'YmdHis' );
4610 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4611
4612 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4613
4614 # Variable replacement
4615 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4616 $text = $this->replaceVariables( $text );
4617
4618 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4619 # which may corrupt this parser instance via its wfMessage()->text() call-
4620
4621 # Signatures
4622 if ( strpos( $text, '~~~' ) !== false ) {
4623 $sigText = $this->getUserSig( $user );
4624 $text = strtr( $text, [
4625 '~~~~~' => $d,
4626 '~~~~' => "$sigText $d",
4627 '~~~' => $sigText
4628 ] );
4629 # The main two signature forms used above are time-sensitive
4630 $this->mOutput->setFlag( 'user-signature' );
4631 }
4632
4633 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4634 $tc = '[' . Title::legalChars() . ']';
4635 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4636
4637 // [[ns:page (context)|]]
4638 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4639 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4640 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4641 // [[ns:page (context), context|]] (using either single or double-width comma)
4642 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4643 // [[|page]] (reverse pipe trick: add context from page title)
4644 $p2 = "/\[\[\\|($tc+)]]/";
4645
4646 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4647 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4648 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4649 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4650
4651 $t = $this->mTitle->getText();
4652 $m = [];
4653 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4654 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4655 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4656 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4657 } else {
4658 # if there's no context, don't bother duplicating the title
4659 $text = preg_replace( $p2, '[[\\1]]', $text );
4660 }
4661
4662 return $text;
4663 }
4664
4665 /**
4666 * Fetch the user's signature text, if any, and normalize to
4667 * validated, ready-to-insert wikitext.
4668 * If you have pre-fetched the nickname or the fancySig option, you can
4669 * specify them here to save a database query.
4670 * Do not reuse this parser instance after calling getUserSig(),
4671 * as it may have changed.
4672 *
4673 * @param User &$user
4674 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4675 * @param bool|null $fancySig whether the nicknname is the complete signature
4676 * or null to use default value
4677 * @return string
4678 */
4679 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4680 $username = $user->getName();
4681
4682 # If not given, retrieve from the user object.
4683 if ( $nickname === false ) {
4684 $nickname = $user->getOption( 'nickname' );
4685 }
4686
4687 if ( is_null( $fancySig ) ) {
4688 $fancySig = $user->getBoolOption( 'fancysig' );
4689 }
4690
4691 $nickname = $nickname == null ? $username : $nickname;
4692
4693 if ( mb_strlen( $nickname ) > $this->siteConfig->get( 'MaxSigChars' ) ) {
4694 $nickname = $username;
4695 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4696 } elseif ( $fancySig !== false ) {
4697 # Sig. might contain markup; validate this
4698 if ( $this->validateSig( $nickname ) !== false ) {
4699 # Validated; clean up (if needed) and return it
4700 return $this->cleanSig( $nickname, true );
4701 } else {
4702 # Failed to validate; fall back to the default
4703 $nickname = $username;
4704 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4705 }
4706 }
4707
4708 # Make sure nickname doesnt get a sig in a sig
4709 $nickname = self::cleanSigInSig( $nickname );
4710
4711 # If we're still here, make it a link to the user page
4712 $userText = wfEscapeWikiText( $username );
4713 $nickText = wfEscapeWikiText( $nickname );
4714 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4715
4716 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4717 ->title( $this->getTitle() )->text();
4718 }
4719
4720 /**
4721 * Check that the user's signature contains no bad XML
4722 *
4723 * @param string $text
4724 * @return string|bool An expanded string, or false if invalid.
4725 */
4726 public function validateSig( $text ) {
4727 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4728 }
4729
4730 /**
4731 * Clean up signature text
4732 *
4733 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4734 * 2) Substitute all transclusions
4735 *
4736 * @param string $text
4737 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4738 * @return string Signature text
4739 */
4740 public function cleanSig( $text, $parsing = false ) {
4741 if ( !$parsing ) {
4742 global $wgTitle;
4743 $magicScopeVariable = $this->lock();
4744 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4745 }
4746
4747 # Option to disable this feature
4748 if ( !$this->mOptions->getCleanSignatures() ) {
4749 return $text;
4750 }
4751
4752 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4753 # => Move this logic to braceSubstitution()
4754 $substWord = $this->magicWordFactory->get( 'subst' );
4755 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4756 $substText = '{{' . $substWord->getSynonym( 0 );
4757
4758 $text = preg_replace( $substRegex, $substText, $text );
4759 $text = self::cleanSigInSig( $text );
4760 $dom = $this->preprocessToDom( $text );
4761 $frame = $this->getPreprocessor()->newFrame();
4762 $text = $frame->expand( $dom );
4763
4764 if ( !$parsing ) {
4765 $text = $this->mStripState->unstripBoth( $text );
4766 }
4767
4768 return $text;
4769 }
4770
4771 /**
4772 * Strip 3, 4 or 5 tildes out of signatures.
4773 *
4774 * @param string $text
4775 * @return string Signature text with /~{3,5}/ removed
4776 */
4777 public static function cleanSigInSig( $text ) {
4778 $text = preg_replace( '/~{3,5}/', '', $text );
4779 return $text;
4780 }
4781
4782 /**
4783 * Set up some variables which are usually set up in parse()
4784 * so that an external function can call some class members with confidence
4785 *
4786 * @param Title|null $title
4787 * @param ParserOptions $options
4788 * @param int $outputType
4789 * @param bool $clearState
4790 */
4791 public function startExternalParse( Title $title = null, ParserOptions $options,
4792 $outputType, $clearState = true
4793 ) {
4794 $this->startParse( $title, $options, $outputType, $clearState );
4795 }
4796
4797 /**
4798 * @param Title|null $title
4799 * @param ParserOptions $options
4800 * @param int $outputType
4801 * @param bool $clearState
4802 */
4803 private function startParse( Title $title = null, ParserOptions $options,
4804 $outputType, $clearState = true
4805 ) {
4806 $this->setTitle( $title );
4807 $this->mOptions = $options;
4808 $this->setOutputType( $outputType );
4809 if ( $clearState ) {
4810 $this->clearState();
4811 }
4812 }
4813
4814 /**
4815 * Wrapper for preprocess()
4816 *
4817 * @param string $text The text to preprocess
4818 * @param ParserOptions $options
4819 * @param Title|null $title Title object or null to use $wgTitle
4820 * @return string
4821 */
4822 public function transformMsg( $text, $options, $title = null ) {
4823 static $executing = false;
4824
4825 # Guard against infinite recursion
4826 if ( $executing ) {
4827 return $text;
4828 }
4829 $executing = true;
4830
4831 if ( !$title ) {
4832 global $wgTitle;
4833 $title = $wgTitle;
4834 }
4835
4836 $text = $this->preprocess( $text, $title, $options );
4837
4838 $executing = false;
4839 return $text;
4840 }
4841
4842 /**
4843 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4844 * The callback should have the following form:
4845 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4846 *
4847 * Transform and return $text. Use $parser for any required context, e.g. use
4848 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4849 *
4850 * Hooks may return extended information by returning an array, of which the
4851 * first numbered element (index 0) must be the return string, and all other
4852 * entries are extracted into local variables within an internal function
4853 * in the Parser class.
4854 *
4855 * This interface (introduced r61913) appears to be undocumented, but
4856 * 'markerType' is used by some core tag hooks to override which strip
4857 * array their results are placed in. **Use great caution if attempting
4858 * this interface, as it is not documented and injudicious use could smash
4859 * private variables.**
4860 *
4861 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4862 * @param callable $callback The callback function (and object) to use for the tag
4863 * @throws MWException
4864 * @return callable|null The old value of the mTagHooks array associated with the hook
4865 */
4866 public function setHook( $tag, callable $callback ) {
4867 $tag = strtolower( $tag );
4868 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4869 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4870 }
4871 $oldVal = $this->mTagHooks[$tag] ?? null;
4872 $this->mTagHooks[$tag] = $callback;
4873 if ( !in_array( $tag, $this->mStripList ) ) {
4874 $this->mStripList[] = $tag;
4875 }
4876
4877 return $oldVal;
4878 }
4879
4880 /**
4881 * As setHook(), but letting the contents be parsed.
4882 *
4883 * Transparent tag hooks are like regular XML-style tag hooks, except they
4884 * operate late in the transformation sequence, on HTML instead of wikitext.
4885 *
4886 * This is probably obsoleted by things dealing with parser frames?
4887 * The only extension currently using it is geoserver.
4888 *
4889 * @since 1.10
4890 * @todo better document or deprecate this
4891 *
4892 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4893 * @param callable $callback The callback function (and object) to use for the tag
4894 * @throws MWException
4895 * @return callable|null The old value of the mTagHooks array associated with the hook
4896 */
4897 public function setTransparentTagHook( $tag, callable $callback ) {
4898 $tag = strtolower( $tag );
4899 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4900 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4901 }
4902 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4903 $this->mTransparentTagHooks[$tag] = $callback;
4904
4905 return $oldVal;
4906 }
4907
4908 /**
4909 * Remove all tag hooks
4910 */
4911 public function clearTagHooks() {
4912 $this->mTagHooks = [];
4913 $this->mFunctionTagHooks = [];
4914 $this->mStripList = $this->mDefaultStripList;
4915 }
4916
4917 /**
4918 * Create a function, e.g. {{sum:1|2|3}}
4919 * The callback function should have the form:
4920 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4921 *
4922 * Or with Parser::SFH_OBJECT_ARGS:
4923 * function myParserFunction( $parser, $frame, $args ) { ... }
4924 *
4925 * The callback may either return the text result of the function, or an array with the text
4926 * in element 0, and a number of flags in the other elements. The names of the flags are
4927 * specified in the keys. Valid flags are:
4928 * found The text returned is valid, stop processing the template. This
4929 * is on by default.
4930 * nowiki Wiki markup in the return value should be escaped
4931 * isHTML The returned text is HTML, armour it against wikitext transformation
4932 *
4933 * @param string $id The magic word ID
4934 * @param callable $callback The callback function (and object) to use
4935 * @param int $flags A combination of the following flags:
4936 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4937 *
4938 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4939 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4940 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4941 * the arguments, and to control the way they are expanded.
4942 *
4943 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4944 * arguments, for instance:
4945 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4946 *
4947 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4948 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4949 * working if/when this is changed.
4950 *
4951 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4952 * expansion.
4953 *
4954 * Please read the documentation in includes/parser/Preprocessor.php for more information
4955 * about the methods available in PPFrame and PPNode.
4956 *
4957 * @throws MWException
4958 * @return string|callable The old callback function for this name, if any
4959 */
4960 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4961 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4962 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4963
4964 # Add to function cache
4965 $mw = $this->magicWordFactory->get( $id );
4966 if ( !$mw ) {
4967 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4968 }
4969
4970 $synonyms = $mw->getSynonyms();
4971 $sensitive = intval( $mw->isCaseSensitive() );
4972
4973 foreach ( $synonyms as $syn ) {
4974 # Case
4975 if ( !$sensitive ) {
4976 $syn = $this->contLang->lc( $syn );
4977 }
4978 # Add leading hash
4979 if ( !( $flags & self::SFH_NO_HASH ) ) {
4980 $syn = '#' . $syn;
4981 }
4982 # Remove trailing colon
4983 if ( substr( $syn, -1, 1 ) === ':' ) {
4984 $syn = substr( $syn, 0, -1 );
4985 }
4986 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4987 }
4988 return $oldVal;
4989 }
4990
4991 /**
4992 * Get all registered function hook identifiers
4993 *
4994 * @return array
4995 */
4996 public function getFunctionHooks() {
4997 $this->firstCallInit();
4998 return array_keys( $this->mFunctionHooks );
4999 }
5000
5001 /**
5002 * Create a tag function, e.g. "<test>some stuff</test>".
5003 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5004 * Unlike parser functions, their content is not preprocessed.
5005 * @param string $tag
5006 * @param callable $callback
5007 * @param int $flags
5008 * @throws MWException
5009 * @return null
5010 */
5011 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
5012 $tag = strtolower( $tag );
5013 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5014 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5015 }
5016 $old = $this->mFunctionTagHooks[$tag] ?? null;
5017 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
5018
5019 if ( !in_array( $tag, $this->mStripList ) ) {
5020 $this->mStripList[] = $tag;
5021 }
5022
5023 return $old;
5024 }
5025
5026 /**
5027 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5028 * Placeholders created in Linker::link()
5029 *
5030 * @param string &$text
5031 * @param int $options
5032 */
5033 public function replaceLinkHolders( &$text, $options = 0 ) {
5034 $this->mLinkHolders->replace( $text );
5035 }
5036
5037 /**
5038 * Replace "<!--LINK-->" link placeholders with plain text of links
5039 * (not HTML-formatted).
5040 *
5041 * @param string $text
5042 * @return string
5043 */
5044 public function replaceLinkHoldersText( $text ) {
5045 return $this->mLinkHolders->replaceText( $text );
5046 }
5047
5048 /**
5049 * Renders an image gallery from a text with one line per image.
5050 * text labels may be given by using |-style alternative text. E.g.
5051 * Image:one.jpg|The number "1"
5052 * Image:tree.jpg|A tree
5053 * given as text will return the HTML of a gallery with two images,
5054 * labeled 'The number "1"' and
5055 * 'A tree'.
5056 *
5057 * @param string $text
5058 * @param array $params
5059 * @return string HTML
5060 */
5061 public function renderImageGallery( $text, $params ) {
5062 $mode = false;
5063 if ( isset( $params['mode'] ) ) {
5064 $mode = $params['mode'];
5065 }
5066
5067 try {
5068 $ig = ImageGalleryBase::factory( $mode );
5069 } catch ( Exception $e ) {
5070 // If invalid type set, fallback to default.
5071 $ig = ImageGalleryBase::factory( false );
5072 }
5073
5074 $ig->setContextTitle( $this->mTitle );
5075 $ig->setShowBytes( false );
5076 $ig->setShowDimensions( false );
5077 $ig->setShowFilename( false );
5078 $ig->setParser( $this );
5079 $ig->setHideBadImages();
5080 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5081
5082 if ( isset( $params['showfilename'] ) ) {
5083 $ig->setShowFilename( true );
5084 } else {
5085 $ig->setShowFilename( false );
5086 }
5087 if ( isset( $params['caption'] ) ) {
5088 // NOTE: We aren't passing a frame here or below. Frame info
5089 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5090 // See T107332#4030581
5091 $caption = $this->recursiveTagParse( $params['caption'] );
5092 $ig->setCaptionHtml( $caption );
5093 }
5094 if ( isset( $params['perrow'] ) ) {
5095 $ig->setPerRow( $params['perrow'] );
5096 }
5097 if ( isset( $params['widths'] ) ) {
5098 $ig->setWidths( $params['widths'] );
5099 }
5100 if ( isset( $params['heights'] ) ) {
5101 $ig->setHeights( $params['heights'] );
5102 }
5103 $ig->setAdditionalOptions( $params );
5104
5105 // Avoid PHP 7.1 warning from passing $this by reference
5106 $parser = $this;
5107 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5108
5109 $lines = StringUtils::explode( "\n", $text );
5110 foreach ( $lines as $line ) {
5111 # match lines like these:
5112 # Image:someimage.jpg|This is some image
5113 $matches = [];
5114 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5115 # Skip empty lines
5116 if ( count( $matches ) == 0 ) {
5117 continue;
5118 }
5119
5120 if ( strpos( $matches[0], '%' ) !== false ) {
5121 $matches[1] = rawurldecode( $matches[1] );
5122 }
5123 $title = Title::newFromText( $matches[1], NS_FILE );
5124 if ( is_null( $title ) ) {
5125 # Bogus title. Ignore these so we don't bomb out later.
5126 continue;
5127 }
5128
5129 # We need to get what handler the file uses, to figure out parameters.
5130 # Note, a hook can overide the file name, and chose an entirely different
5131 # file (which potentially could be of a different type and have different handler).
5132 $options = [];
5133 $descQuery = false;
5134 Hooks::run( 'BeforeParserFetchFileAndTitle',
5135 [ $this, $title, &$options, &$descQuery ] );
5136 # Don't register it now, as TraditionalImageGallery does that later.
5137 $file = $this->fetchFileNoRegister( $title, $options );
5138 $handler = $file ? $file->getHandler() : false;
5139
5140 $paramMap = [
5141 'img_alt' => 'gallery-internal-alt',
5142 'img_link' => 'gallery-internal-link',
5143 ];
5144 if ( $handler ) {
5145 $paramMap += $handler->getParamMap();
5146 // We don't want people to specify per-image widths.
5147 // Additionally the width parameter would need special casing anyhow.
5148 unset( $paramMap['img_width'] );
5149 }
5150
5151 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5152
5153 $label = '';
5154 $alt = '';
5155 $link = '';
5156 $handlerOptions = [];
5157 if ( isset( $matches[3] ) ) {
5158 // look for an |alt= definition while trying not to break existing
5159 // captions with multiple pipes (|) in it, until a more sensible grammar
5160 // is defined for images in galleries
5161
5162 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5163 // splitting on '|' is a bit odd, and different from makeImage.
5164 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5165 // Protect LanguageConverter markup
5166 $parameterMatches = StringUtils::delimiterExplode(
5167 '-{', '}-', '|', $matches[3], true /* nested */
5168 );
5169
5170 foreach ( $parameterMatches as $parameterMatch ) {
5171 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5172 if ( $magicName ) {
5173 $paramName = $paramMap[$magicName];
5174
5175 switch ( $paramName ) {
5176 case 'gallery-internal-alt':
5177 $alt = $this->stripAltText( $match, false );
5178 break;
5179 case 'gallery-internal-link':
5180 $linkValue = $this->stripAltText( $match, false );
5181 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5182 // Result of LanguageConverter::markNoConversion
5183 // invoked on an external link.
5184 $linkValue = substr( $linkValue, 4, -2 );
5185 }
5186 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5187 if ( $type === 'link-url' ) {
5188 $link = $target;
5189 $this->mOutput->addExternalLink( $target );
5190 } elseif ( $type === 'link-title' ) {
5191 $link = $target->getLinkURL();
5192 $this->mOutput->addLink( $target );
5193 }
5194 break;
5195 default:
5196 // Must be a handler specific parameter.
5197 if ( $handler->validateParam( $paramName, $match ) ) {
5198 $handlerOptions[$paramName] = $match;
5199 } else {
5200 // Guess not, consider it as caption.
5201 wfDebug( "$parameterMatch failed parameter validation\n" );
5202 $label = $parameterMatch;
5203 }
5204 }
5205
5206 } else {
5207 // Last pipe wins.
5208 $label = $parameterMatch;
5209 }
5210 }
5211 }
5212
5213 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5214 }
5215 $html = $ig->toHTML();
5216 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5217 return $html;
5218 }
5219
5220 /**
5221 * @param MediaHandler $handler
5222 * @return array
5223 */
5224 public function getImageParams( $handler ) {
5225 if ( $handler ) {
5226 $handlerClass = get_class( $handler );
5227 } else {
5228 $handlerClass = '';
5229 }
5230 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5231 # Initialise static lists
5232 static $internalParamNames = [
5233 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5234 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5235 'bottom', 'text-bottom' ],
5236 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5237 'upright', 'border', 'link', 'alt', 'class' ],
5238 ];
5239 static $internalParamMap;
5240 if ( !$internalParamMap ) {
5241 $internalParamMap = [];
5242 foreach ( $internalParamNames as $type => $names ) {
5243 foreach ( $names as $name ) {
5244 // For grep: img_left, img_right, img_center, img_none,
5245 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5246 // img_bottom, img_text_bottom,
5247 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5248 // img_border, img_link, img_alt, img_class
5249 $magicName = str_replace( '-', '_', "img_$name" );
5250 $internalParamMap[$magicName] = [ $type, $name ];
5251 }
5252 }
5253 }
5254
5255 # Add handler params
5256 $paramMap = $internalParamMap;
5257 if ( $handler ) {
5258 $handlerParamMap = $handler->getParamMap();
5259 foreach ( $handlerParamMap as $magic => $paramName ) {
5260 $paramMap[$magic] = [ 'handler', $paramName ];
5261 }
5262 }
5263 $this->mImageParams[$handlerClass] = $paramMap;
5264 $this->mImageParamsMagicArray[$handlerClass] =
5265 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5266 }
5267 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5268 }
5269
5270 /**
5271 * Parse image options text and use it to make an image
5272 *
5273 * @param Title $title
5274 * @param string $options
5275 * @param LinkHolderArray|bool $holders
5276 * @return string HTML
5277 */
5278 public function makeImage( $title, $options, $holders = false ) {
5279 # Check if the options text is of the form "options|alt text"
5280 # Options are:
5281 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5282 # * left no resizing, just left align. label is used for alt= only
5283 # * right same, but right aligned
5284 # * none same, but not aligned
5285 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5286 # * center center the image
5287 # * frame Keep original image size, no magnify-button.
5288 # * framed Same as "frame"
5289 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5290 # * upright reduce width for upright images, rounded to full __0 px
5291 # * border draw a 1px border around the image
5292 # * alt Text for HTML alt attribute (defaults to empty)
5293 # * class Set a class for img node
5294 # * link Set the target of the image link. Can be external, interwiki, or local
5295 # vertical-align values (no % or length right now):
5296 # * baseline
5297 # * sub
5298 # * super
5299 # * top
5300 # * text-top
5301 # * middle
5302 # * bottom
5303 # * text-bottom
5304
5305 # Protect LanguageConverter markup when splitting into parts
5306 $parts = StringUtils::delimiterExplode(
5307 '-{', '}-', '|', $options, true /* allow nesting */
5308 );
5309
5310 # Give extensions a chance to select the file revision for us
5311 $options = [];
5312 $descQuery = false;
5313 Hooks::run( 'BeforeParserFetchFileAndTitle',
5314 [ $this, $title, &$options, &$descQuery ] );
5315 # Fetch and register the file (file title may be different via hooks)
5316 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5317
5318 # Get parameter map
5319 $handler = $file ? $file->getHandler() : false;
5320
5321 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5322
5323 if ( !$file ) {
5324 $this->addTrackingCategory( 'broken-file-category' );
5325 }
5326
5327 # Process the input parameters
5328 $caption = '';
5329 $params = [ 'frame' => [], 'handler' => [],
5330 'horizAlign' => [], 'vertAlign' => [] ];
5331 $seenformat = false;
5332 foreach ( $parts as $part ) {
5333 $part = trim( $part );
5334 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5335 $validated = false;
5336 if ( isset( $paramMap[$magicName] ) ) {
5337 list( $type, $paramName ) = $paramMap[$magicName];
5338
5339 # Special case; width and height come in one variable together
5340 if ( $type === 'handler' && $paramName === 'width' ) {
5341 $parsedWidthParam = self::parseWidthParam( $value );
5342 if ( isset( $parsedWidthParam['width'] ) ) {
5343 $width = $parsedWidthParam['width'];
5344 if ( $handler->validateParam( 'width', $width ) ) {
5345 $params[$type]['width'] = $width;
5346 $validated = true;
5347 }
5348 }
5349 if ( isset( $parsedWidthParam['height'] ) ) {
5350 $height = $parsedWidthParam['height'];
5351 if ( $handler->validateParam( 'height', $height ) ) {
5352 $params[$type]['height'] = $height;
5353 $validated = true;
5354 }
5355 }
5356 # else no validation -- T15436
5357 } else {
5358 if ( $type === 'handler' ) {
5359 # Validate handler parameter
5360 $validated = $handler->validateParam( $paramName, $value );
5361 } else {
5362 # Validate internal parameters
5363 switch ( $paramName ) {
5364 case 'manualthumb':
5365 case 'alt':
5366 case 'class':
5367 # @todo FIXME: Possibly check validity here for
5368 # manualthumb? downstream behavior seems odd with
5369 # missing manual thumbs.
5370 $validated = true;
5371 $value = $this->stripAltText( $value, $holders );
5372 break;
5373 case 'link':
5374 list( $paramName, $value ) =
5375 $this->parseLinkParameter(
5376 $this->stripAltText( $value, $holders )
5377 );
5378 if ( $paramName ) {
5379 $validated = true;
5380 if ( $paramName === 'no-link' ) {
5381 $value = true;
5382 }
5383 if ( ( $paramName === 'link-url' ) && $this->mOptions->getExternalLinkTarget() ) {
5384 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5385 }
5386 }
5387 break;
5388 case 'frameless':
5389 case 'framed':
5390 case 'thumbnail':
5391 // use first appearing option, discard others.
5392 $validated = !$seenformat;
5393 $seenformat = true;
5394 break;
5395 default:
5396 # Most other things appear to be empty or numeric...
5397 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5398 }
5399 }
5400
5401 if ( $validated ) {
5402 $params[$type][$paramName] = $value;
5403 }
5404 }
5405 }
5406 if ( !$validated ) {
5407 $caption = $part;
5408 }
5409 }
5410
5411 # Process alignment parameters
5412 if ( $params['horizAlign'] ) {
5413 $params['frame']['align'] = key( $params['horizAlign'] );
5414 }
5415 if ( $params['vertAlign'] ) {
5416 $params['frame']['valign'] = key( $params['vertAlign'] );
5417 }
5418
5419 $params['frame']['caption'] = $caption;
5420
5421 # Will the image be presented in a frame, with the caption below?
5422 $imageIsFramed = isset( $params['frame']['frame'] )
5423 || isset( $params['frame']['framed'] )
5424 || isset( $params['frame']['thumbnail'] )
5425 || isset( $params['frame']['manualthumb'] );
5426
5427 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5428 # came to also set the caption, ordinary text after the image -- which
5429 # makes no sense, because that just repeats the text multiple times in
5430 # screen readers. It *also* came to set the title attribute.
5431 # Now that we have an alt attribute, we should not set the alt text to
5432 # equal the caption: that's worse than useless, it just repeats the
5433 # text. This is the framed/thumbnail case. If there's no caption, we
5434 # use the unnamed parameter for alt text as well, just for the time be-
5435 # ing, if the unnamed param is set and the alt param is not.
5436 # For the future, we need to figure out if we want to tweak this more,
5437 # e.g., introducing a title= parameter for the title; ignoring the un-
5438 # named parameter entirely for images without a caption; adding an ex-
5439 # plicit caption= parameter and preserving the old magic unnamed para-
5440 # meter for BC; ...
5441 if ( $imageIsFramed ) { # Framed image
5442 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5443 # No caption or alt text, add the filename as the alt text so
5444 # that screen readers at least get some description of the image
5445 $params['frame']['alt'] = $title->getText();
5446 }
5447 # Do not set $params['frame']['title'] because tooltips don't make sense
5448 # for framed images
5449 } else { # Inline image
5450 if ( !isset( $params['frame']['alt'] ) ) {
5451 # No alt text, use the "caption" for the alt text
5452 if ( $caption !== '' ) {
5453 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5454 } else {
5455 # No caption, fall back to using the filename for the
5456 # alt text
5457 $params['frame']['alt'] = $title->getText();
5458 }
5459 }
5460 # Use the "caption" for the tooltip text
5461 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5462 }
5463 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5464
5465 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5466
5467 # Linker does the rest
5468 $time = $options['time'] ?? false;
5469 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5470 $time, $descQuery, $this->mOptions->getThumbSize() );
5471
5472 # Give the handler a chance to modify the parser object
5473 if ( $handler ) {
5474 $handler->parserTransformHook( $this, $file );
5475 }
5476
5477 return $ret;
5478 }
5479
5480 /**
5481 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5482 *
5483 * Adds an entry to appropriate link tables.
5484 *
5485 * @since 1.32
5486 * @param string $value
5487 * @return array of `[ type, target ]`, where:
5488 * - `type` is one of:
5489 * - `null`: Given value is not a valid link target, use default
5490 * - `'no-link'`: Given value is empty, do not generate a link
5491 * - `'link-url'`: Given value is a valid external link
5492 * - `'link-title'`: Given value is a valid internal link
5493 * - `target` is:
5494 * - When `type` is `null` or `'no-link'`: `false`
5495 * - When `type` is `'link-url'`: URL string corresponding to given value
5496 * - When `type` is `'link-title'`: Title object corresponding to given value
5497 */
5498 public function parseLinkParameter( $value ) {
5499 $chars = self::EXT_LINK_URL_CLASS;
5500 $addr = self::EXT_LINK_ADDR;
5501 $prots = $this->mUrlProtocols;
5502 $type = null;
5503 $target = false;
5504 if ( $value === '' ) {
5505 $type = 'no-link';
5506 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5507 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5508 $this->mOutput->addExternalLink( $value );
5509 $type = 'link-url';
5510 $target = $value;
5511 }
5512 } else {
5513 $linkTitle = Title::newFromText( $value );
5514 if ( $linkTitle ) {
5515 $this->mOutput->addLink( $linkTitle );
5516 $type = 'link-title';
5517 $target = $linkTitle;
5518 }
5519 }
5520 return [ $type, $target ];
5521 }
5522
5523 /**
5524 * @param string $caption
5525 * @param LinkHolderArray|bool $holders
5526 * @return mixed|string
5527 */
5528 protected function stripAltText( $caption, $holders ) {
5529 # Strip bad stuff out of the title (tooltip). We can't just use
5530 # replaceLinkHoldersText() here, because if this function is called
5531 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5532 if ( $holders ) {
5533 $tooltip = $holders->replaceText( $caption );
5534 } else {
5535 $tooltip = $this->replaceLinkHoldersText( $caption );
5536 }
5537
5538 # make sure there are no placeholders in thumbnail attributes
5539 # that are later expanded to html- so expand them now and
5540 # remove the tags
5541 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5542 # Compatibility hack! In HTML certain entity references not terminated
5543 # by a semicolon are decoded (but not if we're in an attribute; that's
5544 # how link URLs get away without properly escaping & in queries).
5545 # But wikitext has always required semicolon-termination of entities,
5546 # so encode & where needed to avoid decode of semicolon-less entities.
5547 # See T209236 and
5548 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5549 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5550 $tooltip = preg_replace( "/
5551 & # 1. entity prefix
5552 (?= # 2. followed by:
5553 (?: # a. one of the legacy semicolon-less named entities
5554 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5555 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5556 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5557 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5558 U(?:acute|circ|grave|uml)|Yacute|
5559 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5560 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5561 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5562 frac(?:1(?:2|4)|34)|
5563 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5564 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5565 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5566 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5567 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5568 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5569 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5570 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5571 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5572 )
5573 (?:[^;]|$)) # b. and not followed by a semicolon
5574 # S = study, for efficiency
5575 /Sx", '&amp;', $tooltip );
5576 $tooltip = Sanitizer::stripAllTags( $tooltip );
5577
5578 return $tooltip;
5579 }
5580
5581 /**
5582 * Set a flag in the output object indicating that the content is dynamic and
5583 * shouldn't be cached.
5584 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5585 */
5586 public function disableCache() {
5587 wfDebug( "Parser output marked as uncacheable.\n" );
5588 if ( !$this->mOutput ) {
5589 throw new MWException( __METHOD__ .
5590 " can only be called when actually parsing something" );
5591 }
5592 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5593 }
5594
5595 /**
5596 * Callback from the Sanitizer for expanding items found in HTML attribute
5597 * values, so they can be safely tested and escaped.
5598 *
5599 * @param string &$text
5600 * @param bool|PPFrame $frame
5601 * @return string
5602 */
5603 public function attributeStripCallback( &$text, $frame = false ) {
5604 $text = $this->replaceVariables( $text, $frame );
5605 $text = $this->mStripState->unstripBoth( $text );
5606 return $text;
5607 }
5608
5609 /**
5610 * Accessor
5611 *
5612 * @return array
5613 */
5614 public function getTags() {
5615 $this->firstCallInit();
5616 return array_merge(
5617 array_keys( $this->mTransparentTagHooks ),
5618 array_keys( $this->mTagHooks ),
5619 array_keys( $this->mFunctionTagHooks )
5620 );
5621 }
5622
5623 /**
5624 * @since 1.32
5625 * @return array
5626 */
5627 public function getFunctionSynonyms() {
5628 $this->firstCallInit();
5629 return $this->mFunctionSynonyms;
5630 }
5631
5632 /**
5633 * @since 1.32
5634 * @return string
5635 */
5636 public function getUrlProtocols() {
5637 return $this->mUrlProtocols;
5638 }
5639
5640 /**
5641 * Replace transparent tags in $text with the values given by the callbacks.
5642 *
5643 * Transparent tag hooks are like regular XML-style tag hooks, except they
5644 * operate late in the transformation sequence, on HTML instead of wikitext.
5645 *
5646 * @param string $text
5647 *
5648 * @return string
5649 */
5650 public function replaceTransparentTags( $text ) {
5651 $matches = [];
5652 $elements = array_keys( $this->mTransparentTagHooks );
5653 $text = self::extractTagsAndParams( $elements, $text, $matches );
5654 $replacements = [];
5655
5656 foreach ( $matches as $marker => $data ) {
5657 list( $element, $content, $params, $tag ) = $data;
5658 $tagName = strtolower( $element );
5659 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5660 $output = call_user_func_array(
5661 $this->mTransparentTagHooks[$tagName],
5662 [ $content, $params, $this ]
5663 );
5664 } else {
5665 $output = $tag;
5666 }
5667 $replacements[$marker] = $output;
5668 }
5669 return strtr( $text, $replacements );
5670 }
5671
5672 /**
5673 * Break wikitext input into sections, and either pull or replace
5674 * some particular section's text.
5675 *
5676 * External callers should use the getSection and replaceSection methods.
5677 *
5678 * @param string $text Page wikitext
5679 * @param string|int $sectionId A section identifier string of the form:
5680 * "<flag1> - <flag2> - ... - <section number>"
5681 *
5682 * Currently the only recognised flag is "T", which means the target section number
5683 * was derived during a template inclusion parse, in other words this is a template
5684 * section edit link. If no flags are given, it was an ordinary section edit link.
5685 * This flag is required to avoid a section numbering mismatch when a section is
5686 * enclosed by "<includeonly>" (T8563).
5687 *
5688 * The section number 0 pulls the text before the first heading; other numbers will
5689 * pull the given section along with its lower-level subsections. If the section is
5690 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5691 *
5692 * Section 0 is always considered to exist, even if it only contains the empty
5693 * string. If $text is the empty string and section 0 is replaced, $newText is
5694 * returned.
5695 *
5696 * @param string $mode One of "get" or "replace"
5697 * @param string $newText Replacement text for section data.
5698 * @return string For "get", the extracted section text.
5699 * for "replace", the whole page with the section replaced.
5700 */
5701 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5702 global $wgTitle; # not generally used but removes an ugly failure mode
5703
5704 $magicScopeVariable = $this->lock();
5705 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5706 $outText = '';
5707 $frame = $this->getPreprocessor()->newFrame();
5708
5709 # Process section extraction flags
5710 $flags = 0;
5711 $sectionParts = explode( '-', $sectionId );
5712 $sectionIndex = array_pop( $sectionParts );
5713 foreach ( $sectionParts as $part ) {
5714 if ( $part === 'T' ) {
5715 $flags |= self::PTD_FOR_INCLUSION;
5716 }
5717 }
5718
5719 # Check for empty input
5720 if ( strval( $text ) === '' ) {
5721 # Only sections 0 and T-0 exist in an empty document
5722 if ( $sectionIndex == 0 ) {
5723 if ( $mode === 'get' ) {
5724 return '';
5725 }
5726
5727 return $newText;
5728 } else {
5729 if ( $mode === 'get' ) {
5730 return $newText;
5731 }
5732
5733 return $text;
5734 }
5735 }
5736
5737 # Preprocess the text
5738 $root = $this->preprocessToDom( $text, $flags );
5739
5740 # <h> nodes indicate section breaks
5741 # They can only occur at the top level, so we can find them by iterating the root's children
5742 $node = $root->getFirstChild();
5743
5744 # Find the target section
5745 if ( $sectionIndex == 0 ) {
5746 # Section zero doesn't nest, level=big
5747 $targetLevel = 1000;
5748 } else {
5749 while ( $node ) {
5750 if ( $node->getName() === 'h' ) {
5751 $bits = $node->splitHeading();
5752 if ( $bits['i'] == $sectionIndex ) {
5753 $targetLevel = $bits['level'];
5754 break;
5755 }
5756 }
5757 if ( $mode === 'replace' ) {
5758 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5759 }
5760 $node = $node->getNextSibling();
5761 }
5762 }
5763
5764 if ( !$node ) {
5765 # Not found
5766 if ( $mode === 'get' ) {
5767 return $newText;
5768 } else {
5769 return $text;
5770 }
5771 }
5772
5773 # Find the end of the section, including nested sections
5774 do {
5775 if ( $node->getName() === 'h' ) {
5776 $bits = $node->splitHeading();
5777 $curLevel = $bits['level'];
5778 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5779 break;
5780 }
5781 }
5782 if ( $mode === 'get' ) {
5783 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5784 }
5785 $node = $node->getNextSibling();
5786 } while ( $node );
5787
5788 # Write out the remainder (in replace mode only)
5789 if ( $mode === 'replace' ) {
5790 # Output the replacement text
5791 # Add two newlines on -- trailing whitespace in $newText is conventionally
5792 # stripped by the editor, so we need both newlines to restore the paragraph gap
5793 # Only add trailing whitespace if there is newText
5794 if ( $newText != "" ) {
5795 $outText .= $newText . "\n\n";
5796 }
5797
5798 while ( $node ) {
5799 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5800 $node = $node->getNextSibling();
5801 }
5802 }
5803
5804 if ( is_string( $outText ) ) {
5805 # Re-insert stripped tags
5806 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5807 }
5808
5809 return $outText;
5810 }
5811
5812 /**
5813 * This function returns the text of a section, specified by a number ($section).
5814 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5815 * the first section before any such heading (section 0).
5816 *
5817 * If a section contains subsections, these are also returned.
5818 *
5819 * @param string $text Text to look in
5820 * @param string|int $sectionId Section identifier as a number or string
5821 * (e.g. 0, 1 or 'T-1').
5822 * @param string $defaultText Default to return if section is not found
5823 *
5824 * @return string Text of the requested section
5825 */
5826 public function getSection( $text, $sectionId, $defaultText = '' ) {
5827 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5828 }
5829
5830 /**
5831 * This function returns $oldtext after the content of the section
5832 * specified by $section has been replaced with $text. If the target
5833 * section does not exist, $oldtext is returned unchanged.
5834 *
5835 * @param string $oldText Former text of the article
5836 * @param string|int $sectionId Section identifier as a number or string
5837 * (e.g. 0, 1 or 'T-1').
5838 * @param string $newText Replacing text
5839 *
5840 * @return string Modified text
5841 */
5842 public function replaceSection( $oldText, $sectionId, $newText ) {
5843 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5844 }
5845
5846 /**
5847 * Get the ID of the revision we are parsing
5848 *
5849 * The return value will be either:
5850 * - a) Positive, indicating a specific revision ID (current or old)
5851 * - b) Zero, meaning the revision ID specified by getCurrentRevisionCallback()
5852 * - c) Null, meaning the parse is for preview mode and there is no revision
5853 *
5854 * @return int|null
5855 */
5856 public function getRevisionId() {
5857 return $this->mRevisionId;
5858 }
5859
5860 /**
5861 * Get the revision object for $this->mRevisionId
5862 *
5863 * @return Revision|null Either a Revision object or null
5864 * @since 1.23 (public since 1.23)
5865 */
5866 public function getRevisionObject() {
5867 if ( !is_null( $this->mRevisionObject ) ) {
5868 return $this->mRevisionObject;
5869 }
5870
5871 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5872 // This is useful when parsing revision that has not yet been saved.
5873 // However, if we get back a saved revision even though we are in
5874 // preview mode, we'll have to ignore it, see below.
5875 // NOTE: This callback may be used to inject an OLD revision that was
5876 // already loaded, so "current" is a bit of a misnomer. We can't just
5877 // skip it if mRevisionId is set.
5878 $rev = call_user_func(
5879 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5880 );
5881
5882 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
5883 // We are in preview mode (mRevisionId is null), and the current revision callback
5884 // returned an existing revision. Ignore it and return null, it's probably the page's
5885 // current revision, which is not what we want here. Note that we do want to call the
5886 // callback to allow the unsaved revision to be injected here, e.g. for
5887 // self-transclusion previews.
5888 return null;
5889 }
5890
5891 // If the parse is for a new revision, then the callback should have
5892 // already been set to force the object and should match mRevisionId.
5893 // If not, try to fetch by mRevisionId for sanity.
5894 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
5895 $rev = Revision::newFromId( $this->mRevisionId );
5896 }
5897
5898 $this->mRevisionObject = $rev;
5899
5900 return $this->mRevisionObject;
5901 }
5902
5903 /**
5904 * Get the timestamp associated with the current revision, adjusted for
5905 * the default server-local timestamp
5906 * @return string
5907 */
5908 public function getRevisionTimestamp() {
5909 if ( is_null( $this->mRevisionTimestamp ) ) {
5910 $revObject = $this->getRevisionObject();
5911 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5912
5913 # The cryptic '' timezone parameter tells to use the site-default
5914 # timezone offset instead of the user settings.
5915 # Since this value will be saved into the parser cache, served
5916 # to other users, and potentially even used inside links and such,
5917 # it needs to be consistent for all visitors.
5918 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
5919 }
5920 return $this->mRevisionTimestamp;
5921 }
5922
5923 /**
5924 * Get the name of the user that edited the last revision
5925 *
5926 * @return string User name
5927 */
5928 public function getRevisionUser() {
5929 if ( is_null( $this->mRevisionUser ) ) {
5930 $revObject = $this->getRevisionObject();
5931
5932 # if this template is subst: the revision id will be blank,
5933 # so just use the current user's name
5934 if ( $revObject ) {
5935 $this->mRevisionUser = $revObject->getUserText();
5936 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5937 $this->mRevisionUser = $this->getUser()->getName();
5938 }
5939 }
5940 return $this->mRevisionUser;
5941 }
5942
5943 /**
5944 * Get the size of the revision
5945 *
5946 * @return int|null Revision size
5947 */
5948 public function getRevisionSize() {
5949 if ( is_null( $this->mRevisionSize ) ) {
5950 $revObject = $this->getRevisionObject();
5951
5952 # if this variable is subst: the revision id will be blank,
5953 # so just use the parser input size, because the own substituation
5954 # will change the size.
5955 if ( $revObject ) {
5956 $this->mRevisionSize = $revObject->getSize();
5957 } else {
5958 $this->mRevisionSize = $this->mInputSize;
5959 }
5960 }
5961 return $this->mRevisionSize;
5962 }
5963
5964 /**
5965 * Mutator for $mDefaultSort
5966 *
5967 * @param string $sort New value
5968 */
5969 public function setDefaultSort( $sort ) {
5970 $this->mDefaultSort = $sort;
5971 $this->mOutput->setProperty( 'defaultsort', $sort );
5972 }
5973
5974 /**
5975 * Accessor for $mDefaultSort
5976 * Will use the empty string if none is set.
5977 *
5978 * This value is treated as a prefix, so the
5979 * empty string is equivalent to sorting by
5980 * page name.
5981 *
5982 * @return string
5983 */
5984 public function getDefaultSort() {
5985 if ( $this->mDefaultSort !== false ) {
5986 return $this->mDefaultSort;
5987 } else {
5988 return '';
5989 }
5990 }
5991
5992 /**
5993 * Accessor for $mDefaultSort
5994 * Unlike getDefaultSort(), will return false if none is set
5995 *
5996 * @return string|bool
5997 */
5998 public function getCustomDefaultSort() {
5999 return $this->mDefaultSort;
6000 }
6001
6002 private static function getSectionNameFromStrippedText( $text ) {
6003 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6004 $text = Sanitizer::decodeCharReferences( $text );
6005 $text = self::normalizeSectionName( $text );
6006 return $text;
6007 }
6008
6009 private static function makeAnchor( $sectionName ) {
6010 return '#' . Sanitizer::escapeIdForLink( $sectionName );
6011 }
6012
6013 private function makeLegacyAnchor( $sectionName ) {
6014 $fragmentMode = $this->siteConfig->get( 'FragmentMode' );
6015 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6016 // ForAttribute() and ForLink() are the same for legacy encoding
6017 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
6018 } else {
6019 $id = Sanitizer::escapeIdForLink( $sectionName );
6020 }
6021
6022 return "#$id";
6023 }
6024
6025 /**
6026 * Try to guess the section anchor name based on a wikitext fragment
6027 * presumably extracted from a heading, for example "Header" from
6028 * "== Header ==".
6029 *
6030 * @param string $text
6031 * @return string Anchor (starting with '#')
6032 */
6033 public function guessSectionNameFromWikiText( $text ) {
6034 # Strip out wikitext links(they break the anchor)
6035 $text = $this->stripSectionName( $text );
6036 $sectionName = self::getSectionNameFromStrippedText( $text );
6037 return self::makeAnchor( $sectionName );
6038 }
6039
6040 /**
6041 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6042 * instead, if possible. For use in redirects, since various versions
6043 * of Microsoft browsers interpret Location: headers as something other
6044 * than UTF-8, resulting in breakage.
6045 *
6046 * @param string $text The section name
6047 * @return string Anchor (starting with '#')
6048 */
6049 public function guessLegacySectionNameFromWikiText( $text ) {
6050 # Strip out wikitext links(they break the anchor)
6051 $text = $this->stripSectionName( $text );
6052 $sectionName = self::getSectionNameFromStrippedText( $text );
6053 return $this->makeLegacyAnchor( $sectionName );
6054 }
6055
6056 /**
6057 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6058 * @param string $text Section name (plain text)
6059 * @return string Anchor (starting with '#')
6060 */
6061 public static function guessSectionNameFromStrippedText( $text ) {
6062 $sectionName = self::getSectionNameFromStrippedText( $text );
6063 return self::makeAnchor( $sectionName );
6064 }
6065
6066 /**
6067 * Apply the same normalization as code making links to this section would
6068 *
6069 * @param string $text
6070 * @return string
6071 */
6072 private static function normalizeSectionName( $text ) {
6073 # T90902: ensure the same normalization is applied for IDs as to links
6074 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
6075 try {
6076
6077 $parts = $titleParser->splitTitleString( "#$text" );
6078 } catch ( MalformedTitleException $ex ) {
6079 return $text;
6080 }
6081 return $parts['fragment'];
6082 }
6083
6084 /**
6085 * Strips a text string of wikitext for use in a section anchor
6086 *
6087 * Accepts a text string and then removes all wikitext from the
6088 * string and leaves only the resultant text (i.e. the result of
6089 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6090 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6091 * to create valid section anchors by mimicing the output of the
6092 * parser when headings are parsed.
6093 *
6094 * @param string $text Text string to be stripped of wikitext
6095 * for use in a Section anchor
6096 * @return string Filtered text string
6097 */
6098 public function stripSectionName( $text ) {
6099 # Strip internal link markup
6100 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6101 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6102
6103 # Strip external link markup
6104 # @todo FIXME: Not tolerant to blank link text
6105 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6106 # on how many empty links there are on the page - need to figure that out.
6107 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6108
6109 # Parse wikitext quotes (italics & bold)
6110 $text = $this->doQuotes( $text );
6111
6112 # Strip HTML tags
6113 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6114 return $text;
6115 }
6116
6117 /**
6118 * strip/replaceVariables/unstrip for preprocessor regression testing
6119 *
6120 * @param string $text
6121 * @param Title $title
6122 * @param ParserOptions $options
6123 * @param int $outputType
6124 *
6125 * @return string
6126 */
6127 public function testSrvus( $text, Title $title, ParserOptions $options,
6128 $outputType = self::OT_HTML
6129 ) {
6130 $magicScopeVariable = $this->lock();
6131 $this->startParse( $title, $options, $outputType, true );
6132
6133 $text = $this->replaceVariables( $text );
6134 $text = $this->mStripState->unstripBoth( $text );
6135 $text = Sanitizer::removeHTMLtags( $text );
6136 return $text;
6137 }
6138
6139 /**
6140 * @param string $text
6141 * @param Title $title
6142 * @param ParserOptions $options
6143 * @return string
6144 */
6145 public function testPst( $text, Title $title, ParserOptions $options ) {
6146 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6147 }
6148
6149 /**
6150 * @param string $text
6151 * @param Title $title
6152 * @param ParserOptions $options
6153 * @return string
6154 */
6155 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6156 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6157 }
6158
6159 /**
6160 * Call a callback function on all regions of the given text that are not
6161 * inside strip markers, and replace those regions with the return value
6162 * of the callback. For example, with input:
6163 *
6164 * aaa<MARKER>bbb
6165 *
6166 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6167 * two strings will be replaced with the value returned by the callback in
6168 * each case.
6169 *
6170 * @param string $s
6171 * @param callable $callback
6172 *
6173 * @return string
6174 */
6175 public function markerSkipCallback( $s, $callback ) {
6176 $i = 0;
6177 $out = '';
6178 while ( $i < strlen( $s ) ) {
6179 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6180 if ( $markerStart === false ) {
6181 $out .= call_user_func( $callback, substr( $s, $i ) );
6182 break;
6183 } else {
6184 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6185 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6186 if ( $markerEnd === false ) {
6187 $out .= substr( $s, $markerStart );
6188 break;
6189 } else {
6190 $markerEnd += strlen( self::MARKER_SUFFIX );
6191 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6192 $i = $markerEnd;
6193 }
6194 }
6195 }
6196 return $out;
6197 }
6198
6199 /**
6200 * Remove any strip markers found in the given text.
6201 *
6202 * @param string $text
6203 * @return string
6204 */
6205 public function killMarkers( $text ) {
6206 return $this->mStripState->killMarkers( $text );
6207 }
6208
6209 /**
6210 * Save the parser state required to convert the given half-parsed text to
6211 * HTML. "Half-parsed" in this context means the output of
6212 * recursiveTagParse() or internalParse(). This output has strip markers
6213 * from replaceVariables (extensionSubstitution() etc.), and link
6214 * placeholders from replaceLinkHolders().
6215 *
6216 * Returns an array which can be serialized and stored persistently. This
6217 * array can later be loaded into another parser instance with
6218 * unserializeHalfParsedText(). The text can then be safely incorporated into
6219 * the return value of a parser hook.
6220 *
6221 * @deprecated since 1.31
6222 * @param string $text
6223 *
6224 * @return array
6225 */
6226 public function serializeHalfParsedText( $text ) {
6227 wfDeprecated( __METHOD__, '1.31' );
6228 $data = [
6229 'text' => $text,
6230 'version' => self::HALF_PARSED_VERSION,
6231 'stripState' => $this->mStripState->getSubState( $text ),
6232 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6233 ];
6234 return $data;
6235 }
6236
6237 /**
6238 * Load the parser state given in the $data array, which is assumed to
6239 * have been generated by serializeHalfParsedText(). The text contents is
6240 * extracted from the array, and its markers are transformed into markers
6241 * appropriate for the current Parser instance. This transformed text is
6242 * returned, and can be safely included in the return value of a parser
6243 * hook.
6244 *
6245 * If the $data array has been stored persistently, the caller should first
6246 * check whether it is still valid, by calling isValidHalfParsedText().
6247 *
6248 * @deprecated since 1.31
6249 * @param array $data Serialized data
6250 * @throws MWException
6251 * @return string
6252 */
6253 public function unserializeHalfParsedText( $data ) {
6254 wfDeprecated( __METHOD__, '1.31' );
6255 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6256 throw new MWException( __METHOD__ . ': invalid version' );
6257 }
6258
6259 # First, extract the strip state.
6260 $texts = [ $data['text'] ];
6261 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6262
6263 # Now renumber links
6264 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6265
6266 # Should be good to go.
6267 return $texts[0];
6268 }
6269
6270 /**
6271 * Returns true if the given array, presumed to be generated by
6272 * serializeHalfParsedText(), is compatible with the current version of the
6273 * parser.
6274 *
6275 * @deprecated since 1.31
6276 * @param array $data
6277 *
6278 * @return bool
6279 */
6280 public function isValidHalfParsedText( $data ) {
6281 wfDeprecated( __METHOD__, '1.31' );
6282 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6283 }
6284
6285 /**
6286 * Parsed a width param of imagelink like 300px or 200x300px
6287 *
6288 * @param string $value
6289 * @param bool $parseHeight
6290 *
6291 * @return array
6292 * @since 1.20
6293 */
6294 public static function parseWidthParam( $value, $parseHeight = true ) {
6295 $parsedWidthParam = [];
6296 if ( $value === '' ) {
6297 return $parsedWidthParam;
6298 }
6299 $m = [];
6300 # (T15500) In both cases (width/height and width only),
6301 # permit trailing "px" for backward compatibility.
6302 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6303 $width = intval( $m[1] );
6304 $height = intval( $m[2] );
6305 $parsedWidthParam['width'] = $width;
6306 $parsedWidthParam['height'] = $height;
6307 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6308 $width = intval( $value );
6309 $parsedWidthParam['width'] = $width;
6310 }
6311 return $parsedWidthParam;
6312 }
6313
6314 /**
6315 * Lock the current instance of the parser.
6316 *
6317 * This is meant to stop someone from calling the parser
6318 * recursively and messing up all the strip state.
6319 *
6320 * @throws MWException If parser is in a parse
6321 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6322 */
6323 protected function lock() {
6324 if ( $this->mInParse ) {
6325 throw new MWException( "Parser state cleared while parsing. "
6326 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6327 }
6328
6329 // Save the backtrace when locking, so that if some code tries locking again,
6330 // we can print the lock owner's backtrace for easier debugging
6331 $e = new Exception;
6332 $this->mInParse = $e->getTraceAsString();
6333
6334 $recursiveCheck = new ScopedCallback( function () {
6335 $this->mInParse = false;
6336 } );
6337
6338 return $recursiveCheck;
6339 }
6340
6341 /**
6342 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6343 *
6344 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6345 * or if there is more than one <p/> tag in the input HTML.
6346 *
6347 * @param string $html
6348 * @return string
6349 * @since 1.24
6350 */
6351 public static function stripOuterParagraph( $html ) {
6352 $m = [];
6353 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6354 $html = $m[1];
6355 }
6356
6357 return $html;
6358 }
6359
6360 /**
6361 * Return this parser if it is not doing anything, otherwise
6362 * get a fresh parser. You can use this method by doing
6363 * $newParser = $oldParser->getFreshParser(), or more simply
6364 * $oldParser->getFreshParser()->parse( ... );
6365 * if you're unsure if $oldParser is safe to use.
6366 *
6367 * @since 1.24
6368 * @return Parser A parser object that is not parsing anything
6369 */
6370 public function getFreshParser() {
6371 if ( $this->mInParse ) {
6372 return $this->factory->create();
6373 } else {
6374 return $this;
6375 }
6376 }
6377
6378 /**
6379 * Set's up the PHP implementation of OOUI for use in this request
6380 * and instructs OutputPage to enable OOUI for itself.
6381 *
6382 * @since 1.26
6383 */
6384 public function enableOOUI() {
6385 OutputPage::setupOOUI();
6386 $this->mOutput->setEnableOOUI( true );
6387 }
6388 }
Wiklou, le Wiki du Wiklou