dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "SlotDiffRenderer: add utility method for parameter type checks"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use MediaWiki\Special\SpecialPageFactory;
26 use Wikimedia\ScopedCallback;
27
28 /**
29 * @defgroup Parser Parser
30 */
31
32 /**
33 * PHP Parser - Processes wiki markup (which uses a more user-friendly
34 * syntax, such as "[[link]]" for making links), and provides a one-way
35 * transformation of that wiki markup it into (X)HTML output / markup
36 * (which in turn the browser understands, and can display).
37 *
38 * There are seven main entry points into the Parser class:
39 *
40 * - Parser::parse()
41 * produces HTML output
42 * - Parser::preSaveTransform()
43 * produces altered wiki markup
44 * - Parser::preprocess()
45 * removes HTML comments and expands templates
46 * - Parser::cleanSig() and Parser::cleanSigInSig()
47 * cleans a signature before saving it to preferences
48 * - Parser::getSection()
49 * return the content of a section from an article for section editing
50 * - Parser::replaceSection()
51 * replaces a section by number inside an article
52 * - Parser::getPreloadText()
53 * removes <noinclude> sections and <includeonly> tags
54 *
55 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
56 *
57 * @par Settings:
58 * $wgNamespacesWithSubpages
59 *
60 * @par Settings only within ParserOptions:
61 * $wgAllowExternalImages
62 * $wgAllowSpecialInclusion
63 * $wgInterwikiMagic
64 * $wgMaxArticleSize
65 *
66 * @ingroup Parser
67 */
68 class Parser {
69 /**
70 * Update this version number when the ParserOutput format
71 * changes in an incompatible way, so the parser cache
72 * can automatically discard old data.
73 */
74 const VERSION = '1.6.4';
75
76 /**
77 * Update this version number when the output of serialiseHalfParsedText()
78 * changes in an incompatible way
79 */
80 const HALF_PARSED_VERSION = 2;
81
82 # Flags for Parser::setFunctionHook
83 const SFH_NO_HASH = 1;
84 const SFH_OBJECT_ARGS = 2;
85
86 # Constants needed for external link processing
87 # Everything except bracket, space, or control characters
88 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
89 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
90 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
91 # uses to replace invalid HTML characters.
92 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
93 # Simplified expression to match an IPv4 or IPv6 address, or
94 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
95 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
96 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
97 // phpcs:ignore Generic.Files.LineLength
98 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
99 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
100
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 /**
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
118 *
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
122 * string constructs.
123 *
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
126 *
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
130 * `'".
131 */
132 const MARKER_SUFFIX = "-QINU`\"'\x7f";
133 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
134
135 # Markers used for wrapping the table of contents
136 const TOC_START = '<mw:toc>';
137 const TOC_END = '</mw:toc>';
138
139 /** @var int Assume that no output will later be saved this many seconds after parsing */
140 const MAX_TTS = 900;
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 /**
155 * @var bool Whether firstCallInit still needs to be called
156 */
157 public $mFirstCall = true;
158
159 # Initialised by initialiseVariables()
160
161 /**
162 * @var MagicWordArray
163 */
164 public $mVariables;
165
166 /**
167 * @var MagicWordArray
168 */
169 public $mSubstWords;
170 # Initialised in constructor
171 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
172
173 # Initialized in getPreprocessor()
174 /** @var Preprocessor */
175 public $mPreprocessor;
176
177 # Cleared with clearState():
178 /**
179 * @var ParserOutput
180 */
181 public $mOutput;
182 public $mAutonumber;
183
184 /**
185 * @var StripState
186 */
187 public $mStripState;
188
189 public $mIncludeCount;
190 /**
191 * @var LinkHolderArray
192 */
193 public $mLinkHolders;
194
195 public $mLinkID;
196 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
197 public $mDefaultSort;
198 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
199 public $mExpensiveFunctionCount; # number of expensive parser function calls
200 public $mShowToc, $mForceTocPosition;
201
202 /**
203 * @var User
204 */
205 public $mUser; # User object; only used when doing pre-save transform
206
207 # Temporary
208 # These are variables reset at least once per parse regardless of $clearState
209
210 /**
211 * @var ParserOptions
212 */
213 public $mOptions;
214
215 /**
216 * @var Title
217 */
218 public $mTitle; # Title context, used for self-link rendering and similar things
219 public $mOutputType; # Output type, one of the OT_xxx constants
220 public $ot; # Shortcut alias, see setOutputType()
221 public $mRevisionObject; # The revision object of the specified revision ID
222 public $mRevisionId; # ID to display in {{REVISIONID}} tags
223 public $mRevisionTimestamp; # The timestamp of the specified revision ID
224 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
225 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
226 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
227 public $mInputSize = false; # For {{PAGESIZE}} on current page.
228
229 /**
230 * @var string Deprecated accessor for the strip marker prefix.
231 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
232 */
233 public $mUniqPrefix = self::MARKER_PREFIX;
234
235 /**
236 * @var array Array with the language name of each language link (i.e. the
237 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
238 * duplicate language links to the ParserOutput.
239 */
240 public $mLangLinkLanguages;
241
242 /**
243 * @var MapCacheLRU|null
244 * @since 1.24
245 *
246 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
247 */
248 public $currentRevisionCache;
249
250 /**
251 * @var bool|string Recursive call protection.
252 * This variable should be treated as if it were private.
253 */
254 public $mInParse = false;
255
256 /** @var SectionProfiler */
257 protected $mProfiler;
258
259 /**
260 * @var LinkRenderer
261 */
262 protected $mLinkRenderer;
263
264 /** @var MagicWordFactory */
265 private $magicWordFactory;
266
267 /** @var Language */
268 private $contLang;
269
270 /** @var ParserFactory */
271 private $factory;
272
273 /** @var SpecialPageFactory */
274 private $specialPageFactory;
275
276 /**
277 * @param array $conf See $wgParserConf documentation
278 * @param MagicWordFactory|null $magicWordFactory
279 * @param Language|null $contLang Content language
280 * @param ParserFactory|null $factory
281 * @param string|null $urlProtocols As returned from wfUrlProtocols()
282 * @param SpecialPageFactory|null $spFactory
283 */
284 public function __construct(
285 array $conf = [], MagicWordFactory $magicWordFactory = null, Language $contLang = null,
286 ParserFactory $factory = null, $urlProtocols = null, SpecialPageFactory $spFactory = null
287 ) {
288 $this->mConf = $conf;
289 $this->mUrlProtocols = $urlProtocols ?? wfUrlProtocols();
290 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
291 self::EXT_LINK_ADDR .
292 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
293 if ( isset( $conf['preprocessorClass'] ) ) {
294 $this->mPreprocessorClass = $conf['preprocessorClass'];
295 } elseif ( wfIsHHVM() ) {
296 # Under HHVM Preprocessor_Hash is much faster than Preprocessor_DOM
297 $this->mPreprocessorClass = Preprocessor_Hash::class;
298 } elseif ( extension_loaded( 'domxml' ) ) {
299 # PECL extension that conflicts with the core DOM extension (T15770)
300 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
301 $this->mPreprocessorClass = Preprocessor_Hash::class;
302 } elseif ( extension_loaded( 'dom' ) ) {
303 $this->mPreprocessorClass = Preprocessor_DOM::class;
304 } else {
305 $this->mPreprocessorClass = Preprocessor_Hash::class;
306 }
307 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
308
309 $services = MediaWikiServices::getInstance();
310 $this->magicWordFactory = $magicWordFactory ??
311 $services->getMagicWordFactory();
312
313 $this->contLang = $contLang ?? $services->getContentLanguage();
314
315 $this->factory = $factory ?? $services->getParserFactory();
316 $this->specialPageFactory = $spFactory ?? $services->getSpecialPageFactory();
317 }
318
319 /**
320 * Reduce memory usage to reduce the impact of circular references
321 */
322 public function __destruct() {
323 if ( isset( $this->mLinkHolders ) ) {
324 unset( $this->mLinkHolders );
325 }
326 foreach ( $this as $name => $value ) {
327 unset( $this->$name );
328 }
329 }
330
331 /**
332 * Allow extensions to clean up when the parser is cloned
333 */
334 public function __clone() {
335 $this->mInParse = false;
336
337 // T58226: When you create a reference "to" an object field, that
338 // makes the object field itself be a reference too (until the other
339 // reference goes out of scope). When cloning, any field that's a
340 // reference is copied as a reference in the new object. Both of these
341 // are defined PHP5 behaviors, as inconvenient as it is for us when old
342 // hooks from PHP4 days are passing fields by reference.
343 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
344 // Make a non-reference copy of the field, then rebind the field to
345 // reference the new copy.
346 $tmp = $this->$k;
347 $this->$k =& $tmp;
348 unset( $tmp );
349 }
350
351 Hooks::run( 'ParserCloned', [ $this ] );
352 }
353
354 /**
355 * Do various kinds of initialisation on the first call of the parser
356 */
357 public function firstCallInit() {
358 if ( !$this->mFirstCall ) {
359 return;
360 }
361 $this->mFirstCall = false;
362
363 CoreParserFunctions::register( $this );
364 CoreTagHooks::register( $this );
365 $this->initialiseVariables();
366
367 // Avoid PHP 7.1 warning from passing $this by reference
368 $parser = $this;
369 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
370 }
371
372 /**
373 * Clear Parser state
374 *
375 * @private
376 */
377 public function clearState() {
378 $this->firstCallInit();
379 $this->mOutput = new ParserOutput;
380 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
381 $this->mAutonumber = 0;
382 $this->mIncludeCount = [];
383 $this->mLinkHolders = new LinkHolderArray( $this );
384 $this->mLinkID = 0;
385 $this->mRevisionObject = $this->mRevisionTimestamp =
386 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
387 $this->mVarCache = [];
388 $this->mUser = null;
389 $this->mLangLinkLanguages = [];
390 $this->currentRevisionCache = null;
391
392 $this->mStripState = new StripState( $this );
393
394 # Clear these on every parse, T6549
395 $this->mTplRedirCache = $this->mTplDomCache = [];
396
397 $this->mShowToc = true;
398 $this->mForceTocPosition = false;
399 $this->mIncludeSizes = [
400 'post-expand' => 0,
401 'arg' => 0,
402 ];
403 $this->mPPNodeCount = 0;
404 $this->mGeneratedPPNodeCount = 0;
405 $this->mHighestExpansionDepth = 0;
406 $this->mDefaultSort = false;
407 $this->mHeadings = [];
408 $this->mDoubleUnderscores = [];
409 $this->mExpensiveFunctionCount = 0;
410
411 # Fix cloning
412 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
413 $this->mPreprocessor = null;
414 }
415
416 $this->mProfiler = new SectionProfiler();
417
418 // Avoid PHP 7.1 warning from passing $this by reference
419 $parser = $this;
420 Hooks::run( 'ParserClearState', [ &$parser ] );
421 }
422
423 /**
424 * Convert wikitext to HTML
425 * Do not call this function recursively.
426 *
427 * @param string $text Text we want to parse
428 * @param-taint $text escapes_htmlnoent
429 * @param Title $title
430 * @param ParserOptions $options
431 * @param bool $linestart
432 * @param bool $clearState
433 * @param int|null $revid Number to pass in {{REVISIONID}}
434 * @return ParserOutput A ParserOutput
435 * @return-taint escaped
436 */
437 public function parse(
438 $text, Title $title, ParserOptions $options,
439 $linestart = true, $clearState = true, $revid = null
440 ) {
441 if ( $clearState ) {
442 // We use U+007F DELETE to construct strip markers, so we have to make
443 // sure that this character does not occur in the input text.
444 $text = strtr( $text, "\x7f", "?" );
445 $magicScopeVariable = $this->lock();
446 }
447 // Strip U+0000 NULL (T159174)
448 $text = str_replace( "\000", '', $text );
449
450 $this->startParse( $title, $options, self::OT_HTML, $clearState );
451
452 $this->currentRevisionCache = null;
453 $this->mInputSize = strlen( $text );
454 if ( $this->mOptions->getEnableLimitReport() ) {
455 $this->mOutput->resetParseStartTime();
456 }
457
458 $oldRevisionId = $this->mRevisionId;
459 $oldRevisionObject = $this->mRevisionObject;
460 $oldRevisionTimestamp = $this->mRevisionTimestamp;
461 $oldRevisionUser = $this->mRevisionUser;
462 $oldRevisionSize = $this->mRevisionSize;
463 if ( $revid !== null ) {
464 $this->mRevisionId = $revid;
465 $this->mRevisionObject = null;
466 $this->mRevisionTimestamp = null;
467 $this->mRevisionUser = null;
468 $this->mRevisionSize = null;
469 }
470
471 // Avoid PHP 7.1 warning from passing $this by reference
472 $parser = $this;
473 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
474 # No more strip!
475 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
476 $text = $this->internalParse( $text );
477 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
478
479 $text = $this->internalParseHalfParsed( $text, true, $linestart );
480
481 /**
482 * A converted title will be provided in the output object if title and
483 * content conversion are enabled, the article text does not contain
484 * a conversion-suppressing double-underscore tag, and no
485 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
486 * automatic link conversion.
487 */
488 if ( !( $options->getDisableTitleConversion()
489 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
490 || isset( $this->mDoubleUnderscores['notitleconvert'] )
491 || $this->mOutput->getDisplayTitle() !== false )
492 ) {
493 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
494 if ( $convruletitle ) {
495 $this->mOutput->setTitleText( $convruletitle );
496 } else {
497 $titleText = $this->getTargetLanguage()->convertTitle( $title );
498 $this->mOutput->setTitleText( $titleText );
499 }
500 }
501
502 # Compute runtime adaptive expiry if set
503 $this->mOutput->finalizeAdaptiveCacheExpiry();
504
505 # Warn if too many heavyweight parser functions were used
506 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
507 $this->limitationWarn( 'expensive-parserfunction',
508 $this->mExpensiveFunctionCount,
509 $this->mOptions->getExpensiveParserFunctionLimit()
510 );
511 }
512
513 # Information on limits, for the benefit of users who try to skirt them
514 if ( $this->mOptions->getEnableLimitReport() ) {
515 $text .= $this->makeLimitReport();
516 }
517
518 # Wrap non-interface parser output in a <div> so it can be targeted
519 # with CSS (T37247)
520 $class = $this->mOptions->getWrapOutputClass();
521 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
522 $this->mOutput->addWrapperDivClass( $class );
523 }
524
525 $this->mOutput->setText( $text );
526
527 $this->mRevisionId = $oldRevisionId;
528 $this->mRevisionObject = $oldRevisionObject;
529 $this->mRevisionTimestamp = $oldRevisionTimestamp;
530 $this->mRevisionUser = $oldRevisionUser;
531 $this->mRevisionSize = $oldRevisionSize;
532 $this->mInputSize = false;
533 $this->currentRevisionCache = null;
534
535 return $this->mOutput;
536 }
537
538 /**
539 * Set the limit report data in the current ParserOutput, and return the
540 * limit report HTML comment.
541 *
542 * @return string
543 */
544 protected function makeLimitReport() {
545 global $wgShowHostnames;
546
547 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
548
549 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
550 if ( $cpuTime !== null ) {
551 $this->mOutput->setLimitReportData( 'limitreport-cputime',
552 sprintf( "%.3f", $cpuTime )
553 );
554 }
555
556 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
557 $this->mOutput->setLimitReportData( 'limitreport-walltime',
558 sprintf( "%.3f", $wallTime )
559 );
560
561 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
562 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
563 );
564 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
565 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
566 );
567 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
568 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
569 );
570 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
571 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
572 );
573 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
574 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
575 );
576 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
577 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
578 );
579
580 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
581 $this->mOutput->setLimitReportData( $key, $value );
582 }
583
584 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
585
586 $limitReport = "NewPP limit report\n";
587 if ( $wgShowHostnames ) {
588 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
589 }
590 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
591 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
592 $limitReport .= 'Dynamic content: ' .
593 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
594 "\n";
595
596 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
597 if ( Hooks::run( 'ParserLimitReportFormat',
598 [ $key, &$value, &$limitReport, false, false ]
599 ) ) {
600 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
601 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
602 ->inLanguage( 'en' )->useDatabase( false );
603 if ( !$valueMsg->exists() ) {
604 $valueMsg = new RawMessage( '$1' );
605 }
606 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
607 $valueMsg->params( $value );
608 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
609 }
610 }
611 }
612 // Since we're not really outputting HTML, decode the entities and
613 // then re-encode the things that need hiding inside HTML comments.
614 $limitReport = htmlspecialchars_decode( $limitReport );
615
616 // Sanitize for comment. Note '‐' in the replacement is U+2010,
617 // which looks much like the problematic '-'.
618 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
619 $text = "\n<!-- \n$limitReport-->\n";
620
621 // Add on template profiling data in human/machine readable way
622 $dataByFunc = $this->mProfiler->getFunctionStats();
623 uasort( $dataByFunc, function ( $a, $b ) {
624 return $b['real'] <=> $a['real']; // descending order
625 } );
626 $profileReport = [];
627 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
628 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
629 $item['%real'], $item['real'], $item['calls'],
630 htmlspecialchars( $item['name'] ) );
631 }
632 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
633 $text .= implode( "\n", $profileReport ) . "\n-->\n";
634
635 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
636
637 // Add other cache related metadata
638 if ( $wgShowHostnames ) {
639 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
640 }
641 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
642 $this->mOutput->getCacheTime() );
643 $this->mOutput->setLimitReportData( 'cachereport-ttl',
644 $this->mOutput->getCacheExpiry() );
645 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
646 $this->mOutput->hasDynamicContent() );
647
648 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
649 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
650 $this->mTitle->getPrefixedDBkey() );
651 }
652 return $text;
653 }
654
655 /**
656 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
657 * can be called from an extension tag hook.
658 *
659 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
660 * instead, which means that lists and links have not been fully parsed yet,
661 * and strip markers are still present.
662 *
663 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
664 *
665 * Use this function if you're a parser tag hook and you want to parse
666 * wikitext before or after applying additional transformations, and you
667 * intend to *return the result as hook output*, which will cause it to go
668 * through the rest of parsing process automatically.
669 *
670 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
671 * $text are not expanded
672 *
673 * @param string $text Text extension wants to have parsed
674 * @param-taint $text escapes_htmlnoent
675 * @param bool|PPFrame $frame The frame to use for expanding any template variables
676 * @return string UNSAFE half-parsed HTML
677 * @return-taint escaped
678 */
679 public function recursiveTagParse( $text, $frame = false ) {
680 // Avoid PHP 7.1 warning from passing $this by reference
681 $parser = $this;
682 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
683 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
684 $text = $this->internalParse( $text, false, $frame );
685 return $text;
686 }
687
688 /**
689 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
690 * point can be called from an extension tag hook.
691 *
692 * The output of this function is fully-parsed HTML that is safe for output.
693 * If you're a parser tag hook, you might want to use recursiveTagParse()
694 * instead.
695 *
696 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
697 * $text are not expanded
698 *
699 * @since 1.25
700 *
701 * @param string $text Text extension wants to have parsed
702 * @param-taint $text escapes_htmlnoent
703 * @param bool|PPFrame $frame The frame to use for expanding any template variables
704 * @return string Fully parsed HTML
705 * @return-taint escaped
706 */
707 public function recursiveTagParseFully( $text, $frame = false ) {
708 $text = $this->recursiveTagParse( $text, $frame );
709 $text = $this->internalParseHalfParsed( $text, false );
710 return $text;
711 }
712
713 /**
714 * Expand templates and variables in the text, producing valid, static wikitext.
715 * Also removes comments.
716 * Do not call this function recursively.
717 * @param string $text
718 * @param Title|null $title
719 * @param ParserOptions $options
720 * @param int|null $revid
721 * @param bool|PPFrame $frame
722 * @return mixed|string
723 */
724 public function preprocess( $text, Title $title = null,
725 ParserOptions $options, $revid = null, $frame = false
726 ) {
727 $magicScopeVariable = $this->lock();
728 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
729 if ( $revid !== null ) {
730 $this->mRevisionId = $revid;
731 }
732 // Avoid PHP 7.1 warning from passing $this by reference
733 $parser = $this;
734 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
735 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
736 $text = $this->replaceVariables( $text, $frame );
737 $text = $this->mStripState->unstripBoth( $text );
738 return $text;
739 }
740
741 /**
742 * Recursive parser entry point that can be called from an extension tag
743 * hook.
744 *
745 * @param string $text Text to be expanded
746 * @param bool|PPFrame $frame The frame to use for expanding any template variables
747 * @return string
748 * @since 1.19
749 */
750 public function recursivePreprocess( $text, $frame = false ) {
751 $text = $this->replaceVariables( $text, $frame );
752 $text = $this->mStripState->unstripBoth( $text );
753 return $text;
754 }
755
756 /**
757 * Process the wikitext for the "?preload=" feature. (T7210)
758 *
759 * "<noinclude>", "<includeonly>" etc. are parsed as for template
760 * transclusion, comments, templates, arguments, tags hooks and parser
761 * functions are untouched.
762 *
763 * @param string $text
764 * @param Title $title
765 * @param ParserOptions $options
766 * @param array $params
767 * @return string
768 */
769 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
770 $msg = new RawMessage( $text );
771 $text = $msg->params( $params )->plain();
772
773 # Parser (re)initialisation
774 $magicScopeVariable = $this->lock();
775 $this->startParse( $title, $options, self::OT_PLAIN, true );
776
777 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
778 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
779 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
780 $text = $this->mStripState->unstripBoth( $text );
781 return $text;
782 }
783
784 /**
785 * Set the current user.
786 * Should only be used when doing pre-save transform.
787 *
788 * @param User|null $user User object or null (to reset)
789 */
790 public function setUser( $user ) {
791 $this->mUser = $user;
792 }
793
794 /**
795 * Set the context title
796 *
797 * @param Title $t
798 */
799 public function setTitle( $t ) {
800 if ( !$t ) {
801 $t = Title::newFromText( 'NO TITLE' );
802 }
803
804 if ( $t->hasFragment() ) {
805 # Strip the fragment to avoid various odd effects
806 $this->mTitle = $t->createFragmentTarget( '' );
807 } else {
808 $this->mTitle = $t;
809 }
810 }
811
812 /**
813 * Accessor for the Title object
814 *
815 * @return Title
816 */
817 public function getTitle() {
818 return $this->mTitle;
819 }
820
821 /**
822 * Accessor/mutator for the Title object
823 *
824 * @param Title|null $x Title object or null to just get the current one
825 * @return Title
826 */
827 public function Title( $x = null ) {
828 return wfSetVar( $this->mTitle, $x );
829 }
830
831 /**
832 * Set the output type
833 *
834 * @param int $ot New value
835 */
836 public function setOutputType( $ot ) {
837 $this->mOutputType = $ot;
838 # Shortcut alias
839 $this->ot = [
840 'html' => $ot == self::OT_HTML,
841 'wiki' => $ot == self::OT_WIKI,
842 'pre' => $ot == self::OT_PREPROCESS,
843 'plain' => $ot == self::OT_PLAIN,
844 ];
845 }
846
847 /**
848 * Accessor/mutator for the output type
849 *
850 * @param int|null $x New value or null to just get the current one
851 * @return int
852 */
853 public function OutputType( $x = null ) {
854 return wfSetVar( $this->mOutputType, $x );
855 }
856
857 /**
858 * Get the ParserOutput object
859 *
860 * @return ParserOutput
861 */
862 public function getOutput() {
863 return $this->mOutput;
864 }
865
866 /**
867 * Get the ParserOptions object
868 *
869 * @return ParserOptions
870 */
871 public function getOptions() {
872 return $this->mOptions;
873 }
874
875 /**
876 * Accessor/mutator for the ParserOptions object
877 *
878 * @param ParserOptions|null $x New value or null to just get the current one
879 * @return ParserOptions Current ParserOptions object
880 */
881 public function Options( $x = null ) {
882 return wfSetVar( $this->mOptions, $x );
883 }
884
885 /**
886 * @return int
887 */
888 public function nextLinkID() {
889 return $this->mLinkID++;
890 }
891
892 /**
893 * @param int $id
894 */
895 public function setLinkID( $id ) {
896 $this->mLinkID = $id;
897 }
898
899 /**
900 * Get a language object for use in parser functions such as {{FORMATNUM:}}
901 * @return Language
902 */
903 public function getFunctionLang() {
904 return $this->getTargetLanguage();
905 }
906
907 /**
908 * Get the target language for the content being parsed. This is usually the
909 * language that the content is in.
910 *
911 * @since 1.19
912 *
913 * @throws MWException
914 * @return Language
915 */
916 public function getTargetLanguage() {
917 $target = $this->mOptions->getTargetLanguage();
918
919 if ( $target !== null ) {
920 return $target;
921 } elseif ( $this->mOptions->getInterfaceMessage() ) {
922 return $this->mOptions->getUserLangObj();
923 } elseif ( is_null( $this->mTitle ) ) {
924 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
925 }
926
927 return $this->mTitle->getPageLanguage();
928 }
929
930 /**
931 * Get the language object for language conversion
932 * @deprecated since 1.32, just use getTargetLanguage()
933 * @return Language|null
934 */
935 public function getConverterLanguage() {
936 return $this->getTargetLanguage();
937 }
938
939 /**
940 * Get a User object either from $this->mUser, if set, or from the
941 * ParserOptions object otherwise
942 *
943 * @return User
944 */
945 public function getUser() {
946 if ( !is_null( $this->mUser ) ) {
947 return $this->mUser;
948 }
949 return $this->mOptions->getUser();
950 }
951
952 /**
953 * Get a preprocessor object
954 *
955 * @return Preprocessor
956 */
957 public function getPreprocessor() {
958 if ( !isset( $this->mPreprocessor ) ) {
959 $class = $this->mPreprocessorClass;
960 $this->mPreprocessor = new $class( $this );
961 }
962 return $this->mPreprocessor;
963 }
964
965 /**
966 * Get a LinkRenderer instance to make links with
967 *
968 * @since 1.28
969 * @return LinkRenderer
970 */
971 public function getLinkRenderer() {
972 if ( !$this->mLinkRenderer ) {
973 $this->mLinkRenderer = MediaWikiServices::getInstance()
974 ->getLinkRendererFactory()->create();
975 $this->mLinkRenderer->setStubThreshold(
976 $this->getOptions()->getStubThreshold()
977 );
978 }
979
980 return $this->mLinkRenderer;
981 }
982
983 /**
984 * Get the MagicWordFactory that this Parser is using
985 *
986 * @since 1.32
987 * @return MagicWordFactory
988 */
989 public function getMagicWordFactory() {
990 return $this->magicWordFactory;
991 }
992
993 /**
994 * Get the content language that this Parser is using
995 *
996 * @since 1.32
997 * @return Language
998 */
999 public function getContentLanguage() {
1000 return $this->contLang;
1001 }
1002
1003 /**
1004 * Replaces all occurrences of HTML-style comments and the given tags
1005 * in the text with a random marker and returns the next text. The output
1006 * parameter $matches will be an associative array filled with data in
1007 * the form:
1008 *
1009 * @code
1010 * 'UNIQ-xxxxx' => [
1011 * 'element',
1012 * 'tag content',
1013 * [ 'param' => 'x' ],
1014 * '<element param="x">tag content</element>' ]
1015 * @endcode
1016 *
1017 * @param array $elements List of element names. Comments are always extracted.
1018 * @param string $text Source text string.
1019 * @param array &$matches Out parameter, Array: extracted tags
1020 * @return string Stripped text
1021 */
1022 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1023 static $n = 1;
1024 $stripped = '';
1025 $matches = [];
1026
1027 $taglist = implode( '|', $elements );
1028 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
1029
1030 while ( $text != '' ) {
1031 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
1032 $stripped .= $p[0];
1033 if ( count( $p ) < 5 ) {
1034 break;
1035 }
1036 if ( count( $p ) > 5 ) {
1037 # comment
1038 $element = $p[4];
1039 $attributes = '';
1040 $close = '';
1041 $inside = $p[5];
1042 } else {
1043 # tag
1044 $element = $p[1];
1045 $attributes = $p[2];
1046 $close = $p[3];
1047 $inside = $p[4];
1048 }
1049
1050 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1051 $stripped .= $marker;
1052
1053 if ( $close === '/>' ) {
1054 # Empty element tag, <tag />
1055 $content = null;
1056 $text = $inside;
1057 $tail = null;
1058 } else {
1059 if ( $element === '!--' ) {
1060 $end = '/(-->)/';
1061 } else {
1062 $end = "/(<\\/$element\\s*>)/i";
1063 }
1064 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1065 $content = $q[0];
1066 if ( count( $q ) < 3 ) {
1067 # No end tag -- let it run out to the end of the text.
1068 $tail = '';
1069 $text = '';
1070 } else {
1071 $tail = $q[1];
1072 $text = $q[2];
1073 }
1074 }
1075
1076 $matches[$marker] = [ $element,
1077 $content,
1078 Sanitizer::decodeTagAttributes( $attributes ),
1079 "<$element$attributes$close$content$tail" ];
1080 }
1081 return $stripped;
1082 }
1083
1084 /**
1085 * Get a list of strippable XML-like elements
1086 *
1087 * @return array
1088 */
1089 public function getStripList() {
1090 return $this->mStripList;
1091 }
1092
1093 /**
1094 * Add an item to the strip state
1095 * Returns the unique tag which must be inserted into the stripped text
1096 * The tag will be replaced with the original text in unstrip()
1097 *
1098 * @param string $text
1099 *
1100 * @return string
1101 */
1102 public function insertStripItem( $text ) {
1103 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1104 $this->mMarkerIndex++;
1105 $this->mStripState->addGeneral( $marker, $text );
1106 return $marker;
1107 }
1108
1109 /**
1110 * parse the wiki syntax used to render tables
1111 *
1112 * @private
1113 * @param string $text
1114 * @return string
1115 */
1116 public function doTableStuff( $text ) {
1117 $lines = StringUtils::explode( "\n", $text );
1118 $out = '';
1119 $td_history = []; # Is currently a td tag open?
1120 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1121 $tr_history = []; # Is currently a tr tag open?
1122 $tr_attributes = []; # history of tr attributes
1123 $has_opened_tr = []; # Did this table open a <tr> element?
1124 $indent_level = 0; # indent level of the table
1125
1126 foreach ( $lines as $outLine ) {
1127 $line = trim( $outLine );
1128
1129 if ( $line === '' ) { # empty line, go to next line
1130 $out .= $outLine . "\n";
1131 continue;
1132 }
1133
1134 $first_character = $line[0];
1135 $first_two = substr( $line, 0, 2 );
1136 $matches = [];
1137
1138 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1139 # First check if we are starting a new table
1140 $indent_level = strlen( $matches[1] );
1141
1142 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1143 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1144
1145 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1146 array_push( $td_history, false );
1147 array_push( $last_tag_history, '' );
1148 array_push( $tr_history, false );
1149 array_push( $tr_attributes, '' );
1150 array_push( $has_opened_tr, false );
1151 } elseif ( count( $td_history ) == 0 ) {
1152 # Don't do any of the following
1153 $out .= $outLine . "\n";
1154 continue;
1155 } elseif ( $first_two === '|}' ) {
1156 # We are ending a table
1157 $line = '</table>' . substr( $line, 2 );
1158 $last_tag = array_pop( $last_tag_history );
1159
1160 if ( !array_pop( $has_opened_tr ) ) {
1161 $line = "<tr><td></td></tr>{$line}";
1162 }
1163
1164 if ( array_pop( $tr_history ) ) {
1165 $line = "</tr>{$line}";
1166 }
1167
1168 if ( array_pop( $td_history ) ) {
1169 $line = "</{$last_tag}>{$line}";
1170 }
1171 array_pop( $tr_attributes );
1172 if ( $indent_level > 0 ) {
1173 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1174 } else {
1175 $outLine = $line;
1176 }
1177 } elseif ( $first_two === '|-' ) {
1178 # Now we have a table row
1179 $line = preg_replace( '#^\|-+#', '', $line );
1180
1181 # Whats after the tag is now only attributes
1182 $attributes = $this->mStripState->unstripBoth( $line );
1183 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1184 array_pop( $tr_attributes );
1185 array_push( $tr_attributes, $attributes );
1186
1187 $line = '';
1188 $last_tag = array_pop( $last_tag_history );
1189 array_pop( $has_opened_tr );
1190 array_push( $has_opened_tr, true );
1191
1192 if ( array_pop( $tr_history ) ) {
1193 $line = '</tr>';
1194 }
1195
1196 if ( array_pop( $td_history ) ) {
1197 $line = "</{$last_tag}>{$line}";
1198 }
1199
1200 $outLine = $line;
1201 array_push( $tr_history, false );
1202 array_push( $td_history, false );
1203 array_push( $last_tag_history, '' );
1204 } elseif ( $first_character === '|'
1205 || $first_character === '!'
1206 || $first_two === '|+'
1207 ) {
1208 # This might be cell elements, td, th or captions
1209 if ( $first_two === '|+' ) {
1210 $first_character = '+';
1211 $line = substr( $line, 2 );
1212 } else {
1213 $line = substr( $line, 1 );
1214 }
1215
1216 // Implies both are valid for table headings.
1217 if ( $first_character === '!' ) {
1218 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1219 }
1220
1221 # Split up multiple cells on the same line.
1222 # FIXME : This can result in improper nesting of tags processed
1223 # by earlier parser steps.
1224 $cells = explode( '||', $line );
1225
1226 $outLine = '';
1227
1228 # Loop through each table cell
1229 foreach ( $cells as $cell ) {
1230 $previous = '';
1231 if ( $first_character !== '+' ) {
1232 $tr_after = array_pop( $tr_attributes );
1233 if ( !array_pop( $tr_history ) ) {
1234 $previous = "<tr{$tr_after}>\n";
1235 }
1236 array_push( $tr_history, true );
1237 array_push( $tr_attributes, '' );
1238 array_pop( $has_opened_tr );
1239 array_push( $has_opened_tr, true );
1240 }
1241
1242 $last_tag = array_pop( $last_tag_history );
1243
1244 if ( array_pop( $td_history ) ) {
1245 $previous = "</{$last_tag}>\n{$previous}";
1246 }
1247
1248 if ( $first_character === '|' ) {
1249 $last_tag = 'td';
1250 } elseif ( $first_character === '!' ) {
1251 $last_tag = 'th';
1252 } elseif ( $first_character === '+' ) {
1253 $last_tag = 'caption';
1254 } else {
1255 $last_tag = '';
1256 }
1257
1258 array_push( $last_tag_history, $last_tag );
1259
1260 # A cell could contain both parameters and data
1261 $cell_data = explode( '|', $cell, 2 );
1262
1263 # T2553: Note that a '|' inside an invalid link should not
1264 # be mistaken as delimiting cell parameters
1265 # Bug T153140: Neither should language converter markup.
1266 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1267 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1268 } elseif ( count( $cell_data ) == 1 ) {
1269 // Whitespace in cells is trimmed
1270 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1271 } else {
1272 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1273 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1274 // Whitespace in cells is trimmed
1275 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1276 }
1277
1278 $outLine .= $cell;
1279 array_push( $td_history, true );
1280 }
1281 }
1282 $out .= $outLine . "\n";
1283 }
1284
1285 # Closing open td, tr && table
1286 while ( count( $td_history ) > 0 ) {
1287 if ( array_pop( $td_history ) ) {
1288 $out .= "</td>\n";
1289 }
1290 if ( array_pop( $tr_history ) ) {
1291 $out .= "</tr>\n";
1292 }
1293 if ( !array_pop( $has_opened_tr ) ) {
1294 $out .= "<tr><td></td></tr>\n";
1295 }
1296
1297 $out .= "</table>\n";
1298 }
1299
1300 # Remove trailing line-ending (b/c)
1301 if ( substr( $out, -1 ) === "\n" ) {
1302 $out = substr( $out, 0, -1 );
1303 }
1304
1305 # special case: don't return empty table
1306 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1307 $out = '';
1308 }
1309
1310 return $out;
1311 }
1312
1313 /**
1314 * Helper function for parse() that transforms wiki markup into half-parsed
1315 * HTML. Only called for $mOutputType == self::OT_HTML.
1316 *
1317 * @private
1318 *
1319 * @param string $text The text to parse
1320 * @param-taint $text escapes_html
1321 * @param bool $isMain Whether this is being called from the main parse() function
1322 * @param PPFrame|bool $frame A pre-processor frame
1323 *
1324 * @return string
1325 */
1326 public function internalParse( $text, $isMain = true, $frame = false ) {
1327 $origText = $text;
1328
1329 // Avoid PHP 7.1 warning from passing $this by reference
1330 $parser = $this;
1331
1332 # Hook to suspend the parser in this state
1333 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1334 return $text;
1335 }
1336
1337 # if $frame is provided, then use $frame for replacing any variables
1338 if ( $frame ) {
1339 # use frame depth to infer how include/noinclude tags should be handled
1340 # depth=0 means this is the top-level document; otherwise it's an included document
1341 if ( !$frame->depth ) {
1342 $flag = 0;
1343 } else {
1344 $flag = self::PTD_FOR_INCLUSION;
1345 }
1346 $dom = $this->preprocessToDom( $text, $flag );
1347 $text = $frame->expand( $dom );
1348 } else {
1349 # if $frame is not provided, then use old-style replaceVariables
1350 $text = $this->replaceVariables( $text );
1351 }
1352
1353 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1354 $text = Sanitizer::removeHTMLtags(
1355 $text,
1356 [ $this, 'attributeStripCallback' ],
1357 false,
1358 array_keys( $this->mTransparentTagHooks ),
1359 [],
1360 [ $this, 'addTrackingCategory' ]
1361 );
1362 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1363
1364 # Tables need to come after variable replacement for things to work
1365 # properly; putting them before other transformations should keep
1366 # exciting things like link expansions from showing up in surprising
1367 # places.
1368 $text = $this->doTableStuff( $text );
1369
1370 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1371
1372 $text = $this->doDoubleUnderscore( $text );
1373
1374 $text = $this->doHeadings( $text );
1375 $text = $this->replaceInternalLinks( $text );
1376 $text = $this->doAllQuotes( $text );
1377 $text = $this->replaceExternalLinks( $text );
1378
1379 # replaceInternalLinks may sometimes leave behind
1380 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1381 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1382
1383 $text = $this->doMagicLinks( $text );
1384 $text = $this->formatHeadings( $text, $origText, $isMain );
1385
1386 return $text;
1387 }
1388
1389 /**
1390 * Helper function for parse() that transforms half-parsed HTML into fully
1391 * parsed HTML.
1392 *
1393 * @param string $text
1394 * @param bool $isMain
1395 * @param bool $linestart
1396 * @return string
1397 */
1398 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1399 $text = $this->mStripState->unstripGeneral( $text );
1400
1401 // Avoid PHP 7.1 warning from passing $this by reference
1402 $parser = $this;
1403
1404 if ( $isMain ) {
1405 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1406 }
1407
1408 # Clean up special characters, only run once, next-to-last before doBlockLevels
1409 $text = Sanitizer::armorFrenchSpaces( $text );
1410
1411 $text = $this->doBlockLevels( $text, $linestart );
1412
1413 $this->replaceLinkHolders( $text );
1414
1415 /**
1416 * The input doesn't get language converted if
1417 * a) It's disabled
1418 * b) Content isn't converted
1419 * c) It's a conversion table
1420 * d) it is an interface message (which is in the user language)
1421 */
1422 if ( !( $this->mOptions->getDisableContentConversion()
1423 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1424 ) {
1425 if ( !$this->mOptions->getInterfaceMessage() ) {
1426 # The position of the convert() call should not be changed. it
1427 # assumes that the links are all replaced and the only thing left
1428 # is the <nowiki> mark.
1429 $text = $this->getTargetLanguage()->convert( $text );
1430 }
1431 }
1432
1433 $text = $this->mStripState->unstripNoWiki( $text );
1434
1435 if ( $isMain ) {
1436 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1437 }
1438
1439 $text = $this->replaceTransparentTags( $text );
1440 $text = $this->mStripState->unstripGeneral( $text );
1441
1442 $text = Sanitizer::normalizeCharReferences( $text );
1443
1444 if ( MWTidy::isEnabled() ) {
1445 if ( $this->mOptions->getTidy() ) {
1446 $text = MWTidy::tidy( $text );
1447 }
1448 } else {
1449 # attempt to sanitize at least some nesting problems
1450 # (T4702 and quite a few others)
1451 $tidyregs = [
1452 # ''Something [http://www.cool.com cool''] -->
1453 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1454 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1455 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1456 # fix up an anchor inside another anchor, only
1457 # at least for a single single nested link (T5695)
1458 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1459 '\\1\\2</a>\\3</a>\\1\\4</a>',
1460 # fix div inside inline elements- doBlockLevels won't wrap a line which
1461 # contains a div, so fix it up here; replace
1462 # div with escaped text
1463 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1464 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1465 # remove empty italic or bold tag pairs, some
1466 # introduced by rules above
1467 '/<([bi])><\/\\1>/' => '',
1468 ];
1469
1470 $text = preg_replace(
1471 array_keys( $tidyregs ),
1472 array_values( $tidyregs ),
1473 $text );
1474 }
1475
1476 if ( $isMain ) {
1477 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1478 }
1479
1480 return $text;
1481 }
1482
1483 /**
1484 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1485 * magic external links.
1486 *
1487 * DML
1488 * @private
1489 *
1490 * @param string $text
1491 *
1492 * @return string
1493 */
1494 public function doMagicLinks( $text ) {
1495 $prots = wfUrlProtocolsWithoutProtRel();
1496 $urlChar = self::EXT_LINK_URL_CLASS;
1497 $addr = self::EXT_LINK_ADDR;
1498 $space = self::SPACE_NOT_NL; # non-newline space
1499 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1500 $spaces = "$space++"; # possessive match of 1 or more spaces
1501 $text = preg_replace_callback(
1502 '!(?: # Start cases
1503 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1504 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1505 (\b # m[3]: Free external links
1506 (?i:$prots)
1507 ($addr$urlChar*) # m[4]: Post-protocol path
1508 ) |
1509 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1510 ([0-9]+)\b |
1511 \bISBN $spaces ( # m[6]: ISBN, capture number
1512 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1513 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1514 [0-9Xx] # check digit
1515 )\b
1516 )!xu", [ $this, 'magicLinkCallback' ], $text );
1517 return $text;
1518 }
1519
1520 /**
1521 * @throws MWException
1522 * @param array $m
1523 * @return string HTML
1524 */
1525 public function magicLinkCallback( $m ) {
1526 if ( isset( $m[1] ) && $m[1] !== '' ) {
1527 # Skip anchor
1528 return $m[0];
1529 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1530 # Skip HTML element
1531 return $m[0];
1532 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1533 # Free external link
1534 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1535 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1536 # RFC or PMID
1537 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1538 if ( !$this->mOptions->getMagicRFCLinks() ) {
1539 return $m[0];
1540 }
1541 $keyword = 'RFC';
1542 $urlmsg = 'rfcurl';
1543 $cssClass = 'mw-magiclink-rfc';
1544 $trackingCat = 'magiclink-tracking-rfc';
1545 $id = $m[5];
1546 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1547 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1548 return $m[0];
1549 }
1550 $keyword = 'PMID';
1551 $urlmsg = 'pubmedurl';
1552 $cssClass = 'mw-magiclink-pmid';
1553 $trackingCat = 'magiclink-tracking-pmid';
1554 $id = $m[5];
1555 } else {
1556 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1557 substr( $m[0], 0, 20 ) . '"' );
1558 }
1559 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1560 $this->addTrackingCategory( $trackingCat );
1561 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1562 } elseif ( isset( $m[6] ) && $m[6] !== ''
1563 && $this->mOptions->getMagicISBNLinks()
1564 ) {
1565 # ISBN
1566 $isbn = $m[6];
1567 $space = self::SPACE_NOT_NL; # non-newline space
1568 $isbn = preg_replace( "/$space/", ' ', $isbn );
1569 $num = strtr( $isbn, [
1570 '-' => '',
1571 ' ' => '',
1572 'x' => 'X',
1573 ] );
1574 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1575 return $this->getLinkRenderer()->makeKnownLink(
1576 SpecialPage::getTitleFor( 'Booksources', $num ),
1577 "ISBN $isbn",
1578 [
1579 'class' => 'internal mw-magiclink-isbn',
1580 'title' => false // suppress title attribute
1581 ]
1582 );
1583 } else {
1584 return $m[0];
1585 }
1586 }
1587
1588 /**
1589 * Make a free external link, given a user-supplied URL
1590 *
1591 * @param string $url
1592 * @param int $numPostProto
1593 * The number of characters after the protocol.
1594 * @return string HTML
1595 * @private
1596 */
1597 public function makeFreeExternalLink( $url, $numPostProto ) {
1598 $trail = '';
1599
1600 # The characters '<' and '>' (which were escaped by
1601 # removeHTMLtags()) should not be included in
1602 # URLs, per RFC 2396.
1603 # Make &nbsp; terminate a URL as well (bug T84937)
1604 $m2 = [];
1605 if ( preg_match(
1606 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1607 $url,
1608 $m2,
1609 PREG_OFFSET_CAPTURE
1610 ) ) {
1611 $trail = substr( $url, $m2[0][1] ) . $trail;
1612 $url = substr( $url, 0, $m2[0][1] );
1613 }
1614
1615 # Move trailing punctuation to $trail
1616 $sep = ',;\.:!?';
1617 # If there is no left bracket, then consider right brackets fair game too
1618 if ( strpos( $url, '(' ) === false ) {
1619 $sep .= ')';
1620 }
1621
1622 $urlRev = strrev( $url );
1623 $numSepChars = strspn( $urlRev, $sep );
1624 # Don't break a trailing HTML entity by moving the ; into $trail
1625 # This is in hot code, so use substr_compare to avoid having to
1626 # create a new string object for the comparison
1627 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1628 # more optimization: instead of running preg_match with a $
1629 # anchor, which can be slow, do the match on the reversed
1630 # string starting at the desired offset.
1631 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1632 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1633 $numSepChars--;
1634 }
1635 }
1636 if ( $numSepChars ) {
1637 $trail = substr( $url, -$numSepChars ) . $trail;
1638 $url = substr( $url, 0, -$numSepChars );
1639 }
1640
1641 # Verify that we still have a real URL after trail removal, and
1642 # not just lone protocol
1643 if ( strlen( $trail ) >= $numPostProto ) {
1644 return $url . $trail;
1645 }
1646
1647 $url = Sanitizer::cleanUrl( $url );
1648
1649 # Is this an external image?
1650 $text = $this->maybeMakeExternalImage( $url );
1651 if ( $text === false ) {
1652 # Not an image, make a link
1653 $text = Linker::makeExternalLink( $url,
1654 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1655 true, 'free',
1656 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1657 # Register it in the output object...
1658 $this->mOutput->addExternalLink( $url );
1659 }
1660 return $text . $trail;
1661 }
1662
1663 /**
1664 * Parse headers and return html
1665 *
1666 * @private
1667 *
1668 * @param string $text
1669 *
1670 * @return string
1671 */
1672 public function doHeadings( $text ) {
1673 for ( $i = 6; $i >= 1; --$i ) {
1674 $h = str_repeat( '=', $i );
1675 // Trim non-newline whitespace from headings
1676 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1677 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1678 }
1679 return $text;
1680 }
1681
1682 /**
1683 * Replace single quotes with HTML markup
1684 * @private
1685 *
1686 * @param string $text
1687 *
1688 * @return string The altered text
1689 */
1690 public function doAllQuotes( $text ) {
1691 $outtext = '';
1692 $lines = StringUtils::explode( "\n", $text );
1693 foreach ( $lines as $line ) {
1694 $outtext .= $this->doQuotes( $line ) . "\n";
1695 }
1696 $outtext = substr( $outtext, 0, -1 );
1697 return $outtext;
1698 }
1699
1700 /**
1701 * Helper function for doAllQuotes()
1702 *
1703 * @param string $text
1704 *
1705 * @return string
1706 */
1707 public function doQuotes( $text ) {
1708 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1709 $countarr = count( $arr );
1710 if ( $countarr == 1 ) {
1711 return $text;
1712 }
1713
1714 // First, do some preliminary work. This may shift some apostrophes from
1715 // being mark-up to being text. It also counts the number of occurrences
1716 // of bold and italics mark-ups.
1717 $numbold = 0;
1718 $numitalics = 0;
1719 for ( $i = 1; $i < $countarr; $i += 2 ) {
1720 $thislen = strlen( $arr[$i] );
1721 // If there are ever four apostrophes, assume the first is supposed to
1722 // be text, and the remaining three constitute mark-up for bold text.
1723 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1724 if ( $thislen == 4 ) {
1725 $arr[$i - 1] .= "'";
1726 $arr[$i] = "'''";
1727 $thislen = 3;
1728 } elseif ( $thislen > 5 ) {
1729 // If there are more than 5 apostrophes in a row, assume they're all
1730 // text except for the last 5.
1731 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1732 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1733 $arr[$i] = "'''''";
1734 $thislen = 5;
1735 }
1736 // Count the number of occurrences of bold and italics mark-ups.
1737 if ( $thislen == 2 ) {
1738 $numitalics++;
1739 } elseif ( $thislen == 3 ) {
1740 $numbold++;
1741 } elseif ( $thislen == 5 ) {
1742 $numitalics++;
1743 $numbold++;
1744 }
1745 }
1746
1747 // If there is an odd number of both bold and italics, it is likely
1748 // that one of the bold ones was meant to be an apostrophe followed
1749 // by italics. Which one we cannot know for certain, but it is more
1750 // likely to be one that has a single-letter word before it.
1751 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1752 $firstsingleletterword = -1;
1753 $firstmultiletterword = -1;
1754 $firstspace = -1;
1755 for ( $i = 1; $i < $countarr; $i += 2 ) {
1756 if ( strlen( $arr[$i] ) == 3 ) {
1757 $x1 = substr( $arr[$i - 1], -1 );
1758 $x2 = substr( $arr[$i - 1], -2, 1 );
1759 if ( $x1 === ' ' ) {
1760 if ( $firstspace == -1 ) {
1761 $firstspace = $i;
1762 }
1763 } elseif ( $x2 === ' ' ) {
1764 $firstsingleletterword = $i;
1765 // if $firstsingleletterword is set, we don't
1766 // look at the other options, so we can bail early.
1767 break;
1768 } else {
1769 if ( $firstmultiletterword == -1 ) {
1770 $firstmultiletterword = $i;
1771 }
1772 }
1773 }
1774 }
1775
1776 // If there is a single-letter word, use it!
1777 if ( $firstsingleletterword > -1 ) {
1778 $arr[$firstsingleletterword] = "''";
1779 $arr[$firstsingleletterword - 1] .= "'";
1780 } elseif ( $firstmultiletterword > -1 ) {
1781 // If not, but there's a multi-letter word, use that one.
1782 $arr[$firstmultiletterword] = "''";
1783 $arr[$firstmultiletterword - 1] .= "'";
1784 } elseif ( $firstspace > -1 ) {
1785 // ... otherwise use the first one that has neither.
1786 // (notice that it is possible for all three to be -1 if, for example,
1787 // there is only one pentuple-apostrophe in the line)
1788 $arr[$firstspace] = "''";
1789 $arr[$firstspace - 1] .= "'";
1790 }
1791 }
1792
1793 // Now let's actually convert our apostrophic mush to HTML!
1794 $output = '';
1795 $buffer = '';
1796 $state = '';
1797 $i = 0;
1798 foreach ( $arr as $r ) {
1799 if ( ( $i % 2 ) == 0 ) {
1800 if ( $state === 'both' ) {
1801 $buffer .= $r;
1802 } else {
1803 $output .= $r;
1804 }
1805 } else {
1806 $thislen = strlen( $r );
1807 if ( $thislen == 2 ) {
1808 if ( $state === 'i' ) {
1809 $output .= '</i>';
1810 $state = '';
1811 } elseif ( $state === 'bi' ) {
1812 $output .= '</i>';
1813 $state = 'b';
1814 } elseif ( $state === 'ib' ) {
1815 $output .= '</b></i><b>';
1816 $state = 'b';
1817 } elseif ( $state === 'both' ) {
1818 $output .= '<b><i>' . $buffer . '</i>';
1819 $state = 'b';
1820 } else { // $state can be 'b' or ''
1821 $output .= '<i>';
1822 $state .= 'i';
1823 }
1824 } elseif ( $thislen == 3 ) {
1825 if ( $state === 'b' ) {
1826 $output .= '</b>';
1827 $state = '';
1828 } elseif ( $state === 'bi' ) {
1829 $output .= '</i></b><i>';
1830 $state = 'i';
1831 } elseif ( $state === 'ib' ) {
1832 $output .= '</b>';
1833 $state = 'i';
1834 } elseif ( $state === 'both' ) {
1835 $output .= '<i><b>' . $buffer . '</b>';
1836 $state = 'i';
1837 } else { // $state can be 'i' or ''
1838 $output .= '<b>';
1839 $state .= 'b';
1840 }
1841 } elseif ( $thislen == 5 ) {
1842 if ( $state === 'b' ) {
1843 $output .= '</b><i>';
1844 $state = 'i';
1845 } elseif ( $state === 'i' ) {
1846 $output .= '</i><b>';
1847 $state = 'b';
1848 } elseif ( $state === 'bi' ) {
1849 $output .= '</i></b>';
1850 $state = '';
1851 } elseif ( $state === 'ib' ) {
1852 $output .= '</b></i>';
1853 $state = '';
1854 } elseif ( $state === 'both' ) {
1855 $output .= '<i><b>' . $buffer . '</b></i>';
1856 $state = '';
1857 } else { // ($state == '')
1858 $buffer = '';
1859 $state = 'both';
1860 }
1861 }
1862 }
1863 $i++;
1864 }
1865 // Now close all remaining tags. Notice that the order is important.
1866 if ( $state === 'b' || $state === 'ib' ) {
1867 $output .= '</b>';
1868 }
1869 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1870 $output .= '</i>';
1871 }
1872 if ( $state === 'bi' ) {
1873 $output .= '</b>';
1874 }
1875 // There might be lonely ''''', so make sure we have a buffer
1876 if ( $state === 'both' && $buffer ) {
1877 $output .= '<b><i>' . $buffer . '</i></b>';
1878 }
1879 return $output;
1880 }
1881
1882 /**
1883 * Replace external links (REL)
1884 *
1885 * Note: this is all very hackish and the order of execution matters a lot.
1886 * Make sure to run tests/parser/parserTests.php if you change this code.
1887 *
1888 * @private
1889 *
1890 * @param string $text
1891 *
1892 * @throws MWException
1893 * @return string
1894 */
1895 public function replaceExternalLinks( $text ) {
1896 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1897 if ( $bits === false ) {
1898 throw new MWException( "PCRE needs to be compiled with "
1899 . "--enable-unicode-properties in order for MediaWiki to function" );
1900 }
1901 $s = array_shift( $bits );
1902
1903 $i = 0;
1904 while ( $i < count( $bits ) ) {
1905 $url = $bits[$i++];
1906 $i++; // protocol
1907 $text = $bits[$i++];
1908 $trail = $bits[$i++];
1909
1910 # The characters '<' and '>' (which were escaped by
1911 # removeHTMLtags()) should not be included in
1912 # URLs, per RFC 2396.
1913 $m2 = [];
1914 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1915 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1916 $url = substr( $url, 0, $m2[0][1] );
1917 }
1918
1919 # If the link text is an image URL, replace it with an <img> tag
1920 # This happened by accident in the original parser, but some people used it extensively
1921 $img = $this->maybeMakeExternalImage( $text );
1922 if ( $img !== false ) {
1923 $text = $img;
1924 }
1925
1926 $dtrail = '';
1927
1928 # Set linktype for CSS
1929 $linktype = 'text';
1930
1931 # No link text, e.g. [http://domain.tld/some.link]
1932 if ( $text == '' ) {
1933 # Autonumber
1934 $langObj = $this->getTargetLanguage();
1935 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1936 $linktype = 'autonumber';
1937 } else {
1938 # Have link text, e.g. [http://domain.tld/some.link text]s
1939 # Check for trail
1940 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1941 }
1942
1943 // Excluding protocol-relative URLs may avoid many false positives.
1944 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1945 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1946 }
1947
1948 $url = Sanitizer::cleanUrl( $url );
1949
1950 # Use the encoded URL
1951 # This means that users can paste URLs directly into the text
1952 # Funny characters like ö aren't valid in URLs anyway
1953 # This was changed in August 2004
1954 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1955 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1956
1957 # Register link in the output object.
1958 $this->mOutput->addExternalLink( $url );
1959 }
1960
1961 return $s;
1962 }
1963
1964 /**
1965 * Get the rel attribute for a particular external link.
1966 *
1967 * @since 1.21
1968 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1969 * nofollow if appropriate
1970 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1971 * @return string|null Rel attribute for $url
1972 */
1973 public static function getExternalLinkRel( $url = false, $title = null ) {
1974 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1975 $ns = $title ? $title->getNamespace() : false;
1976 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1977 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1978 ) {
1979 return 'nofollow';
1980 }
1981 return null;
1982 }
1983
1984 /**
1985 * Get an associative array of additional HTML attributes appropriate for a
1986 * particular external link. This currently may include rel => nofollow
1987 * (depending on configuration, namespace, and the URL's domain) and/or a
1988 * target attribute (depending on configuration).
1989 *
1990 * @param string $url URL to extract the domain from for rel =>
1991 * nofollow if appropriate
1992 * @return array Associative array of HTML attributes
1993 */
1994 public function getExternalLinkAttribs( $url ) {
1995 $attribs = [];
1996 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1997
1998 $target = $this->mOptions->getExternalLinkTarget();
1999 if ( $target ) {
2000 $attribs['target'] = $target;
2001 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2002 // T133507. New windows can navigate parent cross-origin.
2003 // Including noreferrer due to lacking browser
2004 // support of noopener. Eventually noreferrer should be removed.
2005 if ( $rel !== '' ) {
2006 $rel .= ' ';
2007 }
2008 $rel .= 'noreferrer noopener';
2009 }
2010 }
2011 $attribs['rel'] = $rel;
2012 return $attribs;
2013 }
2014
2015 /**
2016 * Replace unusual escape codes in a URL with their equivalent characters
2017 *
2018 * This generally follows the syntax defined in RFC 3986, with special
2019 * consideration for HTTP query strings.
2020 *
2021 * @param string $url
2022 * @return string
2023 */
2024 public static function normalizeLinkUrl( $url ) {
2025 # First, make sure unsafe characters are encoded
2026 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2027 function ( $m ) {
2028 return rawurlencode( $m[0] );
2029 },
2030 $url
2031 );
2032
2033 $ret = '';
2034 $end = strlen( $url );
2035
2036 # Fragment part - 'fragment'
2037 $start = strpos( $url, '#' );
2038 if ( $start !== false && $start < $end ) {
2039 $ret = self::normalizeUrlComponent(
2040 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2041 $end = $start;
2042 }
2043
2044 # Query part - 'query' minus &=+;
2045 $start = strpos( $url, '?' );
2046 if ( $start !== false && $start < $end ) {
2047 $ret = self::normalizeUrlComponent(
2048 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2049 $end = $start;
2050 }
2051
2052 # Scheme and path part - 'pchar'
2053 # (we assume no userinfo or encoded colons in the host)
2054 $ret = self::normalizeUrlComponent(
2055 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2056
2057 return $ret;
2058 }
2059
2060 private static function normalizeUrlComponent( $component, $unsafe ) {
2061 $callback = function ( $matches ) use ( $unsafe ) {
2062 $char = urldecode( $matches[0] );
2063 $ord = ord( $char );
2064 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2065 # Unescape it
2066 return $char;
2067 } else {
2068 # Leave it escaped, but use uppercase for a-f
2069 return strtoupper( $matches[0] );
2070 }
2071 };
2072 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2073 }
2074
2075 /**
2076 * make an image if it's allowed, either through the global
2077 * option, through the exception, or through the on-wiki whitelist
2078 *
2079 * @param string $url
2080 *
2081 * @return string
2082 */
2083 private function maybeMakeExternalImage( $url ) {
2084 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2085 $imagesexception = !empty( $imagesfrom );
2086 $text = false;
2087 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2088 if ( $imagesexception && is_array( $imagesfrom ) ) {
2089 $imagematch = false;
2090 foreach ( $imagesfrom as $match ) {
2091 if ( strpos( $url, $match ) === 0 ) {
2092 $imagematch = true;
2093 break;
2094 }
2095 }
2096 } elseif ( $imagesexception ) {
2097 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2098 } else {
2099 $imagematch = false;
2100 }
2101
2102 if ( $this->mOptions->getAllowExternalImages()
2103 || ( $imagesexception && $imagematch )
2104 ) {
2105 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2106 # Image found
2107 $text = Linker::makeExternalImage( $url );
2108 }
2109 }
2110 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2111 && preg_match( self::EXT_IMAGE_REGEX, $url )
2112 ) {
2113 $whitelist = explode(
2114 "\n",
2115 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2116 );
2117
2118 foreach ( $whitelist as $entry ) {
2119 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2120 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2121 continue;
2122 }
2123 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2124 # Image matches a whitelist entry
2125 $text = Linker::makeExternalImage( $url );
2126 break;
2127 }
2128 }
2129 }
2130 return $text;
2131 }
2132
2133 /**
2134 * Process [[ ]] wikilinks
2135 *
2136 * @param string $s
2137 *
2138 * @return string Processed text
2139 *
2140 * @private
2141 */
2142 public function replaceInternalLinks( $s ) {
2143 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2144 return $s;
2145 }
2146
2147 /**
2148 * Process [[ ]] wikilinks (RIL)
2149 * @param string &$s
2150 * @throws MWException
2151 * @return LinkHolderArray
2152 *
2153 * @private
2154 */
2155 public function replaceInternalLinks2( &$s ) {
2156 global $wgExtraInterlanguageLinkPrefixes;
2157
2158 static $tc = false, $e1, $e1_img;
2159 # the % is needed to support urlencoded titles as well
2160 if ( !$tc ) {
2161 $tc = Title::legalChars() . '#%';
2162 # Match a link having the form [[namespace:link|alternate]]trail
2163 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2164 # Match cases where there is no "]]", which might still be images
2165 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2166 }
2167
2168 $holders = new LinkHolderArray( $this );
2169
2170 # split the entire text string on occurrences of [[
2171 $a = StringUtils::explode( '[[', ' ' . $s );
2172 # get the first element (all text up to first [[), and remove the space we added
2173 $s = $a->current();
2174 $a->next();
2175 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2176 $s = substr( $s, 1 );
2177
2178 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2179 $e2 = null;
2180 if ( $useLinkPrefixExtension ) {
2181 # Match the end of a line for a word that's not followed by whitespace,
2182 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2183 $charset = $this->contLang->linkPrefixCharset();
2184 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2185 }
2186
2187 if ( is_null( $this->mTitle ) ) {
2188 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2189 }
2190 $nottalk = !$this->mTitle->isTalkPage();
2191
2192 if ( $useLinkPrefixExtension ) {
2193 $m = [];
2194 if ( preg_match( $e2, $s, $m ) ) {
2195 $first_prefix = $m[2];
2196 } else {
2197 $first_prefix = false;
2198 }
2199 } else {
2200 $prefix = '';
2201 }
2202
2203 $useSubpages = $this->areSubpagesAllowed();
2204
2205 # Loop for each link
2206 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2207 # Check for excessive memory usage
2208 if ( $holders->isBig() ) {
2209 # Too big
2210 # Do the existence check, replace the link holders and clear the array
2211 $holders->replace( $s );
2212 $holders->clear();
2213 }
2214
2215 if ( $useLinkPrefixExtension ) {
2216 if ( preg_match( $e2, $s, $m ) ) {
2217 $prefix = $m[2];
2218 $s = $m[1];
2219 } else {
2220 $prefix = '';
2221 }
2222 # first link
2223 if ( $first_prefix ) {
2224 $prefix = $first_prefix;
2225 $first_prefix = false;
2226 }
2227 }
2228
2229 $might_be_img = false;
2230
2231 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2232 $text = $m[2];
2233 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2234 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2235 # the real problem is with the $e1 regex
2236 # See T1500.
2237 # Still some problems for cases where the ] is meant to be outside punctuation,
2238 # and no image is in sight. See T4095.
2239 if ( $text !== ''
2240 && substr( $m[3], 0, 1 ) === ']'
2241 && strpos( $text, '[' ) !== false
2242 ) {
2243 $text .= ']'; # so that replaceExternalLinks($text) works later
2244 $m[3] = substr( $m[3], 1 );
2245 }
2246 # fix up urlencoded title texts
2247 if ( strpos( $m[1], '%' ) !== false ) {
2248 # Should anchors '#' also be rejected?
2249 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2250 }
2251 $trail = $m[3];
2252 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2253 # Invalid, but might be an image with a link in its caption
2254 $might_be_img = true;
2255 $text = $m[2];
2256 if ( strpos( $m[1], '%' ) !== false ) {
2257 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2258 }
2259 $trail = "";
2260 } else { # Invalid form; output directly
2261 $s .= $prefix . '[[' . $line;
2262 continue;
2263 }
2264
2265 $origLink = ltrim( $m[1], ' ' );
2266
2267 # Don't allow internal links to pages containing
2268 # PROTO: where PROTO is a valid URL protocol; these
2269 # should be external links.
2270 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2271 $s .= $prefix . '[[' . $line;
2272 continue;
2273 }
2274
2275 # Make subpage if necessary
2276 if ( $useSubpages ) {
2277 $link = $this->maybeDoSubpageLink( $origLink, $text );
2278 } else {
2279 $link = $origLink;
2280 }
2281
2282 // \x7f isn't a default legal title char, so most likely strip
2283 // markers will force us into the "invalid form" path above. But,
2284 // just in case, let's assert that xmlish tags aren't valid in
2285 // the title position.
2286 $unstrip = $this->mStripState->killMarkers( $link );
2287 $noMarkers = ( $unstrip === $link );
2288
2289 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2290 if ( $nt === null ) {
2291 $s .= $prefix . '[[' . $line;
2292 continue;
2293 }
2294
2295 $ns = $nt->getNamespace();
2296 $iw = $nt->getInterwiki();
2297
2298 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2299
2300 if ( $might_be_img ) { # if this is actually an invalid link
2301 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2302 $found = false;
2303 while ( true ) {
2304 # look at the next 'line' to see if we can close it there
2305 $a->next();
2306 $next_line = $a->current();
2307 if ( $next_line === false || $next_line === null ) {
2308 break;
2309 }
2310 $m = explode( ']]', $next_line, 3 );
2311 if ( count( $m ) == 3 ) {
2312 # the first ]] closes the inner link, the second the image
2313 $found = true;
2314 $text .= "[[{$m[0]}]]{$m[1]}";
2315 $trail = $m[2];
2316 break;
2317 } elseif ( count( $m ) == 2 ) {
2318 # if there's exactly one ]] that's fine, we'll keep looking
2319 $text .= "[[{$m[0]}]]{$m[1]}";
2320 } else {
2321 # if $next_line is invalid too, we need look no further
2322 $text .= '[[' . $next_line;
2323 break;
2324 }
2325 }
2326 if ( !$found ) {
2327 # we couldn't find the end of this imageLink, so output it raw
2328 # but don't ignore what might be perfectly normal links in the text we've examined
2329 $holders->merge( $this->replaceInternalLinks2( $text ) );
2330 $s .= "{$prefix}[[$link|$text";
2331 # note: no $trail, because without an end, there *is* no trail
2332 continue;
2333 }
2334 } else { # it's not an image, so output it raw
2335 $s .= "{$prefix}[[$link|$text";
2336 # note: no $trail, because without an end, there *is* no trail
2337 continue;
2338 }
2339 }
2340
2341 $wasblank = ( $text == '' );
2342 if ( $wasblank ) {
2343 $text = $link;
2344 if ( !$noforce ) {
2345 # Strip off leading ':'
2346 $text = substr( $text, 1 );
2347 }
2348 } else {
2349 # T6598 madness. Handle the quotes only if they come from the alternate part
2350 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2351 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2352 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2353 $text = $this->doQuotes( $text );
2354 }
2355
2356 # Link not escaped by : , create the various objects
2357 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2358 # Interwikis
2359 if (
2360 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2361 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2362 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2363 )
2364 ) {
2365 # T26502: filter duplicates
2366 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2367 $this->mLangLinkLanguages[$iw] = true;
2368 $this->mOutput->addLanguageLink( $nt->getFullText() );
2369 }
2370
2371 /**
2372 * Strip the whitespace interwiki links produce, see T10897
2373 */
2374 $s = rtrim( $s . $prefix ) . $trail; # T175416
2375 continue;
2376 }
2377
2378 if ( $ns == NS_FILE ) {
2379 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2380 if ( $wasblank ) {
2381 # if no parameters were passed, $text
2382 # becomes something like "File:Foo.png",
2383 # which we don't want to pass on to the
2384 # image generator
2385 $text = '';
2386 } else {
2387 # recursively parse links inside the image caption
2388 # actually, this will parse them in any other parameters, too,
2389 # but it might be hard to fix that, and it doesn't matter ATM
2390 $text = $this->replaceExternalLinks( $text );
2391 $holders->merge( $this->replaceInternalLinks2( $text ) );
2392 }
2393 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2394 $s .= $prefix . $this->armorLinks(
2395 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2396 continue;
2397 }
2398 } elseif ( $ns == NS_CATEGORY ) {
2399 /**
2400 * Strip the whitespace Category links produce, see T2087
2401 */
2402 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2403
2404 if ( $wasblank ) {
2405 $sortkey = $this->getDefaultSort();
2406 } else {
2407 $sortkey = $text;
2408 }
2409 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2410 $sortkey = str_replace( "\n", '', $sortkey );
2411 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2412 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2413
2414 continue;
2415 }
2416 }
2417
2418 # Self-link checking. For some languages, variants of the title are checked in
2419 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2420 # for linking to a different variant.
2421 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2422 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2423 continue;
2424 }
2425
2426 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2427 # @todo FIXME: Should do batch file existence checks, see comment below
2428 if ( $ns == NS_MEDIA ) {
2429 # Give extensions a chance to select the file revision for us
2430 $options = [];
2431 $descQuery = false;
2432 Hooks::run( 'BeforeParserFetchFileAndTitle',
2433 [ $this, $nt, &$options, &$descQuery ] );
2434 # Fetch and register the file (file title may be different via hooks)
2435 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2436 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2437 $s .= $prefix . $this->armorLinks(
2438 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2439 continue;
2440 }
2441
2442 # Some titles, such as valid special pages or files in foreign repos, should
2443 # be shown as bluelinks even though they're not included in the page table
2444 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2445 # batch file existence checks for NS_FILE and NS_MEDIA
2446 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2447 $this->mOutput->addLink( $nt );
2448 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2449 } else {
2450 # Links will be added to the output link list after checking
2451 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2452 }
2453 }
2454 return $holders;
2455 }
2456
2457 /**
2458 * Render a forced-blue link inline; protect against double expansion of
2459 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2460 * Since this little disaster has to split off the trail text to avoid
2461 * breaking URLs in the following text without breaking trails on the
2462 * wiki links, it's been made into a horrible function.
2463 *
2464 * @param Title $nt
2465 * @param string $text
2466 * @param string $trail
2467 * @param string $prefix
2468 * @return string HTML-wikitext mix oh yuck
2469 */
2470 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2471 list( $inside, $trail ) = Linker::splitTrail( $trail );
2472
2473 if ( $text == '' ) {
2474 $text = htmlspecialchars( $nt->getPrefixedText() );
2475 }
2476
2477 $link = $this->getLinkRenderer()->makeKnownLink(
2478 $nt, new HtmlArmor( "$prefix$text$inside" )
2479 );
2480
2481 return $this->armorLinks( $link ) . $trail;
2482 }
2483
2484 /**
2485 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2486 * going to go through further parsing steps before inline URL expansion.
2487 *
2488 * Not needed quite as much as it used to be since free links are a bit
2489 * more sensible these days. But bracketed links are still an issue.
2490 *
2491 * @param string $text More-or-less HTML
2492 * @return string Less-or-more HTML with NOPARSE bits
2493 */
2494 public function armorLinks( $text ) {
2495 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2496 self::MARKER_PREFIX . "NOPARSE$1", $text );
2497 }
2498
2499 /**
2500 * Return true if subpage links should be expanded on this page.
2501 * @return bool
2502 */
2503 public function areSubpagesAllowed() {
2504 # Some namespaces don't allow subpages
2505 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2506 }
2507
2508 /**
2509 * Handle link to subpage if necessary
2510 *
2511 * @param string $target The source of the link
2512 * @param string &$text The link text, modified as necessary
2513 * @return string The full name of the link
2514 * @private
2515 */
2516 public function maybeDoSubpageLink( $target, &$text ) {
2517 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2518 }
2519
2520 /**
2521 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2522 *
2523 * @param string $text
2524 * @param bool $linestart Whether or not this is at the start of a line.
2525 * @private
2526 * @return string The lists rendered as HTML
2527 */
2528 public function doBlockLevels( $text, $linestart ) {
2529 return BlockLevelPass::doBlockLevels( $text, $linestart );
2530 }
2531
2532 /**
2533 * Return value of a magic variable (like PAGENAME)
2534 *
2535 * @private
2536 *
2537 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2538 * @param bool|PPFrame $frame
2539 *
2540 * @throws MWException
2541 * @return string
2542 */
2543 public function getVariableValue( $index, $frame = false ) {
2544 global $wgSitename, $wgServer, $wgServerName;
2545 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2546
2547 if ( is_null( $this->mTitle ) ) {
2548 // If no title set, bad things are going to happen
2549 // later. Title should always be set since this
2550 // should only be called in the middle of a parse
2551 // operation (but the unit-tests do funky stuff)
2552 throw new MWException( __METHOD__ . ' Should only be '
2553 . ' called while parsing (no title set)' );
2554 }
2555
2556 // Avoid PHP 7.1 warning from passing $this by reference
2557 $parser = $this;
2558
2559 /**
2560 * Some of these require message or data lookups and can be
2561 * expensive to check many times.
2562 */
2563 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2564 if ( isset( $this->mVarCache[$index] ) ) {
2565 return $this->mVarCache[$index];
2566 }
2567 }
2568
2569 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2570 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2571
2572 $pageLang = $this->getFunctionLang();
2573
2574 switch ( $index ) {
2575 case '!':
2576 $value = '|';
2577 break;
2578 case 'currentmonth':
2579 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2580 break;
2581 case 'currentmonth1':
2582 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2583 break;
2584 case 'currentmonthname':
2585 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2586 break;
2587 case 'currentmonthnamegen':
2588 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2589 break;
2590 case 'currentmonthabbrev':
2591 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2592 break;
2593 case 'currentday':
2594 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2595 break;
2596 case 'currentday2':
2597 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2598 break;
2599 case 'localmonth':
2600 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2601 break;
2602 case 'localmonth1':
2603 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2604 break;
2605 case 'localmonthname':
2606 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2607 break;
2608 case 'localmonthnamegen':
2609 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2610 break;
2611 case 'localmonthabbrev':
2612 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2613 break;
2614 case 'localday':
2615 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2616 break;
2617 case 'localday2':
2618 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2619 break;
2620 case 'pagename':
2621 $value = wfEscapeWikiText( $this->mTitle->getText() );
2622 break;
2623 case 'pagenamee':
2624 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2625 break;
2626 case 'fullpagename':
2627 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2628 break;
2629 case 'fullpagenamee':
2630 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2631 break;
2632 case 'subpagename':
2633 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2634 break;
2635 case 'subpagenamee':
2636 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2637 break;
2638 case 'rootpagename':
2639 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2640 break;
2641 case 'rootpagenamee':
2642 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2643 ' ',
2644 '_',
2645 $this->mTitle->getRootText()
2646 ) ) );
2647 break;
2648 case 'basepagename':
2649 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2650 break;
2651 case 'basepagenamee':
2652 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2653 ' ',
2654 '_',
2655 $this->mTitle->getBaseText()
2656 ) ) );
2657 break;
2658 case 'talkpagename':
2659 if ( $this->mTitle->canHaveTalkPage() ) {
2660 $talkPage = $this->mTitle->getTalkPage();
2661 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2662 } else {
2663 $value = '';
2664 }
2665 break;
2666 case 'talkpagenamee':
2667 if ( $this->mTitle->canHaveTalkPage() ) {
2668 $talkPage = $this->mTitle->getTalkPage();
2669 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2670 } else {
2671 $value = '';
2672 }
2673 break;
2674 case 'subjectpagename':
2675 $subjPage = $this->mTitle->getSubjectPage();
2676 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2677 break;
2678 case 'subjectpagenamee':
2679 $subjPage = $this->mTitle->getSubjectPage();
2680 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2681 break;
2682 case 'pageid': // requested in T25427
2683 $pageid = $this->getTitle()->getArticleID();
2684 if ( $pageid == 0 ) {
2685 # 0 means the page doesn't exist in the database,
2686 # which means the user is previewing a new page.
2687 # The vary-revision flag must be set, because the magic word
2688 # will have a different value once the page is saved.
2689 $this->mOutput->setFlag( 'vary-revision' );
2690 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2691 }
2692 $value = $pageid ?: null;
2693 break;
2694 case 'revisionid':
2695 # Let the edit saving system know we should parse the page
2696 # *after* a revision ID has been assigned.
2697 $this->mOutput->setFlag( 'vary-revision-id' );
2698 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2699 $value = $this->mRevisionId;
2700
2701 if ( !$value ) {
2702 $rev = $this->getRevisionObject();
2703 if ( $rev ) {
2704 $value = $rev->getId();
2705 }
2706 }
2707
2708 if ( !$value ) {
2709 $value = $this->mOptions->getSpeculativeRevId();
2710 if ( $value ) {
2711 $this->mOutput->setSpeculativeRevIdUsed( $value );
2712 }
2713 }
2714 break;
2715 case 'revisionday':
2716 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2717 break;
2718 case 'revisionday2':
2719 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2720 break;
2721 case 'revisionmonth':
2722 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2723 break;
2724 case 'revisionmonth1':
2725 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2726 break;
2727 case 'revisionyear':
2728 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2729 break;
2730 case 'revisiontimestamp':
2731 # Let the edit saving system know we should parse the page
2732 # *after* a revision ID has been assigned. This is for null edits.
2733 $this->mOutput->setFlag( 'vary-revision' );
2734 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2735 $value = $this->getRevisionTimestamp();
2736 break;
2737 case 'revisionuser':
2738 # Let the edit saving system know we should parse the page
2739 # *after* a revision ID has been assigned for null edits.
2740 $this->mOutput->setFlag( 'vary-user' );
2741 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2742 $value = $this->getRevisionUser();
2743 break;
2744 case 'revisionsize':
2745 $value = $this->getRevisionSize();
2746 break;
2747 case 'namespace':
2748 $value = str_replace( '_', ' ',
2749 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2750 break;
2751 case 'namespacee':
2752 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2753 break;
2754 case 'namespacenumber':
2755 $value = $this->mTitle->getNamespace();
2756 break;
2757 case 'talkspace':
2758 $value = $this->mTitle->canHaveTalkPage()
2759 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2760 : '';
2761 break;
2762 case 'talkspacee':
2763 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2764 break;
2765 case 'subjectspace':
2766 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2767 break;
2768 case 'subjectspacee':
2769 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2770 break;
2771 case 'currentdayname':
2772 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2773 break;
2774 case 'currentyear':
2775 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2776 break;
2777 case 'currenttime':
2778 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2779 break;
2780 case 'currenthour':
2781 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2782 break;
2783 case 'currentweek':
2784 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2785 # int to remove the padding
2786 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2787 break;
2788 case 'currentdow':
2789 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2790 break;
2791 case 'localdayname':
2792 $value = $pageLang->getWeekdayName(
2793 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2794 );
2795 break;
2796 case 'localyear':
2797 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2798 break;
2799 case 'localtime':
2800 $value = $pageLang->time(
2801 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2802 false,
2803 false
2804 );
2805 break;
2806 case 'localhour':
2807 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2808 break;
2809 case 'localweek':
2810 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2811 # int to remove the padding
2812 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2813 break;
2814 case 'localdow':
2815 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2816 break;
2817 case 'numberofarticles':
2818 $value = $pageLang->formatNum( SiteStats::articles() );
2819 break;
2820 case 'numberoffiles':
2821 $value = $pageLang->formatNum( SiteStats::images() );
2822 break;
2823 case 'numberofusers':
2824 $value = $pageLang->formatNum( SiteStats::users() );
2825 break;
2826 case 'numberofactiveusers':
2827 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2828 break;
2829 case 'numberofpages':
2830 $value = $pageLang->formatNum( SiteStats::pages() );
2831 break;
2832 case 'numberofadmins':
2833 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2834 break;
2835 case 'numberofedits':
2836 $value = $pageLang->formatNum( SiteStats::edits() );
2837 break;
2838 case 'currenttimestamp':
2839 $value = wfTimestamp( TS_MW, $ts );
2840 break;
2841 case 'localtimestamp':
2842 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2843 break;
2844 case 'currentversion':
2845 $value = SpecialVersion::getVersion();
2846 break;
2847 case 'articlepath':
2848 return $wgArticlePath;
2849 case 'sitename':
2850 return $wgSitename;
2851 case 'server':
2852 return $wgServer;
2853 case 'servername':
2854 return $wgServerName;
2855 case 'scriptpath':
2856 return $wgScriptPath;
2857 case 'stylepath':
2858 return $wgStylePath;
2859 case 'directionmark':
2860 return $pageLang->getDirMark();
2861 case 'contentlanguage':
2862 global $wgLanguageCode;
2863 return $wgLanguageCode;
2864 case 'pagelanguage':
2865 $value = $pageLang->getCode();
2866 break;
2867 case 'cascadingsources':
2868 $value = CoreParserFunctions::cascadingsources( $this );
2869 break;
2870 default:
2871 $ret = null;
2872 Hooks::run(
2873 'ParserGetVariableValueSwitch',
2874 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2875 );
2876
2877 return $ret;
2878 }
2879
2880 if ( $index ) {
2881 $this->mVarCache[$index] = $value;
2882 }
2883
2884 return $value;
2885 }
2886
2887 /**
2888 * @param int $start
2889 * @param int $len
2890 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2891 * @param string $variable Parser variable name
2892 * @return string
2893 */
2894 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2895 # Get the timezone-adjusted timestamp to be used for this revision
2896 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2897 # Possibly set vary-revision if there is not yet an associated revision
2898 if ( !$this->getRevisionObject() ) {
2899 # Get the timezone-adjusted timestamp $mtts seconds in the future
2900 $resThen = substr(
2901 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2902 $start,
2903 $len
2904 );
2905
2906 if ( $resNow !== $resThen ) {
2907 # Let the edit saving system know we should parse the page
2908 # *after* a revision ID has been assigned. This is for null edits.
2909 $this->mOutput->setFlag( 'vary-revision' );
2910 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2911 }
2912 }
2913
2914 return $resNow;
2915 }
2916
2917 /**
2918 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2919 *
2920 * @private
2921 */
2922 public function initialiseVariables() {
2923 $variableIDs = $this->magicWordFactory->getVariableIDs();
2924 $substIDs = $this->magicWordFactory->getSubstIDs();
2925
2926 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
2927 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
2928 }
2929
2930 /**
2931 * Preprocess some wikitext and return the document tree.
2932 * This is the ghost of replace_variables().
2933 *
2934 * @param string $text The text to parse
2935 * @param int $flags Bitwise combination of:
2936 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2937 * included. Default is to assume a direct page view.
2938 *
2939 * The generated DOM tree must depend only on the input text and the flags.
2940 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2941 *
2942 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2943 * change in the DOM tree for a given text, must be passed through the section identifier
2944 * in the section edit link and thus back to extractSections().
2945 *
2946 * The output of this function is currently only cached in process memory, but a persistent
2947 * cache may be implemented at a later date which takes further advantage of these strict
2948 * dependency requirements.
2949 *
2950 * @return PPNode
2951 */
2952 public function preprocessToDom( $text, $flags = 0 ) {
2953 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2954 return $dom;
2955 }
2956
2957 /**
2958 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2959 *
2960 * @param string $s
2961 *
2962 * @return array
2963 */
2964 public static function splitWhitespace( $s ) {
2965 $ltrimmed = ltrim( $s );
2966 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2967 $trimmed = rtrim( $ltrimmed );
2968 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2969 if ( $diff > 0 ) {
2970 $w2 = substr( $ltrimmed, -$diff );
2971 } else {
2972 $w2 = '';
2973 }
2974 return [ $w1, $trimmed, $w2 ];
2975 }
2976
2977 /**
2978 * Replace magic variables, templates, and template arguments
2979 * with the appropriate text. Templates are substituted recursively,
2980 * taking care to avoid infinite loops.
2981 *
2982 * Note that the substitution depends on value of $mOutputType:
2983 * self::OT_WIKI: only {{subst:}} templates
2984 * self::OT_PREPROCESS: templates but not extension tags
2985 * self::OT_HTML: all templates and extension tags
2986 *
2987 * @param string $text The text to transform
2988 * @param bool|PPFrame $frame Object describing the arguments passed to the
2989 * template. Arguments may also be provided as an associative array, as
2990 * was the usual case before MW1.12. Providing arguments this way may be
2991 * useful for extensions wishing to perform variable replacement
2992 * explicitly.
2993 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2994 * double-brace expansion.
2995 * @return string
2996 */
2997 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2998 # Is there any text? Also, Prevent too big inclusions!
2999 $textSize = strlen( $text );
3000 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3001 return $text;
3002 }
3003
3004 if ( $frame === false ) {
3005 $frame = $this->getPreprocessor()->newFrame();
3006 } elseif ( !( $frame instanceof PPFrame ) ) {
3007 wfDebug( __METHOD__ . " called using plain parameters instead of "
3008 . "a PPFrame instance. Creating custom frame.\n" );
3009 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3010 }
3011
3012 $dom = $this->preprocessToDom( $text );
3013 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3014 $text = $frame->expand( $dom, $flags );
3015
3016 return $text;
3017 }
3018
3019 /**
3020 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3021 *
3022 * @param array $args
3023 *
3024 * @return array
3025 */
3026 public static function createAssocArgs( $args ) {
3027 $assocArgs = [];
3028 $index = 1;
3029 foreach ( $args as $arg ) {
3030 $eqpos = strpos( $arg, '=' );
3031 if ( $eqpos === false ) {
3032 $assocArgs[$index++] = $arg;
3033 } else {
3034 $name = trim( substr( $arg, 0, $eqpos ) );
3035 $value = trim( substr( $arg, $eqpos + 1 ) );
3036 if ( $value === false ) {
3037 $value = '';
3038 }
3039 if ( $name !== false ) {
3040 $assocArgs[$name] = $value;
3041 }
3042 }
3043 }
3044
3045 return $assocArgs;
3046 }
3047
3048 /**
3049 * Warn the user when a parser limitation is reached
3050 * Will warn at most once the user per limitation type
3051 *
3052 * The results are shown during preview and run through the Parser (See EditPage.php)
3053 *
3054 * @param string $limitationType Should be one of:
3055 * 'expensive-parserfunction' (corresponding messages:
3056 * 'expensive-parserfunction-warning',
3057 * 'expensive-parserfunction-category')
3058 * 'post-expand-template-argument' (corresponding messages:
3059 * 'post-expand-template-argument-warning',
3060 * 'post-expand-template-argument-category')
3061 * 'post-expand-template-inclusion' (corresponding messages:
3062 * 'post-expand-template-inclusion-warning',
3063 * 'post-expand-template-inclusion-category')
3064 * 'node-count-exceeded' (corresponding messages:
3065 * 'node-count-exceeded-warning',
3066 * 'node-count-exceeded-category')
3067 * 'expansion-depth-exceeded' (corresponding messages:
3068 * 'expansion-depth-exceeded-warning',
3069 * 'expansion-depth-exceeded-category')
3070 * @param string|int|null $current Current value
3071 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3072 * exceeded, provide the values (optional)
3073 */
3074 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3075 # does no harm if $current and $max are present but are unnecessary for the message
3076 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3077 # only during preview, and that would split the parser cache unnecessarily.
3078 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3079 ->text();
3080 $this->mOutput->addWarning( $warning );
3081 $this->addTrackingCategory( "$limitationType-category" );
3082 }
3083
3084 /**
3085 * Return the text of a template, after recursively
3086 * replacing any variables or templates within the template.
3087 *
3088 * @param array $piece The parts of the template
3089 * $piece['title']: the title, i.e. the part before the |
3090 * $piece['parts']: the parameter array
3091 * $piece['lineStart']: whether the brace was at the start of a line
3092 * @param PPFrame $frame The current frame, contains template arguments
3093 * @throws Exception
3094 * @return string The text of the template
3095 */
3096 public function braceSubstitution( $piece, $frame ) {
3097 // Flags
3098
3099 // $text has been filled
3100 $found = false;
3101 // wiki markup in $text should be escaped
3102 $nowiki = false;
3103 // $text is HTML, armour it against wikitext transformation
3104 $isHTML = false;
3105 // Force interwiki transclusion to be done in raw mode not rendered
3106 $forceRawInterwiki = false;
3107 // $text is a DOM node needing expansion in a child frame
3108 $isChildObj = false;
3109 // $text is a DOM node needing expansion in the current frame
3110 $isLocalObj = false;
3111
3112 # Title object, where $text came from
3113 $title = false;
3114
3115 # $part1 is the bit before the first |, and must contain only title characters.
3116 # Various prefixes will be stripped from it later.
3117 $titleWithSpaces = $frame->expand( $piece['title'] );
3118 $part1 = trim( $titleWithSpaces );
3119 $titleText = false;
3120
3121 # Original title text preserved for various purposes
3122 $originalTitle = $part1;
3123
3124 # $args is a list of argument nodes, starting from index 0, not including $part1
3125 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3126 # below won't work b/c this $args isn't an object
3127 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3128
3129 $profileSection = null; // profile templates
3130
3131 # SUBST
3132 if ( !$found ) {
3133 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3134
3135 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3136 # Decide whether to expand template or keep wikitext as-is.
3137 if ( $this->ot['wiki'] ) {
3138 if ( $substMatch === false ) {
3139 $literal = true; # literal when in PST with no prefix
3140 } else {
3141 $literal = false; # expand when in PST with subst: or safesubst:
3142 }
3143 } else {
3144 if ( $substMatch == 'subst' ) {
3145 $literal = true; # literal when not in PST with plain subst:
3146 } else {
3147 $literal = false; # expand when not in PST with safesubst: or no prefix
3148 }
3149 }
3150 if ( $literal ) {
3151 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3152 $isLocalObj = true;
3153 $found = true;
3154 }
3155 }
3156
3157 # Variables
3158 if ( !$found && $args->getLength() == 0 ) {
3159 $id = $this->mVariables->matchStartToEnd( $part1 );
3160 if ( $id !== false ) {
3161 $text = $this->getVariableValue( $id, $frame );
3162 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3163 $this->mOutput->updateCacheExpiry(
3164 $this->magicWordFactory->getCacheTTL( $id ) );
3165 }
3166 $found = true;
3167 }
3168 }
3169
3170 # MSG, MSGNW and RAW
3171 if ( !$found ) {
3172 # Check for MSGNW:
3173 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3174 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3175 $nowiki = true;
3176 } else {
3177 # Remove obsolete MSG:
3178 $mwMsg = $this->magicWordFactory->get( 'msg' );
3179 $mwMsg->matchStartAndRemove( $part1 );
3180 }
3181
3182 # Check for RAW:
3183 $mwRaw = $this->magicWordFactory->get( 'raw' );
3184 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3185 $forceRawInterwiki = true;
3186 }
3187 }
3188
3189 # Parser functions
3190 if ( !$found ) {
3191 $colonPos = strpos( $part1, ':' );
3192 if ( $colonPos !== false ) {
3193 $func = substr( $part1, 0, $colonPos );
3194 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3195 $argsLength = $args->getLength();
3196 for ( $i = 0; $i < $argsLength; $i++ ) {
3197 $funcArgs[] = $args->item( $i );
3198 }
3199
3200 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3201
3202 // Extract any forwarded flags
3203 if ( isset( $result['title'] ) ) {
3204 $title = $result['title'];
3205 }
3206 if ( isset( $result['found'] ) ) {
3207 $found = $result['found'];
3208 }
3209 if ( array_key_exists( 'text', $result ) ) {
3210 // a string or null
3211 $text = $result['text'];
3212 }
3213 if ( isset( $result['nowiki'] ) ) {
3214 $nowiki = $result['nowiki'];
3215 }
3216 if ( isset( $result['isHTML'] ) ) {
3217 $isHTML = $result['isHTML'];
3218 }
3219 if ( isset( $result['forceRawInterwiki'] ) ) {
3220 $forceRawInterwiki = $result['forceRawInterwiki'];
3221 }
3222 if ( isset( $result['isChildObj'] ) ) {
3223 $isChildObj = $result['isChildObj'];
3224 }
3225 if ( isset( $result['isLocalObj'] ) ) {
3226 $isLocalObj = $result['isLocalObj'];
3227 }
3228 }
3229 }
3230
3231 # Finish mangling title and then check for loops.
3232 # Set $title to a Title object and $titleText to the PDBK
3233 if ( !$found ) {
3234 $ns = NS_TEMPLATE;
3235 # Split the title into page and subpage
3236 $subpage = '';
3237 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3238 if ( $part1 !== $relative ) {
3239 $part1 = $relative;
3240 $ns = $this->mTitle->getNamespace();
3241 }
3242 $title = Title::newFromText( $part1, $ns );
3243 if ( $title ) {
3244 $titleText = $title->getPrefixedText();
3245 # Check for language variants if the template is not found
3246 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3247 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3248 }
3249 # Do recursion depth check
3250 $limit = $this->mOptions->getMaxTemplateDepth();
3251 if ( $frame->depth >= $limit ) {
3252 $found = true;
3253 $text = '<span class="error">'
3254 . wfMessage( 'parser-template-recursion-depth-warning' )
3255 ->numParams( $limit )->inContentLanguage()->text()
3256 . '</span>';
3257 }
3258 }
3259 }
3260
3261 # Load from database
3262 if ( !$found && $title ) {
3263 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3264 if ( !$title->isExternal() ) {
3265 if ( $title->isSpecialPage()
3266 && $this->mOptions->getAllowSpecialInclusion()
3267 && $this->ot['html']
3268 ) {
3269 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3270 // Pass the template arguments as URL parameters.
3271 // "uselang" will have no effect since the Language object
3272 // is forced to the one defined in ParserOptions.
3273 $pageArgs = [];
3274 $argsLength = $args->getLength();
3275 for ( $i = 0; $i < $argsLength; $i++ ) {
3276 $bits = $args->item( $i )->splitArg();
3277 if ( strval( $bits['index'] ) === '' ) {
3278 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3279 $value = trim( $frame->expand( $bits['value'] ) );
3280 $pageArgs[$name] = $value;
3281 }
3282 }
3283
3284 // Create a new context to execute the special page
3285 $context = new RequestContext;
3286 $context->setTitle( $title );
3287 $context->setRequest( new FauxRequest( $pageArgs ) );
3288 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3289 $context->setUser( $this->getUser() );
3290 } else {
3291 // If this page is cached, then we better not be per user.
3292 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3293 }
3294 $context->setLanguage( $this->mOptions->getUserLangObj() );
3295 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3296 if ( $ret ) {
3297 $text = $context->getOutput()->getHTML();
3298 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3299 $found = true;
3300 $isHTML = true;
3301 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3302 $this->mOutput->updateRuntimeAdaptiveExpiry(
3303 $specialPage->maxIncludeCacheTime()
3304 );
3305 }
3306 }
3307 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3308 $found = false; # access denied
3309 wfDebug( __METHOD__ . ": template inclusion denied for " .
3310 $title->getPrefixedDBkey() . "\n" );
3311 } else {
3312 list( $text, $title ) = $this->getTemplateDom( $title );
3313 if ( $text !== false ) {
3314 $found = true;
3315 $isChildObj = true;
3316 }
3317 }
3318
3319 # If the title is valid but undisplayable, make a link to it
3320 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3321 $text = "[[:$titleText]]";
3322 $found = true;
3323 }
3324 } elseif ( $title->isTrans() ) {
3325 # Interwiki transclusion
3326 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3327 $text = $this->interwikiTransclude( $title, 'render' );
3328 $isHTML = true;
3329 } else {
3330 $text = $this->interwikiTransclude( $title, 'raw' );
3331 # Preprocess it like a template
3332 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3333 $isChildObj = true;
3334 }
3335 $found = true;
3336 }
3337
3338 # Do infinite loop check
3339 # This has to be done after redirect resolution to avoid infinite loops via redirects
3340 if ( !$frame->loopCheck( $title ) ) {
3341 $found = true;
3342 $text = '<span class="error">'
3343 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3344 . '</span>';
3345 $this->addTrackingCategory( 'template-loop-category' );
3346 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3347 wfEscapeWikiText( $titleText ) )->text() );
3348 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3349 }
3350 }
3351
3352 # If we haven't found text to substitute by now, we're done
3353 # Recover the source wikitext and return it
3354 if ( !$found ) {
3355 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3356 if ( $profileSection ) {
3357 $this->mProfiler->scopedProfileOut( $profileSection );
3358 }
3359 return [ 'object' => $text ];
3360 }
3361
3362 # Expand DOM-style return values in a child frame
3363 if ( $isChildObj ) {
3364 # Clean up argument array
3365 $newFrame = $frame->newChild( $args, $title );
3366
3367 if ( $nowiki ) {
3368 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3369 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3370 # Expansion is eligible for the empty-frame cache
3371 $text = $newFrame->cachedExpand( $titleText, $text );
3372 } else {
3373 # Uncached expansion
3374 $text = $newFrame->expand( $text );
3375 }
3376 }
3377 if ( $isLocalObj && $nowiki ) {
3378 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3379 $isLocalObj = false;
3380 }
3381
3382 if ( $profileSection ) {
3383 $this->mProfiler->scopedProfileOut( $profileSection );
3384 }
3385
3386 # Replace raw HTML by a placeholder
3387 if ( $isHTML ) {
3388 $text = $this->insertStripItem( $text );
3389 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3390 # Escape nowiki-style return values
3391 $text = wfEscapeWikiText( $text );
3392 } elseif ( is_string( $text )
3393 && !$piece['lineStart']
3394 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3395 ) {
3396 # T2529: if the template begins with a table or block-level
3397 # element, it should be treated as beginning a new line.
3398 # This behavior is somewhat controversial.
3399 $text = "\n" . $text;
3400 }
3401
3402 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3403 # Error, oversize inclusion
3404 if ( $titleText !== false ) {
3405 # Make a working, properly escaped link if possible (T25588)
3406 $text = "[[:$titleText]]";
3407 } else {
3408 # This will probably not be a working link, but at least it may
3409 # provide some hint of where the problem is
3410 preg_replace( '/^:/', '', $originalTitle );
3411 $text = "[[:$originalTitle]]";
3412 }
3413 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3414 . 'post-expand include size too large -->' );
3415 $this->limitationWarn( 'post-expand-template-inclusion' );
3416 }
3417
3418 if ( $isLocalObj ) {
3419 $ret = [ 'object' => $text ];
3420 } else {
3421 $ret = [ 'text' => $text ];
3422 }
3423
3424 return $ret;
3425 }
3426
3427 /**
3428 * Call a parser function and return an array with text and flags.
3429 *
3430 * The returned array will always contain a boolean 'found', indicating
3431 * whether the parser function was found or not. It may also contain the
3432 * following:
3433 * text: string|object, resulting wikitext or PP DOM object
3434 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3435 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3436 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3437 * nowiki: bool, wiki markup in $text should be escaped
3438 *
3439 * @since 1.21
3440 * @param PPFrame $frame The current frame, contains template arguments
3441 * @param string $function Function name
3442 * @param array $args Arguments to the function
3443 * @throws MWException
3444 * @return array
3445 */
3446 public function callParserFunction( $frame, $function, array $args = [] ) {
3447 # Case sensitive functions
3448 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3449 $function = $this->mFunctionSynonyms[1][$function];
3450 } else {
3451 # Case insensitive functions
3452 $function = $this->contLang->lc( $function );
3453 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3454 $function = $this->mFunctionSynonyms[0][$function];
3455 } else {
3456 return [ 'found' => false ];
3457 }
3458 }
3459
3460 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3461
3462 // Avoid PHP 7.1 warning from passing $this by reference
3463 $parser = $this;
3464
3465 $allArgs = [ &$parser ];
3466 if ( $flags & self::SFH_OBJECT_ARGS ) {
3467 # Convert arguments to PPNodes and collect for appending to $allArgs
3468 $funcArgs = [];
3469 foreach ( $args as $k => $v ) {
3470 if ( $v instanceof PPNode || $k === 0 ) {
3471 $funcArgs[] = $v;
3472 } else {
3473 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3474 }
3475 }
3476
3477 # Add a frame parameter, and pass the arguments as an array
3478 $allArgs[] = $frame;
3479 $allArgs[] = $funcArgs;
3480 } else {
3481 # Convert arguments to plain text and append to $allArgs
3482 foreach ( $args as $k => $v ) {
3483 if ( $v instanceof PPNode ) {
3484 $allArgs[] = trim( $frame->expand( $v ) );
3485 } elseif ( is_int( $k ) && $k >= 0 ) {
3486 $allArgs[] = trim( $v );
3487 } else {
3488 $allArgs[] = trim( "$k=$v" );
3489 }
3490 }
3491 }
3492
3493 $result = $callback( ...$allArgs );
3494
3495 # The interface for function hooks allows them to return a wikitext
3496 # string or an array containing the string and any flags. This mungs
3497 # things around to match what this method should return.
3498 if ( !is_array( $result ) ) {
3499 $result = [
3500 'found' => true,
3501 'text' => $result,
3502 ];
3503 } else {
3504 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3505 $result['text'] = $result[0];
3506 }
3507 unset( $result[0] );
3508 $result += [
3509 'found' => true,
3510 ];
3511 }
3512
3513 $noparse = true;
3514 $preprocessFlags = 0;
3515 if ( isset( $result['noparse'] ) ) {
3516 $noparse = $result['noparse'];
3517 }
3518 if ( isset( $result['preprocessFlags'] ) ) {
3519 $preprocessFlags = $result['preprocessFlags'];
3520 }
3521
3522 if ( !$noparse ) {
3523 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3524 $result['isChildObj'] = true;
3525 }
3526
3527 return $result;
3528 }
3529
3530 /**
3531 * Get the semi-parsed DOM representation of a template with a given title,
3532 * and its redirect destination title. Cached.
3533 *
3534 * @param Title $title
3535 *
3536 * @return array
3537 */
3538 public function getTemplateDom( $title ) {
3539 $cacheTitle = $title;
3540 $titleText = $title->getPrefixedDBkey();
3541
3542 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3543 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3544 $title = Title::makeTitle( $ns, $dbk );
3545 $titleText = $title->getPrefixedDBkey();
3546 }
3547 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3548 return [ $this->mTplDomCache[$titleText], $title ];
3549 }
3550
3551 # Cache miss, go to the database
3552 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3553
3554 if ( $text === false ) {
3555 $this->mTplDomCache[$titleText] = false;
3556 return [ false, $title ];
3557 }
3558
3559 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3560 $this->mTplDomCache[$titleText] = $dom;
3561
3562 if ( !$title->equals( $cacheTitle ) ) {
3563 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3564 [ $title->getNamespace(), $title->getDBkey() ];
3565 }
3566
3567 return [ $dom, $title ];
3568 }
3569
3570 /**
3571 * Fetch the current revision of a given title. Note that the revision
3572 * (and even the title) may not exist in the database, so everything
3573 * contributing to the output of the parser should use this method
3574 * where possible, rather than getting the revisions themselves. This
3575 * method also caches its results, so using it benefits performance.
3576 *
3577 * @since 1.24
3578 * @param Title $title
3579 * @return Revision
3580 */
3581 public function fetchCurrentRevisionOfTitle( $title ) {
3582 $cacheKey = $title->getPrefixedDBkey();
3583 if ( !$this->currentRevisionCache ) {
3584 $this->currentRevisionCache = new MapCacheLRU( 100 );
3585 }
3586 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3587 $this->currentRevisionCache->set( $cacheKey,
3588 // Defaults to Parser::statelessFetchRevision()
3589 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3590 );
3591 }
3592 return $this->currentRevisionCache->get( $cacheKey );
3593 }
3594
3595 /**
3596 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3597 * without passing them on to it.
3598 *
3599 * @since 1.24
3600 * @param Title $title
3601 * @param Parser|bool $parser
3602 * @return Revision|bool False if missing
3603 */
3604 public static function statelessFetchRevision( Title $title, $parser = false ) {
3605 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3606
3607 return $rev;
3608 }
3609
3610 /**
3611 * Fetch the unparsed text of a template and register a reference to it.
3612 * @param Title $title
3613 * @return array ( string or false, Title )
3614 */
3615 public function fetchTemplateAndTitle( $title ) {
3616 // Defaults to Parser::statelessFetchTemplate()
3617 $templateCb = $this->mOptions->getTemplateCallback();
3618 $stuff = call_user_func( $templateCb, $title, $this );
3619 // We use U+007F DELETE to distinguish strip markers from regular text.
3620 $text = $stuff['text'];
3621 if ( is_string( $stuff['text'] ) ) {
3622 $text = strtr( $text, "\x7f", "?" );
3623 }
3624 $finalTitle = $stuff['finalTitle'] ?? $title;
3625 if ( isset( $stuff['deps'] ) ) {
3626 foreach ( $stuff['deps'] as $dep ) {
3627 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3628 if ( $dep['title']->equals( $this->getTitle() ) ) {
3629 // If we transclude ourselves, the final result
3630 // will change based on the new version of the page
3631 $this->mOutput->setFlag( 'vary-revision' );
3632 }
3633 }
3634 }
3635 return [ $text, $finalTitle ];
3636 }
3637
3638 /**
3639 * Fetch the unparsed text of a template and register a reference to it.
3640 * @param Title $title
3641 * @return string|bool
3642 */
3643 public function fetchTemplate( $title ) {
3644 return $this->fetchTemplateAndTitle( $title )[0];
3645 }
3646
3647 /**
3648 * Static function to get a template
3649 * Can be overridden via ParserOptions::setTemplateCallback().
3650 *
3651 * @param Title $title
3652 * @param bool|Parser $parser
3653 *
3654 * @return array
3655 */
3656 public static function statelessFetchTemplate( $title, $parser = false ) {
3657 $text = $skip = false;
3658 $finalTitle = $title;
3659 $deps = [];
3660
3661 # Loop to fetch the article, with up to 1 redirect
3662 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
3663 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3664 # Give extensions a chance to select the revision instead
3665 $id = false; # Assume current
3666 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3667 [ $parser, $title, &$skip, &$id ] );
3668
3669 if ( $skip ) {
3670 $text = false;
3671 $deps[] = [
3672 'title' => $title,
3673 'page_id' => $title->getArticleID(),
3674 'rev_id' => null
3675 ];
3676 break;
3677 }
3678 # Get the revision
3679 if ( $id ) {
3680 $rev = Revision::newFromId( $id );
3681 } elseif ( $parser ) {
3682 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3683 } else {
3684 $rev = Revision::newFromTitle( $title );
3685 }
3686 $rev_id = $rev ? $rev->getId() : 0;
3687 # If there is no current revision, there is no page
3688 if ( $id === false && !$rev ) {
3689 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3690 $linkCache->addBadLinkObj( $title );
3691 }
3692
3693 $deps[] = [
3694 'title' => $title,
3695 'page_id' => $title->getArticleID(),
3696 'rev_id' => $rev_id ];
3697 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3698 # We fetched a rev from a different title; register it too...
3699 $deps[] = [
3700 'title' => $rev->getTitle(),
3701 'page_id' => $rev->getPage(),
3702 'rev_id' => $rev_id ];
3703 }
3704
3705 if ( $rev ) {
3706 $content = $rev->getContent();
3707 $text = $content ? $content->getWikitextForTransclusion() : null;
3708
3709 Hooks::run( 'ParserFetchTemplate',
3710 [ $parser, $title, $rev, &$text, &$deps ] );
3711
3712 if ( $text === false || $text === null ) {
3713 $text = false;
3714 break;
3715 }
3716 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3717 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
3718 lcfirst( $title->getText() ) )->inContentLanguage();
3719 if ( !$message->exists() ) {
3720 $text = false;
3721 break;
3722 }
3723 $content = $message->content();
3724 $text = $message->plain();
3725 } else {
3726 break;
3727 }
3728 if ( !$content ) {
3729 break;
3730 }
3731 # Redirect?
3732 $finalTitle = $title;
3733 $title = $content->getRedirectTarget();
3734 }
3735 return [
3736 'text' => $text,
3737 'finalTitle' => $finalTitle,
3738 'deps' => $deps ];
3739 }
3740
3741 /**
3742 * Fetch a file and its title and register a reference to it.
3743 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3744 * @param Title $title
3745 * @param array $options Array of options to RepoGroup::findFile
3746 * @return File|bool
3747 * @deprecated since 1.32, use fetchFileAndTitle instead
3748 */
3749 public function fetchFile( $title, $options = [] ) {
3750 wfDeprecated( __METHOD__, '1.32' );
3751 return $this->fetchFileAndTitle( $title, $options )[0];
3752 }
3753
3754 /**
3755 * Fetch a file and its title and register a reference to it.
3756 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3757 * @param Title $title
3758 * @param array $options Array of options to RepoGroup::findFile
3759 * @return array ( File or false, Title of file )
3760 */
3761 public function fetchFileAndTitle( $title, $options = [] ) {
3762 $file = $this->fetchFileNoRegister( $title, $options );
3763
3764 $time = $file ? $file->getTimestamp() : false;
3765 $sha1 = $file ? $file->getSha1() : false;
3766 # Register the file as a dependency...
3767 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3768 if ( $file && !$title->equals( $file->getTitle() ) ) {
3769 # Update fetched file title
3770 $title = $file->getTitle();
3771 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3772 }
3773 return [ $file, $title ];
3774 }
3775
3776 /**
3777 * Helper function for fetchFileAndTitle.
3778 *
3779 * Also useful if you need to fetch a file but not use it yet,
3780 * for example to get the file's handler.
3781 *
3782 * @param Title $title
3783 * @param array $options Array of options to RepoGroup::findFile
3784 * @return File|bool
3785 */
3786 protected function fetchFileNoRegister( $title, $options = [] ) {
3787 if ( isset( $options['broken'] ) ) {
3788 $file = false; // broken thumbnail forced by hook
3789 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3790 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3791 } else { // get by (name,timestamp)
3792 $file = wfFindFile( $title, $options );
3793 }
3794 return $file;
3795 }
3796
3797 /**
3798 * Transclude an interwiki link.
3799 *
3800 * @param Title $title
3801 * @param string $action Usually one of (raw, render)
3802 *
3803 * @return string
3804 */
3805 public function interwikiTransclude( $title, $action ) {
3806 global $wgEnableScaryTranscluding, $wgTranscludeCacheExpiry;
3807
3808 if ( !$wgEnableScaryTranscluding ) {
3809 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3810 }
3811
3812 $url = $title->getFullURL( [ 'action' => $action ] );
3813 if ( strlen( $url ) > 1024 ) {
3814 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3815 }
3816
3817 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3818
3819 $fname = __METHOD__;
3820 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3821
3822 $data = $cache->getWithSetCallback(
3823 $cache->makeGlobalKey(
3824 'interwiki-transclude',
3825 ( $wikiId !== false ) ? $wikiId : 'external',
3826 sha1( $url )
3827 ),
3828 $wgTranscludeCacheExpiry,
3829 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3830 $req = MWHttpRequest::factory( $url, [], $fname );
3831
3832 $status = $req->execute(); // Status object
3833 if ( !$status->isOK() ) {
3834 $ttl = $cache::TTL_UNCACHEABLE;
3835 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3836 $ttl = min( $cache::TTL_LAGGED, $ttl );
3837 }
3838
3839 return [
3840 'text' => $status->isOK() ? $req->getContent() : null,
3841 'code' => $req->getStatus()
3842 ];
3843 },
3844 [
3845 'checkKeys' => ( $wikiId !== false )
3846 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3847 : [],
3848 'pcGroup' => 'interwiki-transclude:5',
3849 'pcTTL' => $cache::TTL_PROC_LONG
3850 ]
3851 );
3852
3853 if ( is_string( $data['text'] ) ) {
3854 $text = $data['text'];
3855 } elseif ( $data['code'] != 200 ) {
3856 // Though we failed to fetch the content, this status is useless.
3857 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3858 ->params( $url, $data['code'] )->inContentLanguage()->text();
3859 } else {
3860 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3861 }
3862
3863 return $text;
3864 }
3865
3866 /**
3867 * Triple brace replacement -- used for template arguments
3868 * @private
3869 *
3870 * @param array $piece
3871 * @param PPFrame $frame
3872 *
3873 * @return array
3874 */
3875 public function argSubstitution( $piece, $frame ) {
3876 $error = false;
3877 $parts = $piece['parts'];
3878 $nameWithSpaces = $frame->expand( $piece['title'] );
3879 $argName = trim( $nameWithSpaces );
3880 $object = false;
3881 $text = $frame->getArgument( $argName );
3882 if ( $text === false && $parts->getLength() > 0
3883 && ( $this->ot['html']
3884 || $this->ot['pre']
3885 || ( $this->ot['wiki'] && $frame->isTemplate() )
3886 )
3887 ) {
3888 # No match in frame, use the supplied default
3889 $object = $parts->item( 0 )->getChildren();
3890 }
3891 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3892 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3893 $this->limitationWarn( 'post-expand-template-argument' );
3894 }
3895
3896 if ( $text === false && $object === false ) {
3897 # No match anywhere
3898 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3899 }
3900 if ( $error !== false ) {
3901 $text .= $error;
3902 }
3903 if ( $object !== false ) {
3904 $ret = [ 'object' => $object ];
3905 } else {
3906 $ret = [ 'text' => $text ];
3907 }
3908
3909 return $ret;
3910 }
3911
3912 /**
3913 * Return the text to be used for a given extension tag.
3914 * This is the ghost of strip().
3915 *
3916 * @param array $params Associative array of parameters:
3917 * name PPNode for the tag name
3918 * attr PPNode for unparsed text where tag attributes are thought to be
3919 * attributes Optional associative array of parsed attributes
3920 * inner Contents of extension element
3921 * noClose Original text did not have a close tag
3922 * @param PPFrame $frame
3923 *
3924 * @throws MWException
3925 * @return string
3926 */
3927 public function extensionSubstitution( $params, $frame ) {
3928 static $errorStr = '<span class="error">';
3929 static $errorLen = 20;
3930
3931 $name = $frame->expand( $params['name'] );
3932 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3933 // Probably expansion depth or node count exceeded. Just punt the
3934 // error up.
3935 return $name;
3936 }
3937
3938 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3939 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3940 // See above
3941 return $attrText;
3942 }
3943
3944 // We can't safely check if the expansion for $content resulted in an
3945 // error, because the content could happen to be the error string
3946 // (T149622).
3947 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3948
3949 $marker = self::MARKER_PREFIX . "-$name-"
3950 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3951
3952 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3953 ( $this->ot['html'] || $this->ot['pre'] );
3954 if ( $isFunctionTag ) {
3955 $markerType = 'none';
3956 } else {
3957 $markerType = 'general';
3958 }
3959 if ( $this->ot['html'] || $isFunctionTag ) {
3960 $name = strtolower( $name );
3961 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3962 if ( isset( $params['attributes'] ) ) {
3963 $attributes = $attributes + $params['attributes'];
3964 }
3965
3966 if ( isset( $this->mTagHooks[$name] ) ) {
3967 $output = call_user_func_array( $this->mTagHooks[$name],
3968 [ $content, $attributes, $this, $frame ] );
3969 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3970 list( $callback, ) = $this->mFunctionTagHooks[$name];
3971
3972 // Avoid PHP 7.1 warning from passing $this by reference
3973 $parser = $this;
3974 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3975 } else {
3976 $output = '<span class="error">Invalid tag extension name: ' .
3977 htmlspecialchars( $name ) . '</span>';
3978 }
3979
3980 if ( is_array( $output ) ) {
3981 // Extract flags
3982 $flags = $output;
3983 $output = $flags[0];
3984 if ( isset( $flags['markerType'] ) ) {
3985 $markerType = $flags['markerType'];
3986 }
3987 }
3988 } else {
3989 if ( is_null( $attrText ) ) {
3990 $attrText = '';
3991 }
3992 if ( isset( $params['attributes'] ) ) {
3993 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3994 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3995 htmlspecialchars( $attrValue ) . '"';
3996 }
3997 }
3998 if ( $content === null ) {
3999 $output = "<$name$attrText/>";
4000 } else {
4001 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4002 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4003 // See above
4004 return $close;
4005 }
4006 $output = "<$name$attrText>$content$close";
4007 }
4008 }
4009
4010 if ( $markerType === 'none' ) {
4011 return $output;
4012 } elseif ( $markerType === 'nowiki' ) {
4013 $this->mStripState->addNoWiki( $marker, $output );
4014 } elseif ( $markerType === 'general' ) {
4015 $this->mStripState->addGeneral( $marker, $output );
4016 } else {
4017 throw new MWException( __METHOD__ . ': invalid marker type' );
4018 }
4019 return $marker;
4020 }
4021
4022 /**
4023 * Increment an include size counter
4024 *
4025 * @param string $type The type of expansion
4026 * @param int $size The size of the text
4027 * @return bool False if this inclusion would take it over the maximum, true otherwise
4028 */
4029 public function incrementIncludeSize( $type, $size ) {
4030 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4031 return false;
4032 } else {
4033 $this->mIncludeSizes[$type] += $size;
4034 return true;
4035 }
4036 }
4037
4038 /**
4039 * Increment the expensive function count
4040 *
4041 * @return bool False if the limit has been exceeded
4042 */
4043 public function incrementExpensiveFunctionCount() {
4044 $this->mExpensiveFunctionCount++;
4045 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4046 }
4047
4048 /**
4049 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4050 * Fills $this->mDoubleUnderscores, returns the modified text
4051 *
4052 * @param string $text
4053 *
4054 * @return string
4055 */
4056 public function doDoubleUnderscore( $text ) {
4057 # The position of __TOC__ needs to be recorded
4058 $mw = $this->magicWordFactory->get( 'toc' );
4059 if ( $mw->match( $text ) ) {
4060 $this->mShowToc = true;
4061 $this->mForceTocPosition = true;
4062
4063 # Set a placeholder. At the end we'll fill it in with the TOC.
4064 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4065
4066 # Only keep the first one.
4067 $text = $mw->replace( '', $text );
4068 }
4069
4070 # Now match and remove the rest of them
4071 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4072 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4073
4074 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4075 $this->mOutput->mNoGallery = true;
4076 }
4077 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4078 $this->mShowToc = false;
4079 }
4080 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4081 && $this->mTitle->getNamespace() == NS_CATEGORY
4082 ) {
4083 $this->addTrackingCategory( 'hidden-category-category' );
4084 }
4085 # (T10068) Allow control over whether robots index a page.
4086 # __INDEX__ always overrides __NOINDEX__, see T16899
4087 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4088 $this->mOutput->setIndexPolicy( 'noindex' );
4089 $this->addTrackingCategory( 'noindex-category' );
4090 }
4091 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4092 $this->mOutput->setIndexPolicy( 'index' );
4093 $this->addTrackingCategory( 'index-category' );
4094 }
4095
4096 # Cache all double underscores in the database
4097 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4098 $this->mOutput->setProperty( $key, '' );
4099 }
4100
4101 return $text;
4102 }
4103
4104 /**
4105 * @see ParserOutput::addTrackingCategory()
4106 * @param string $msg Message key
4107 * @return bool Whether the addition was successful
4108 */
4109 public function addTrackingCategory( $msg ) {
4110 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4111 }
4112
4113 /**
4114 * This function accomplishes several tasks:
4115 * 1) Auto-number headings if that option is enabled
4116 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4117 * 3) Add a Table of contents on the top for users who have enabled the option
4118 * 4) Auto-anchor headings
4119 *
4120 * It loops through all headlines, collects the necessary data, then splits up the
4121 * string and re-inserts the newly formatted headlines.
4122 *
4123 * @param string $text
4124 * @param string $origText Original, untouched wikitext
4125 * @param bool $isMain
4126 * @return mixed|string
4127 * @private
4128 */
4129 public function formatHeadings( $text, $origText, $isMain = true ) {
4130 global $wgMaxTocLevel;
4131
4132 # Inhibit editsection links if requested in the page
4133 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4134 $maybeShowEditLink = false;
4135 } else {
4136 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4137 }
4138
4139 # Get all headlines for numbering them and adding funky stuff like [edit]
4140 # links - this is for later, but we need the number of headlines right now
4141 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4142 # be trimmed here since whitespace in HTML headings is significant.
4143 $matches = [];
4144 $numMatches = preg_match_all(
4145 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4146 $text,
4147 $matches
4148 );
4149
4150 # if there are fewer than 4 headlines in the article, do not show TOC
4151 # unless it's been explicitly enabled.
4152 $enoughToc = $this->mShowToc &&
4153 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4154
4155 # Allow user to stipulate that a page should have a "new section"
4156 # link added via __NEWSECTIONLINK__
4157 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4158 $this->mOutput->setNewSection( true );
4159 }
4160
4161 # Allow user to remove the "new section"
4162 # link via __NONEWSECTIONLINK__
4163 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4164 $this->mOutput->hideNewSection( true );
4165 }
4166
4167 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4168 # override above conditions and always show TOC above first header
4169 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4170 $this->mShowToc = true;
4171 $enoughToc = true;
4172 }
4173
4174 # headline counter
4175 $headlineCount = 0;
4176 $numVisible = 0;
4177
4178 # Ugh .. the TOC should have neat indentation levels which can be
4179 # passed to the skin functions. These are determined here
4180 $toc = '';
4181 $full = '';
4182 $head = [];
4183 $sublevelCount = [];
4184 $levelCount = [];
4185 $level = 0;
4186 $prevlevel = 0;
4187 $toclevel = 0;
4188 $prevtoclevel = 0;
4189 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4190 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4191 $oldType = $this->mOutputType;
4192 $this->setOutputType( self::OT_WIKI );
4193 $frame = $this->getPreprocessor()->newFrame();
4194 $root = $this->preprocessToDom( $origText );
4195 $node = $root->getFirstChild();
4196 $byteOffset = 0;
4197 $tocraw = [];
4198 $refers = [];
4199
4200 $headlines = $numMatches !== false ? $matches[3] : [];
4201
4202 foreach ( $headlines as $headline ) {
4203 $isTemplate = false;
4204 $titleText = false;
4205 $sectionIndex = false;
4206 $numbering = '';
4207 $markerMatches = [];
4208 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4209 $serial = $markerMatches[1];
4210 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4211 $isTemplate = ( $titleText != $baseTitleText );
4212 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4213 }
4214
4215 if ( $toclevel ) {
4216 $prevlevel = $level;
4217 }
4218 $level = $matches[1][$headlineCount];
4219
4220 if ( $level > $prevlevel ) {
4221 # Increase TOC level
4222 $toclevel++;
4223 $sublevelCount[$toclevel] = 0;
4224 if ( $toclevel < $wgMaxTocLevel ) {
4225 $prevtoclevel = $toclevel;
4226 $toc .= Linker::tocIndent();
4227 $numVisible++;
4228 }
4229 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4230 # Decrease TOC level, find level to jump to
4231
4232 for ( $i = $toclevel; $i > 0; $i-- ) {
4233 if ( $levelCount[$i] == $level ) {
4234 # Found last matching level
4235 $toclevel = $i;
4236 break;
4237 } elseif ( $levelCount[$i] < $level ) {
4238 # Found first matching level below current level
4239 $toclevel = $i + 1;
4240 break;
4241 }
4242 }
4243 if ( $i == 0 ) {
4244 $toclevel = 1;
4245 }
4246 if ( $toclevel < $wgMaxTocLevel ) {
4247 if ( $prevtoclevel < $wgMaxTocLevel ) {
4248 # Unindent only if the previous toc level was shown :p
4249 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4250 $prevtoclevel = $toclevel;
4251 } else {
4252 $toc .= Linker::tocLineEnd();
4253 }
4254 }
4255 } else {
4256 # No change in level, end TOC line
4257 if ( $toclevel < $wgMaxTocLevel ) {
4258 $toc .= Linker::tocLineEnd();
4259 }
4260 }
4261
4262 $levelCount[$toclevel] = $level;
4263
4264 # count number of headlines for each level
4265 $sublevelCount[$toclevel]++;
4266 $dot = 0;
4267 for ( $i = 1; $i <= $toclevel; $i++ ) {
4268 if ( !empty( $sublevelCount[$i] ) ) {
4269 if ( $dot ) {
4270 $numbering .= '.';
4271 }
4272 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4273 $dot = 1;
4274 }
4275 }
4276
4277 # The safe header is a version of the header text safe to use for links
4278
4279 # Remove link placeholders by the link text.
4280 # <!--LINK number-->
4281 # turns into
4282 # link text with suffix
4283 # Do this before unstrip since link text can contain strip markers
4284 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4285
4286 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4287 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4288
4289 # Remove any <style> or <script> tags (T198618)
4290 $safeHeadline = preg_replace(
4291 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4292 '',
4293 $safeHeadline
4294 );
4295
4296 # Strip out HTML (first regex removes any tag not allowed)
4297 # Allowed tags are:
4298 # * <sup> and <sub> (T10393)
4299 # * <i> (T28375)
4300 # * <b> (r105284)
4301 # * <bdi> (T74884)
4302 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4303 # * <s> and <strike> (T35715)
4304 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4305 # to allow setting directionality in toc items.
4306 $tocline = preg_replace(
4307 [
4308 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4309 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4310 ],
4311 [ '', '<$1>' ],
4312 $safeHeadline
4313 );
4314
4315 # Strip '<span></span>', which is the result from the above if
4316 # <span id="foo"></span> is used to produce an additional anchor
4317 # for a section.
4318 $tocline = str_replace( '<span></span>', '', $tocline );
4319
4320 $tocline = trim( $tocline );
4321
4322 # For the anchor, strip out HTML-y stuff period
4323 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4324 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4325
4326 # Save headline for section edit hint before it's escaped
4327 $headlineHint = $safeHeadline;
4328
4329 # Decode HTML entities
4330 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4331
4332 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4333
4334 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4335 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4336 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4337 if ( $fallbackHeadline === $safeHeadline ) {
4338 # No reason to have both (in fact, we can't)
4339 $fallbackHeadline = false;
4340 }
4341
4342 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4343 # @todo FIXME: We may be changing them depending on the current locale.
4344 $arrayKey = strtolower( $safeHeadline );
4345 if ( $fallbackHeadline === false ) {
4346 $fallbackArrayKey = false;
4347 } else {
4348 $fallbackArrayKey = strtolower( $fallbackHeadline );
4349 }
4350
4351 # Create the anchor for linking from the TOC to the section
4352 $anchor = $safeHeadline;
4353 $fallbackAnchor = $fallbackHeadline;
4354 if ( isset( $refers[$arrayKey] ) ) {
4355 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4356 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4357 $anchor .= "_$i";
4358 $linkAnchor .= "_$i";
4359 $refers["${arrayKey}_$i"] = true;
4360 } else {
4361 $refers[$arrayKey] = true;
4362 }
4363 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4364 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4365 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4366 $fallbackAnchor .= "_$i";
4367 $refers["${fallbackArrayKey}_$i"] = true;
4368 } else {
4369 $refers[$fallbackArrayKey] = true;
4370 }
4371
4372 # Don't number the heading if it is the only one (looks silly)
4373 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4374 # the two are different if the line contains a link
4375 $headline = Html::element(
4376 'span',
4377 [ 'class' => 'mw-headline-number' ],
4378 $numbering
4379 ) . ' ' . $headline;
4380 }
4381
4382 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4383 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4384 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4385 }
4386
4387 # Add the section to the section tree
4388 # Find the DOM node for this header
4389 $noOffset = ( $isTemplate || $sectionIndex === false );
4390 while ( $node && !$noOffset ) {
4391 if ( $node->getName() === 'h' ) {
4392 $bits = $node->splitHeading();
4393 if ( $bits['i'] == $sectionIndex ) {
4394 break;
4395 }
4396 }
4397 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4398 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4399 $node = $node->getNextSibling();
4400 }
4401 $tocraw[] = [
4402 'toclevel' => $toclevel,
4403 'level' => $level,
4404 'line' => $tocline,
4405 'number' => $numbering,
4406 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4407 'fromtitle' => $titleText,
4408 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4409 'anchor' => $anchor,
4410 ];
4411
4412 # give headline the correct <h#> tag
4413 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4414 // Output edit section links as markers with styles that can be customized by skins
4415 if ( $isTemplate ) {
4416 # Put a T flag in the section identifier, to indicate to extractSections()
4417 # that sections inside <includeonly> should be counted.
4418 $editsectionPage = $titleText;
4419 $editsectionSection = "T-$sectionIndex";
4420 $editsectionContent = null;
4421 } else {
4422 $editsectionPage = $this->mTitle->getPrefixedText();
4423 $editsectionSection = $sectionIndex;
4424 $editsectionContent = $headlineHint;
4425 }
4426 // We use a bit of pesudo-xml for editsection markers. The
4427 // language converter is run later on. Using a UNIQ style marker
4428 // leads to the converter screwing up the tokens when it
4429 // converts stuff. And trying to insert strip tags fails too. At
4430 // this point all real inputted tags have already been escaped,
4431 // so we don't have to worry about a user trying to input one of
4432 // these markers directly. We use a page and section attribute
4433 // to stop the language converter from converting these
4434 // important bits of data, but put the headline hint inside a
4435 // content block because the language converter is supposed to
4436 // be able to convert that piece of data.
4437 // Gets replaced with html in ParserOutput::getText
4438 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4439 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4440 if ( $editsectionContent !== null ) {
4441 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4442 } else {
4443 $editlink .= '/>';
4444 }
4445 } else {
4446 $editlink = '';
4447 }
4448 $head[$headlineCount] = Linker::makeHeadline( $level,
4449 $matches['attrib'][$headlineCount], $anchor, $headline,
4450 $editlink, $fallbackAnchor );
4451
4452 $headlineCount++;
4453 }
4454
4455 $this->setOutputType( $oldType );
4456
4457 # Never ever show TOC if no headers
4458 if ( $numVisible < 1 ) {
4459 $enoughToc = false;
4460 }
4461
4462 if ( $enoughToc ) {
4463 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4464 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4465 }
4466 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4467 $this->mOutput->setTOCHTML( $toc );
4468 $toc = self::TOC_START . $toc . self::TOC_END;
4469 }
4470
4471 if ( $isMain ) {
4472 $this->mOutput->setSections( $tocraw );
4473 }
4474
4475 # split up and insert constructed headlines
4476 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4477 $i = 0;
4478
4479 // build an array of document sections
4480 $sections = [];
4481 foreach ( $blocks as $block ) {
4482 // $head is zero-based, sections aren't.
4483 if ( empty( $head[$i - 1] ) ) {
4484 $sections[$i] = $block;
4485 } else {
4486 $sections[$i] = $head[$i - 1] . $block;
4487 }
4488
4489 /**
4490 * Send a hook, one per section.
4491 * The idea here is to be able to make section-level DIVs, but to do so in a
4492 * lower-impact, more correct way than r50769
4493 *
4494 * $this : caller
4495 * $section : the section number
4496 * &$sectionContent : ref to the content of the section
4497 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4498 */
4499 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4500
4501 $i++;
4502 }
4503
4504 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4505 // append the TOC at the beginning
4506 // Top anchor now in skin
4507 $sections[0] = $sections[0] . $toc . "\n";
4508 }
4509
4510 $full .= implode( '', $sections );
4511
4512 if ( $this->mForceTocPosition ) {
4513 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4514 } else {
4515 return $full;
4516 }
4517 }
4518
4519 /**
4520 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4521 * conversion, substituting signatures, {{subst:}} templates, etc.
4522 *
4523 * @param string $text The text to transform
4524 * @param Title $title The Title object for the current article
4525 * @param User $user The User object describing the current user
4526 * @param ParserOptions $options Parsing options
4527 * @param bool $clearState Whether to clear the parser state first
4528 * @return string The altered wiki markup
4529 */
4530 public function preSaveTransform( $text, Title $title, User $user,
4531 ParserOptions $options, $clearState = true
4532 ) {
4533 if ( $clearState ) {
4534 $magicScopeVariable = $this->lock();
4535 }
4536 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4537 $this->setUser( $user );
4538
4539 // Strip U+0000 NULL (T159174)
4540 $text = str_replace( "\000", '', $text );
4541
4542 // We still normalize line endings for backwards-compatibility
4543 // with other code that just calls PST, but this should already
4544 // be handled in TextContent subclasses
4545 $text = TextContent::normalizeLineEndings( $text );
4546
4547 if ( $options->getPreSaveTransform() ) {
4548 $text = $this->pstPass2( $text, $user );
4549 }
4550 $text = $this->mStripState->unstripBoth( $text );
4551
4552 $this->setUser( null ); # Reset
4553
4554 return $text;
4555 }
4556
4557 /**
4558 * Pre-save transform helper function
4559 *
4560 * @param string $text
4561 * @param User $user
4562 *
4563 * @return string
4564 */
4565 private function pstPass2( $text, $user ) {
4566 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4567 # $this->contLang here in order to give everyone the same signature and use the default one
4568 # rather than the one selected in each user's preferences. (see also T14815)
4569 $ts = $this->mOptions->getTimestamp();
4570 $timestamp = MWTimestamp::getLocalInstance( $ts );
4571 $ts = $timestamp->format( 'YmdHis' );
4572 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4573
4574 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4575
4576 # Variable replacement
4577 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4578 $text = $this->replaceVariables( $text );
4579
4580 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4581 # which may corrupt this parser instance via its wfMessage()->text() call-
4582
4583 # Signatures
4584 if ( strpos( $text, '~~~' ) !== false ) {
4585 $sigText = $this->getUserSig( $user );
4586 $text = strtr( $text, [
4587 '~~~~~' => $d,
4588 '~~~~' => "$sigText $d",
4589 '~~~' => $sigText
4590 ] );
4591 # The main two signature forms used above are time-sensitive
4592 $this->mOutput->setFlag( 'user-signature' );
4593 }
4594
4595 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4596 $tc = '[' . Title::legalChars() . ']';
4597 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4598
4599 // [[ns:page (context)|]]
4600 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4601 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4602 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4603 // [[ns:page (context), context|]] (using either single or double-width comma)
4604 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4605 // [[|page]] (reverse pipe trick: add context from page title)
4606 $p2 = "/\[\[\\|($tc+)]]/";
4607
4608 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4609 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4610 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4611 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4612
4613 $t = $this->mTitle->getText();
4614 $m = [];
4615 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4616 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4617 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4618 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4619 } else {
4620 # if there's no context, don't bother duplicating the title
4621 $text = preg_replace( $p2, '[[\\1]]', $text );
4622 }
4623
4624 return $text;
4625 }
4626
4627 /**
4628 * Fetch the user's signature text, if any, and normalize to
4629 * validated, ready-to-insert wikitext.
4630 * If you have pre-fetched the nickname or the fancySig option, you can
4631 * specify them here to save a database query.
4632 * Do not reuse this parser instance after calling getUserSig(),
4633 * as it may have changed if it's the $wgParser.
4634 *
4635 * @param User &$user
4636 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4637 * @param bool|null $fancySig whether the nicknname is the complete signature
4638 * or null to use default value
4639 * @return string
4640 */
4641 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4642 global $wgMaxSigChars;
4643
4644 $username = $user->getName();
4645
4646 # If not given, retrieve from the user object.
4647 if ( $nickname === false ) {
4648 $nickname = $user->getOption( 'nickname' );
4649 }
4650
4651 if ( is_null( $fancySig ) ) {
4652 $fancySig = $user->getBoolOption( 'fancysig' );
4653 }
4654
4655 $nickname = $nickname == null ? $username : $nickname;
4656
4657 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4658 $nickname = $username;
4659 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4660 } elseif ( $fancySig !== false ) {
4661 # Sig. might contain markup; validate this
4662 if ( $this->validateSig( $nickname ) !== false ) {
4663 # Validated; clean up (if needed) and return it
4664 return $this->cleanSig( $nickname, true );
4665 } else {
4666 # Failed to validate; fall back to the default
4667 $nickname = $username;
4668 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4669 }
4670 }
4671
4672 # Make sure nickname doesnt get a sig in a sig
4673 $nickname = self::cleanSigInSig( $nickname );
4674
4675 # If we're still here, make it a link to the user page
4676 $userText = wfEscapeWikiText( $username );
4677 $nickText = wfEscapeWikiText( $nickname );
4678 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4679
4680 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4681 ->title( $this->getTitle() )->text();
4682 }
4683
4684 /**
4685 * Check that the user's signature contains no bad XML
4686 *
4687 * @param string $text
4688 * @return string|bool An expanded string, or false if invalid.
4689 */
4690 public function validateSig( $text ) {
4691 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4692 }
4693
4694 /**
4695 * Clean up signature text
4696 *
4697 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4698 * 2) Substitute all transclusions
4699 *
4700 * @param string $text
4701 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4702 * @return string Signature text
4703 */
4704 public function cleanSig( $text, $parsing = false ) {
4705 if ( !$parsing ) {
4706 global $wgTitle;
4707 $magicScopeVariable = $this->lock();
4708 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4709 }
4710
4711 # Option to disable this feature
4712 if ( !$this->mOptions->getCleanSignatures() ) {
4713 return $text;
4714 }
4715
4716 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4717 # => Move this logic to braceSubstitution()
4718 $substWord = $this->magicWordFactory->get( 'subst' );
4719 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4720 $substText = '{{' . $substWord->getSynonym( 0 );
4721
4722 $text = preg_replace( $substRegex, $substText, $text );
4723 $text = self::cleanSigInSig( $text );
4724 $dom = $this->preprocessToDom( $text );
4725 $frame = $this->getPreprocessor()->newFrame();
4726 $text = $frame->expand( $dom );
4727
4728 if ( !$parsing ) {
4729 $text = $this->mStripState->unstripBoth( $text );
4730 }
4731
4732 return $text;
4733 }
4734
4735 /**
4736 * Strip 3, 4 or 5 tildes out of signatures.
4737 *
4738 * @param string $text
4739 * @return string Signature text with /~{3,5}/ removed
4740 */
4741 public static function cleanSigInSig( $text ) {
4742 $text = preg_replace( '/~{3,5}/', '', $text );
4743 return $text;
4744 }
4745
4746 /**
4747 * Set up some variables which are usually set up in parse()
4748 * so that an external function can call some class members with confidence
4749 *
4750 * @param Title|null $title
4751 * @param ParserOptions $options
4752 * @param int $outputType
4753 * @param bool $clearState
4754 */
4755 public function startExternalParse( Title $title = null, ParserOptions $options,
4756 $outputType, $clearState = true
4757 ) {
4758 $this->startParse( $title, $options, $outputType, $clearState );
4759 }
4760
4761 /**
4762 * @param Title|null $title
4763 * @param ParserOptions $options
4764 * @param int $outputType
4765 * @param bool $clearState
4766 */
4767 private function startParse( Title $title = null, ParserOptions $options,
4768 $outputType, $clearState = true
4769 ) {
4770 $this->setTitle( $title );
4771 $this->mOptions = $options;
4772 $this->setOutputType( $outputType );
4773 if ( $clearState ) {
4774 $this->clearState();
4775 }
4776 }
4777
4778 /**
4779 * Wrapper for preprocess()
4780 *
4781 * @param string $text The text to preprocess
4782 * @param ParserOptions $options
4783 * @param Title|null $title Title object or null to use $wgTitle
4784 * @return string
4785 */
4786 public function transformMsg( $text, $options, $title = null ) {
4787 static $executing = false;
4788
4789 # Guard against infinite recursion
4790 if ( $executing ) {
4791 return $text;
4792 }
4793 $executing = true;
4794
4795 if ( !$title ) {
4796 global $wgTitle;
4797 $title = $wgTitle;
4798 }
4799
4800 $text = $this->preprocess( $text, $title, $options );
4801
4802 $executing = false;
4803 return $text;
4804 }
4805
4806 /**
4807 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4808 * The callback should have the following form:
4809 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4810 *
4811 * Transform and return $text. Use $parser for any required context, e.g. use
4812 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4813 *
4814 * Hooks may return extended information by returning an array, of which the
4815 * first numbered element (index 0) must be the return string, and all other
4816 * entries are extracted into local variables within an internal function
4817 * in the Parser class.
4818 *
4819 * This interface (introduced r61913) appears to be undocumented, but
4820 * 'markerType' is used by some core tag hooks to override which strip
4821 * array their results are placed in. **Use great caution if attempting
4822 * this interface, as it is not documented and injudicious use could smash
4823 * private variables.**
4824 *
4825 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4826 * @param callable $callback The callback function (and object) to use for the tag
4827 * @throws MWException
4828 * @return callable|null The old value of the mTagHooks array associated with the hook
4829 */
4830 public function setHook( $tag, callable $callback ) {
4831 $tag = strtolower( $tag );
4832 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4833 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4834 }
4835 $oldVal = $this->mTagHooks[$tag] ?? null;
4836 $this->mTagHooks[$tag] = $callback;
4837 if ( !in_array( $tag, $this->mStripList ) ) {
4838 $this->mStripList[] = $tag;
4839 }
4840
4841 return $oldVal;
4842 }
4843
4844 /**
4845 * As setHook(), but letting the contents be parsed.
4846 *
4847 * Transparent tag hooks are like regular XML-style tag hooks, except they
4848 * operate late in the transformation sequence, on HTML instead of wikitext.
4849 *
4850 * This is probably obsoleted by things dealing with parser frames?
4851 * The only extension currently using it is geoserver.
4852 *
4853 * @since 1.10
4854 * @todo better document or deprecate this
4855 *
4856 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4857 * @param callable $callback The callback function (and object) to use for the tag
4858 * @throws MWException
4859 * @return callable|null The old value of the mTagHooks array associated with the hook
4860 */
4861 public function setTransparentTagHook( $tag, callable $callback ) {
4862 $tag = strtolower( $tag );
4863 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4864 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4865 }
4866 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4867 $this->mTransparentTagHooks[$tag] = $callback;
4868
4869 return $oldVal;
4870 }
4871
4872 /**
4873 * Remove all tag hooks
4874 */
4875 public function clearTagHooks() {
4876 $this->mTagHooks = [];
4877 $this->mFunctionTagHooks = [];
4878 $this->mStripList = $this->mDefaultStripList;
4879 }
4880
4881 /**
4882 * Create a function, e.g. {{sum:1|2|3}}
4883 * The callback function should have the form:
4884 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4885 *
4886 * Or with Parser::SFH_OBJECT_ARGS:
4887 * function myParserFunction( $parser, $frame, $args ) { ... }
4888 *
4889 * The callback may either return the text result of the function, or an array with the text
4890 * in element 0, and a number of flags in the other elements. The names of the flags are
4891 * specified in the keys. Valid flags are:
4892 * found The text returned is valid, stop processing the template. This
4893 * is on by default.
4894 * nowiki Wiki markup in the return value should be escaped
4895 * isHTML The returned text is HTML, armour it against wikitext transformation
4896 *
4897 * @param string $id The magic word ID
4898 * @param callable $callback The callback function (and object) to use
4899 * @param int $flags A combination of the following flags:
4900 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4901 *
4902 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4903 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4904 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4905 * the arguments, and to control the way they are expanded.
4906 *
4907 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4908 * arguments, for instance:
4909 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4910 *
4911 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4912 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4913 * working if/when this is changed.
4914 *
4915 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4916 * expansion.
4917 *
4918 * Please read the documentation in includes/parser/Preprocessor.php for more information
4919 * about the methods available in PPFrame and PPNode.
4920 *
4921 * @throws MWException
4922 * @return string|callable The old callback function for this name, if any
4923 */
4924 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4925 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4926 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4927
4928 # Add to function cache
4929 $mw = $this->magicWordFactory->get( $id );
4930 if ( !$mw ) {
4931 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4932 }
4933
4934 $synonyms = $mw->getSynonyms();
4935 $sensitive = intval( $mw->isCaseSensitive() );
4936
4937 foreach ( $synonyms as $syn ) {
4938 # Case
4939 if ( !$sensitive ) {
4940 $syn = $this->contLang->lc( $syn );
4941 }
4942 # Add leading hash
4943 if ( !( $flags & self::SFH_NO_HASH ) ) {
4944 $syn = '#' . $syn;
4945 }
4946 # Remove trailing colon
4947 if ( substr( $syn, -1, 1 ) === ':' ) {
4948 $syn = substr( $syn, 0, -1 );
4949 }
4950 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4951 }
4952 return $oldVal;
4953 }
4954
4955 /**
4956 * Get all registered function hook identifiers
4957 *
4958 * @return array
4959 */
4960 public function getFunctionHooks() {
4961 $this->firstCallInit();
4962 return array_keys( $this->mFunctionHooks );
4963 }
4964
4965 /**
4966 * Create a tag function, e.g. "<test>some stuff</test>".
4967 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4968 * Unlike parser functions, their content is not preprocessed.
4969 * @param string $tag
4970 * @param callable $callback
4971 * @param int $flags
4972 * @throws MWException
4973 * @return null
4974 */
4975 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4976 $tag = strtolower( $tag );
4977 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4978 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4979 }
4980 $old = $this->mFunctionTagHooks[$tag] ?? null;
4981 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4982
4983 if ( !in_array( $tag, $this->mStripList ) ) {
4984 $this->mStripList[] = $tag;
4985 }
4986
4987 return $old;
4988 }
4989
4990 /**
4991 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4992 * Placeholders created in Linker::link()
4993 *
4994 * @param string &$text
4995 * @param int $options
4996 */
4997 public function replaceLinkHolders( &$text, $options = 0 ) {
4998 $this->mLinkHolders->replace( $text );
4999 }
5000
5001 /**
5002 * Replace "<!--LINK-->" link placeholders with plain text of links
5003 * (not HTML-formatted).
5004 *
5005 * @param string $text
5006 * @return string
5007 */
5008 public function replaceLinkHoldersText( $text ) {
5009 return $this->mLinkHolders->replaceText( $text );
5010 }
5011
5012 /**
5013 * Renders an image gallery from a text with one line per image.
5014 * text labels may be given by using |-style alternative text. E.g.
5015 * Image:one.jpg|The number "1"
5016 * Image:tree.jpg|A tree
5017 * given as text will return the HTML of a gallery with two images,
5018 * labeled 'The number "1"' and
5019 * 'A tree'.
5020 *
5021 * @param string $text
5022 * @param array $params
5023 * @return string HTML
5024 */
5025 public function renderImageGallery( $text, $params ) {
5026 $mode = false;
5027 if ( isset( $params['mode'] ) ) {
5028 $mode = $params['mode'];
5029 }
5030
5031 try {
5032 $ig = ImageGalleryBase::factory( $mode );
5033 } catch ( Exception $e ) {
5034 // If invalid type set, fallback to default.
5035 $ig = ImageGalleryBase::factory( false );
5036 }
5037
5038 $ig->setContextTitle( $this->mTitle );
5039 $ig->setShowBytes( false );
5040 $ig->setShowDimensions( false );
5041 $ig->setShowFilename( false );
5042 $ig->setParser( $this );
5043 $ig->setHideBadImages();
5044 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5045
5046 if ( isset( $params['showfilename'] ) ) {
5047 $ig->setShowFilename( true );
5048 } else {
5049 $ig->setShowFilename( false );
5050 }
5051 if ( isset( $params['caption'] ) ) {
5052 $caption = $params['caption'];
5053 $caption = htmlspecialchars( $caption );
5054 $caption = $this->replaceInternalLinks( $caption );
5055 $ig->setCaptionHtml( $caption );
5056 }
5057 if ( isset( $params['perrow'] ) ) {
5058 $ig->setPerRow( $params['perrow'] );
5059 }
5060 if ( isset( $params['widths'] ) ) {
5061 $ig->setWidths( $params['widths'] );
5062 }
5063 if ( isset( $params['heights'] ) ) {
5064 $ig->setHeights( $params['heights'] );
5065 }
5066 $ig->setAdditionalOptions( $params );
5067
5068 // Avoid PHP 7.1 warning from passing $this by reference
5069 $parser = $this;
5070 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5071
5072 $lines = StringUtils::explode( "\n", $text );
5073 foreach ( $lines as $line ) {
5074 # match lines like these:
5075 # Image:someimage.jpg|This is some image
5076 $matches = [];
5077 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5078 # Skip empty lines
5079 if ( count( $matches ) == 0 ) {
5080 continue;
5081 }
5082
5083 if ( strpos( $matches[0], '%' ) !== false ) {
5084 $matches[1] = rawurldecode( $matches[1] );
5085 }
5086 $title = Title::newFromText( $matches[1], NS_FILE );
5087 if ( is_null( $title ) ) {
5088 # Bogus title. Ignore these so we don't bomb out later.
5089 continue;
5090 }
5091
5092 # We need to get what handler the file uses, to figure out parameters.
5093 # Note, a hook can overide the file name, and chose an entirely different
5094 # file (which potentially could be of a different type and have different handler).
5095 $options = [];
5096 $descQuery = false;
5097 Hooks::run( 'BeforeParserFetchFileAndTitle',
5098 [ $this, $title, &$options, &$descQuery ] );
5099 # Don't register it now, as TraditionalImageGallery does that later.
5100 $file = $this->fetchFileNoRegister( $title, $options );
5101 $handler = $file ? $file->getHandler() : false;
5102
5103 $paramMap = [
5104 'img_alt' => 'gallery-internal-alt',
5105 'img_link' => 'gallery-internal-link',
5106 ];
5107 if ( $handler ) {
5108 $paramMap = $paramMap + $handler->getParamMap();
5109 // We don't want people to specify per-image widths.
5110 // Additionally the width parameter would need special casing anyhow.
5111 unset( $paramMap['img_width'] );
5112 }
5113
5114 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5115
5116 $label = '';
5117 $alt = '';
5118 $link = '';
5119 $handlerOptions = [];
5120 if ( isset( $matches[3] ) ) {
5121 // look for an |alt= definition while trying not to break existing
5122 // captions with multiple pipes (|) in it, until a more sensible grammar
5123 // is defined for images in galleries
5124
5125 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5126 // splitting on '|' is a bit odd, and different from makeImage.
5127 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5128 // Protect LanguageConverter markup
5129 $parameterMatches = StringUtils::delimiterExplode(
5130 '-{', '}-', '|', $matches[3], true /* nested */
5131 );
5132
5133 foreach ( $parameterMatches as $parameterMatch ) {
5134 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5135 if ( $magicName ) {
5136 $paramName = $paramMap[$magicName];
5137
5138 switch ( $paramName ) {
5139 case 'gallery-internal-alt':
5140 $alt = $this->stripAltText( $match, false );
5141 break;
5142 case 'gallery-internal-link':
5143 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5144 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5145 // Result of LanguageConverter::markNoConversion
5146 // invoked on an external link.
5147 $linkValue = substr( $linkValue, 4, -2 );
5148 }
5149 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5150 if ( $type === 'link-url' ) {
5151 $link = $target;
5152 $this->mOutput->addExternalLink( $target );
5153 } elseif ( $type === 'link-title' ) {
5154 $link = $target->getLinkURL();
5155 $this->mOutput->addLink( $target );
5156 }
5157 break;
5158 default:
5159 // Must be a handler specific parameter.
5160 if ( $handler->validateParam( $paramName, $match ) ) {
5161 $handlerOptions[$paramName] = $match;
5162 } else {
5163 // Guess not, consider it as caption.
5164 wfDebug( "$parameterMatch failed parameter validation\n" );
5165 $label = $parameterMatch;
5166 }
5167 }
5168
5169 } else {
5170 // Last pipe wins.
5171 $label = $parameterMatch;
5172 }
5173 }
5174 }
5175
5176 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5177 }
5178 $html = $ig->toHTML();
5179 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5180 return $html;
5181 }
5182
5183 /**
5184 * @param MediaHandler $handler
5185 * @return array
5186 */
5187 public function getImageParams( $handler ) {
5188 if ( $handler ) {
5189 $handlerClass = get_class( $handler );
5190 } else {
5191 $handlerClass = '';
5192 }
5193 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5194 # Initialise static lists
5195 static $internalParamNames = [
5196 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5197 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5198 'bottom', 'text-bottom' ],
5199 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5200 'upright', 'border', 'link', 'alt', 'class' ],
5201 ];
5202 static $internalParamMap;
5203 if ( !$internalParamMap ) {
5204 $internalParamMap = [];
5205 foreach ( $internalParamNames as $type => $names ) {
5206 foreach ( $names as $name ) {
5207 // For grep: img_left, img_right, img_center, img_none,
5208 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5209 // img_bottom, img_text_bottom,
5210 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5211 // img_border, img_link, img_alt, img_class
5212 $magicName = str_replace( '-', '_', "img_$name" );
5213 $internalParamMap[$magicName] = [ $type, $name ];
5214 }
5215 }
5216 }
5217
5218 # Add handler params
5219 $paramMap = $internalParamMap;
5220 if ( $handler ) {
5221 $handlerParamMap = $handler->getParamMap();
5222 foreach ( $handlerParamMap as $magic => $paramName ) {
5223 $paramMap[$magic] = [ 'handler', $paramName ];
5224 }
5225 }
5226 $this->mImageParams[$handlerClass] = $paramMap;
5227 $this->mImageParamsMagicArray[$handlerClass] =
5228 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5229 }
5230 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5231 }
5232
5233 /**
5234 * Parse image options text and use it to make an image
5235 *
5236 * @param Title $title
5237 * @param string $options
5238 * @param LinkHolderArray|bool $holders
5239 * @return string HTML
5240 */
5241 public function makeImage( $title, $options, $holders = false ) {
5242 # Check if the options text is of the form "options|alt text"
5243 # Options are:
5244 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5245 # * left no resizing, just left align. label is used for alt= only
5246 # * right same, but right aligned
5247 # * none same, but not aligned
5248 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5249 # * center center the image
5250 # * frame Keep original image size, no magnify-button.
5251 # * framed Same as "frame"
5252 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5253 # * upright reduce width for upright images, rounded to full __0 px
5254 # * border draw a 1px border around the image
5255 # * alt Text for HTML alt attribute (defaults to empty)
5256 # * class Set a class for img node
5257 # * link Set the target of the image link. Can be external, interwiki, or local
5258 # vertical-align values (no % or length right now):
5259 # * baseline
5260 # * sub
5261 # * super
5262 # * top
5263 # * text-top
5264 # * middle
5265 # * bottom
5266 # * text-bottom
5267
5268 # Protect LanguageConverter markup when splitting into parts
5269 $parts = StringUtils::delimiterExplode(
5270 '-{', '}-', '|', $options, true /* allow nesting */
5271 );
5272
5273 # Give extensions a chance to select the file revision for us
5274 $options = [];
5275 $descQuery = false;
5276 Hooks::run( 'BeforeParserFetchFileAndTitle',
5277 [ $this, $title, &$options, &$descQuery ] );
5278 # Fetch and register the file (file title may be different via hooks)
5279 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5280
5281 # Get parameter map
5282 $handler = $file ? $file->getHandler() : false;
5283
5284 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5285
5286 if ( !$file ) {
5287 $this->addTrackingCategory( 'broken-file-category' );
5288 }
5289
5290 # Process the input parameters
5291 $caption = '';
5292 $params = [ 'frame' => [], 'handler' => [],
5293 'horizAlign' => [], 'vertAlign' => [] ];
5294 $seenformat = false;
5295 foreach ( $parts as $part ) {
5296 $part = trim( $part );
5297 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5298 $validated = false;
5299 if ( isset( $paramMap[$magicName] ) ) {
5300 list( $type, $paramName ) = $paramMap[$magicName];
5301
5302 # Special case; width and height come in one variable together
5303 if ( $type === 'handler' && $paramName === 'width' ) {
5304 $parsedWidthParam = self::parseWidthParam( $value );
5305 if ( isset( $parsedWidthParam['width'] ) ) {
5306 $width = $parsedWidthParam['width'];
5307 if ( $handler->validateParam( 'width', $width ) ) {
5308 $params[$type]['width'] = $width;
5309 $validated = true;
5310 }
5311 }
5312 if ( isset( $parsedWidthParam['height'] ) ) {
5313 $height = $parsedWidthParam['height'];
5314 if ( $handler->validateParam( 'height', $height ) ) {
5315 $params[$type]['height'] = $height;
5316 $validated = true;
5317 }
5318 }
5319 # else no validation -- T15436
5320 } else {
5321 if ( $type === 'handler' ) {
5322 # Validate handler parameter
5323 $validated = $handler->validateParam( $paramName, $value );
5324 } else {
5325 # Validate internal parameters
5326 switch ( $paramName ) {
5327 case 'manualthumb':
5328 case 'alt':
5329 case 'class':
5330 # @todo FIXME: Possibly check validity here for
5331 # manualthumb? downstream behavior seems odd with
5332 # missing manual thumbs.
5333 $validated = true;
5334 $value = $this->stripAltText( $value, $holders );
5335 break;
5336 case 'link':
5337 list( $paramName, $value ) = $this->parseLinkParameter( $value );
5338 if ( $paramName ) {
5339 $validated = true;
5340 if ( $paramName === 'no-link' ) {
5341 $value = true;
5342 }
5343 if ( $paramName === 'link-url' ) {
5344 if ( $this->mOptions->getExternalLinkTarget() ) {
5345 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5346 }
5347 }
5348 }
5349 break;
5350 case 'frameless':
5351 case 'framed':
5352 case 'thumbnail':
5353 // use first appearing option, discard others.
5354 $validated = !$seenformat;
5355 $seenformat = true;
5356 break;
5357 default:
5358 # Most other things appear to be empty or numeric...
5359 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5360 }
5361 }
5362
5363 if ( $validated ) {
5364 $params[$type][$paramName] = $value;
5365 }
5366 }
5367 }
5368 if ( !$validated ) {
5369 $caption = $part;
5370 }
5371 }
5372
5373 # Process alignment parameters
5374 if ( $params['horizAlign'] ) {
5375 $params['frame']['align'] = key( $params['horizAlign'] );
5376 }
5377 if ( $params['vertAlign'] ) {
5378 $params['frame']['valign'] = key( $params['vertAlign'] );
5379 }
5380
5381 $params['frame']['caption'] = $caption;
5382
5383 # Will the image be presented in a frame, with the caption below?
5384 $imageIsFramed = isset( $params['frame']['frame'] )
5385 || isset( $params['frame']['framed'] )
5386 || isset( $params['frame']['thumbnail'] )
5387 || isset( $params['frame']['manualthumb'] );
5388
5389 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5390 # came to also set the caption, ordinary text after the image -- which
5391 # makes no sense, because that just repeats the text multiple times in
5392 # screen readers. It *also* came to set the title attribute.
5393 # Now that we have an alt attribute, we should not set the alt text to
5394 # equal the caption: that's worse than useless, it just repeats the
5395 # text. This is the framed/thumbnail case. If there's no caption, we
5396 # use the unnamed parameter for alt text as well, just for the time be-
5397 # ing, if the unnamed param is set and the alt param is not.
5398 # For the future, we need to figure out if we want to tweak this more,
5399 # e.g., introducing a title= parameter for the title; ignoring the un-
5400 # named parameter entirely for images without a caption; adding an ex-
5401 # plicit caption= parameter and preserving the old magic unnamed para-
5402 # meter for BC; ...
5403 if ( $imageIsFramed ) { # Framed image
5404 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5405 # No caption or alt text, add the filename as the alt text so
5406 # that screen readers at least get some description of the image
5407 $params['frame']['alt'] = $title->getText();
5408 }
5409 # Do not set $params['frame']['title'] because tooltips don't make sense
5410 # for framed images
5411 } else { # Inline image
5412 if ( !isset( $params['frame']['alt'] ) ) {
5413 # No alt text, use the "caption" for the alt text
5414 if ( $caption !== '' ) {
5415 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5416 } else {
5417 # No caption, fall back to using the filename for the
5418 # alt text
5419 $params['frame']['alt'] = $title->getText();
5420 }
5421 }
5422 # Use the "caption" for the tooltip text
5423 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5424 }
5425
5426 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5427
5428 # Linker does the rest
5429 $time = $options['time'] ?? false;
5430 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5431 $time, $descQuery, $this->mOptions->getThumbSize() );
5432
5433 # Give the handler a chance to modify the parser object
5434 if ( $handler ) {
5435 $handler->parserTransformHook( $this, $file );
5436 }
5437
5438 return $ret;
5439 }
5440
5441 /**
5442 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5443 *
5444 * Adds an entry to appropriate link tables.
5445 *
5446 * @since 1.32
5447 * @return array of `[ type, target ]`, where:
5448 * - `type` is one of:
5449 * - `null`: Given value is not a valid link target, use default
5450 * - `'no-link'`: Given value is empty, do not generate a link
5451 * - `'link-url'`: Given value is a valid external link
5452 * - `'link-title'`: Given value is a valid internal link
5453 * - `target` is:
5454 * - When `type` is `null` or `'no-link'`: `false`
5455 * - When `type` is `'link-url'`: URL string corresponding to given value
5456 * - When `type` is `'link-title'`: Title object corresponding to given value
5457 */
5458 public function parseLinkParameter( $value ) {
5459 $chars = self::EXT_LINK_URL_CLASS;
5460 $addr = self::EXT_LINK_ADDR;
5461 $prots = $this->mUrlProtocols;
5462 $type = null;
5463 $target = false;
5464 if ( $value === '' ) {
5465 $type = 'no-link';
5466 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5467 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5468 $this->mOutput->addExternalLink( $value );
5469 $type = 'link-url';
5470 $target = $value;
5471 }
5472 } else {
5473 $linkTitle = Title::newFromText( $value );
5474 if ( $linkTitle ) {
5475 $this->mOutput->addLink( $linkTitle );
5476 $type = 'link-title';
5477 $target = $linkTitle;
5478 }
5479 }
5480 return [ $type, $target ];
5481 }
5482
5483 /**
5484 * @param string $caption
5485 * @param LinkHolderArray|bool $holders
5486 * @return mixed|string
5487 */
5488 protected function stripAltText( $caption, $holders ) {
5489 # Strip bad stuff out of the title (tooltip). We can't just use
5490 # replaceLinkHoldersText() here, because if this function is called
5491 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5492 if ( $holders ) {
5493 $tooltip = $holders->replaceText( $caption );
5494 } else {
5495 $tooltip = $this->replaceLinkHoldersText( $caption );
5496 }
5497
5498 # make sure there are no placeholders in thumbnail attributes
5499 # that are later expanded to html- so expand them now and
5500 # remove the tags
5501 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5502 $tooltip = Sanitizer::stripAllTags( $tooltip );
5503
5504 return $tooltip;
5505 }
5506
5507 /**
5508 * Set a flag in the output object indicating that the content is dynamic and
5509 * shouldn't be cached.
5510 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5511 */
5512 public function disableCache() {
5513 wfDebug( "Parser output marked as uncacheable.\n" );
5514 if ( !$this->mOutput ) {
5515 throw new MWException( __METHOD__ .
5516 " can only be called when actually parsing something" );
5517 }
5518 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5519 }
5520
5521 /**
5522 * Callback from the Sanitizer for expanding items found in HTML attribute
5523 * values, so they can be safely tested and escaped.
5524 *
5525 * @param string &$text
5526 * @param bool|PPFrame $frame
5527 * @return string
5528 */
5529 public function attributeStripCallback( &$text, $frame = false ) {
5530 $text = $this->replaceVariables( $text, $frame );
5531 $text = $this->mStripState->unstripBoth( $text );
5532 return $text;
5533 }
5534
5535 /**
5536 * Accessor
5537 *
5538 * @return array
5539 */
5540 public function getTags() {
5541 $this->firstCallInit();
5542 return array_merge(
5543 array_keys( $this->mTransparentTagHooks ),
5544 array_keys( $this->mTagHooks ),
5545 array_keys( $this->mFunctionTagHooks )
5546 );
5547 }
5548
5549 /**
5550 * @since 1.32
5551 * @return array
5552 */
5553 public function getFunctionSynonyms() {
5554 $this->firstCallInit();
5555 return $this->mFunctionSynonyms;
5556 }
5557
5558 /**
5559 * @since 1.32
5560 * @return string
5561 */
5562 public function getUrlProtocols() {
5563 return $this->mUrlProtocols;
5564 }
5565
5566 /**
5567 * Replace transparent tags in $text with the values given by the callbacks.
5568 *
5569 * Transparent tag hooks are like regular XML-style tag hooks, except they
5570 * operate late in the transformation sequence, on HTML instead of wikitext.
5571 *
5572 * @param string $text
5573 *
5574 * @return string
5575 */
5576 public function replaceTransparentTags( $text ) {
5577 $matches = [];
5578 $elements = array_keys( $this->mTransparentTagHooks );
5579 $text = self::extractTagsAndParams( $elements, $text, $matches );
5580 $replacements = [];
5581
5582 foreach ( $matches as $marker => $data ) {
5583 list( $element, $content, $params, $tag ) = $data;
5584 $tagName = strtolower( $element );
5585 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5586 $output = call_user_func_array(
5587 $this->mTransparentTagHooks[$tagName],
5588 [ $content, $params, $this ]
5589 );
5590 } else {
5591 $output = $tag;
5592 }
5593 $replacements[$marker] = $output;
5594 }
5595 return strtr( $text, $replacements );
5596 }
5597
5598 /**
5599 * Break wikitext input into sections, and either pull or replace
5600 * some particular section's text.
5601 *
5602 * External callers should use the getSection and replaceSection methods.
5603 *
5604 * @param string $text Page wikitext
5605 * @param string|int $sectionId A section identifier string of the form:
5606 * "<flag1> - <flag2> - ... - <section number>"
5607 *
5608 * Currently the only recognised flag is "T", which means the target section number
5609 * was derived during a template inclusion parse, in other words this is a template
5610 * section edit link. If no flags are given, it was an ordinary section edit link.
5611 * This flag is required to avoid a section numbering mismatch when a section is
5612 * enclosed by "<includeonly>" (T8563).
5613 *
5614 * The section number 0 pulls the text before the first heading; other numbers will
5615 * pull the given section along with its lower-level subsections. If the section is
5616 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5617 *
5618 * Section 0 is always considered to exist, even if it only contains the empty
5619 * string. If $text is the empty string and section 0 is replaced, $newText is
5620 * returned.
5621 *
5622 * @param string $mode One of "get" or "replace"
5623 * @param string $newText Replacement text for section data.
5624 * @return string For "get", the extracted section text.
5625 * for "replace", the whole page with the section replaced.
5626 */
5627 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5628 global $wgTitle; # not generally used but removes an ugly failure mode
5629
5630 $magicScopeVariable = $this->lock();
5631 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5632 $outText = '';
5633 $frame = $this->getPreprocessor()->newFrame();
5634
5635 # Process section extraction flags
5636 $flags = 0;
5637 $sectionParts = explode( '-', $sectionId );
5638 $sectionIndex = array_pop( $sectionParts );
5639 foreach ( $sectionParts as $part ) {
5640 if ( $part === 'T' ) {
5641 $flags |= self::PTD_FOR_INCLUSION;
5642 }
5643 }
5644
5645 # Check for empty input
5646 if ( strval( $text ) === '' ) {
5647 # Only sections 0 and T-0 exist in an empty document
5648 if ( $sectionIndex == 0 ) {
5649 if ( $mode === 'get' ) {
5650 return '';
5651 } else {
5652 return $newText;
5653 }
5654 } else {
5655 if ( $mode === 'get' ) {
5656 return $newText;
5657 } else {
5658 return $text;
5659 }
5660 }
5661 }
5662
5663 # Preprocess the text
5664 $root = $this->preprocessToDom( $text, $flags );
5665
5666 # <h> nodes indicate section breaks
5667 # They can only occur at the top level, so we can find them by iterating the root's children
5668 $node = $root->getFirstChild();
5669
5670 # Find the target section
5671 if ( $sectionIndex == 0 ) {
5672 # Section zero doesn't nest, level=big
5673 $targetLevel = 1000;
5674 } else {
5675 while ( $node ) {
5676 if ( $node->getName() === 'h' ) {
5677 $bits = $node->splitHeading();
5678 if ( $bits['i'] == $sectionIndex ) {
5679 $targetLevel = $bits['level'];
5680 break;
5681 }
5682 }
5683 if ( $mode === 'replace' ) {
5684 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5685 }
5686 $node = $node->getNextSibling();
5687 }
5688 }
5689
5690 if ( !$node ) {
5691 # Not found
5692 if ( $mode === 'get' ) {
5693 return $newText;
5694 } else {
5695 return $text;
5696 }
5697 }
5698
5699 # Find the end of the section, including nested sections
5700 do {
5701 if ( $node->getName() === 'h' ) {
5702 $bits = $node->splitHeading();
5703 $curLevel = $bits['level'];
5704 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5705 break;
5706 }
5707 }
5708 if ( $mode === 'get' ) {
5709 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5710 }
5711 $node = $node->getNextSibling();
5712 } while ( $node );
5713
5714 # Write out the remainder (in replace mode only)
5715 if ( $mode === 'replace' ) {
5716 # Output the replacement text
5717 # Add two newlines on -- trailing whitespace in $newText is conventionally
5718 # stripped by the editor, so we need both newlines to restore the paragraph gap
5719 # Only add trailing whitespace if there is newText
5720 if ( $newText != "" ) {
5721 $outText .= $newText . "\n\n";
5722 }
5723
5724 while ( $node ) {
5725 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5726 $node = $node->getNextSibling();
5727 }
5728 }
5729
5730 if ( is_string( $outText ) ) {
5731 # Re-insert stripped tags
5732 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5733 }
5734
5735 return $outText;
5736 }
5737
5738 /**
5739 * This function returns the text of a section, specified by a number ($section).
5740 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5741 * the first section before any such heading (section 0).
5742 *
5743 * If a section contains subsections, these are also returned.
5744 *
5745 * @param string $text Text to look in
5746 * @param string|int $sectionId Section identifier as a number or string
5747 * (e.g. 0, 1 or 'T-1').
5748 * @param string $defaultText Default to return if section is not found
5749 *
5750 * @return string Text of the requested section
5751 */
5752 public function getSection( $text, $sectionId, $defaultText = '' ) {
5753 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5754 }
5755
5756 /**
5757 * This function returns $oldtext after the content of the section
5758 * specified by $section has been replaced with $text. If the target
5759 * section does not exist, $oldtext is returned unchanged.
5760 *
5761 * @param string $oldText Former text of the article
5762 * @param string|int $sectionId Section identifier as a number or string
5763 * (e.g. 0, 1 or 'T-1').
5764 * @param string $newText Replacing text
5765 *
5766 * @return string Modified text
5767 */
5768 public function replaceSection( $oldText, $sectionId, $newText ) {
5769 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5770 }
5771
5772 /**
5773 * Get the ID of the revision we are parsing
5774 *
5775 * @return int|null
5776 */
5777 public function getRevisionId() {
5778 return $this->mRevisionId;
5779 }
5780
5781 /**
5782 * Get the revision object for $this->mRevisionId
5783 *
5784 * @return Revision|null Either a Revision object or null
5785 * @since 1.23 (public since 1.23)
5786 */
5787 public function getRevisionObject() {
5788 if ( !is_null( $this->mRevisionObject ) ) {
5789 return $this->mRevisionObject;
5790 }
5791
5792 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5793 // This is useful when parsing revision that has not yet been saved.
5794 // However, if we get back a saved revision even though we are in
5795 // preview mode, we'll have to ignore it, see below.
5796 // NOTE: This callback may be used to inject an OLD revision that was
5797 // already loaded, so "current" is a bit of a misnomer. We can't just
5798 // skip it if mRevisionId is set.
5799 $rev = call_user_func(
5800 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5801 );
5802
5803 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
5804 // We are in preview mode (mRevisionId is null), and the current revision callback
5805 // returned an existing revision. Ignore it and return null, it's probably the page's
5806 // current revision, which is not what we want here. Note that we do want to call the
5807 // callback to allow the unsaved revision to be injected here, e.g. for
5808 // self-transclusion previews.
5809 return null;
5810 }
5811
5812 // If the parse is for a new revision, then the callback should have
5813 // already been set to force the object and should match mRevisionId.
5814 // If not, try to fetch by mRevisionId for sanity.
5815 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
5816 $rev = Revision::newFromId( $this->mRevisionId );
5817 }
5818
5819 $this->mRevisionObject = $rev;
5820
5821 return $this->mRevisionObject;
5822 }
5823
5824 /**
5825 * Get the timestamp associated with the current revision, adjusted for
5826 * the default server-local timestamp
5827 * @return string
5828 */
5829 public function getRevisionTimestamp() {
5830 if ( is_null( $this->mRevisionTimestamp ) ) {
5831 $revObject = $this->getRevisionObject();
5832 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5833
5834 # The cryptic '' timezone parameter tells to use the site-default
5835 # timezone offset instead of the user settings.
5836 # Since this value will be saved into the parser cache, served
5837 # to other users, and potentially even used inside links and such,
5838 # it needs to be consistent for all visitors.
5839 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
5840 }
5841 return $this->mRevisionTimestamp;
5842 }
5843
5844 /**
5845 * Get the name of the user that edited the last revision
5846 *
5847 * @return string User name
5848 */
5849 public function getRevisionUser() {
5850 if ( is_null( $this->mRevisionUser ) ) {
5851 $revObject = $this->getRevisionObject();
5852
5853 # if this template is subst: the revision id will be blank,
5854 # so just use the current user's name
5855 if ( $revObject ) {
5856 $this->mRevisionUser = $revObject->getUserText();
5857 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5858 $this->mRevisionUser = $this->getUser()->getName();
5859 }
5860 }
5861 return $this->mRevisionUser;
5862 }
5863
5864 /**
5865 * Get the size of the revision
5866 *
5867 * @return int|null Revision size
5868 */
5869 public function getRevisionSize() {
5870 if ( is_null( $this->mRevisionSize ) ) {
5871 $revObject = $this->getRevisionObject();
5872
5873 # if this variable is subst: the revision id will be blank,
5874 # so just use the parser input size, because the own substituation
5875 # will change the size.
5876 if ( $revObject ) {
5877 $this->mRevisionSize = $revObject->getSize();
5878 } else {
5879 $this->mRevisionSize = $this->mInputSize;
5880 }
5881 }
5882 return $this->mRevisionSize;
5883 }
5884
5885 /**
5886 * Mutator for $mDefaultSort
5887 *
5888 * @param string $sort New value
5889 */
5890 public function setDefaultSort( $sort ) {
5891 $this->mDefaultSort = $sort;
5892 $this->mOutput->setProperty( 'defaultsort', $sort );
5893 }
5894
5895 /**
5896 * Accessor for $mDefaultSort
5897 * Will use the empty string if none is set.
5898 *
5899 * This value is treated as a prefix, so the
5900 * empty string is equivalent to sorting by
5901 * page name.
5902 *
5903 * @return string
5904 */
5905 public function getDefaultSort() {
5906 if ( $this->mDefaultSort !== false ) {
5907 return $this->mDefaultSort;
5908 } else {
5909 return '';
5910 }
5911 }
5912
5913 /**
5914 * Accessor for $mDefaultSort
5915 * Unlike getDefaultSort(), will return false if none is set
5916 *
5917 * @return string|bool
5918 */
5919 public function getCustomDefaultSort() {
5920 return $this->mDefaultSort;
5921 }
5922
5923 private static function getSectionNameFromStrippedText( $text ) {
5924 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5925 $text = Sanitizer::decodeCharReferences( $text );
5926 $text = self::normalizeSectionName( $text );
5927 return $text;
5928 }
5929
5930 private static function makeAnchor( $sectionName ) {
5931 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5932 }
5933
5934 private static function makeLegacyAnchor( $sectionName ) {
5935 global $wgFragmentMode;
5936 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5937 // ForAttribute() and ForLink() are the same for legacy encoding
5938 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5939 } else {
5940 $id = Sanitizer::escapeIdForLink( $sectionName );
5941 }
5942
5943 return "#$id";
5944 }
5945
5946 /**
5947 * Try to guess the section anchor name based on a wikitext fragment
5948 * presumably extracted from a heading, for example "Header" from
5949 * "== Header ==".
5950 *
5951 * @param string $text
5952 * @return string Anchor (starting with '#')
5953 */
5954 public function guessSectionNameFromWikiText( $text ) {
5955 # Strip out wikitext links(they break the anchor)
5956 $text = $this->stripSectionName( $text );
5957 $sectionName = self::getSectionNameFromStrippedText( $text );
5958 return self::makeAnchor( $sectionName );
5959 }
5960
5961 /**
5962 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5963 * instead, if possible. For use in redirects, since various versions
5964 * of Microsoft browsers interpret Location: headers as something other
5965 * than UTF-8, resulting in breakage.
5966 *
5967 * @param string $text The section name
5968 * @return string Anchor (starting with '#')
5969 */
5970 public function guessLegacySectionNameFromWikiText( $text ) {
5971 # Strip out wikitext links(they break the anchor)
5972 $text = $this->stripSectionName( $text );
5973 $sectionName = self::getSectionNameFromStrippedText( $text );
5974 return self::makeLegacyAnchor( $sectionName );
5975 }
5976
5977 /**
5978 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5979 * @param string $text Section name (plain text)
5980 * @return string Anchor (starting with '#')
5981 */
5982 public static function guessSectionNameFromStrippedText( $text ) {
5983 $sectionName = self::getSectionNameFromStrippedText( $text );
5984 return self::makeAnchor( $sectionName );
5985 }
5986
5987 /**
5988 * Apply the same normalization as code making links to this section would
5989 *
5990 * @param string $text
5991 * @return string
5992 */
5993 private static function normalizeSectionName( $text ) {
5994 # T90902: ensure the same normalization is applied for IDs as to links
5995 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5996 try {
5997
5998 $parts = $titleParser->splitTitleString( "#$text" );
5999 } catch ( MalformedTitleException $ex ) {
6000 return $text;
6001 }
6002 return $parts['fragment'];
6003 }
6004
6005 /**
6006 * Strips a text string of wikitext for use in a section anchor
6007 *
6008 * Accepts a text string and then removes all wikitext from the
6009 * string and leaves only the resultant text (i.e. the result of
6010 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6011 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6012 * to create valid section anchors by mimicing the output of the
6013 * parser when headings are parsed.
6014 *
6015 * @param string $text Text string to be stripped of wikitext
6016 * for use in a Section anchor
6017 * @return string Filtered text string
6018 */
6019 public function stripSectionName( $text ) {
6020 # Strip internal link markup
6021 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6022 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6023
6024 # Strip external link markup
6025 # @todo FIXME: Not tolerant to blank link text
6026 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6027 # on how many empty links there are on the page - need to figure that out.
6028 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6029
6030 # Parse wikitext quotes (italics & bold)
6031 $text = $this->doQuotes( $text );
6032
6033 # Strip HTML tags
6034 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6035 return $text;
6036 }
6037
6038 /**
6039 * strip/replaceVariables/unstrip for preprocessor regression testing
6040 *
6041 * @param string $text
6042 * @param Title $title
6043 * @param ParserOptions $options
6044 * @param int $outputType
6045 *
6046 * @return string
6047 */
6048 public function testSrvus( $text, Title $title, ParserOptions $options,
6049 $outputType = self::OT_HTML
6050 ) {
6051 $magicScopeVariable = $this->lock();
6052 $this->startParse( $title, $options, $outputType, true );
6053
6054 $text = $this->replaceVariables( $text );
6055 $text = $this->mStripState->unstripBoth( $text );
6056 $text = Sanitizer::removeHTMLtags( $text );
6057 return $text;
6058 }
6059
6060 /**
6061 * @param string $text
6062 * @param Title $title
6063 * @param ParserOptions $options
6064 * @return string
6065 */
6066 public function testPst( $text, Title $title, ParserOptions $options ) {
6067 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6068 }
6069
6070 /**
6071 * @param string $text
6072 * @param Title $title
6073 * @param ParserOptions $options
6074 * @return string
6075 */
6076 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6077 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6078 }
6079
6080 /**
6081 * Call a callback function on all regions of the given text that are not
6082 * inside strip markers, and replace those regions with the return value
6083 * of the callback. For example, with input:
6084 *
6085 * aaa<MARKER>bbb
6086 *
6087 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6088 * two strings will be replaced with the value returned by the callback in
6089 * each case.
6090 *
6091 * @param string $s
6092 * @param callable $callback
6093 *
6094 * @return string
6095 */
6096 public function markerSkipCallback( $s, $callback ) {
6097 $i = 0;
6098 $out = '';
6099 while ( $i < strlen( $s ) ) {
6100 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6101 if ( $markerStart === false ) {
6102 $out .= call_user_func( $callback, substr( $s, $i ) );
6103 break;
6104 } else {
6105 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6106 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6107 if ( $markerEnd === false ) {
6108 $out .= substr( $s, $markerStart );
6109 break;
6110 } else {
6111 $markerEnd += strlen( self::MARKER_SUFFIX );
6112 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6113 $i = $markerEnd;
6114 }
6115 }
6116 }
6117 return $out;
6118 }
6119
6120 /**
6121 * Remove any strip markers found in the given text.
6122 *
6123 * @param string $text
6124 * @return string
6125 */
6126 public function killMarkers( $text ) {
6127 return $this->mStripState->killMarkers( $text );
6128 }
6129
6130 /**
6131 * Save the parser state required to convert the given half-parsed text to
6132 * HTML. "Half-parsed" in this context means the output of
6133 * recursiveTagParse() or internalParse(). This output has strip markers
6134 * from replaceVariables (extensionSubstitution() etc.), and link
6135 * placeholders from replaceLinkHolders().
6136 *
6137 * Returns an array which can be serialized and stored persistently. This
6138 * array can later be loaded into another parser instance with
6139 * unserializeHalfParsedText(). The text can then be safely incorporated into
6140 * the return value of a parser hook.
6141 *
6142 * @deprecated since 1.31
6143 * @param string $text
6144 *
6145 * @return array
6146 */
6147 public function serializeHalfParsedText( $text ) {
6148 wfDeprecated( __METHOD__, '1.31' );
6149 $data = [
6150 'text' => $text,
6151 'version' => self::HALF_PARSED_VERSION,
6152 'stripState' => $this->mStripState->getSubState( $text ),
6153 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6154 ];
6155 return $data;
6156 }
6157
6158 /**
6159 * Load the parser state given in the $data array, which is assumed to
6160 * have been generated by serializeHalfParsedText(). The text contents is
6161 * extracted from the array, and its markers are transformed into markers
6162 * appropriate for the current Parser instance. This transformed text is
6163 * returned, and can be safely included in the return value of a parser
6164 * hook.
6165 *
6166 * If the $data array has been stored persistently, the caller should first
6167 * check whether it is still valid, by calling isValidHalfParsedText().
6168 *
6169 * @deprecated since 1.31
6170 * @param array $data Serialized data
6171 * @throws MWException
6172 * @return string
6173 */
6174 public function unserializeHalfParsedText( $data ) {
6175 wfDeprecated( __METHOD__, '1.31' );
6176 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6177 throw new MWException( __METHOD__ . ': invalid version' );
6178 }
6179
6180 # First, extract the strip state.
6181 $texts = [ $data['text'] ];
6182 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6183
6184 # Now renumber links
6185 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6186
6187 # Should be good to go.
6188 return $texts[0];
6189 }
6190
6191 /**
6192 * Returns true if the given array, presumed to be generated by
6193 * serializeHalfParsedText(), is compatible with the current version of the
6194 * parser.
6195 *
6196 * @deprecated since 1.31
6197 * @param array $data
6198 *
6199 * @return bool
6200 */
6201 public function isValidHalfParsedText( $data ) {
6202 wfDeprecated( __METHOD__, '1.31' );
6203 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6204 }
6205
6206 /**
6207 * Parsed a width param of imagelink like 300px or 200x300px
6208 *
6209 * @param string $value
6210 * @param bool $parseHeight
6211 *
6212 * @return array
6213 * @since 1.20
6214 */
6215 public static function parseWidthParam( $value, $parseHeight = true ) {
6216 $parsedWidthParam = [];
6217 if ( $value === '' ) {
6218 return $parsedWidthParam;
6219 }
6220 $m = [];
6221 # (T15500) In both cases (width/height and width only),
6222 # permit trailing "px" for backward compatibility.
6223 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6224 $width = intval( $m[1] );
6225 $height = intval( $m[2] );
6226 $parsedWidthParam['width'] = $width;
6227 $parsedWidthParam['height'] = $height;
6228 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6229 $width = intval( $value );
6230 $parsedWidthParam['width'] = $width;
6231 }
6232 return $parsedWidthParam;
6233 }
6234
6235 /**
6236 * Lock the current instance of the parser.
6237 *
6238 * This is meant to stop someone from calling the parser
6239 * recursively and messing up all the strip state.
6240 *
6241 * @throws MWException If parser is in a parse
6242 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6243 */
6244 protected function lock() {
6245 if ( $this->mInParse ) {
6246 throw new MWException( "Parser state cleared while parsing. "
6247 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6248 }
6249
6250 // Save the backtrace when locking, so that if some code tries locking again,
6251 // we can print the lock owner's backtrace for easier debugging
6252 $e = new Exception;
6253 $this->mInParse = $e->getTraceAsString();
6254
6255 $recursiveCheck = new ScopedCallback( function () {
6256 $this->mInParse = false;
6257 } );
6258
6259 return $recursiveCheck;
6260 }
6261
6262 /**
6263 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6264 *
6265 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6266 * or if there is more than one <p/> tag in the input HTML.
6267 *
6268 * @param string $html
6269 * @return string
6270 * @since 1.24
6271 */
6272 public static function stripOuterParagraph( $html ) {
6273 $m = [];
6274 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6275 if ( strpos( $m[1], '</p>' ) === false ) {
6276 $html = $m[1];
6277 }
6278 }
6279
6280 return $html;
6281 }
6282
6283 /**
6284 * Return this parser if it is not doing anything, otherwise
6285 * get a fresh parser. You can use this method by doing
6286 * $myParser = $wgParser->getFreshParser(), or more simply
6287 * $wgParser->getFreshParser()->parse( ... );
6288 * if you're unsure if $wgParser is safe to use.
6289 *
6290 * @since 1.24
6291 * @return Parser A parser object that is not parsing anything
6292 */
6293 public function getFreshParser() {
6294 if ( $this->mInParse ) {
6295 return $this->factory->create();
6296 } else {
6297 return $this;
6298 }
6299 }
6300
6301 /**
6302 * Set's up the PHP implementation of OOUI for use in this request
6303 * and instructs OutputPage to enable OOUI for itself.
6304 *
6305 * @since 1.26
6306 */
6307 public function enableOOUI() {
6308 OutputPage::setupOOUI();
6309 $this->mOutput->setEnableOOUI( true );
6310 }
6311 }
Wiklou, le Wiki du Wiklou