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