dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Inject Config into Parser instead of using globals
[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 // Run deprecated hook
620 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
621
622 // Sanitize for comment. Note '‐' in the replacement is U+2010,
623 // which looks much like the problematic '-'.
624 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
625 $text = "\n<!-- \n$limitReport-->\n";
626
627 // Add on template profiling data in human/machine readable way
628 $dataByFunc = $this->mProfiler->getFunctionStats();
629 uasort( $dataByFunc, function ( $a, $b ) {
630 return $b['real'] <=> $a['real']; // descending order
631 } );
632 $profileReport = [];
633 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
634 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
635 $item['%real'], $item['real'], $item['calls'],
636 htmlspecialchars( $item['name'] ) );
637 }
638 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
639 $text .= implode( "\n", $profileReport ) . "\n-->\n";
640
641 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
642
643 // Add other cache related metadata
644 if ( $this->siteConfig->get( 'ShowHostnames' ) ) {
645 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
646 }
647 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
648 $this->mOutput->getCacheTime() );
649 $this->mOutput->setLimitReportData( 'cachereport-ttl',
650 $this->mOutput->getCacheExpiry() );
651 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
652 $this->mOutput->hasDynamicContent() );
653
654 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
655 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
656 $this->mTitle->getPrefixedDBkey() );
657 }
658 return $text;
659 }
660
661 /**
662 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
663 * can be called from an extension tag hook.
664 *
665 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
666 * instead, which means that lists and links have not been fully parsed yet,
667 * and strip markers are still present.
668 *
669 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
670 *
671 * Use this function if you're a parser tag hook and you want to parse
672 * wikitext before or after applying additional transformations, and you
673 * intend to *return the result as hook output*, which will cause it to go
674 * through the rest of parsing process automatically.
675 *
676 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
677 * $text are not expanded
678 *
679 * @param string $text Text extension wants to have parsed
680 * @param-taint $text escapes_htmlnoent
681 * @param bool|PPFrame $frame The frame to use for expanding any template variables
682 * @return string UNSAFE half-parsed HTML
683 * @return-taint escaped
684 */
685 public function recursiveTagParse( $text, $frame = false ) {
686 // Avoid PHP 7.1 warning from passing $this by reference
687 $parser = $this;
688 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
689 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
690 $text = $this->internalParse( $text, false, $frame );
691 return $text;
692 }
693
694 /**
695 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
696 * point can be called from an extension tag hook.
697 *
698 * The output of this function is fully-parsed HTML that is safe for output.
699 * If you're a parser tag hook, you might want to use recursiveTagParse()
700 * instead.
701 *
702 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
703 * $text are not expanded
704 *
705 * @since 1.25
706 *
707 * @param string $text Text extension wants to have parsed
708 * @param-taint $text escapes_htmlnoent
709 * @param bool|PPFrame $frame The frame to use for expanding any template variables
710 * @return string Fully parsed HTML
711 * @return-taint escaped
712 */
713 public function recursiveTagParseFully( $text, $frame = false ) {
714 $text = $this->recursiveTagParse( $text, $frame );
715 $text = $this->internalParseHalfParsed( $text, false );
716 return $text;
717 }
718
719 /**
720 * Expand templates and variables in the text, producing valid, static wikitext.
721 * Also removes comments.
722 * Do not call this function recursively.
723 * @param string $text
724 * @param Title|null $title
725 * @param ParserOptions $options
726 * @param int|null $revid
727 * @param bool|PPFrame $frame
728 * @return mixed|string
729 */
730 public function preprocess( $text, Title $title = null,
731 ParserOptions $options, $revid = null, $frame = false
732 ) {
733 $magicScopeVariable = $this->lock();
734 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
735 if ( $revid !== null ) {
736 $this->mRevisionId = $revid;
737 }
738 // Avoid PHP 7.1 warning from passing $this by reference
739 $parser = $this;
740 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
741 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
742 $text = $this->replaceVariables( $text, $frame );
743 $text = $this->mStripState->unstripBoth( $text );
744 return $text;
745 }
746
747 /**
748 * Recursive parser entry point that can be called from an extension tag
749 * hook.
750 *
751 * @param string $text Text to be expanded
752 * @param bool|PPFrame $frame The frame to use for expanding any template variables
753 * @return string
754 * @since 1.19
755 */
756 public function recursivePreprocess( $text, $frame = false ) {
757 $text = $this->replaceVariables( $text, $frame );
758 $text = $this->mStripState->unstripBoth( $text );
759 return $text;
760 }
761
762 /**
763 * Process the wikitext for the "?preload=" feature. (T7210)
764 *
765 * "<noinclude>", "<includeonly>" etc. are parsed as for template
766 * transclusion, comments, templates, arguments, tags hooks and parser
767 * functions are untouched.
768 *
769 * @param string $text
770 * @param Title $title
771 * @param ParserOptions $options
772 * @param array $params
773 * @return string
774 */
775 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
776 $msg = new RawMessage( $text );
777 $text = $msg->params( $params )->plain();
778
779 # Parser (re)initialisation
780 $magicScopeVariable = $this->lock();
781 $this->startParse( $title, $options, self::OT_PLAIN, true );
782
783 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
784 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
785 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
786 $text = $this->mStripState->unstripBoth( $text );
787 return $text;
788 }
789
790 /**
791 * Set the current user.
792 * Should only be used when doing pre-save transform.
793 *
794 * @param User|null $user User object or null (to reset)
795 */
796 public function setUser( $user ) {
797 $this->mUser = $user;
798 }
799
800 /**
801 * Set the context title
802 *
803 * @param Title $t
804 */
805 public function setTitle( $t ) {
806 if ( !$t ) {
807 $t = Title::newFromText( 'NO TITLE' );
808 }
809
810 if ( $t->hasFragment() ) {
811 # Strip the fragment to avoid various odd effects
812 $this->mTitle = $t->createFragmentTarget( '' );
813 } else {
814 $this->mTitle = $t;
815 }
816 }
817
818 /**
819 * Accessor for the Title object
820 *
821 * @return Title
822 */
823 public function getTitle() {
824 return $this->mTitle;
825 }
826
827 /**
828 * Accessor/mutator for the Title object
829 *
830 * @param Title|null $x Title object or null to just get the current one
831 * @return Title
832 */
833 public function Title( $x = null ) {
834 return wfSetVar( $this->mTitle, $x );
835 }
836
837 /**
838 * Set the output type
839 *
840 * @param int $ot New value
841 */
842 public function setOutputType( $ot ) {
843 $this->mOutputType = $ot;
844 # Shortcut alias
845 $this->ot = [
846 'html' => $ot == self::OT_HTML,
847 'wiki' => $ot == self::OT_WIKI,
848 'pre' => $ot == self::OT_PREPROCESS,
849 'plain' => $ot == self::OT_PLAIN,
850 ];
851 }
852
853 /**
854 * Accessor/mutator for the output type
855 *
856 * @param int|null $x New value or null to just get the current one
857 * @return int
858 */
859 public function OutputType( $x = null ) {
860 return wfSetVar( $this->mOutputType, $x );
861 }
862
863 /**
864 * Get the ParserOutput object
865 *
866 * @return ParserOutput
867 */
868 public function getOutput() {
869 return $this->mOutput;
870 }
871
872 /**
873 * Get the ParserOptions object
874 *
875 * @return ParserOptions
876 */
877 public function getOptions() {
878 return $this->mOptions;
879 }
880
881 /**
882 * Accessor/mutator for the ParserOptions object
883 *
884 * @param ParserOptions|null $x New value or null to just get the current one
885 * @return ParserOptions Current ParserOptions object
886 */
887 public function Options( $x = null ) {
888 return wfSetVar( $this->mOptions, $x );
889 }
890
891 /**
892 * @return int
893 */
894 public function nextLinkID() {
895 return $this->mLinkID++;
896 }
897
898 /**
899 * @param int $id
900 */
901 public function setLinkID( $id ) {
902 $this->mLinkID = $id;
903 }
904
905 /**
906 * Get a language object for use in parser functions such as {{FORMATNUM:}}
907 * @return Language
908 */
909 public function getFunctionLang() {
910 return $this->getTargetLanguage();
911 }
912
913 /**
914 * Get the target language for the content being parsed. This is usually the
915 * language that the content is in.
916 *
917 * @since 1.19
918 *
919 * @throws MWException
920 * @return Language
921 */
922 public function getTargetLanguage() {
923 $target = $this->mOptions->getTargetLanguage();
924
925 if ( $target !== null ) {
926 return $target;
927 } elseif ( $this->mOptions->getInterfaceMessage() ) {
928 return $this->mOptions->getUserLangObj();
929 } elseif ( is_null( $this->mTitle ) ) {
930 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
931 }
932
933 return $this->mTitle->getPageLanguage();
934 }
935
936 /**
937 * Get the language object for language conversion
938 * @deprecated since 1.32, just use getTargetLanguage()
939 * @return Language|null
940 */
941 public function getConverterLanguage() {
942 return $this->getTargetLanguage();
943 }
944
945 /**
946 * Get a User object either from $this->mUser, if set, or from the
947 * ParserOptions object otherwise
948 *
949 * @return User
950 */
951 public function getUser() {
952 if ( !is_null( $this->mUser ) ) {
953 return $this->mUser;
954 }
955 return $this->mOptions->getUser();
956 }
957
958 /**
959 * Get a preprocessor object
960 *
961 * @return Preprocessor
962 */
963 public function getPreprocessor() {
964 if ( !isset( $this->mPreprocessor ) ) {
965 $class = $this->mPreprocessorClass;
966 $this->mPreprocessor = new $class( $this );
967 }
968 return $this->mPreprocessor;
969 }
970
971 /**
972 * Get a LinkRenderer instance to make links with
973 *
974 * @since 1.28
975 * @return LinkRenderer
976 */
977 public function getLinkRenderer() {
978 if ( !$this->mLinkRenderer ) {
979 $this->mLinkRenderer = MediaWikiServices::getInstance()
980 ->getLinkRendererFactory()->create();
981 $this->mLinkRenderer->setStubThreshold(
982 $this->getOptions()->getStubThreshold()
983 );
984 }
985
986 return $this->mLinkRenderer;
987 }
988
989 /**
990 * Get the MagicWordFactory that this Parser is using
991 *
992 * @since 1.32
993 * @return MagicWordFactory
994 */
995 public function getMagicWordFactory() {
996 return $this->magicWordFactory;
997 }
998
999 /**
1000 * Get the content language that this Parser is using
1001 *
1002 * @since 1.32
1003 * @return Language
1004 */
1005 public function getContentLanguage() {
1006 return $this->contLang;
1007 }
1008
1009 /**
1010 * Replaces all occurrences of HTML-style comments and the given tags
1011 * in the text with a random marker and returns the next text. The output
1012 * parameter $matches will be an associative array filled with data in
1013 * the form:
1014 *
1015 * @code
1016 * 'UNIQ-xxxxx' => [
1017 * 'element',
1018 * 'tag content',
1019 * [ 'param' => 'x' ],
1020 * '<element param="x">tag content</element>' ]
1021 * @endcode
1022 *
1023 * @param array $elements List of element names. Comments are always extracted.
1024 * @param string $text Source text string.
1025 * @param array &$matches Out parameter, Array: extracted tags
1026 * @return string Stripped text
1027 */
1028 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1029 static $n = 1;
1030 $stripped = '';
1031 $matches = [];
1032
1033 $taglist = implode( '|', $elements );
1034 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
1035
1036 while ( $text != '' ) {
1037 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
1038 $stripped .= $p[0];
1039 if ( count( $p ) < 5 ) {
1040 break;
1041 }
1042 if ( count( $p ) > 5 ) {
1043 # comment
1044 $element = $p[4];
1045 $attributes = '';
1046 $close = '';
1047 $inside = $p[5];
1048 } else {
1049 # tag
1050 $element = $p[1];
1051 $attributes = $p[2];
1052 $close = $p[3];
1053 $inside = $p[4];
1054 }
1055
1056 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1057 $stripped .= $marker;
1058
1059 if ( $close === '/>' ) {
1060 # Empty element tag, <tag />
1061 $content = null;
1062 $text = $inside;
1063 $tail = null;
1064 } else {
1065 if ( $element === '!--' ) {
1066 $end = '/(-->)/';
1067 } else {
1068 $end = "/(<\\/$element\\s*>)/i";
1069 }
1070 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1071 $content = $q[0];
1072 if ( count( $q ) < 3 ) {
1073 # No end tag -- let it run out to the end of the text.
1074 $tail = '';
1075 $text = '';
1076 } else {
1077 $tail = $q[1];
1078 $text = $q[2];
1079 }
1080 }
1081
1082 $matches[$marker] = [ $element,
1083 $content,
1084 Sanitizer::decodeTagAttributes( $attributes ),
1085 "<$element$attributes$close$content$tail" ];
1086 }
1087 return $stripped;
1088 }
1089
1090 /**
1091 * Get a list of strippable XML-like elements
1092 *
1093 * @return array
1094 */
1095 public function getStripList() {
1096 return $this->mStripList;
1097 }
1098
1099 /**
1100 * Add an item to the strip state
1101 * Returns the unique tag which must be inserted into the stripped text
1102 * The tag will be replaced with the original text in unstrip()
1103 *
1104 * @param string $text
1105 *
1106 * @return string
1107 */
1108 public function insertStripItem( $text ) {
1109 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1110 $this->mMarkerIndex++;
1111 $this->mStripState->addGeneral( $marker, $text );
1112 return $marker;
1113 }
1114
1115 /**
1116 * parse the wiki syntax used to render tables
1117 *
1118 * @private
1119 * @param string $text
1120 * @return string
1121 */
1122 public function doTableStuff( $text ) {
1123 $lines = StringUtils::explode( "\n", $text );
1124 $out = '';
1125 $td_history = []; # Is currently a td tag open?
1126 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1127 $tr_history = []; # Is currently a tr tag open?
1128 $tr_attributes = []; # history of tr attributes
1129 $has_opened_tr = []; # Did this table open a <tr> element?
1130 $indent_level = 0; # indent level of the table
1131
1132 foreach ( $lines as $outLine ) {
1133 $line = trim( $outLine );
1134
1135 if ( $line === '' ) { # empty line, go to next line
1136 $out .= $outLine . "\n";
1137 continue;
1138 }
1139
1140 $first_character = $line[0];
1141 $first_two = substr( $line, 0, 2 );
1142 $matches = [];
1143
1144 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1145 # First check if we are starting a new table
1146 $indent_level = strlen( $matches[1] );
1147
1148 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1149 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1150
1151 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1152 array_push( $td_history, false );
1153 array_push( $last_tag_history, '' );
1154 array_push( $tr_history, false );
1155 array_push( $tr_attributes, '' );
1156 array_push( $has_opened_tr, false );
1157 } elseif ( count( $td_history ) == 0 ) {
1158 # Don't do any of the following
1159 $out .= $outLine . "\n";
1160 continue;
1161 } elseif ( $first_two === '|}' ) {
1162 # We are ending a table
1163 $line = '</table>' . substr( $line, 2 );
1164 $last_tag = array_pop( $last_tag_history );
1165
1166 if ( !array_pop( $has_opened_tr ) ) {
1167 $line = "<tr><td></td></tr>{$line}";
1168 }
1169
1170 if ( array_pop( $tr_history ) ) {
1171 $line = "</tr>{$line}";
1172 }
1173
1174 if ( array_pop( $td_history ) ) {
1175 $line = "</{$last_tag}>{$line}";
1176 }
1177 array_pop( $tr_attributes );
1178 if ( $indent_level > 0 ) {
1179 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1180 } else {
1181 $outLine = $line;
1182 }
1183 } elseif ( $first_two === '|-' ) {
1184 # Now we have a table row
1185 $line = preg_replace( '#^\|-+#', '', $line );
1186
1187 # Whats after the tag is now only attributes
1188 $attributes = $this->mStripState->unstripBoth( $line );
1189 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1190 array_pop( $tr_attributes );
1191 array_push( $tr_attributes, $attributes );
1192
1193 $line = '';
1194 $last_tag = array_pop( $last_tag_history );
1195 array_pop( $has_opened_tr );
1196 array_push( $has_opened_tr, true );
1197
1198 if ( array_pop( $tr_history ) ) {
1199 $line = '</tr>';
1200 }
1201
1202 if ( array_pop( $td_history ) ) {
1203 $line = "</{$last_tag}>{$line}";
1204 }
1205
1206 $outLine = $line;
1207 array_push( $tr_history, false );
1208 array_push( $td_history, false );
1209 array_push( $last_tag_history, '' );
1210 } elseif ( $first_character === '|'
1211 || $first_character === '!'
1212 || $first_two === '|+'
1213 ) {
1214 # This might be cell elements, td, th or captions
1215 if ( $first_two === '|+' ) {
1216 $first_character = '+';
1217 $line = substr( $line, 2 );
1218 } else {
1219 $line = substr( $line, 1 );
1220 }
1221
1222 // Implies both are valid for table headings.
1223 if ( $first_character === '!' ) {
1224 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1225 }
1226
1227 # Split up multiple cells on the same line.
1228 # FIXME : This can result in improper nesting of tags processed
1229 # by earlier parser steps.
1230 $cells = explode( '||', $line );
1231
1232 $outLine = '';
1233
1234 # Loop through each table cell
1235 foreach ( $cells as $cell ) {
1236 $previous = '';
1237 if ( $first_character !== '+' ) {
1238 $tr_after = array_pop( $tr_attributes );
1239 if ( !array_pop( $tr_history ) ) {
1240 $previous = "<tr{$tr_after}>\n";
1241 }
1242 array_push( $tr_history, true );
1243 array_push( $tr_attributes, '' );
1244 array_pop( $has_opened_tr );
1245 array_push( $has_opened_tr, true );
1246 }
1247
1248 $last_tag = array_pop( $last_tag_history );
1249
1250 if ( array_pop( $td_history ) ) {
1251 $previous = "</{$last_tag}>\n{$previous}";
1252 }
1253
1254 if ( $first_character === '|' ) {
1255 $last_tag = 'td';
1256 } elseif ( $first_character === '!' ) {
1257 $last_tag = 'th';
1258 } elseif ( $first_character === '+' ) {
1259 $last_tag = 'caption';
1260 } else {
1261 $last_tag = '';
1262 }
1263
1264 array_push( $last_tag_history, $last_tag );
1265
1266 # A cell could contain both parameters and data
1267 $cell_data = explode( '|', $cell, 2 );
1268
1269 # T2553: Note that a '|' inside an invalid link should not
1270 # be mistaken as delimiting cell parameters
1271 # Bug T153140: Neither should language converter markup.
1272 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1273 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1274 } elseif ( count( $cell_data ) == 1 ) {
1275 // Whitespace in cells is trimmed
1276 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1277 } else {
1278 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1279 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1280 // Whitespace in cells is trimmed
1281 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1282 }
1283
1284 $outLine .= $cell;
1285 array_push( $td_history, true );
1286 }
1287 }
1288 $out .= $outLine . "\n";
1289 }
1290
1291 # Closing open td, tr && table
1292 while ( count( $td_history ) > 0 ) {
1293 if ( array_pop( $td_history ) ) {
1294 $out .= "</td>\n";
1295 }
1296 if ( array_pop( $tr_history ) ) {
1297 $out .= "</tr>\n";
1298 }
1299 if ( !array_pop( $has_opened_tr ) ) {
1300 $out .= "<tr><td></td></tr>\n";
1301 }
1302
1303 $out .= "</table>\n";
1304 }
1305
1306 # Remove trailing line-ending (b/c)
1307 if ( substr( $out, -1 ) === "\n" ) {
1308 $out = substr( $out, 0, -1 );
1309 }
1310
1311 # special case: don't return empty table
1312 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1313 $out = '';
1314 }
1315
1316 return $out;
1317 }
1318
1319 /**
1320 * Helper function for parse() that transforms wiki markup into half-parsed
1321 * HTML. Only called for $mOutputType == self::OT_HTML.
1322 *
1323 * @private
1324 *
1325 * @param string $text The text to parse
1326 * @param-taint $text escapes_html
1327 * @param bool $isMain Whether this is being called from the main parse() function
1328 * @param PPFrame|bool $frame A pre-processor frame
1329 *
1330 * @return string
1331 */
1332 public function internalParse( $text, $isMain = true, $frame = false ) {
1333 $origText = $text;
1334
1335 // Avoid PHP 7.1 warning from passing $this by reference
1336 $parser = $this;
1337
1338 # Hook to suspend the parser in this state
1339 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1340 return $text;
1341 }
1342
1343 # if $frame is provided, then use $frame for replacing any variables
1344 if ( $frame ) {
1345 # use frame depth to infer how include/noinclude tags should be handled
1346 # depth=0 means this is the top-level document; otherwise it's an included document
1347 if ( !$frame->depth ) {
1348 $flag = 0;
1349 } else {
1350 $flag = self::PTD_FOR_INCLUSION;
1351 }
1352 $dom = $this->preprocessToDom( $text, $flag );
1353 $text = $frame->expand( $dom );
1354 } else {
1355 # if $frame is not provided, then use old-style replaceVariables
1356 $text = $this->replaceVariables( $text );
1357 }
1358
1359 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1360 $text = Sanitizer::removeHTMLtags(
1361 $text,
1362 [ $this, 'attributeStripCallback' ],
1363 false,
1364 array_keys( $this->mTransparentTagHooks ),
1365 [],
1366 [ $this, 'addTrackingCategory' ]
1367 );
1368 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1369
1370 # Tables need to come after variable replacement for things to work
1371 # properly; putting them before other transformations should keep
1372 # exciting things like link expansions from showing up in surprising
1373 # places.
1374 $text = $this->doTableStuff( $text );
1375
1376 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1377
1378 $text = $this->doDoubleUnderscore( $text );
1379
1380 $text = $this->doHeadings( $text );
1381 $text = $this->replaceInternalLinks( $text );
1382 $text = $this->doAllQuotes( $text );
1383 $text = $this->replaceExternalLinks( $text );
1384
1385 # replaceInternalLinks may sometimes leave behind
1386 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1387 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1388
1389 $text = $this->doMagicLinks( $text );
1390 $text = $this->formatHeadings( $text, $origText, $isMain );
1391
1392 return $text;
1393 }
1394
1395 /**
1396 * Helper function for parse() that transforms half-parsed HTML into fully
1397 * parsed HTML.
1398 *
1399 * @param string $text
1400 * @param bool $isMain
1401 * @param bool $linestart
1402 * @return string
1403 */
1404 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1405 $text = $this->mStripState->unstripGeneral( $text );
1406
1407 // Avoid PHP 7.1 warning from passing $this by reference
1408 $parser = $this;
1409
1410 if ( $isMain ) {
1411 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1412 }
1413
1414 # Clean up special characters, only run once, next-to-last before doBlockLevels
1415 $text = Sanitizer::armorFrenchSpaces( $text );
1416
1417 $text = $this->doBlockLevels( $text, $linestart );
1418
1419 $this->replaceLinkHolders( $text );
1420
1421 /**
1422 * The input doesn't get language converted if
1423 * a) It's disabled
1424 * b) Content isn't converted
1425 * c) It's a conversion table
1426 * d) it is an interface message (which is in the user language)
1427 */
1428 if ( !( $this->mOptions->getDisableContentConversion()
1429 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1430 ) {
1431 if ( !$this->mOptions->getInterfaceMessage() ) {
1432 # The position of the convert() call should not be changed. it
1433 # assumes that the links are all replaced and the only thing left
1434 # is the <nowiki> mark.
1435 $text = $this->getTargetLanguage()->convert( $text );
1436 }
1437 }
1438
1439 $text = $this->mStripState->unstripNoWiki( $text );
1440
1441 if ( $isMain ) {
1442 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1443 }
1444
1445 $text = $this->replaceTransparentTags( $text );
1446 $text = $this->mStripState->unstripGeneral( $text );
1447
1448 $text = Sanitizer::normalizeCharReferences( $text );
1449
1450 if ( MWTidy::isEnabled() ) {
1451 if ( $this->mOptions->getTidy() ) {
1452 $text = MWTidy::tidy( $text );
1453 }
1454 } else {
1455 # attempt to sanitize at least some nesting problems
1456 # (T4702 and quite a few others)
1457 $tidyregs = [
1458 # ''Something [http://www.cool.com cool''] -->
1459 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1460 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1461 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1462 # fix up an anchor inside another anchor, only
1463 # at least for a single single nested link (T5695)
1464 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1465 '\\1\\2</a>\\3</a>\\1\\4</a>',
1466 # fix div inside inline elements- doBlockLevels won't wrap a line which
1467 # contains a div, so fix it up here; replace
1468 # div with escaped text
1469 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1470 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1471 # remove empty italic or bold tag pairs, some
1472 # introduced by rules above
1473 '/<([bi])><\/\\1>/' => '',
1474 ];
1475
1476 $text = preg_replace(
1477 array_keys( $tidyregs ),
1478 array_values( $tidyregs ),
1479 $text );
1480 }
1481
1482 if ( $isMain ) {
1483 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1484 }
1485
1486 return $text;
1487 }
1488
1489 /**
1490 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1491 * magic external links.
1492 *
1493 * DML
1494 * @private
1495 *
1496 * @param string $text
1497 *
1498 * @return string
1499 */
1500 public function doMagicLinks( $text ) {
1501 $prots = wfUrlProtocolsWithoutProtRel();
1502 $urlChar = self::EXT_LINK_URL_CLASS;
1503 $addr = self::EXT_LINK_ADDR;
1504 $space = self::SPACE_NOT_NL; # non-newline space
1505 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1506 $spaces = "$space++"; # possessive match of 1 or more spaces
1507 $text = preg_replace_callback(
1508 '!(?: # Start cases
1509 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1510 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1511 (\b # m[3]: Free external links
1512 (?i:$prots)
1513 ($addr$urlChar*) # m[4]: Post-protocol path
1514 ) |
1515 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1516 ([0-9]+)\b |
1517 \bISBN $spaces ( # m[6]: ISBN, capture number
1518 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1519 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1520 [0-9Xx] # check digit
1521 )\b
1522 )!xu", [ $this, 'magicLinkCallback' ], $text );
1523 return $text;
1524 }
1525
1526 /**
1527 * @throws MWException
1528 * @param array $m
1529 * @return string HTML
1530 */
1531 public function magicLinkCallback( $m ) {
1532 if ( isset( $m[1] ) && $m[1] !== '' ) {
1533 # Skip anchor
1534 return $m[0];
1535 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1536 # Skip HTML element
1537 return $m[0];
1538 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1539 # Free external link
1540 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1541 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1542 # RFC or PMID
1543 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1544 if ( !$this->mOptions->getMagicRFCLinks() ) {
1545 return $m[0];
1546 }
1547 $keyword = 'RFC';
1548 $urlmsg = 'rfcurl';
1549 $cssClass = 'mw-magiclink-rfc';
1550 $trackingCat = 'magiclink-tracking-rfc';
1551 $id = $m[5];
1552 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1553 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1554 return $m[0];
1555 }
1556 $keyword = 'PMID';
1557 $urlmsg = 'pubmedurl';
1558 $cssClass = 'mw-magiclink-pmid';
1559 $trackingCat = 'magiclink-tracking-pmid';
1560 $id = $m[5];
1561 } else {
1562 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1563 substr( $m[0], 0, 20 ) . '"' );
1564 }
1565 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1566 $this->addTrackingCategory( $trackingCat );
1567 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1568 } elseif ( isset( $m[6] ) && $m[6] !== ''
1569 && $this->mOptions->getMagicISBNLinks()
1570 ) {
1571 # ISBN
1572 $isbn = $m[6];
1573 $space = self::SPACE_NOT_NL; # non-newline space
1574 $isbn = preg_replace( "/$space/", ' ', $isbn );
1575 $num = strtr( $isbn, [
1576 '-' => '',
1577 ' ' => '',
1578 'x' => 'X',
1579 ] );
1580 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1581 return $this->getLinkRenderer()->makeKnownLink(
1582 SpecialPage::getTitleFor( 'Booksources', $num ),
1583 "ISBN $isbn",
1584 [
1585 'class' => 'internal mw-magiclink-isbn',
1586 'title' => false // suppress title attribute
1587 ]
1588 );
1589 } else {
1590 return $m[0];
1591 }
1592 }
1593
1594 /**
1595 * Make a free external link, given a user-supplied URL
1596 *
1597 * @param string $url
1598 * @param int $numPostProto
1599 * The number of characters after the protocol.
1600 * @return string HTML
1601 * @private
1602 */
1603 public function makeFreeExternalLink( $url, $numPostProto ) {
1604 $trail = '';
1605
1606 # The characters '<' and '>' (which were escaped by
1607 # removeHTMLtags()) should not be included in
1608 # URLs, per RFC 2396.
1609 # Make &nbsp; terminate a URL as well (bug T84937)
1610 $m2 = [];
1611 if ( preg_match(
1612 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1613 $url,
1614 $m2,
1615 PREG_OFFSET_CAPTURE
1616 ) ) {
1617 $trail = substr( $url, $m2[0][1] ) . $trail;
1618 $url = substr( $url, 0, $m2[0][1] );
1619 }
1620
1621 # Move trailing punctuation to $trail
1622 $sep = ',;\.:!?';
1623 # If there is no left bracket, then consider right brackets fair game too
1624 if ( strpos( $url, '(' ) === false ) {
1625 $sep .= ')';
1626 }
1627
1628 $urlRev = strrev( $url );
1629 $numSepChars = strspn( $urlRev, $sep );
1630 # Don't break a trailing HTML entity by moving the ; into $trail
1631 # This is in hot code, so use substr_compare to avoid having to
1632 # create a new string object for the comparison
1633 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1634 # more optimization: instead of running preg_match with a $
1635 # anchor, which can be slow, do the match on the reversed
1636 # string starting at the desired offset.
1637 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1638 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1639 $numSepChars--;
1640 }
1641 }
1642 if ( $numSepChars ) {
1643 $trail = substr( $url, -$numSepChars ) . $trail;
1644 $url = substr( $url, 0, -$numSepChars );
1645 }
1646
1647 # Verify that we still have a real URL after trail removal, and
1648 # not just lone protocol
1649 if ( strlen( $trail ) >= $numPostProto ) {
1650 return $url . $trail;
1651 }
1652
1653 $url = Sanitizer::cleanUrl( $url );
1654
1655 # Is this an external image?
1656 $text = $this->maybeMakeExternalImage( $url );
1657 if ( $text === false ) {
1658 # Not an image, make a link
1659 $text = Linker::makeExternalLink( $url,
1660 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1661 true, 'free',
1662 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1663 # Register it in the output object...
1664 $this->mOutput->addExternalLink( $url );
1665 }
1666 return $text . $trail;
1667 }
1668
1669 /**
1670 * Parse headers and return html
1671 *
1672 * @private
1673 *
1674 * @param string $text
1675 *
1676 * @return string
1677 */
1678 public function doHeadings( $text ) {
1679 for ( $i = 6; $i >= 1; --$i ) {
1680 $h = str_repeat( '=', $i );
1681 // Trim non-newline whitespace from headings
1682 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1683 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1684 }
1685 return $text;
1686 }
1687
1688 /**
1689 * Replace single quotes with HTML markup
1690 * @private
1691 *
1692 * @param string $text
1693 *
1694 * @return string The altered text
1695 */
1696 public function doAllQuotes( $text ) {
1697 $outtext = '';
1698 $lines = StringUtils::explode( "\n", $text );
1699 foreach ( $lines as $line ) {
1700 $outtext .= $this->doQuotes( $line ) . "\n";
1701 }
1702 $outtext = substr( $outtext, 0, -1 );
1703 return $outtext;
1704 }
1705
1706 /**
1707 * Helper function for doAllQuotes()
1708 *
1709 * @param string $text
1710 *
1711 * @return string
1712 */
1713 public function doQuotes( $text ) {
1714 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1715 $countarr = count( $arr );
1716 if ( $countarr == 1 ) {
1717 return $text;
1718 }
1719
1720 // First, do some preliminary work. This may shift some apostrophes from
1721 // being mark-up to being text. It also counts the number of occurrences
1722 // of bold and italics mark-ups.
1723 $numbold = 0;
1724 $numitalics = 0;
1725 for ( $i = 1; $i < $countarr; $i += 2 ) {
1726 $thislen = strlen( $arr[$i] );
1727 // If there are ever four apostrophes, assume the first is supposed to
1728 // be text, and the remaining three constitute mark-up for bold text.
1729 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1730 if ( $thislen == 4 ) {
1731 $arr[$i - 1] .= "'";
1732 $arr[$i] = "'''";
1733 $thislen = 3;
1734 } elseif ( $thislen > 5 ) {
1735 // If there are more than 5 apostrophes in a row, assume they're all
1736 // text except for the last 5.
1737 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1738 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1739 $arr[$i] = "'''''";
1740 $thislen = 5;
1741 }
1742 // Count the number of occurrences of bold and italics mark-ups.
1743 if ( $thislen == 2 ) {
1744 $numitalics++;
1745 } elseif ( $thislen == 3 ) {
1746 $numbold++;
1747 } elseif ( $thislen == 5 ) {
1748 $numitalics++;
1749 $numbold++;
1750 }
1751 }
1752
1753 // If there is an odd number of both bold and italics, it is likely
1754 // that one of the bold ones was meant to be an apostrophe followed
1755 // by italics. Which one we cannot know for certain, but it is more
1756 // likely to be one that has a single-letter word before it.
1757 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1758 $firstsingleletterword = -1;
1759 $firstmultiletterword = -1;
1760 $firstspace = -1;
1761 for ( $i = 1; $i < $countarr; $i += 2 ) {
1762 if ( strlen( $arr[$i] ) == 3 ) {
1763 $x1 = substr( $arr[$i - 1], -1 );
1764 $x2 = substr( $arr[$i - 1], -2, 1 );
1765 if ( $x1 === ' ' ) {
1766 if ( $firstspace == -1 ) {
1767 $firstspace = $i;
1768 }
1769 } elseif ( $x2 === ' ' ) {
1770 $firstsingleletterword = $i;
1771 // if $firstsingleletterword is set, we don't
1772 // look at the other options, so we can bail early.
1773 break;
1774 } else {
1775 if ( $firstmultiletterword == -1 ) {
1776 $firstmultiletterword = $i;
1777 }
1778 }
1779 }
1780 }
1781
1782 // If there is a single-letter word, use it!
1783 if ( $firstsingleletterword > -1 ) {
1784 $arr[$firstsingleletterword] = "''";
1785 $arr[$firstsingleletterword - 1] .= "'";
1786 } elseif ( $firstmultiletterword > -1 ) {
1787 // If not, but there's a multi-letter word, use that one.
1788 $arr[$firstmultiletterword] = "''";
1789 $arr[$firstmultiletterword - 1] .= "'";
1790 } elseif ( $firstspace > -1 ) {
1791 // ... otherwise use the first one that has neither.
1792 // (notice that it is possible for all three to be -1 if, for example,
1793 // there is only one pentuple-apostrophe in the line)
1794 $arr[$firstspace] = "''";
1795 $arr[$firstspace - 1] .= "'";
1796 }
1797 }
1798
1799 // Now let's actually convert our apostrophic mush to HTML!
1800 $output = '';
1801 $buffer = '';
1802 $state = '';
1803 $i = 0;
1804 foreach ( $arr as $r ) {
1805 if ( ( $i % 2 ) == 0 ) {
1806 if ( $state === 'both' ) {
1807 $buffer .= $r;
1808 } else {
1809 $output .= $r;
1810 }
1811 } else {
1812 $thislen = strlen( $r );
1813 if ( $thislen == 2 ) {
1814 if ( $state === 'i' ) {
1815 $output .= '</i>';
1816 $state = '';
1817 } elseif ( $state === 'bi' ) {
1818 $output .= '</i>';
1819 $state = 'b';
1820 } elseif ( $state === 'ib' ) {
1821 $output .= '</b></i><b>';
1822 $state = 'b';
1823 } elseif ( $state === 'both' ) {
1824 $output .= '<b><i>' . $buffer . '</i>';
1825 $state = 'b';
1826 } else { // $state can be 'b' or ''
1827 $output .= '<i>';
1828 $state .= 'i';
1829 }
1830 } elseif ( $thislen == 3 ) {
1831 if ( $state === 'b' ) {
1832 $output .= '</b>';
1833 $state = '';
1834 } elseif ( $state === 'bi' ) {
1835 $output .= '</i></b><i>';
1836 $state = 'i';
1837 } elseif ( $state === 'ib' ) {
1838 $output .= '</b>';
1839 $state = 'i';
1840 } elseif ( $state === 'both' ) {
1841 $output .= '<i><b>' . $buffer . '</b>';
1842 $state = 'i';
1843 } else { // $state can be 'i' or ''
1844 $output .= '<b>';
1845 $state .= 'b';
1846 }
1847 } elseif ( $thislen == 5 ) {
1848 if ( $state === 'b' ) {
1849 $output .= '</b><i>';
1850 $state = 'i';
1851 } elseif ( $state === 'i' ) {
1852 $output .= '</i><b>';
1853 $state = 'b';
1854 } elseif ( $state === 'bi' ) {
1855 $output .= '</i></b>';
1856 $state = '';
1857 } elseif ( $state === 'ib' ) {
1858 $output .= '</b></i>';
1859 $state = '';
1860 } elseif ( $state === 'both' ) {
1861 $output .= '<i><b>' . $buffer . '</b></i>';
1862 $state = '';
1863 } else { // ($state == '')
1864 $buffer = '';
1865 $state = 'both';
1866 }
1867 }
1868 }
1869 $i++;
1870 }
1871 // Now close all remaining tags. Notice that the order is important.
1872 if ( $state === 'b' || $state === 'ib' ) {
1873 $output .= '</b>';
1874 }
1875 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1876 $output .= '</i>';
1877 }
1878 if ( $state === 'bi' ) {
1879 $output .= '</b>';
1880 }
1881 // There might be lonely ''''', so make sure we have a buffer
1882 if ( $state === 'both' && $buffer ) {
1883 $output .= '<b><i>' . $buffer . '</i></b>';
1884 }
1885 return $output;
1886 }
1887
1888 /**
1889 * Replace external links (REL)
1890 *
1891 * Note: this is all very hackish and the order of execution matters a lot.
1892 * Make sure to run tests/parser/parserTests.php if you change this code.
1893 *
1894 * @private
1895 *
1896 * @param string $text
1897 *
1898 * @throws MWException
1899 * @return string
1900 */
1901 public function replaceExternalLinks( $text ) {
1902 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1903 if ( $bits === false ) {
1904 throw new MWException( "PCRE needs to be compiled with "
1905 . "--enable-unicode-properties in order for MediaWiki to function" );
1906 }
1907 $s = array_shift( $bits );
1908
1909 $i = 0;
1910 while ( $i < count( $bits ) ) {
1911 $url = $bits[$i++];
1912 $i++; // protocol
1913 $text = $bits[$i++];
1914 $trail = $bits[$i++];
1915
1916 # The characters '<' and '>' (which were escaped by
1917 # removeHTMLtags()) should not be included in
1918 # URLs, per RFC 2396.
1919 $m2 = [];
1920 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1921 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1922 $url = substr( $url, 0, $m2[0][1] );
1923 }
1924
1925 # If the link text is an image URL, replace it with an <img> tag
1926 # This happened by accident in the original parser, but some people used it extensively
1927 $img = $this->maybeMakeExternalImage( $text );
1928 if ( $img !== false ) {
1929 $text = $img;
1930 }
1931
1932 $dtrail = '';
1933
1934 # Set linktype for CSS
1935 $linktype = 'text';
1936
1937 # No link text, e.g. [http://domain.tld/some.link]
1938 if ( $text == '' ) {
1939 # Autonumber
1940 $langObj = $this->getTargetLanguage();
1941 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1942 $linktype = 'autonumber';
1943 } else {
1944 # Have link text, e.g. [http://domain.tld/some.link text]s
1945 # Check for trail
1946 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1947 }
1948
1949 // Excluding protocol-relative URLs may avoid many false positives.
1950 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1951 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1952 }
1953
1954 $url = Sanitizer::cleanUrl( $url );
1955
1956 # Use the encoded URL
1957 # This means that users can paste URLs directly into the text
1958 # Funny characters like ö aren't valid in URLs anyway
1959 # This was changed in August 2004
1960 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1961 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1962
1963 # Register link in the output object.
1964 $this->mOutput->addExternalLink( $url );
1965 }
1966
1967 return $s;
1968 }
1969
1970 /**
1971 * Get the rel attribute for a particular external link.
1972 *
1973 * @since 1.21
1974 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1975 * nofollow if appropriate
1976 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1977 * @return string|null Rel attribute for $url
1978 */
1979 public static function getExternalLinkRel( $url = false, $title = null ) {
1980 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1981 $ns = $title ? $title->getNamespace() : false;
1982 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1983 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1984 ) {
1985 return 'nofollow';
1986 }
1987 return null;
1988 }
1989
1990 /**
1991 * Get an associative array of additional HTML attributes appropriate for a
1992 * particular external link. This currently may include rel => nofollow
1993 * (depending on configuration, namespace, and the URL's domain) and/or a
1994 * target attribute (depending on configuration).
1995 *
1996 * @param string $url URL to extract the domain from for rel =>
1997 * nofollow if appropriate
1998 * @return array Associative array of HTML attributes
1999 */
2000 public function getExternalLinkAttribs( $url ) {
2001 $attribs = [];
2002 $rel = self::getExternalLinkRel( $url, $this->mTitle );
2003
2004 $target = $this->mOptions->getExternalLinkTarget();
2005 if ( $target ) {
2006 $attribs['target'] = $target;
2007 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2008 // T133507. New windows can navigate parent cross-origin.
2009 // Including noreferrer due to lacking browser
2010 // support of noopener. Eventually noreferrer should be removed.
2011 if ( $rel !== '' ) {
2012 $rel .= ' ';
2013 }
2014 $rel .= 'noreferrer noopener';
2015 }
2016 }
2017 $attribs['rel'] = $rel;
2018 return $attribs;
2019 }
2020
2021 /**
2022 * Replace unusual escape codes in a URL with their equivalent characters
2023 *
2024 * This generally follows the syntax defined in RFC 3986, with special
2025 * consideration for HTTP query strings.
2026 *
2027 * @param string $url
2028 * @return string
2029 */
2030 public static function normalizeLinkUrl( $url ) {
2031 # First, make sure unsafe characters are encoded
2032 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2033 function ( $m ) {
2034 return rawurlencode( $m[0] );
2035 },
2036 $url
2037 );
2038
2039 $ret = '';
2040 $end = strlen( $url );
2041
2042 # Fragment part - 'fragment'
2043 $start = strpos( $url, '#' );
2044 if ( $start !== false && $start < $end ) {
2045 $ret = self::normalizeUrlComponent(
2046 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2047 $end = $start;
2048 }
2049
2050 # Query part - 'query' minus &=+;
2051 $start = strpos( $url, '?' );
2052 if ( $start !== false && $start < $end ) {
2053 $ret = self::normalizeUrlComponent(
2054 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2055 $end = $start;
2056 }
2057
2058 # Scheme and path part - 'pchar'
2059 # (we assume no userinfo or encoded colons in the host)
2060 $ret = self::normalizeUrlComponent(
2061 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2062
2063 return $ret;
2064 }
2065
2066 private static function normalizeUrlComponent( $component, $unsafe ) {
2067 $callback = function ( $matches ) use ( $unsafe ) {
2068 $char = urldecode( $matches[0] );
2069 $ord = ord( $char );
2070 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2071 # Unescape it
2072 return $char;
2073 } else {
2074 # Leave it escaped, but use uppercase for a-f
2075 return strtoupper( $matches[0] );
2076 }
2077 };
2078 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2079 }
2080
2081 /**
2082 * make an image if it's allowed, either through the global
2083 * option, through the exception, or through the on-wiki whitelist
2084 *
2085 * @param string $url
2086 *
2087 * @return string
2088 */
2089 private function maybeMakeExternalImage( $url ) {
2090 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2091 $imagesexception = !empty( $imagesfrom );
2092 $text = false;
2093 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2094 if ( $imagesexception && is_array( $imagesfrom ) ) {
2095 $imagematch = false;
2096 foreach ( $imagesfrom as $match ) {
2097 if ( strpos( $url, $match ) === 0 ) {
2098 $imagematch = true;
2099 break;
2100 }
2101 }
2102 } elseif ( $imagesexception ) {
2103 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2104 } else {
2105 $imagematch = false;
2106 }
2107
2108 if ( $this->mOptions->getAllowExternalImages()
2109 || ( $imagesexception && $imagematch )
2110 ) {
2111 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2112 # Image found
2113 $text = Linker::makeExternalImage( $url );
2114 }
2115 }
2116 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2117 && preg_match( self::EXT_IMAGE_REGEX, $url )
2118 ) {
2119 $whitelist = explode(
2120 "\n",
2121 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2122 );
2123
2124 foreach ( $whitelist as $entry ) {
2125 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2126 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2127 continue;
2128 }
2129 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2130 # Image matches a whitelist entry
2131 $text = Linker::makeExternalImage( $url );
2132 break;
2133 }
2134 }
2135 }
2136 return $text;
2137 }
2138
2139 /**
2140 * Process [[ ]] wikilinks
2141 *
2142 * @param string $s
2143 *
2144 * @return string Processed text
2145 *
2146 * @private
2147 */
2148 public function replaceInternalLinks( $s ) {
2149 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2150 return $s;
2151 }
2152
2153 /**
2154 * Process [[ ]] wikilinks (RIL)
2155 * @param string &$s
2156 * @throws MWException
2157 * @return LinkHolderArray
2158 *
2159 * @private
2160 */
2161 public function replaceInternalLinks2( &$s ) {
2162 static $tc = false, $e1, $e1_img;
2163 # the % is needed to support urlencoded titles as well
2164 if ( !$tc ) {
2165 $tc = Title::legalChars() . '#%';
2166 # Match a link having the form [[namespace:link|alternate]]trail
2167 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2168 # Match cases where there is no "]]", which might still be images
2169 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2170 }
2171
2172 $holders = new LinkHolderArray( $this );
2173
2174 # split the entire text string on occurrences of [[
2175 $a = StringUtils::explode( '[[', ' ' . $s );
2176 # get the first element (all text up to first [[), and remove the space we added
2177 $s = $a->current();
2178 $a->next();
2179 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2180 $s = substr( $s, 1 );
2181
2182 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2183 $e2 = null;
2184 if ( $useLinkPrefixExtension ) {
2185 # Match the end of a line for a word that's not followed by whitespace,
2186 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2187 $charset = $this->contLang->linkPrefixCharset();
2188 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2189 }
2190
2191 if ( is_null( $this->mTitle ) ) {
2192 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2193 }
2194 $nottalk = !$this->mTitle->isTalkPage();
2195
2196 if ( $useLinkPrefixExtension ) {
2197 $m = [];
2198 if ( preg_match( $e2, $s, $m ) ) {
2199 $first_prefix = $m[2];
2200 } else {
2201 $first_prefix = false;
2202 }
2203 } else {
2204 $prefix = '';
2205 }
2206
2207 $useSubpages = $this->areSubpagesAllowed();
2208
2209 # Loop for each link
2210 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2211 # Check for excessive memory usage
2212 if ( $holders->isBig() ) {
2213 # Too big
2214 # Do the existence check, replace the link holders and clear the array
2215 $holders->replace( $s );
2216 $holders->clear();
2217 }
2218
2219 if ( $useLinkPrefixExtension ) {
2220 if ( preg_match( $e2, $s, $m ) ) {
2221 $prefix = $m[2];
2222 $s = $m[1];
2223 } else {
2224 $prefix = '';
2225 }
2226 # first link
2227 if ( $first_prefix ) {
2228 $prefix = $first_prefix;
2229 $first_prefix = false;
2230 }
2231 }
2232
2233 $might_be_img = false;
2234
2235 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2236 $text = $m[2];
2237 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2238 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2239 # the real problem is with the $e1 regex
2240 # See T1500.
2241 # Still some problems for cases where the ] is meant to be outside punctuation,
2242 # and no image is in sight. See T4095.
2243 if ( $text !== ''
2244 && substr( $m[3], 0, 1 ) === ']'
2245 && strpos( $text, '[' ) !== false
2246 ) {
2247 $text .= ']'; # so that replaceExternalLinks($text) works later
2248 $m[3] = substr( $m[3], 1 );
2249 }
2250 # fix up urlencoded title texts
2251 if ( strpos( $m[1], '%' ) !== false ) {
2252 # Should anchors '#' also be rejected?
2253 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2254 }
2255 $trail = $m[3];
2256 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2257 # Invalid, but might be an image with a link in its caption
2258 $might_be_img = true;
2259 $text = $m[2];
2260 if ( strpos( $m[1], '%' ) !== false ) {
2261 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2262 }
2263 $trail = "";
2264 } else { # Invalid form; output directly
2265 $s .= $prefix . '[[' . $line;
2266 continue;
2267 }
2268
2269 $origLink = ltrim( $m[1], ' ' );
2270
2271 # Don't allow internal links to pages containing
2272 # PROTO: where PROTO is a valid URL protocol; these
2273 # should be external links.
2274 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2275 $s .= $prefix . '[[' . $line;
2276 continue;
2277 }
2278
2279 # Make subpage if necessary
2280 if ( $useSubpages ) {
2281 $link = $this->maybeDoSubpageLink( $origLink, $text );
2282 } else {
2283 $link = $origLink;
2284 }
2285
2286 // \x7f isn't a default legal title char, so most likely strip
2287 // markers will force us into the "invalid form" path above. But,
2288 // just in case, let's assert that xmlish tags aren't valid in
2289 // the title position.
2290 $unstrip = $this->mStripState->killMarkers( $link );
2291 $noMarkers = ( $unstrip === $link );
2292
2293 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2294 if ( $nt === null ) {
2295 $s .= $prefix . '[[' . $line;
2296 continue;
2297 }
2298
2299 $ns = $nt->getNamespace();
2300 $iw = $nt->getInterwiki();
2301
2302 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2303
2304 if ( $might_be_img ) { # if this is actually an invalid link
2305 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2306 $found = false;
2307 while ( true ) {
2308 # look at the next 'line' to see if we can close it there
2309 $a->next();
2310 $next_line = $a->current();
2311 if ( $next_line === false || $next_line === null ) {
2312 break;
2313 }
2314 $m = explode( ']]', $next_line, 3 );
2315 if ( count( $m ) == 3 ) {
2316 # the first ]] closes the inner link, the second the image
2317 $found = true;
2318 $text .= "[[{$m[0]}]]{$m[1]}";
2319 $trail = $m[2];
2320 break;
2321 } elseif ( count( $m ) == 2 ) {
2322 # if there's exactly one ]] that's fine, we'll keep looking
2323 $text .= "[[{$m[0]}]]{$m[1]}";
2324 } else {
2325 # if $next_line is invalid too, we need look no further
2326 $text .= '[[' . $next_line;
2327 break;
2328 }
2329 }
2330 if ( !$found ) {
2331 # we couldn't find the end of this imageLink, so output it raw
2332 # but don't ignore what might be perfectly normal links in the text we've examined
2333 $holders->merge( $this->replaceInternalLinks2( $text ) );
2334 $s .= "{$prefix}[[$link|$text";
2335 # note: no $trail, because without an end, there *is* no trail
2336 continue;
2337 }
2338 } else { # it's not an image, so output it raw
2339 $s .= "{$prefix}[[$link|$text";
2340 # note: no $trail, because without an end, there *is* no trail
2341 continue;
2342 }
2343 }
2344
2345 $wasblank = ( $text == '' );
2346 if ( $wasblank ) {
2347 $text = $link;
2348 if ( !$noforce ) {
2349 # Strip off leading ':'
2350 $text = substr( $text, 1 );
2351 }
2352 } else {
2353 # T6598 madness. Handle the quotes only if they come from the alternate part
2354 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2355 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2356 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2357 $text = $this->doQuotes( $text );
2358 }
2359
2360 # Link not escaped by : , create the various objects
2361 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2362 # Interwikis
2363 if (
2364 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2365 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2366 in_array( $iw, $this->siteConfig->get( 'ExtraInterlanguageLinkPrefixes' ) )
2367 )
2368 ) {
2369 # T26502: filter duplicates
2370 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2371 $this->mLangLinkLanguages[$iw] = true;
2372 $this->mOutput->addLanguageLink( $nt->getFullText() );
2373 }
2374
2375 /**
2376 * Strip the whitespace interwiki links produce, see T10897
2377 */
2378 $s = rtrim( $s . $prefix ) . $trail; # T175416
2379 continue;
2380 }
2381
2382 if ( $ns == NS_FILE ) {
2383 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2384 if ( $wasblank ) {
2385 # if no parameters were passed, $text
2386 # becomes something like "File:Foo.png",
2387 # which we don't want to pass on to the
2388 # image generator
2389 $text = '';
2390 } else {
2391 # recursively parse links inside the image caption
2392 # actually, this will parse them in any other parameters, too,
2393 # but it might be hard to fix that, and it doesn't matter ATM
2394 $text = $this->replaceExternalLinks( $text );
2395 $holders->merge( $this->replaceInternalLinks2( $text ) );
2396 }
2397 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2398 $s .= $prefix . $this->armorLinks(
2399 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2400 continue;
2401 }
2402 } elseif ( $ns == NS_CATEGORY ) {
2403 /**
2404 * Strip the whitespace Category links produce, see T2087
2405 */
2406 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2407
2408 if ( $wasblank ) {
2409 $sortkey = $this->getDefaultSort();
2410 } else {
2411 $sortkey = $text;
2412 }
2413 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2414 $sortkey = str_replace( "\n", '', $sortkey );
2415 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2416 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2417
2418 continue;
2419 }
2420 }
2421
2422 # Self-link checking. For some languages, variants of the title are checked in
2423 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2424 # for linking to a different variant.
2425 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2426 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2427 continue;
2428 }
2429
2430 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2431 # @todo FIXME: Should do batch file existence checks, see comment below
2432 if ( $ns == NS_MEDIA ) {
2433 # Give extensions a chance to select the file revision for us
2434 $options = [];
2435 $descQuery = false;
2436 Hooks::run( 'BeforeParserFetchFileAndTitle',
2437 [ $this, $nt, &$options, &$descQuery ] );
2438 # Fetch and register the file (file title may be different via hooks)
2439 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2440 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2441 $s .= $prefix . $this->armorLinks(
2442 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2443 continue;
2444 }
2445
2446 # Some titles, such as valid special pages or files in foreign repos, should
2447 # be shown as bluelinks even though they're not included in the page table
2448 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2449 # batch file existence checks for NS_FILE and NS_MEDIA
2450 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2451 $this->mOutput->addLink( $nt );
2452 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2453 } else {
2454 # Links will be added to the output link list after checking
2455 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2456 }
2457 }
2458 return $holders;
2459 }
2460
2461 /**
2462 * Render a forced-blue link inline; protect against double expansion of
2463 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2464 * Since this little disaster has to split off the trail text to avoid
2465 * breaking URLs in the following text without breaking trails on the
2466 * wiki links, it's been made into a horrible function.
2467 *
2468 * @param Title $nt
2469 * @param string $text
2470 * @param string $trail
2471 * @param string $prefix
2472 * @return string HTML-wikitext mix oh yuck
2473 */
2474 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2475 list( $inside, $trail ) = Linker::splitTrail( $trail );
2476
2477 if ( $text == '' ) {
2478 $text = htmlspecialchars( $nt->getPrefixedText() );
2479 }
2480
2481 $link = $this->getLinkRenderer()->makeKnownLink(
2482 $nt, new HtmlArmor( "$prefix$text$inside" )
2483 );
2484
2485 return $this->armorLinks( $link ) . $trail;
2486 }
2487
2488 /**
2489 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2490 * going to go through further parsing steps before inline URL expansion.
2491 *
2492 * Not needed quite as much as it used to be since free links are a bit
2493 * more sensible these days. But bracketed links are still an issue.
2494 *
2495 * @param string $text More-or-less HTML
2496 * @return string Less-or-more HTML with NOPARSE bits
2497 */
2498 public function armorLinks( $text ) {
2499 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2500 self::MARKER_PREFIX . "NOPARSE$1", $text );
2501 }
2502
2503 /**
2504 * Return true if subpage links should be expanded on this page.
2505 * @return bool
2506 */
2507 public function areSubpagesAllowed() {
2508 # Some namespaces don't allow subpages
2509 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2510 }
2511
2512 /**
2513 * Handle link to subpage if necessary
2514 *
2515 * @param string $target The source of the link
2516 * @param string &$text The link text, modified as necessary
2517 * @return string The full name of the link
2518 * @private
2519 */
2520 public function maybeDoSubpageLink( $target, &$text ) {
2521 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2522 }
2523
2524 /**
2525 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2526 *
2527 * @param string $text
2528 * @param bool $linestart Whether or not this is at the start of a line.
2529 * @private
2530 * @return string The lists rendered as HTML
2531 */
2532 public function doBlockLevels( $text, $linestart ) {
2533 return BlockLevelPass::doBlockLevels( $text, $linestart );
2534 }
2535
2536 /**
2537 * Return value of a magic variable (like PAGENAME)
2538 *
2539 * @private
2540 *
2541 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2542 * @param bool|PPFrame $frame
2543 *
2544 * @throws MWException
2545 * @return string
2546 */
2547 public function getVariableValue( $index, $frame = false ) {
2548 if ( is_null( $this->mTitle ) ) {
2549 // If no title set, bad things are going to happen
2550 // later. Title should always be set since this
2551 // should only be called in the middle of a parse
2552 // operation (but the unit-tests do funky stuff)
2553 throw new MWException( __METHOD__ . ' Should only be '
2554 . ' called while parsing (no title set)' );
2555 }
2556
2557 // Avoid PHP 7.1 warning from passing $this by reference
2558 $parser = $this;
2559
2560 /**
2561 * Some of these require message or data lookups and can be
2562 * expensive to check many times.
2563 */
2564 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2565 if ( isset( $this->mVarCache[$index] ) ) {
2566 return $this->mVarCache[$index];
2567 }
2568 }
2569
2570 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2571 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2572
2573 $pageLang = $this->getFunctionLang();
2574
2575 switch ( $index ) {
2576 case '!':
2577 $value = '|';
2578 break;
2579 case 'currentmonth':
2580 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2581 break;
2582 case 'currentmonth1':
2583 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2584 break;
2585 case 'currentmonthname':
2586 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2587 break;
2588 case 'currentmonthnamegen':
2589 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2590 break;
2591 case 'currentmonthabbrev':
2592 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2593 break;
2594 case 'currentday':
2595 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2596 break;
2597 case 'currentday2':
2598 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2599 break;
2600 case 'localmonth':
2601 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2602 break;
2603 case 'localmonth1':
2604 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2605 break;
2606 case 'localmonthname':
2607 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2608 break;
2609 case 'localmonthnamegen':
2610 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2611 break;
2612 case 'localmonthabbrev':
2613 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2614 break;
2615 case 'localday':
2616 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2617 break;
2618 case 'localday2':
2619 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2620 break;
2621 case 'pagename':
2622 $value = wfEscapeWikiText( $this->mTitle->getText() );
2623 break;
2624 case 'pagenamee':
2625 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2626 break;
2627 case 'fullpagename':
2628 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2629 break;
2630 case 'fullpagenamee':
2631 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2632 break;
2633 case 'subpagename':
2634 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2635 break;
2636 case 'subpagenamee':
2637 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2638 break;
2639 case 'rootpagename':
2640 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2641 break;
2642 case 'rootpagenamee':
2643 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2644 ' ',
2645 '_',
2646 $this->mTitle->getRootText()
2647 ) ) );
2648 break;
2649 case 'basepagename':
2650 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2651 break;
2652 case 'basepagenamee':
2653 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2654 ' ',
2655 '_',
2656 $this->mTitle->getBaseText()
2657 ) ) );
2658 break;
2659 case 'talkpagename':
2660 if ( $this->mTitle->canHaveTalkPage() ) {
2661 $talkPage = $this->mTitle->getTalkPage();
2662 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2663 } else {
2664 $value = '';
2665 }
2666 break;
2667 case 'talkpagenamee':
2668 if ( $this->mTitle->canHaveTalkPage() ) {
2669 $talkPage = $this->mTitle->getTalkPage();
2670 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2671 } else {
2672 $value = '';
2673 }
2674 break;
2675 case 'subjectpagename':
2676 $subjPage = $this->mTitle->getSubjectPage();
2677 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2678 break;
2679 case 'subjectpagenamee':
2680 $subjPage = $this->mTitle->getSubjectPage();
2681 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2682 break;
2683 case 'pageid': // requested in T25427
2684 $pageid = $this->getTitle()->getArticleID();
2685 if ( $pageid == 0 ) {
2686 # 0 means the page doesn't exist in the database,
2687 # which means the user is previewing a new page.
2688 # The vary-revision flag must be set, because the magic word
2689 # will have a different value once the page is saved.
2690 $this->mOutput->setFlag( 'vary-revision' );
2691 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2692 }
2693 $value = $pageid ?: null;
2694 break;
2695 case 'revisionid':
2696 # Let the edit saving system know we should parse the page
2697 # *after* a revision ID has been assigned.
2698 $this->mOutput->setFlag( 'vary-revision-id' );
2699 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2700 $value = $this->mRevisionId;
2701
2702 if ( !$value ) {
2703 $rev = $this->getRevisionObject();
2704 if ( $rev ) {
2705 $value = $rev->getId();
2706 }
2707 }
2708
2709 if ( !$value ) {
2710 $value = $this->mOptions->getSpeculativeRevId();
2711 if ( $value ) {
2712 $this->mOutput->setSpeculativeRevIdUsed( $value );
2713 }
2714 }
2715 break;
2716 case 'revisionday':
2717 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2718 break;
2719 case 'revisionday2':
2720 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2721 break;
2722 case 'revisionmonth':
2723 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2724 break;
2725 case 'revisionmonth1':
2726 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2727 break;
2728 case 'revisionyear':
2729 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2730 break;
2731 case 'revisiontimestamp':
2732 # Let the edit saving system know we should parse the page
2733 # *after* a revision ID has been assigned. This is for null edits.
2734 $this->mOutput->setFlag( 'vary-revision' );
2735 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2736 $value = $this->getRevisionTimestamp();
2737 break;
2738 case 'revisionuser':
2739 # Let the edit saving system know we should parse the page
2740 # *after* a revision ID has been assigned for null edits.
2741 $this->mOutput->setFlag( 'vary-user' );
2742 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2743 $value = $this->getRevisionUser();
2744 break;
2745 case 'revisionsize':
2746 $value = $this->getRevisionSize();
2747 break;
2748 case 'namespace':
2749 $value = str_replace( '_', ' ',
2750 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2751 break;
2752 case 'namespacee':
2753 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
2754 break;
2755 case 'namespacenumber':
2756 $value = $this->mTitle->getNamespace();
2757 break;
2758 case 'talkspace':
2759 $value = $this->mTitle->canHaveTalkPage()
2760 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2761 : '';
2762 break;
2763 case 'talkspacee':
2764 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2765 break;
2766 case 'subjectspace':
2767 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2768 break;
2769 case 'subjectspacee':
2770 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2771 break;
2772 case 'currentdayname':
2773 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2774 break;
2775 case 'currentyear':
2776 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2777 break;
2778 case 'currenttime':
2779 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2780 break;
2781 case 'currenthour':
2782 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2783 break;
2784 case 'currentweek':
2785 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2786 # int to remove the padding
2787 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2788 break;
2789 case 'currentdow':
2790 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2791 break;
2792 case 'localdayname':
2793 $value = $pageLang->getWeekdayName(
2794 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2795 );
2796 break;
2797 case 'localyear':
2798 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2799 break;
2800 case 'localtime':
2801 $value = $pageLang->time(
2802 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2803 false,
2804 false
2805 );
2806 break;
2807 case 'localhour':
2808 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2809 break;
2810 case 'localweek':
2811 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2812 # int to remove the padding
2813 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2814 break;
2815 case 'localdow':
2816 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2817 break;
2818 case 'numberofarticles':
2819 $value = $pageLang->formatNum( SiteStats::articles() );
2820 break;
2821 case 'numberoffiles':
2822 $value = $pageLang->formatNum( SiteStats::images() );
2823 break;
2824 case 'numberofusers':
2825 $value = $pageLang->formatNum( SiteStats::users() );
2826 break;
2827 case 'numberofactiveusers':
2828 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2829 break;
2830 case 'numberofpages':
2831 $value = $pageLang->formatNum( SiteStats::pages() );
2832 break;
2833 case 'numberofadmins':
2834 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2835 break;
2836 case 'numberofedits':
2837 $value = $pageLang->formatNum( SiteStats::edits() );
2838 break;
2839 case 'currenttimestamp':
2840 $value = wfTimestamp( TS_MW, $ts );
2841 break;
2842 case 'localtimestamp':
2843 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2844 break;
2845 case 'currentversion':
2846 $value = SpecialVersion::getVersion();
2847 break;
2848 case 'articlepath':
2849 return $this->siteConfig->get( 'ArticlePath' );
2850 case 'sitename':
2851 return $this->siteConfig->get( 'Sitename' );
2852 case 'server':
2853 return $this->siteConfig->get( 'Server' );
2854 case 'servername':
2855 return $this->siteConfig->get( 'ServerName' );
2856 case 'scriptpath':
2857 return $this->siteConfig->get( 'ScriptPath' );
2858 case 'stylepath':
2859 return $this->siteConfig->get( 'StylePath' );
2860 case 'directionmark':
2861 return $pageLang->getDirMark();
2862 case 'contentlanguage':
2863 return $this->siteConfig->get( 'LanguageCode' );
2864 case 'pagelanguage':
2865 $value = $pageLang->getCode();
2866 break;
2867 case 'cascadingsources':
2868 $value = CoreParserFunctions::cascadingsources( $this );
2869 break;
2870 default:
2871 $ret = null;
2872 Hooks::run(
2873 'ParserGetVariableValueSwitch',
2874 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2875 );
2876
2877 return $ret;
2878 }
2879
2880 if ( $index ) {
2881 $this->mVarCache[$index] = $value;
2882 }
2883
2884 return $value;
2885 }
2886
2887 /**
2888 * @param int $start
2889 * @param int $len
2890 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2891 * @param string $variable Parser variable name
2892 * @return string
2893 */
2894 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2895 # Get the timezone-adjusted timestamp to be used for this revision
2896 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2897 # Possibly set vary-revision if there is not yet an associated revision
2898 if ( !$this->getRevisionObject() ) {
2899 # Get the timezone-adjusted timestamp $mtts seconds in the future
2900 $resThen = substr(
2901 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2902 $start,
2903 $len
2904 );
2905
2906 if ( $resNow !== $resThen ) {
2907 # Let the edit saving system know we should parse the page
2908 # *after* a revision ID has been assigned. This is for null edits.
2909 $this->mOutput->setFlag( 'vary-revision' );
2910 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2911 }
2912 }
2913
2914 return $resNow;
2915 }
2916
2917 /**
2918 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2919 *
2920 * @private
2921 */
2922 public function initialiseVariables() {
2923 $variableIDs = $this->magicWordFactory->getVariableIDs();
2924 $substIDs = $this->magicWordFactory->getSubstIDs();
2925
2926 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
2927 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
2928 }
2929
2930 /**
2931 * Preprocess some wikitext and return the document tree.
2932 * This is the ghost of replace_variables().
2933 *
2934 * @param string $text The text to parse
2935 * @param int $flags Bitwise combination of:
2936 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2937 * included. Default is to assume a direct page view.
2938 *
2939 * The generated DOM tree must depend only on the input text and the flags.
2940 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2941 *
2942 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2943 * change in the DOM tree for a given text, must be passed through the section identifier
2944 * in the section edit link and thus back to extractSections().
2945 *
2946 * The output of this function is currently only cached in process memory, but a persistent
2947 * cache may be implemented at a later date which takes further advantage of these strict
2948 * dependency requirements.
2949 *
2950 * @return PPNode
2951 */
2952 public function preprocessToDom( $text, $flags = 0 ) {
2953 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2954 return $dom;
2955 }
2956
2957 /**
2958 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2959 *
2960 * @param string $s
2961 *
2962 * @return array
2963 */
2964 public static function splitWhitespace( $s ) {
2965 $ltrimmed = ltrim( $s );
2966 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2967 $trimmed = rtrim( $ltrimmed );
2968 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2969 if ( $diff > 0 ) {
2970 $w2 = substr( $ltrimmed, -$diff );
2971 } else {
2972 $w2 = '';
2973 }
2974 return [ $w1, $trimmed, $w2 ];
2975 }
2976
2977 /**
2978 * Replace magic variables, templates, and template arguments
2979 * with the appropriate text. Templates are substituted recursively,
2980 * taking care to avoid infinite loops.
2981 *
2982 * Note that the substitution depends on value of $mOutputType:
2983 * self::OT_WIKI: only {{subst:}} templates
2984 * self::OT_PREPROCESS: templates but not extension tags
2985 * self::OT_HTML: all templates and extension tags
2986 *
2987 * @param string $text The text to transform
2988 * @param bool|PPFrame $frame Object describing the arguments passed to the
2989 * template. Arguments may also be provided as an associative array, as
2990 * was the usual case before MW1.12. Providing arguments this way may be
2991 * useful for extensions wishing to perform variable replacement
2992 * explicitly.
2993 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2994 * double-brace expansion.
2995 * @return string
2996 */
2997 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2998 # Is there any text? Also, Prevent too big inclusions!
2999 $textSize = strlen( $text );
3000 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3001 return $text;
3002 }
3003
3004 if ( $frame === false ) {
3005 $frame = $this->getPreprocessor()->newFrame();
3006 } elseif ( !( $frame instanceof PPFrame ) ) {
3007 wfDebug( __METHOD__ . " called using plain parameters instead of "
3008 . "a PPFrame instance. Creating custom frame.\n" );
3009 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3010 }
3011
3012 $dom = $this->preprocessToDom( $text );
3013 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3014 $text = $frame->expand( $dom, $flags );
3015
3016 return $text;
3017 }
3018
3019 /**
3020 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3021 *
3022 * @param array $args
3023 *
3024 * @return array
3025 */
3026 public static function createAssocArgs( $args ) {
3027 $assocArgs = [];
3028 $index = 1;
3029 foreach ( $args as $arg ) {
3030 $eqpos = strpos( $arg, '=' );
3031 if ( $eqpos === false ) {
3032 $assocArgs[$index++] = $arg;
3033 } else {
3034 $name = trim( substr( $arg, 0, $eqpos ) );
3035 $value = trim( substr( $arg, $eqpos + 1 ) );
3036 if ( $value === false ) {
3037 $value = '';
3038 }
3039 if ( $name !== false ) {
3040 $assocArgs[$name] = $value;
3041 }
3042 }
3043 }
3044
3045 return $assocArgs;
3046 }
3047
3048 /**
3049 * Warn the user when a parser limitation is reached
3050 * Will warn at most once the user per limitation type
3051 *
3052 * The results are shown during preview and run through the Parser (See EditPage.php)
3053 *
3054 * @param string $limitationType Should be one of:
3055 * 'expensive-parserfunction' (corresponding messages:
3056 * 'expensive-parserfunction-warning',
3057 * 'expensive-parserfunction-category')
3058 * 'post-expand-template-argument' (corresponding messages:
3059 * 'post-expand-template-argument-warning',
3060 * 'post-expand-template-argument-category')
3061 * 'post-expand-template-inclusion' (corresponding messages:
3062 * 'post-expand-template-inclusion-warning',
3063 * 'post-expand-template-inclusion-category')
3064 * 'node-count-exceeded' (corresponding messages:
3065 * 'node-count-exceeded-warning',
3066 * 'node-count-exceeded-category')
3067 * 'expansion-depth-exceeded' (corresponding messages:
3068 * 'expansion-depth-exceeded-warning',
3069 * 'expansion-depth-exceeded-category')
3070 * @param string|int|null $current Current value
3071 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3072 * exceeded, provide the values (optional)
3073 */
3074 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3075 # does no harm if $current and $max are present but are unnecessary for the message
3076 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3077 # only during preview, and that would split the parser cache unnecessarily.
3078 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3079 ->text();
3080 $this->mOutput->addWarning( $warning );
3081 $this->addTrackingCategory( "$limitationType-category" );
3082 }
3083
3084 /**
3085 * Return the text of a template, after recursively
3086 * replacing any variables or templates within the template.
3087 *
3088 * @param array $piece The parts of the template
3089 * $piece['title']: the title, i.e. the part before the |
3090 * $piece['parts']: the parameter array
3091 * $piece['lineStart']: whether the brace was at the start of a line
3092 * @param PPFrame $frame The current frame, contains template arguments
3093 * @throws Exception
3094 * @return string The text of the template
3095 */
3096 public function braceSubstitution( $piece, $frame ) {
3097 // Flags
3098
3099 // $text has been filled
3100 $found = false;
3101 // wiki markup in $text should be escaped
3102 $nowiki = false;
3103 // $text is HTML, armour it against wikitext transformation
3104 $isHTML = false;
3105 // Force interwiki transclusion to be done in raw mode not rendered
3106 $forceRawInterwiki = false;
3107 // $text is a DOM node needing expansion in a child frame
3108 $isChildObj = false;
3109 // $text is a DOM node needing expansion in the current frame
3110 $isLocalObj = false;
3111
3112 # Title object, where $text came from
3113 $title = false;
3114
3115 # $part1 is the bit before the first |, and must contain only title characters.
3116 # Various prefixes will be stripped from it later.
3117 $titleWithSpaces = $frame->expand( $piece['title'] );
3118 $part1 = trim( $titleWithSpaces );
3119 $titleText = false;
3120
3121 # Original title text preserved for various purposes
3122 $originalTitle = $part1;
3123
3124 # $args is a list of argument nodes, starting from index 0, not including $part1
3125 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3126 # below won't work b/c this $args isn't an object
3127 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3128
3129 $profileSection = null; // profile templates
3130
3131 # SUBST
3132 if ( !$found ) {
3133 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3134
3135 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3136 # Decide whether to expand template or keep wikitext as-is.
3137 if ( $this->ot['wiki'] ) {
3138 if ( $substMatch === false ) {
3139 $literal = true; # literal when in PST with no prefix
3140 } else {
3141 $literal = false; # expand when in PST with subst: or safesubst:
3142 }
3143 } else {
3144 if ( $substMatch == 'subst' ) {
3145 $literal = true; # literal when not in PST with plain subst:
3146 } else {
3147 $literal = false; # expand when not in PST with safesubst: or no prefix
3148 }
3149 }
3150 if ( $literal ) {
3151 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3152 $isLocalObj = true;
3153 $found = true;
3154 }
3155 }
3156
3157 # Variables
3158 if ( !$found && $args->getLength() == 0 ) {
3159 $id = $this->mVariables->matchStartToEnd( $part1 );
3160 if ( $id !== false ) {
3161 $text = $this->getVariableValue( $id, $frame );
3162 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3163 $this->mOutput->updateCacheExpiry(
3164 $this->magicWordFactory->getCacheTTL( $id ) );
3165 }
3166 $found = true;
3167 }
3168 }
3169
3170 # MSG, MSGNW and RAW
3171 if ( !$found ) {
3172 # Check for MSGNW:
3173 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3174 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3175 $nowiki = true;
3176 } else {
3177 # Remove obsolete MSG:
3178 $mwMsg = $this->magicWordFactory->get( 'msg' );
3179 $mwMsg->matchStartAndRemove( $part1 );
3180 }
3181
3182 # Check for RAW:
3183 $mwRaw = $this->magicWordFactory->get( 'raw' );
3184 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3185 $forceRawInterwiki = true;
3186 }
3187 }
3188
3189 # Parser functions
3190 if ( !$found ) {
3191 $colonPos = strpos( $part1, ':' );
3192 if ( $colonPos !== false ) {
3193 $func = substr( $part1, 0, $colonPos );
3194 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3195 $argsLength = $args->getLength();
3196 for ( $i = 0; $i < $argsLength; $i++ ) {
3197 $funcArgs[] = $args->item( $i );
3198 }
3199
3200 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3201
3202 // Extract any forwarded flags
3203 if ( isset( $result['title'] ) ) {
3204 $title = $result['title'];
3205 }
3206 if ( isset( $result['found'] ) ) {
3207 $found = $result['found'];
3208 }
3209 if ( array_key_exists( 'text', $result ) ) {
3210 // a string or null
3211 $text = $result['text'];
3212 }
3213 if ( isset( $result['nowiki'] ) ) {
3214 $nowiki = $result['nowiki'];
3215 }
3216 if ( isset( $result['isHTML'] ) ) {
3217 $isHTML = $result['isHTML'];
3218 }
3219 if ( isset( $result['forceRawInterwiki'] ) ) {
3220 $forceRawInterwiki = $result['forceRawInterwiki'];
3221 }
3222 if ( isset( $result['isChildObj'] ) ) {
3223 $isChildObj = $result['isChildObj'];
3224 }
3225 if ( isset( $result['isLocalObj'] ) ) {
3226 $isLocalObj = $result['isLocalObj'];
3227 }
3228 }
3229 }
3230
3231 # Finish mangling title and then check for loops.
3232 # Set $title to a Title object and $titleText to the PDBK
3233 if ( !$found ) {
3234 $ns = NS_TEMPLATE;
3235 # Split the title into page and subpage
3236 $subpage = '';
3237 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3238 if ( $part1 !== $relative ) {
3239 $part1 = $relative;
3240 $ns = $this->mTitle->getNamespace();
3241 }
3242 $title = Title::newFromText( $part1, $ns );
3243 if ( $title ) {
3244 $titleText = $title->getPrefixedText();
3245 # Check for language variants if the template is not found
3246 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3247 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3248 }
3249 # Do recursion depth check
3250 $limit = $this->mOptions->getMaxTemplateDepth();
3251 if ( $frame->depth >= $limit ) {
3252 $found = true;
3253 $text = '<span class="error">'
3254 . wfMessage( 'parser-template-recursion-depth-warning' )
3255 ->numParams( $limit )->inContentLanguage()->text()
3256 . '</span>';
3257 }
3258 }
3259 }
3260
3261 # Load from database
3262 if ( !$found && $title ) {
3263 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3264 if ( !$title->isExternal() ) {
3265 if ( $title->isSpecialPage()
3266 && $this->mOptions->getAllowSpecialInclusion()
3267 && $this->ot['html']
3268 ) {
3269 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3270 // Pass the template arguments as URL parameters.
3271 // "uselang" will have no effect since the Language object
3272 // is forced to the one defined in ParserOptions.
3273 $pageArgs = [];
3274 $argsLength = $args->getLength();
3275 for ( $i = 0; $i < $argsLength; $i++ ) {
3276 $bits = $args->item( $i )->splitArg();
3277 if ( strval( $bits['index'] ) === '' ) {
3278 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3279 $value = trim( $frame->expand( $bits['value'] ) );
3280 $pageArgs[$name] = $value;
3281 }
3282 }
3283
3284 // Create a new context to execute the special page
3285 $context = new RequestContext;
3286 $context->setTitle( $title );
3287 $context->setRequest( new FauxRequest( $pageArgs ) );
3288 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3289 $context->setUser( $this->getUser() );
3290 } else {
3291 // If this page is cached, then we better not be per user.
3292 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3293 }
3294 $context->setLanguage( $this->mOptions->getUserLangObj() );
3295 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3296 if ( $ret ) {
3297 $text = $context->getOutput()->getHTML();
3298 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3299 $found = true;
3300 $isHTML = true;
3301 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3302 $this->mOutput->updateRuntimeAdaptiveExpiry(
3303 $specialPage->maxIncludeCacheTime()
3304 );
3305 }
3306 }
3307 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3308 $found = false; # access denied
3309 wfDebug( __METHOD__ . ": template inclusion denied for " .
3310 $title->getPrefixedDBkey() . "\n" );
3311 } else {
3312 list( $text, $title ) = $this->getTemplateDom( $title );
3313 if ( $text !== false ) {
3314 $found = true;
3315 $isChildObj = true;
3316 }
3317 }
3318
3319 # If the title is valid but undisplayable, make a link to it
3320 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3321 $text = "[[:$titleText]]";
3322 $found = true;
3323 }
3324 } elseif ( $title->isTrans() ) {
3325 # Interwiki transclusion
3326 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3327 $text = $this->interwikiTransclude( $title, 'render' );
3328 $isHTML = true;
3329 } else {
3330 $text = $this->interwikiTransclude( $title, 'raw' );
3331 # Preprocess it like a template
3332 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3333 $isChildObj = true;
3334 }
3335 $found = true;
3336 }
3337
3338 # Do infinite loop check
3339 # This has to be done after redirect resolution to avoid infinite loops via redirects
3340 if ( !$frame->loopCheck( $title ) ) {
3341 $found = true;
3342 $text = '<span class="error">'
3343 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3344 . '</span>';
3345 $this->addTrackingCategory( 'template-loop-category' );
3346 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3347 wfEscapeWikiText( $titleText ) )->text() );
3348 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3349 }
3350 }
3351
3352 # If we haven't found text to substitute by now, we're done
3353 # Recover the source wikitext and return it
3354 if ( !$found ) {
3355 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3356 if ( $profileSection ) {
3357 $this->mProfiler->scopedProfileOut( $profileSection );
3358 }
3359 return [ 'object' => $text ];
3360 }
3361
3362 # Expand DOM-style return values in a child frame
3363 if ( $isChildObj ) {
3364 # Clean up argument array
3365 $newFrame = $frame->newChild( $args, $title );
3366
3367 if ( $nowiki ) {
3368 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3369 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3370 # Expansion is eligible for the empty-frame cache
3371 $text = $newFrame->cachedExpand( $titleText, $text );
3372 } else {
3373 # Uncached expansion
3374 $text = $newFrame->expand( $text );
3375 }
3376 }
3377 if ( $isLocalObj && $nowiki ) {
3378 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3379 $isLocalObj = false;
3380 }
3381
3382 if ( $profileSection ) {
3383 $this->mProfiler->scopedProfileOut( $profileSection );
3384 }
3385
3386 # Replace raw HTML by a placeholder
3387 if ( $isHTML ) {
3388 $text = $this->insertStripItem( $text );
3389 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3390 # Escape nowiki-style return values
3391 $text = wfEscapeWikiText( $text );
3392 } elseif ( is_string( $text )
3393 && !$piece['lineStart']
3394 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3395 ) {
3396 # T2529: if the template begins with a table or block-level
3397 # element, it should be treated as beginning a new line.
3398 # This behavior is somewhat controversial.
3399 $text = "\n" . $text;
3400 }
3401
3402 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3403 # Error, oversize inclusion
3404 if ( $titleText !== false ) {
3405 # Make a working, properly escaped link if possible (T25588)
3406 $text = "[[:$titleText]]";
3407 } else {
3408 # This will probably not be a working link, but at least it may
3409 # provide some hint of where the problem is
3410 preg_replace( '/^:/', '', $originalTitle );
3411 $text = "[[:$originalTitle]]";
3412 }
3413 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3414 . 'post-expand include size too large -->' );
3415 $this->limitationWarn( 'post-expand-template-inclusion' );
3416 }
3417
3418 if ( $isLocalObj ) {
3419 $ret = [ 'object' => $text ];
3420 } else {
3421 $ret = [ 'text' => $text ];
3422 }
3423
3424 return $ret;
3425 }
3426
3427 /**
3428 * Call a parser function and return an array with text and flags.
3429 *
3430 * The returned array will always contain a boolean 'found', indicating
3431 * whether the parser function was found or not. It may also contain the
3432 * following:
3433 * text: string|object, resulting wikitext or PP DOM object
3434 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3435 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3436 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3437 * nowiki: bool, wiki markup in $text should be escaped
3438 *
3439 * @since 1.21
3440 * @param PPFrame $frame The current frame, contains template arguments
3441 * @param string $function Function name
3442 * @param array $args Arguments to the function
3443 * @throws MWException
3444 * @return array
3445 */
3446 public function callParserFunction( $frame, $function, array $args = [] ) {
3447 # Case sensitive functions
3448 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3449 $function = $this->mFunctionSynonyms[1][$function];
3450 } else {
3451 # Case insensitive functions
3452 $function = $this->contLang->lc( $function );
3453 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3454 $function = $this->mFunctionSynonyms[0][$function];
3455 } else {
3456 return [ 'found' => false ];
3457 }
3458 }
3459
3460 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3461
3462 // Avoid PHP 7.1 warning from passing $this by reference
3463 $parser = $this;
3464
3465 $allArgs = [ &$parser ];
3466 if ( $flags & self::SFH_OBJECT_ARGS ) {
3467 # Convert arguments to PPNodes and collect for appending to $allArgs
3468 $funcArgs = [];
3469 foreach ( $args as $k => $v ) {
3470 if ( $v instanceof PPNode || $k === 0 ) {
3471 $funcArgs[] = $v;
3472 } else {
3473 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3474 }
3475 }
3476
3477 # Add a frame parameter, and pass the arguments as an array
3478 $allArgs[] = $frame;
3479 $allArgs[] = $funcArgs;
3480 } else {
3481 # Convert arguments to plain text and append to $allArgs
3482 foreach ( $args as $k => $v ) {
3483 if ( $v instanceof PPNode ) {
3484 $allArgs[] = trim( $frame->expand( $v ) );
3485 } elseif ( is_int( $k ) && $k >= 0 ) {
3486 $allArgs[] = trim( $v );
3487 } else {
3488 $allArgs[] = trim( "$k=$v" );
3489 }
3490 }
3491 }
3492
3493 $result = $callback( ...$allArgs );
3494
3495 # The interface for function hooks allows them to return a wikitext
3496 # string or an array containing the string and any flags. This mungs
3497 # things around to match what this method should return.
3498 if ( !is_array( $result ) ) {
3499 $result = [
3500 'found' => true,
3501 'text' => $result,
3502 ];
3503 } else {
3504 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3505 $result['text'] = $result[0];
3506 }
3507 unset( $result[0] );
3508 $result += [
3509 'found' => true,
3510 ];
3511 }
3512
3513 $noparse = true;
3514 $preprocessFlags = 0;
3515 if ( isset( $result['noparse'] ) ) {
3516 $noparse = $result['noparse'];
3517 }
3518 if ( isset( $result['preprocessFlags'] ) ) {
3519 $preprocessFlags = $result['preprocessFlags'];
3520 }
3521
3522 if ( !$noparse ) {
3523 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3524 $result['isChildObj'] = true;
3525 }
3526
3527 return $result;
3528 }
3529
3530 /**
3531 * Get the semi-parsed DOM representation of a template with a given title,
3532 * and its redirect destination title. Cached.
3533 *
3534 * @param Title $title
3535 *
3536 * @return array
3537 */
3538 public function getTemplateDom( $title ) {
3539 $cacheTitle = $title;
3540 $titleText = $title->getPrefixedDBkey();
3541
3542 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3543 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3544 $title = Title::makeTitle( $ns, $dbk );
3545 $titleText = $title->getPrefixedDBkey();
3546 }
3547 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3548 return [ $this->mTplDomCache[$titleText], $title ];
3549 }
3550
3551 # Cache miss, go to the database
3552 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3553
3554 if ( $text === false ) {
3555 $this->mTplDomCache[$titleText] = false;
3556 return [ false, $title ];
3557 }
3558
3559 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3560 $this->mTplDomCache[$titleText] = $dom;
3561
3562 if ( !$title->equals( $cacheTitle ) ) {
3563 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3564 [ $title->getNamespace(), $title->getDBkey() ];
3565 }
3566
3567 return [ $dom, $title ];
3568 }
3569
3570 /**
3571 * Fetch the current revision of a given title. Note that the revision
3572 * (and even the title) may not exist in the database, so everything
3573 * contributing to the output of the parser should use this method
3574 * where possible, rather than getting the revisions themselves. This
3575 * method also caches its results, so using it benefits performance.
3576 *
3577 * @since 1.24
3578 * @param Title $title
3579 * @return Revision
3580 */
3581 public function fetchCurrentRevisionOfTitle( $title ) {
3582 $cacheKey = $title->getPrefixedDBkey();
3583 if ( !$this->currentRevisionCache ) {
3584 $this->currentRevisionCache = new MapCacheLRU( 100 );
3585 }
3586 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3587 $this->currentRevisionCache->set( $cacheKey,
3588 // Defaults to Parser::statelessFetchRevision()
3589 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3590 );
3591 }
3592 return $this->currentRevisionCache->get( $cacheKey );
3593 }
3594
3595 /**
3596 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3597 * without passing them on to it.
3598 *
3599 * @since 1.24
3600 * @param Title $title
3601 * @param Parser|bool $parser
3602 * @return Revision|bool False if missing
3603 */
3604 public static function statelessFetchRevision( Title $title, $parser = false ) {
3605 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3606
3607 return $rev;
3608 }
3609
3610 /**
3611 * Fetch the unparsed text of a template and register a reference to it.
3612 * @param Title $title
3613 * @return array ( string or false, Title )
3614 */
3615 public function fetchTemplateAndTitle( $title ) {
3616 // Defaults to Parser::statelessFetchTemplate()
3617 $templateCb = $this->mOptions->getTemplateCallback();
3618 $stuff = call_user_func( $templateCb, $title, $this );
3619 // We use U+007F DELETE to distinguish strip markers from regular text.
3620 $text = $stuff['text'];
3621 if ( is_string( $stuff['text'] ) ) {
3622 $text = strtr( $text, "\x7f", "?" );
3623 }
3624 $finalTitle = $stuff['finalTitle'] ?? $title;
3625 if ( isset( $stuff['deps'] ) ) {
3626 foreach ( $stuff['deps'] as $dep ) {
3627 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3628 if ( $dep['title']->equals( $this->getTitle() ) ) {
3629 // If we transclude ourselves, the final result
3630 // will change based on the new version of the page
3631 $this->mOutput->setFlag( 'vary-revision' );
3632 }
3633 }
3634 }
3635 return [ $text, $finalTitle ];
3636 }
3637
3638 /**
3639 * Fetch the unparsed text of a template and register a reference to it.
3640 * @param Title $title
3641 * @return string|bool
3642 */
3643 public function fetchTemplate( $title ) {
3644 return $this->fetchTemplateAndTitle( $title )[0];
3645 }
3646
3647 /**
3648 * Static function to get a template
3649 * Can be overridden via ParserOptions::setTemplateCallback().
3650 *
3651 * @param Title $title
3652 * @param bool|Parser $parser
3653 *
3654 * @return array
3655 */
3656 public static function statelessFetchTemplate( $title, $parser = false ) {
3657 $text = $skip = false;
3658 $finalTitle = $title;
3659 $deps = [];
3660
3661 # Loop to fetch the article, with up to 1 redirect
3662 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
3663 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3664 # Give extensions a chance to select the revision instead
3665 $id = false; # Assume current
3666 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3667 [ $parser, $title, &$skip, &$id ] );
3668
3669 if ( $skip ) {
3670 $text = false;
3671 $deps[] = [
3672 'title' => $title,
3673 'page_id' => $title->getArticleID(),
3674 'rev_id' => null
3675 ];
3676 break;
3677 }
3678 # Get the revision
3679 if ( $id ) {
3680 $rev = Revision::newFromId( $id );
3681 } elseif ( $parser ) {
3682 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3683 } else {
3684 $rev = Revision::newFromTitle( $title );
3685 }
3686 $rev_id = $rev ? $rev->getId() : 0;
3687 # If there is no current revision, there is no page
3688 if ( $id === false && !$rev ) {
3689 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3690 $linkCache->addBadLinkObj( $title );
3691 }
3692
3693 $deps[] = [
3694 'title' => $title,
3695 'page_id' => $title->getArticleID(),
3696 'rev_id' => $rev_id ];
3697 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3698 # We fetched a rev from a different title; register it too...
3699 $deps[] = [
3700 'title' => $rev->getTitle(),
3701 'page_id' => $rev->getPage(),
3702 'rev_id' => $rev_id ];
3703 }
3704
3705 if ( $rev ) {
3706 $content = $rev->getContent();
3707 $text = $content ? $content->getWikitextForTransclusion() : null;
3708
3709 Hooks::run( 'ParserFetchTemplate',
3710 [ $parser, $title, $rev, &$text, &$deps ] );
3711
3712 if ( $text === false || $text === null ) {
3713 $text = false;
3714 break;
3715 }
3716 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3717 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
3718 lcfirst( $title->getText() ) )->inContentLanguage();
3719 if ( !$message->exists() ) {
3720 $text = false;
3721 break;
3722 }
3723 $content = $message->content();
3724 $text = $message->plain();
3725 } else {
3726 break;
3727 }
3728 if ( !$content ) {
3729 break;
3730 }
3731 # Redirect?
3732 $finalTitle = $title;
3733 $title = $content->getRedirectTarget();
3734 }
3735 return [
3736 'text' => $text,
3737 'finalTitle' => $finalTitle,
3738 'deps' => $deps ];
3739 }
3740
3741 /**
3742 * Fetch a file and its title and register a reference to it.
3743 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3744 * @param Title $title
3745 * @param array $options Array of options to RepoGroup::findFile
3746 * @return File|bool
3747 * @deprecated since 1.32, use fetchFileAndTitle instead
3748 */
3749 public function fetchFile( $title, $options = [] ) {
3750 wfDeprecated( __METHOD__, '1.32' );
3751 return $this->fetchFileAndTitle( $title, $options )[0];
3752 }
3753
3754 /**
3755 * Fetch a file and its title and register a reference to it.
3756 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3757 * @param Title $title
3758 * @param array $options Array of options to RepoGroup::findFile
3759 * @return array ( File or false, Title of file )
3760 */
3761 public function fetchFileAndTitle( $title, $options = [] ) {
3762 $file = $this->fetchFileNoRegister( $title, $options );
3763
3764 $time = $file ? $file->getTimestamp() : false;
3765 $sha1 = $file ? $file->getSha1() : false;
3766 # Register the file as a dependency...
3767 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3768 if ( $file && !$title->equals( $file->getTitle() ) ) {
3769 # Update fetched file title
3770 $title = $file->getTitle();
3771 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3772 }
3773 return [ $file, $title ];
3774 }
3775
3776 /**
3777 * Helper function for fetchFileAndTitle.
3778 *
3779 * Also useful if you need to fetch a file but not use it yet,
3780 * for example to get the file's handler.
3781 *
3782 * @param Title $title
3783 * @param array $options Array of options to RepoGroup::findFile
3784 * @return File|bool
3785 */
3786 protected function fetchFileNoRegister( $title, $options = [] ) {
3787 if ( isset( $options['broken'] ) ) {
3788 $file = false; // broken thumbnail forced by hook
3789 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3790 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3791 } else { // get by (name,timestamp)
3792 $file = wfFindFile( $title, $options );
3793 }
3794 return $file;
3795 }
3796
3797 /**
3798 * Transclude an interwiki link.
3799 *
3800 * @param Title $title
3801 * @param string $action Usually one of (raw, render)
3802 *
3803 * @return string
3804 */
3805 public function interwikiTransclude( $title, $action ) {
3806 if ( !$this->siteConfig->get( 'EnableScaryTranscluding' ) ) {
3807 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3808 }
3809
3810 $url = $title->getFullURL( [ 'action' => $action ] );
3811 if ( strlen( $url ) > 1024 ) {
3812 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3813 }
3814
3815 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3816
3817 $fname = __METHOD__;
3818 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3819
3820 $data = $cache->getWithSetCallback(
3821 $cache->makeGlobalKey(
3822 'interwiki-transclude',
3823 ( $wikiId !== false ) ? $wikiId : 'external',
3824 sha1( $url )
3825 ),
3826 $this->siteConfig->get( 'TranscludeCacheExpiry' ),
3827 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3828 $req = MWHttpRequest::factory( $url, [], $fname );
3829
3830 $status = $req->execute(); // Status object
3831 if ( !$status->isOK() ) {
3832 $ttl = $cache::TTL_UNCACHEABLE;
3833 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3834 $ttl = min( $cache::TTL_LAGGED, $ttl );
3835 }
3836
3837 return [
3838 'text' => $status->isOK() ? $req->getContent() : null,
3839 'code' => $req->getStatus()
3840 ];
3841 },
3842 [
3843 'checkKeys' => ( $wikiId !== false )
3844 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3845 : [],
3846 'pcGroup' => 'interwiki-transclude:5',
3847 'pcTTL' => $cache::TTL_PROC_LONG
3848 ]
3849 );
3850
3851 if ( is_string( $data['text'] ) ) {
3852 $text = $data['text'];
3853 } elseif ( $data['code'] != 200 ) {
3854 // Though we failed to fetch the content, this status is useless.
3855 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3856 ->params( $url, $data['code'] )->inContentLanguage()->text();
3857 } else {
3858 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3859 }
3860
3861 return $text;
3862 }
3863
3864 /**
3865 * Triple brace replacement -- used for template arguments
3866 * @private
3867 *
3868 * @param array $piece
3869 * @param PPFrame $frame
3870 *
3871 * @return array
3872 */
3873 public function argSubstitution( $piece, $frame ) {
3874 $error = false;
3875 $parts = $piece['parts'];
3876 $nameWithSpaces = $frame->expand( $piece['title'] );
3877 $argName = trim( $nameWithSpaces );
3878 $object = false;
3879 $text = $frame->getArgument( $argName );
3880 if ( $text === false && $parts->getLength() > 0
3881 && ( $this->ot['html']
3882 || $this->ot['pre']
3883 || ( $this->ot['wiki'] && $frame->isTemplate() )
3884 )
3885 ) {
3886 # No match in frame, use the supplied default
3887 $object = $parts->item( 0 )->getChildren();
3888 }
3889 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3890 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3891 $this->limitationWarn( 'post-expand-template-argument' );
3892 }
3893
3894 if ( $text === false && $object === false ) {
3895 # No match anywhere
3896 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3897 }
3898 if ( $error !== false ) {
3899 $text .= $error;
3900 }
3901 if ( $object !== false ) {
3902 $ret = [ 'object' => $object ];
3903 } else {
3904 $ret = [ 'text' => $text ];
3905 }
3906
3907 return $ret;
3908 }
3909
3910 /**
3911 * Return the text to be used for a given extension tag.
3912 * This is the ghost of strip().
3913 *
3914 * @param array $params Associative array of parameters:
3915 * name PPNode for the tag name
3916 * attr PPNode for unparsed text where tag attributes are thought to be
3917 * attributes Optional associative array of parsed attributes
3918 * inner Contents of extension element
3919 * noClose Original text did not have a close tag
3920 * @param PPFrame $frame
3921 *
3922 * @throws MWException
3923 * @return string
3924 */
3925 public function extensionSubstitution( $params, $frame ) {
3926 static $errorStr = '<span class="error">';
3927 static $errorLen = 20;
3928
3929 $name = $frame->expand( $params['name'] );
3930 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3931 // Probably expansion depth or node count exceeded. Just punt the
3932 // error up.
3933 return $name;
3934 }
3935
3936 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3937 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3938 // See above
3939 return $attrText;
3940 }
3941
3942 // We can't safely check if the expansion for $content resulted in an
3943 // error, because the content could happen to be the error string
3944 // (T149622).
3945 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3946
3947 $marker = self::MARKER_PREFIX . "-$name-"
3948 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3949
3950 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3951 ( $this->ot['html'] || $this->ot['pre'] );
3952 if ( $isFunctionTag ) {
3953 $markerType = 'none';
3954 } else {
3955 $markerType = 'general';
3956 }
3957 if ( $this->ot['html'] || $isFunctionTag ) {
3958 $name = strtolower( $name );
3959 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3960 if ( isset( $params['attributes'] ) ) {
3961 $attributes = $attributes + $params['attributes'];
3962 }
3963
3964 if ( isset( $this->mTagHooks[$name] ) ) {
3965 $output = call_user_func_array( $this->mTagHooks[$name],
3966 [ $content, $attributes, $this, $frame ] );
3967 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3968 list( $callback, ) = $this->mFunctionTagHooks[$name];
3969
3970 // Avoid PHP 7.1 warning from passing $this by reference
3971 $parser = $this;
3972 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3973 } else {
3974 $output = '<span class="error">Invalid tag extension name: ' .
3975 htmlspecialchars( $name ) . '</span>';
3976 }
3977
3978 if ( is_array( $output ) ) {
3979 // Extract flags
3980 $flags = $output;
3981 $output = $flags[0];
3982 if ( isset( $flags['markerType'] ) ) {
3983 $markerType = $flags['markerType'];
3984 }
3985 }
3986 } else {
3987 if ( is_null( $attrText ) ) {
3988 $attrText = '';
3989 }
3990 if ( isset( $params['attributes'] ) ) {
3991 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3992 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3993 htmlspecialchars( $attrValue ) . '"';
3994 }
3995 }
3996 if ( $content === null ) {
3997 $output = "<$name$attrText/>";
3998 } else {
3999 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4000 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4001 // See above
4002 return $close;
4003 }
4004 $output = "<$name$attrText>$content$close";
4005 }
4006 }
4007
4008 if ( $markerType === 'none' ) {
4009 return $output;
4010 } elseif ( $markerType === 'nowiki' ) {
4011 $this->mStripState->addNoWiki( $marker, $output );
4012 } elseif ( $markerType === 'general' ) {
4013 $this->mStripState->addGeneral( $marker, $output );
4014 } else {
4015 throw new MWException( __METHOD__ . ': invalid marker type' );
4016 }
4017 return $marker;
4018 }
4019
4020 /**
4021 * Increment an include size counter
4022 *
4023 * @param string $type The type of expansion
4024 * @param int $size The size of the text
4025 * @return bool False if this inclusion would take it over the maximum, true otherwise
4026 */
4027 public function incrementIncludeSize( $type, $size ) {
4028 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4029 return false;
4030 } else {
4031 $this->mIncludeSizes[$type] += $size;
4032 return true;
4033 }
4034 }
4035
4036 /**
4037 * Increment the expensive function count
4038 *
4039 * @return bool False if the limit has been exceeded
4040 */
4041 public function incrementExpensiveFunctionCount() {
4042 $this->mExpensiveFunctionCount++;
4043 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4044 }
4045
4046 /**
4047 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4048 * Fills $this->mDoubleUnderscores, returns the modified text
4049 *
4050 * @param string $text
4051 *
4052 * @return string
4053 */
4054 public function doDoubleUnderscore( $text ) {
4055 # The position of __TOC__ needs to be recorded
4056 $mw = $this->magicWordFactory->get( 'toc' );
4057 if ( $mw->match( $text ) ) {
4058 $this->mShowToc = true;
4059 $this->mForceTocPosition = true;
4060
4061 # Set a placeholder. At the end we'll fill it in with the TOC.
4062 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4063
4064 # Only keep the first one.
4065 $text = $mw->replace( '', $text );
4066 }
4067
4068 # Now match and remove the rest of them
4069 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4070 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4071
4072 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4073 $this->mOutput->mNoGallery = true;
4074 }
4075 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4076 $this->mShowToc = false;
4077 }
4078 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4079 && $this->mTitle->getNamespace() == NS_CATEGORY
4080 ) {
4081 $this->addTrackingCategory( 'hidden-category-category' );
4082 }
4083 # (T10068) Allow control over whether robots index a page.
4084 # __INDEX__ always overrides __NOINDEX__, see T16899
4085 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4086 $this->mOutput->setIndexPolicy( 'noindex' );
4087 $this->addTrackingCategory( 'noindex-category' );
4088 }
4089 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4090 $this->mOutput->setIndexPolicy( 'index' );
4091 $this->addTrackingCategory( 'index-category' );
4092 }
4093
4094 # Cache all double underscores in the database
4095 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4096 $this->mOutput->setProperty( $key, '' );
4097 }
4098
4099 return $text;
4100 }
4101
4102 /**
4103 * @see ParserOutput::addTrackingCategory()
4104 * @param string $msg Message key
4105 * @return bool Whether the addition was successful
4106 */
4107 public function addTrackingCategory( $msg ) {
4108 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4109 }
4110
4111 /**
4112 * This function accomplishes several tasks:
4113 * 1) Auto-number headings if that option is enabled
4114 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4115 * 3) Add a Table of contents on the top for users who have enabled the option
4116 * 4) Auto-anchor headings
4117 *
4118 * It loops through all headlines, collects the necessary data, then splits up the
4119 * string and re-inserts the newly formatted headlines.
4120 *
4121 * @param string $text
4122 * @param string $origText Original, untouched wikitext
4123 * @param bool $isMain
4124 * @return mixed|string
4125 * @private
4126 */
4127 public function formatHeadings( $text, $origText, $isMain = true ) {
4128 # Inhibit editsection links if requested in the page
4129 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4130 $maybeShowEditLink = false;
4131 } else {
4132 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4133 }
4134
4135 # Get all headlines for numbering them and adding funky stuff like [edit]
4136 # links - this is for later, but we need the number of headlines right now
4137 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4138 # be trimmed here since whitespace in HTML headings is significant.
4139 $matches = [];
4140 $numMatches = preg_match_all(
4141 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4142 $text,
4143 $matches
4144 );
4145
4146 # if there are fewer than 4 headlines in the article, do not show TOC
4147 # unless it's been explicitly enabled.
4148 $enoughToc = $this->mShowToc &&
4149 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4150
4151 # Allow user to stipulate that a page should have a "new section"
4152 # link added via __NEWSECTIONLINK__
4153 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4154 $this->mOutput->setNewSection( true );
4155 }
4156
4157 # Allow user to remove the "new section"
4158 # link via __NONEWSECTIONLINK__
4159 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4160 $this->mOutput->hideNewSection( true );
4161 }
4162
4163 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4164 # override above conditions and always show TOC above first header
4165 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4166 $this->mShowToc = true;
4167 $enoughToc = true;
4168 }
4169
4170 # headline counter
4171 $headlineCount = 0;
4172 $numVisible = 0;
4173
4174 # Ugh .. the TOC should have neat indentation levels which can be
4175 # passed to the skin functions. These are determined here
4176 $toc = '';
4177 $full = '';
4178 $head = [];
4179 $sublevelCount = [];
4180 $levelCount = [];
4181 $level = 0;
4182 $prevlevel = 0;
4183 $toclevel = 0;
4184 $prevtoclevel = 0;
4185 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4186 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4187 $oldType = $this->mOutputType;
4188 $this->setOutputType( self::OT_WIKI );
4189 $frame = $this->getPreprocessor()->newFrame();
4190 $root = $this->preprocessToDom( $origText );
4191 $node = $root->getFirstChild();
4192 $byteOffset = 0;
4193 $tocraw = [];
4194 $refers = [];
4195
4196 $headlines = $numMatches !== false ? $matches[3] : [];
4197
4198 $maxTocLevel = $this->siteConfig->get( 'MaxTocLevel' );
4199 foreach ( $headlines as $headline ) {
4200 $isTemplate = false;
4201 $titleText = false;
4202 $sectionIndex = false;
4203 $numbering = '';
4204 $markerMatches = [];
4205 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4206 $serial = $markerMatches[1];
4207 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4208 $isTemplate = ( $titleText != $baseTitleText );
4209 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4210 }
4211
4212 if ( $toclevel ) {
4213 $prevlevel = $level;
4214 }
4215 $level = $matches[1][$headlineCount];
4216
4217 if ( $level > $prevlevel ) {
4218 # Increase TOC level
4219 $toclevel++;
4220 $sublevelCount[$toclevel] = 0;
4221 if ( $toclevel < $maxTocLevel ) {
4222 $prevtoclevel = $toclevel;
4223 $toc .= Linker::tocIndent();
4224 $numVisible++;
4225 }
4226 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4227 # Decrease TOC level, find level to jump to
4228
4229 for ( $i = $toclevel; $i > 0; $i-- ) {
4230 if ( $levelCount[$i] == $level ) {
4231 # Found last matching level
4232 $toclevel = $i;
4233 break;
4234 } elseif ( $levelCount[$i] < $level ) {
4235 # Found first matching level below current level
4236 $toclevel = $i + 1;
4237 break;
4238 }
4239 }
4240 if ( $i == 0 ) {
4241 $toclevel = 1;
4242 }
4243 if ( $toclevel < $maxTocLevel ) {
4244 if ( $prevtoclevel < $maxTocLevel ) {
4245 # Unindent only if the previous toc level was shown :p
4246 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4247 $prevtoclevel = $toclevel;
4248 } else {
4249 $toc .= Linker::tocLineEnd();
4250 }
4251 }
4252 } else {
4253 # No change in level, end TOC line
4254 if ( $toclevel < $maxTocLevel ) {
4255 $toc .= Linker::tocLineEnd();
4256 }
4257 }
4258
4259 $levelCount[$toclevel] = $level;
4260
4261 # count number of headlines for each level
4262 $sublevelCount[$toclevel]++;
4263 $dot = 0;
4264 for ( $i = 1; $i <= $toclevel; $i++ ) {
4265 if ( !empty( $sublevelCount[$i] ) ) {
4266 if ( $dot ) {
4267 $numbering .= '.';
4268 }
4269 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4270 $dot = 1;
4271 }
4272 }
4273
4274 # The safe header is a version of the header text safe to use for links
4275
4276 # Remove link placeholders by the link text.
4277 # <!--LINK number-->
4278 # turns into
4279 # link text with suffix
4280 # Do this before unstrip since link text can contain strip markers
4281 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4282
4283 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4284 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4285
4286 # Remove any <style> or <script> tags (T198618)
4287 $safeHeadline = preg_replace(
4288 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4289 '',
4290 $safeHeadline
4291 );
4292
4293 # Strip out HTML (first regex removes any tag not allowed)
4294 # Allowed tags are:
4295 # * <sup> and <sub> (T10393)
4296 # * <i> (T28375)
4297 # * <b> (r105284)
4298 # * <bdi> (T74884)
4299 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4300 # * <s> and <strike> (T35715)
4301 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4302 # to allow setting directionality in toc items.
4303 $tocline = preg_replace(
4304 [
4305 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4306 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4307 ],
4308 [ '', '<$1>' ],
4309 $safeHeadline
4310 );
4311
4312 # Strip '<span></span>', which is the result from the above if
4313 # <span id="foo"></span> is used to produce an additional anchor
4314 # for a section.
4315 $tocline = str_replace( '<span></span>', '', $tocline );
4316
4317 $tocline = trim( $tocline );
4318
4319 # For the anchor, strip out HTML-y stuff period
4320 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4321 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4322
4323 # Save headline for section edit hint before it's escaped
4324 $headlineHint = $safeHeadline;
4325
4326 # Decode HTML entities
4327 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4328
4329 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4330
4331 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4332 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4333 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4334 if ( $fallbackHeadline === $safeHeadline ) {
4335 # No reason to have both (in fact, we can't)
4336 $fallbackHeadline = false;
4337 }
4338
4339 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4340 # @todo FIXME: We may be changing them depending on the current locale.
4341 $arrayKey = strtolower( $safeHeadline );
4342 if ( $fallbackHeadline === false ) {
4343 $fallbackArrayKey = false;
4344 } else {
4345 $fallbackArrayKey = strtolower( $fallbackHeadline );
4346 }
4347
4348 # Create the anchor for linking from the TOC to the section
4349 $anchor = $safeHeadline;
4350 $fallbackAnchor = $fallbackHeadline;
4351 if ( isset( $refers[$arrayKey] ) ) {
4352 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4353 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4354 $anchor .= "_$i";
4355 $linkAnchor .= "_$i";
4356 $refers["${arrayKey}_$i"] = true;
4357 } else {
4358 $refers[$arrayKey] = true;
4359 }
4360 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4361 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4362 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4363 $fallbackAnchor .= "_$i";
4364 $refers["${fallbackArrayKey}_$i"] = true;
4365 } else {
4366 $refers[$fallbackArrayKey] = true;
4367 }
4368
4369 # Don't number the heading if it is the only one (looks silly)
4370 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4371 # the two are different if the line contains a link
4372 $headline = Html::element(
4373 'span',
4374 [ 'class' => 'mw-headline-number' ],
4375 $numbering
4376 ) . ' ' . $headline;
4377 }
4378
4379 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4380 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4381 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4382 }
4383
4384 # Add the section to the section tree
4385 # Find the DOM node for this header
4386 $noOffset = ( $isTemplate || $sectionIndex === false );
4387 while ( $node && !$noOffset ) {
4388 if ( $node->getName() === 'h' ) {
4389 $bits = $node->splitHeading();
4390 if ( $bits['i'] == $sectionIndex ) {
4391 break;
4392 }
4393 }
4394 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4395 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4396 $node = $node->getNextSibling();
4397 }
4398 $tocraw[] = [
4399 'toclevel' => $toclevel,
4400 'level' => $level,
4401 'line' => $tocline,
4402 'number' => $numbering,
4403 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4404 'fromtitle' => $titleText,
4405 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4406 'anchor' => $anchor,
4407 ];
4408
4409 # give headline the correct <h#> tag
4410 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4411 // Output edit section links as markers with styles that can be customized by skins
4412 if ( $isTemplate ) {
4413 # Put a T flag in the section identifier, to indicate to extractSections()
4414 # that sections inside <includeonly> should be counted.
4415 $editsectionPage = $titleText;
4416 $editsectionSection = "T-$sectionIndex";
4417 $editsectionContent = null;
4418 } else {
4419 $editsectionPage = $this->mTitle->getPrefixedText();
4420 $editsectionSection = $sectionIndex;
4421 $editsectionContent = $headlineHint;
4422 }
4423 // We use a bit of pesudo-xml for editsection markers. The
4424 // language converter is run later on. Using a UNIQ style marker
4425 // leads to the converter screwing up the tokens when it
4426 // converts stuff. And trying to insert strip tags fails too. At
4427 // this point all real inputted tags have already been escaped,
4428 // so we don't have to worry about a user trying to input one of
4429 // these markers directly. We use a page and section attribute
4430 // to stop the language converter from converting these
4431 // important bits of data, but put the headline hint inside a
4432 // content block because the language converter is supposed to
4433 // be able to convert that piece of data.
4434 // Gets replaced with html in ParserOutput::getText
4435 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4436 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4437 if ( $editsectionContent !== null ) {
4438 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4439 } else {
4440 $editlink .= '/>';
4441 }
4442 } else {
4443 $editlink = '';
4444 }
4445 $head[$headlineCount] = Linker::makeHeadline( $level,
4446 $matches['attrib'][$headlineCount], $anchor, $headline,
4447 $editlink, $fallbackAnchor );
4448
4449 $headlineCount++;
4450 }
4451
4452 $this->setOutputType( $oldType );
4453
4454 # Never ever show TOC if no headers
4455 if ( $numVisible < 1 ) {
4456 $enoughToc = false;
4457 }
4458
4459 if ( $enoughToc ) {
4460 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4461 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4462 }
4463 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4464 $this->mOutput->setTOCHTML( $toc );
4465 $toc = self::TOC_START . $toc . self::TOC_END;
4466 }
4467
4468 if ( $isMain ) {
4469 $this->mOutput->setSections( $tocraw );
4470 }
4471
4472 # split up and insert constructed headlines
4473 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4474 $i = 0;
4475
4476 // build an array of document sections
4477 $sections = [];
4478 foreach ( $blocks as $block ) {
4479 // $head is zero-based, sections aren't.
4480 if ( empty( $head[$i - 1] ) ) {
4481 $sections[$i] = $block;
4482 } else {
4483 $sections[$i] = $head[$i - 1] . $block;
4484 }
4485
4486 /**
4487 * Send a hook, one per section.
4488 * The idea here is to be able to make section-level DIVs, but to do so in a
4489 * lower-impact, more correct way than r50769
4490 *
4491 * $this : caller
4492 * $section : the section number
4493 * &$sectionContent : ref to the content of the section
4494 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4495 */
4496 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4497
4498 $i++;
4499 }
4500
4501 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4502 // append the TOC at the beginning
4503 // Top anchor now in skin
4504 $sections[0] = $sections[0] . $toc . "\n";
4505 }
4506
4507 $full .= implode( '', $sections );
4508
4509 if ( $this->mForceTocPosition ) {
4510 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4511 } else {
4512 return $full;
4513 }
4514 }
4515
4516 /**
4517 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4518 * conversion, substituting signatures, {{subst:}} templates, etc.
4519 *
4520 * @param string $text The text to transform
4521 * @param Title $title The Title object for the current article
4522 * @param User $user The User object describing the current user
4523 * @param ParserOptions $options Parsing options
4524 * @param bool $clearState Whether to clear the parser state first
4525 * @return string The altered wiki markup
4526 */
4527 public function preSaveTransform( $text, Title $title, User $user,
4528 ParserOptions $options, $clearState = true
4529 ) {
4530 if ( $clearState ) {
4531 $magicScopeVariable = $this->lock();
4532 }
4533 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4534 $this->setUser( $user );
4535
4536 // Strip U+0000 NULL (T159174)
4537 $text = str_replace( "\000", '', $text );
4538
4539 // We still normalize line endings for backwards-compatibility
4540 // with other code that just calls PST, but this should already
4541 // be handled in TextContent subclasses
4542 $text = TextContent::normalizeLineEndings( $text );
4543
4544 if ( $options->getPreSaveTransform() ) {
4545 $text = $this->pstPass2( $text, $user );
4546 }
4547 $text = $this->mStripState->unstripBoth( $text );
4548
4549 $this->setUser( null ); # Reset
4550
4551 return $text;
4552 }
4553
4554 /**
4555 * Pre-save transform helper function
4556 *
4557 * @param string $text
4558 * @param User $user
4559 *
4560 * @return string
4561 */
4562 private function pstPass2( $text, $user ) {
4563 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4564 # $this->contLang here in order to give everyone the same signature and use the default one
4565 # rather than the one selected in each user's preferences. (see also T14815)
4566 $ts = $this->mOptions->getTimestamp();
4567 $timestamp = MWTimestamp::getLocalInstance( $ts );
4568 $ts = $timestamp->format( 'YmdHis' );
4569 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4570
4571 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4572
4573 # Variable replacement
4574 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4575 $text = $this->replaceVariables( $text );
4576
4577 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4578 # which may corrupt this parser instance via its wfMessage()->text() call-
4579
4580 # Signatures
4581 if ( strpos( $text, '~~~' ) !== false ) {
4582 $sigText = $this->getUserSig( $user );
4583 $text = strtr( $text, [
4584 '~~~~~' => $d,
4585 '~~~~' => "$sigText $d",
4586 '~~~' => $sigText
4587 ] );
4588 # The main two signature forms used above are time-sensitive
4589 $this->mOutput->setFlag( 'user-signature' );
4590 }
4591
4592 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4593 $tc = '[' . Title::legalChars() . ']';
4594 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4595
4596 // [[ns:page (context)|]]
4597 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4598 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4599 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4600 // [[ns:page (context), context|]] (using either single or double-width comma)
4601 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4602 // [[|page]] (reverse pipe trick: add context from page title)
4603 $p2 = "/\[\[\\|($tc+)]]/";
4604
4605 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4606 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4607 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4608 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4609
4610 $t = $this->mTitle->getText();
4611 $m = [];
4612 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4613 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4614 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4615 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4616 } else {
4617 # if there's no context, don't bother duplicating the title
4618 $text = preg_replace( $p2, '[[\\1]]', $text );
4619 }
4620
4621 return $text;
4622 }
4623
4624 /**
4625 * Fetch the user's signature text, if any, and normalize to
4626 * validated, ready-to-insert wikitext.
4627 * If you have pre-fetched the nickname or the fancySig option, you can
4628 * specify them here to save a database query.
4629 * Do not reuse this parser instance after calling getUserSig(),
4630 * as it may have changed if it's the $wgParser.
4631 *
4632 * @param User &$user
4633 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4634 * @param bool|null $fancySig whether the nicknname is the complete signature
4635 * or null to use default value
4636 * @return string
4637 */
4638 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4639 $username = $user->getName();
4640
4641 # If not given, retrieve from the user object.
4642 if ( $nickname === false ) {
4643 $nickname = $user->getOption( 'nickname' );
4644 }
4645
4646 if ( is_null( $fancySig ) ) {
4647 $fancySig = $user->getBoolOption( 'fancysig' );
4648 }
4649
4650 $nickname = $nickname == null ? $username : $nickname;
4651
4652 if ( mb_strlen( $nickname ) > $this->siteConfig->get( 'MaxSigChars' ) ) {
4653 $nickname = $username;
4654 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4655 } elseif ( $fancySig !== false ) {
4656 # Sig. might contain markup; validate this
4657 if ( $this->validateSig( $nickname ) !== false ) {
4658 # Validated; clean up (if needed) and return it
4659 return $this->cleanSig( $nickname, true );
4660 } else {
4661 # Failed to validate; fall back to the default
4662 $nickname = $username;
4663 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4664 }
4665 }
4666
4667 # Make sure nickname doesnt get a sig in a sig
4668 $nickname = self::cleanSigInSig( $nickname );
4669
4670 # If we're still here, make it a link to the user page
4671 $userText = wfEscapeWikiText( $username );
4672 $nickText = wfEscapeWikiText( $nickname );
4673 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4674
4675 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4676 ->title( $this->getTitle() )->text();
4677 }
4678
4679 /**
4680 * Check that the user's signature contains no bad XML
4681 *
4682 * @param string $text
4683 * @return string|bool An expanded string, or false if invalid.
4684 */
4685 public function validateSig( $text ) {
4686 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4687 }
4688
4689 /**
4690 * Clean up signature text
4691 *
4692 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4693 * 2) Substitute all transclusions
4694 *
4695 * @param string $text
4696 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4697 * @return string Signature text
4698 */
4699 public function cleanSig( $text, $parsing = false ) {
4700 if ( !$parsing ) {
4701 global $wgTitle;
4702 $magicScopeVariable = $this->lock();
4703 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4704 }
4705
4706 # Option to disable this feature
4707 if ( !$this->mOptions->getCleanSignatures() ) {
4708 return $text;
4709 }
4710
4711 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4712 # => Move this logic to braceSubstitution()
4713 $substWord = $this->magicWordFactory->get( 'subst' );
4714 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4715 $substText = '{{' . $substWord->getSynonym( 0 );
4716
4717 $text = preg_replace( $substRegex, $substText, $text );
4718 $text = self::cleanSigInSig( $text );
4719 $dom = $this->preprocessToDom( $text );
4720 $frame = $this->getPreprocessor()->newFrame();
4721 $text = $frame->expand( $dom );
4722
4723 if ( !$parsing ) {
4724 $text = $this->mStripState->unstripBoth( $text );
4725 }
4726
4727 return $text;
4728 }
4729
4730 /**
4731 * Strip 3, 4 or 5 tildes out of signatures.
4732 *
4733 * @param string $text
4734 * @return string Signature text with /~{3,5}/ removed
4735 */
4736 public static function cleanSigInSig( $text ) {
4737 $text = preg_replace( '/~{3,5}/', '', $text );
4738 return $text;
4739 }
4740
4741 /**
4742 * Set up some variables which are usually set up in parse()
4743 * so that an external function can call some class members with confidence
4744 *
4745 * @param Title|null $title
4746 * @param ParserOptions $options
4747 * @param int $outputType
4748 * @param bool $clearState
4749 */
4750 public function startExternalParse( Title $title = null, ParserOptions $options,
4751 $outputType, $clearState = true
4752 ) {
4753 $this->startParse( $title, $options, $outputType, $clearState );
4754 }
4755
4756 /**
4757 * @param Title|null $title
4758 * @param ParserOptions $options
4759 * @param int $outputType
4760 * @param bool $clearState
4761 */
4762 private function startParse( Title $title = null, ParserOptions $options,
4763 $outputType, $clearState = true
4764 ) {
4765 $this->setTitle( $title );
4766 $this->mOptions = $options;
4767 $this->setOutputType( $outputType );
4768 if ( $clearState ) {
4769 $this->clearState();
4770 }
4771 }
4772
4773 /**
4774 * Wrapper for preprocess()
4775 *
4776 * @param string $text The text to preprocess
4777 * @param ParserOptions $options
4778 * @param Title|null $title Title object or null to use $wgTitle
4779 * @return string
4780 */
4781 public function transformMsg( $text, $options, $title = null ) {
4782 static $executing = false;
4783
4784 # Guard against infinite recursion
4785 if ( $executing ) {
4786 return $text;
4787 }
4788 $executing = true;
4789
4790 if ( !$title ) {
4791 global $wgTitle;
4792 $title = $wgTitle;
4793 }
4794
4795 $text = $this->preprocess( $text, $title, $options );
4796
4797 $executing = false;
4798 return $text;
4799 }
4800
4801 /**
4802 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4803 * The callback should have the following form:
4804 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4805 *
4806 * Transform and return $text. Use $parser for any required context, e.g. use
4807 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4808 *
4809 * Hooks may return extended information by returning an array, of which the
4810 * first numbered element (index 0) must be the return string, and all other
4811 * entries are extracted into local variables within an internal function
4812 * in the Parser class.
4813 *
4814 * This interface (introduced r61913) appears to be undocumented, but
4815 * 'markerType' is used by some core tag hooks to override which strip
4816 * array their results are placed in. **Use great caution if attempting
4817 * this interface, as it is not documented and injudicious use could smash
4818 * private variables.**
4819 *
4820 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4821 * @param callable $callback The callback function (and object) to use for the tag
4822 * @throws MWException
4823 * @return callable|null The old value of the mTagHooks array associated with the hook
4824 */
4825 public function setHook( $tag, callable $callback ) {
4826 $tag = strtolower( $tag );
4827 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4828 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4829 }
4830 $oldVal = $this->mTagHooks[$tag] ?? null;
4831 $this->mTagHooks[$tag] = $callback;
4832 if ( !in_array( $tag, $this->mStripList ) ) {
4833 $this->mStripList[] = $tag;
4834 }
4835
4836 return $oldVal;
4837 }
4838
4839 /**
4840 * As setHook(), but letting the contents be parsed.
4841 *
4842 * Transparent tag hooks are like regular XML-style tag hooks, except they
4843 * operate late in the transformation sequence, on HTML instead of wikitext.
4844 *
4845 * This is probably obsoleted by things dealing with parser frames?
4846 * The only extension currently using it is geoserver.
4847 *
4848 * @since 1.10
4849 * @todo better document or deprecate this
4850 *
4851 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4852 * @param callable $callback The callback function (and object) to use for the tag
4853 * @throws MWException
4854 * @return callable|null The old value of the mTagHooks array associated with the hook
4855 */
4856 public function setTransparentTagHook( $tag, callable $callback ) {
4857 $tag = strtolower( $tag );
4858 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4859 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4860 }
4861 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4862 $this->mTransparentTagHooks[$tag] = $callback;
4863
4864 return $oldVal;
4865 }
4866
4867 /**
4868 * Remove all tag hooks
4869 */
4870 public function clearTagHooks() {
4871 $this->mTagHooks = [];
4872 $this->mFunctionTagHooks = [];
4873 $this->mStripList = $this->mDefaultStripList;
4874 }
4875
4876 /**
4877 * Create a function, e.g. {{sum:1|2|3}}
4878 * The callback function should have the form:
4879 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4880 *
4881 * Or with Parser::SFH_OBJECT_ARGS:
4882 * function myParserFunction( $parser, $frame, $args ) { ... }
4883 *
4884 * The callback may either return the text result of the function, or an array with the text
4885 * in element 0, and a number of flags in the other elements. The names of the flags are
4886 * specified in the keys. Valid flags are:
4887 * found The text returned is valid, stop processing the template. This
4888 * is on by default.
4889 * nowiki Wiki markup in the return value should be escaped
4890 * isHTML The returned text is HTML, armour it against wikitext transformation
4891 *
4892 * @param string $id The magic word ID
4893 * @param callable $callback The callback function (and object) to use
4894 * @param int $flags A combination of the following flags:
4895 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4896 *
4897 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4898 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4899 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4900 * the arguments, and to control the way they are expanded.
4901 *
4902 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4903 * arguments, for instance:
4904 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4905 *
4906 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4907 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4908 * working if/when this is changed.
4909 *
4910 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4911 * expansion.
4912 *
4913 * Please read the documentation in includes/parser/Preprocessor.php for more information
4914 * about the methods available in PPFrame and PPNode.
4915 *
4916 * @throws MWException
4917 * @return string|callable The old callback function for this name, if any
4918 */
4919 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4920 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4921 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4922
4923 # Add to function cache
4924 $mw = $this->magicWordFactory->get( $id );
4925 if ( !$mw ) {
4926 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4927 }
4928
4929 $synonyms = $mw->getSynonyms();
4930 $sensitive = intval( $mw->isCaseSensitive() );
4931
4932 foreach ( $synonyms as $syn ) {
4933 # Case
4934 if ( !$sensitive ) {
4935 $syn = $this->contLang->lc( $syn );
4936 }
4937 # Add leading hash
4938 if ( !( $flags & self::SFH_NO_HASH ) ) {
4939 $syn = '#' . $syn;
4940 }
4941 # Remove trailing colon
4942 if ( substr( $syn, -1, 1 ) === ':' ) {
4943 $syn = substr( $syn, 0, -1 );
4944 }
4945 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4946 }
4947 return $oldVal;
4948 }
4949
4950 /**
4951 * Get all registered function hook identifiers
4952 *
4953 * @return array
4954 */
4955 public function getFunctionHooks() {
4956 $this->firstCallInit();
4957 return array_keys( $this->mFunctionHooks );
4958 }
4959
4960 /**
4961 * Create a tag function, e.g. "<test>some stuff</test>".
4962 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4963 * Unlike parser functions, their content is not preprocessed.
4964 * @param string $tag
4965 * @param callable $callback
4966 * @param int $flags
4967 * @throws MWException
4968 * @return null
4969 */
4970 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4971 $tag = strtolower( $tag );
4972 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4973 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4974 }
4975 $old = $this->mFunctionTagHooks[$tag] ?? null;
4976 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4977
4978 if ( !in_array( $tag, $this->mStripList ) ) {
4979 $this->mStripList[] = $tag;
4980 }
4981
4982 return $old;
4983 }
4984
4985 /**
4986 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4987 * Placeholders created in Linker::link()
4988 *
4989 * @param string &$text
4990 * @param int $options
4991 */
4992 public function replaceLinkHolders( &$text, $options = 0 ) {
4993 $this->mLinkHolders->replace( $text );
4994 }
4995
4996 /**
4997 * Replace "<!--LINK-->" link placeholders with plain text of links
4998 * (not HTML-formatted).
4999 *
5000 * @param string $text
5001 * @return string
5002 */
5003 public function replaceLinkHoldersText( $text ) {
5004 return $this->mLinkHolders->replaceText( $text );
5005 }
5006
5007 /**
5008 * Renders an image gallery from a text with one line per image.
5009 * text labels may be given by using |-style alternative text. E.g.
5010 * Image:one.jpg|The number "1"
5011 * Image:tree.jpg|A tree
5012 * given as text will return the HTML of a gallery with two images,
5013 * labeled 'The number "1"' and
5014 * 'A tree'.
5015 *
5016 * @param string $text
5017 * @param array $params
5018 * @return string HTML
5019 */
5020 public function renderImageGallery( $text, $params ) {
5021 $mode = false;
5022 if ( isset( $params['mode'] ) ) {
5023 $mode = $params['mode'];
5024 }
5025
5026 try {
5027 $ig = ImageGalleryBase::factory( $mode );
5028 } catch ( Exception $e ) {
5029 // If invalid type set, fallback to default.
5030 $ig = ImageGalleryBase::factory( false );
5031 }
5032
5033 $ig->setContextTitle( $this->mTitle );
5034 $ig->setShowBytes( false );
5035 $ig->setShowDimensions( false );
5036 $ig->setShowFilename( false );
5037 $ig->setParser( $this );
5038 $ig->setHideBadImages();
5039 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5040
5041 if ( isset( $params['showfilename'] ) ) {
5042 $ig->setShowFilename( true );
5043 } else {
5044 $ig->setShowFilename( false );
5045 }
5046 if ( isset( $params['caption'] ) ) {
5047 $caption = $params['caption'];
5048 $caption = htmlspecialchars( $caption );
5049 $caption = $this->replaceInternalLinks( $caption );
5050 $ig->setCaptionHtml( $caption );
5051 }
5052 if ( isset( $params['perrow'] ) ) {
5053 $ig->setPerRow( $params['perrow'] );
5054 }
5055 if ( isset( $params['widths'] ) ) {
5056 $ig->setWidths( $params['widths'] );
5057 }
5058 if ( isset( $params['heights'] ) ) {
5059 $ig->setHeights( $params['heights'] );
5060 }
5061 $ig->setAdditionalOptions( $params );
5062
5063 // Avoid PHP 7.1 warning from passing $this by reference
5064 $parser = $this;
5065 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5066
5067 $lines = StringUtils::explode( "\n", $text );
5068 foreach ( $lines as $line ) {
5069 # match lines like these:
5070 # Image:someimage.jpg|This is some image
5071 $matches = [];
5072 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5073 # Skip empty lines
5074 if ( count( $matches ) == 0 ) {
5075 continue;
5076 }
5077
5078 if ( strpos( $matches[0], '%' ) !== false ) {
5079 $matches[1] = rawurldecode( $matches[1] );
5080 }
5081 $title = Title::newFromText( $matches[1], NS_FILE );
5082 if ( is_null( $title ) ) {
5083 # Bogus title. Ignore these so we don't bomb out later.
5084 continue;
5085 }
5086
5087 # We need to get what handler the file uses, to figure out parameters.
5088 # Note, a hook can overide the file name, and chose an entirely different
5089 # file (which potentially could be of a different type and have different handler).
5090 $options = [];
5091 $descQuery = false;
5092 Hooks::run( 'BeforeParserFetchFileAndTitle',
5093 [ $this, $title, &$options, &$descQuery ] );
5094 # Don't register it now, as TraditionalImageGallery does that later.
5095 $file = $this->fetchFileNoRegister( $title, $options );
5096 $handler = $file ? $file->getHandler() : false;
5097
5098 $paramMap = [
5099 'img_alt' => 'gallery-internal-alt',
5100 'img_link' => 'gallery-internal-link',
5101 ];
5102 if ( $handler ) {
5103 $paramMap = $paramMap + $handler->getParamMap();
5104 // We don't want people to specify per-image widths.
5105 // Additionally the width parameter would need special casing anyhow.
5106 unset( $paramMap['img_width'] );
5107 }
5108
5109 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5110
5111 $label = '';
5112 $alt = '';
5113 $link = '';
5114 $handlerOptions = [];
5115 if ( isset( $matches[3] ) ) {
5116 // look for an |alt= definition while trying not to break existing
5117 // captions with multiple pipes (|) in it, until a more sensible grammar
5118 // is defined for images in galleries
5119
5120 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5121 // splitting on '|' is a bit odd, and different from makeImage.
5122 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5123 // Protect LanguageConverter markup
5124 $parameterMatches = StringUtils::delimiterExplode(
5125 '-{', '}-', '|', $matches[3], true /* nested */
5126 );
5127
5128 foreach ( $parameterMatches as $parameterMatch ) {
5129 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5130 if ( $magicName ) {
5131 $paramName = $paramMap[$magicName];
5132
5133 switch ( $paramName ) {
5134 case 'gallery-internal-alt':
5135 $alt = $this->stripAltText( $match, false );
5136 break;
5137 case 'gallery-internal-link':
5138 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5139 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5140 // Result of LanguageConverter::markNoConversion
5141 // invoked on an external link.
5142 $linkValue = substr( $linkValue, 4, -2 );
5143 }
5144 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5145 if ( $type === 'link-url' ) {
5146 $link = $target;
5147 $this->mOutput->addExternalLink( $target );
5148 } elseif ( $type === 'link-title' ) {
5149 $link = $target->getLinkURL();
5150 $this->mOutput->addLink( $target );
5151 }
5152 break;
5153 default:
5154 // Must be a handler specific parameter.
5155 if ( $handler->validateParam( $paramName, $match ) ) {
5156 $handlerOptions[$paramName] = $match;
5157 } else {
5158 // Guess not, consider it as caption.
5159 wfDebug( "$parameterMatch failed parameter validation\n" );
5160 $label = $parameterMatch;
5161 }
5162 }
5163
5164 } else {
5165 // Last pipe wins.
5166 $label = $parameterMatch;
5167 }
5168 }
5169 }
5170
5171 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5172 }
5173 $html = $ig->toHTML();
5174 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5175 return $html;
5176 }
5177
5178 /**
5179 * @param MediaHandler $handler
5180 * @return array
5181 */
5182 public function getImageParams( $handler ) {
5183 if ( $handler ) {
5184 $handlerClass = get_class( $handler );
5185 } else {
5186 $handlerClass = '';
5187 }
5188 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5189 # Initialise static lists
5190 static $internalParamNames = [
5191 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5192 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5193 'bottom', 'text-bottom' ],
5194 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5195 'upright', 'border', 'link', 'alt', 'class' ],
5196 ];
5197 static $internalParamMap;
5198 if ( !$internalParamMap ) {
5199 $internalParamMap = [];
5200 foreach ( $internalParamNames as $type => $names ) {
5201 foreach ( $names as $name ) {
5202 // For grep: img_left, img_right, img_center, img_none,
5203 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5204 // img_bottom, img_text_bottom,
5205 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5206 // img_border, img_link, img_alt, img_class
5207 $magicName = str_replace( '-', '_', "img_$name" );
5208 $internalParamMap[$magicName] = [ $type, $name ];
5209 }
5210 }
5211 }
5212
5213 # Add handler params
5214 $paramMap = $internalParamMap;
5215 if ( $handler ) {
5216 $handlerParamMap = $handler->getParamMap();
5217 foreach ( $handlerParamMap as $magic => $paramName ) {
5218 $paramMap[$magic] = [ 'handler', $paramName ];
5219 }
5220 }
5221 $this->mImageParams[$handlerClass] = $paramMap;
5222 $this->mImageParamsMagicArray[$handlerClass] =
5223 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5224 }
5225 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5226 }
5227
5228 /**
5229 * Parse image options text and use it to make an image
5230 *
5231 * @param Title $title
5232 * @param string $options
5233 * @param LinkHolderArray|bool $holders
5234 * @return string HTML
5235 */
5236 public function makeImage( $title, $options, $holders = false ) {
5237 # Check if the options text is of the form "options|alt text"
5238 # Options are:
5239 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5240 # * left no resizing, just left align. label is used for alt= only
5241 # * right same, but right aligned
5242 # * none same, but not aligned
5243 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5244 # * center center the image
5245 # * frame Keep original image size, no magnify-button.
5246 # * framed Same as "frame"
5247 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5248 # * upright reduce width for upright images, rounded to full __0 px
5249 # * border draw a 1px border around the image
5250 # * alt Text for HTML alt attribute (defaults to empty)
5251 # * class Set a class for img node
5252 # * link Set the target of the image link. Can be external, interwiki, or local
5253 # vertical-align values (no % or length right now):
5254 # * baseline
5255 # * sub
5256 # * super
5257 # * top
5258 # * text-top
5259 # * middle
5260 # * bottom
5261 # * text-bottom
5262
5263 # Protect LanguageConverter markup when splitting into parts
5264 $parts = StringUtils::delimiterExplode(
5265 '-{', '}-', '|', $options, true /* allow nesting */
5266 );
5267
5268 # Give extensions a chance to select the file revision for us
5269 $options = [];
5270 $descQuery = false;
5271 Hooks::run( 'BeforeParserFetchFileAndTitle',
5272 [ $this, $title, &$options, &$descQuery ] );
5273 # Fetch and register the file (file title may be different via hooks)
5274 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5275
5276 # Get parameter map
5277 $handler = $file ? $file->getHandler() : false;
5278
5279 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5280
5281 if ( !$file ) {
5282 $this->addTrackingCategory( 'broken-file-category' );
5283 }
5284
5285 # Process the input parameters
5286 $caption = '';
5287 $params = [ 'frame' => [], 'handler' => [],
5288 'horizAlign' => [], 'vertAlign' => [] ];
5289 $seenformat = false;
5290 foreach ( $parts as $part ) {
5291 $part = trim( $part );
5292 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5293 $validated = false;
5294 if ( isset( $paramMap[$magicName] ) ) {
5295 list( $type, $paramName ) = $paramMap[$magicName];
5296
5297 # Special case; width and height come in one variable together
5298 if ( $type === 'handler' && $paramName === 'width' ) {
5299 $parsedWidthParam = self::parseWidthParam( $value );
5300 if ( isset( $parsedWidthParam['width'] ) ) {
5301 $width = $parsedWidthParam['width'];
5302 if ( $handler->validateParam( 'width', $width ) ) {
5303 $params[$type]['width'] = $width;
5304 $validated = true;
5305 }
5306 }
5307 if ( isset( $parsedWidthParam['height'] ) ) {
5308 $height = $parsedWidthParam['height'];
5309 if ( $handler->validateParam( 'height', $height ) ) {
5310 $params[$type]['height'] = $height;
5311 $validated = true;
5312 }
5313 }
5314 # else no validation -- T15436
5315 } else {
5316 if ( $type === 'handler' ) {
5317 # Validate handler parameter
5318 $validated = $handler->validateParam( $paramName, $value );
5319 } else {
5320 # Validate internal parameters
5321 switch ( $paramName ) {
5322 case 'manualthumb':
5323 case 'alt':
5324 case 'class':
5325 # @todo FIXME: Possibly check validity here for
5326 # manualthumb? downstream behavior seems odd with
5327 # missing manual thumbs.
5328 $validated = true;
5329 $value = $this->stripAltText( $value, $holders );
5330 break;
5331 case 'link':
5332 list( $paramName, $value ) = $this->parseLinkParameter( $value );
5333 if ( $paramName ) {
5334 $validated = true;
5335 if ( $paramName === 'no-link' ) {
5336 $value = true;
5337 }
5338 if ( $paramName === 'link-url' ) {
5339 if ( $this->mOptions->getExternalLinkTarget() ) {
5340 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5341 }
5342 }
5343 }
5344 break;
5345 case 'frameless':
5346 case 'framed':
5347 case 'thumbnail':
5348 // use first appearing option, discard others.
5349 $validated = !$seenformat;
5350 $seenformat = true;
5351 break;
5352 default:
5353 # Most other things appear to be empty or numeric...
5354 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5355 }
5356 }
5357
5358 if ( $validated ) {
5359 $params[$type][$paramName] = $value;
5360 }
5361 }
5362 }
5363 if ( !$validated ) {
5364 $caption = $part;
5365 }
5366 }
5367
5368 # Process alignment parameters
5369 if ( $params['horizAlign'] ) {
5370 $params['frame']['align'] = key( $params['horizAlign'] );
5371 }
5372 if ( $params['vertAlign'] ) {
5373 $params['frame']['valign'] = key( $params['vertAlign'] );
5374 }
5375
5376 $params['frame']['caption'] = $caption;
5377
5378 # Will the image be presented in a frame, with the caption below?
5379 $imageIsFramed = isset( $params['frame']['frame'] )
5380 || isset( $params['frame']['framed'] )
5381 || isset( $params['frame']['thumbnail'] )
5382 || isset( $params['frame']['manualthumb'] );
5383
5384 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5385 # came to also set the caption, ordinary text after the image -- which
5386 # makes no sense, because that just repeats the text multiple times in
5387 # screen readers. It *also* came to set the title attribute.
5388 # Now that we have an alt attribute, we should not set the alt text to
5389 # equal the caption: that's worse than useless, it just repeats the
5390 # text. This is the framed/thumbnail case. If there's no caption, we
5391 # use the unnamed parameter for alt text as well, just for the time be-
5392 # ing, if the unnamed param is set and the alt param is not.
5393 # For the future, we need to figure out if we want to tweak this more,
5394 # e.g., introducing a title= parameter for the title; ignoring the un-
5395 # named parameter entirely for images without a caption; adding an ex-
5396 # plicit caption= parameter and preserving the old magic unnamed para-
5397 # meter for BC; ...
5398 if ( $imageIsFramed ) { # Framed image
5399 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5400 # No caption or alt text, add the filename as the alt text so
5401 # that screen readers at least get some description of the image
5402 $params['frame']['alt'] = $title->getText();
5403 }
5404 # Do not set $params['frame']['title'] because tooltips don't make sense
5405 # for framed images
5406 } else { # Inline image
5407 if ( !isset( $params['frame']['alt'] ) ) {
5408 # No alt text, use the "caption" for the alt text
5409 if ( $caption !== '' ) {
5410 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5411 } else {
5412 # No caption, fall back to using the filename for the
5413 # alt text
5414 $params['frame']['alt'] = $title->getText();
5415 }
5416 }
5417 # Use the "caption" for the tooltip text
5418 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5419 }
5420
5421 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5422
5423 # Linker does the rest
5424 $time = $options['time'] ?? false;
5425 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5426 $time, $descQuery, $this->mOptions->getThumbSize() );
5427
5428 # Give the handler a chance to modify the parser object
5429 if ( $handler ) {
5430 $handler->parserTransformHook( $this, $file );
5431 }
5432
5433 return $ret;
5434 }
5435
5436 /**
5437 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5438 *
5439 * Adds an entry to appropriate link tables.
5440 *
5441 * @since 1.32
5442 * @return array of `[ type, target ]`, where:
5443 * - `type` is one of:
5444 * - `null`: Given value is not a valid link target, use default
5445 * - `'no-link'`: Given value is empty, do not generate a link
5446 * - `'link-url'`: Given value is a valid external link
5447 * - `'link-title'`: Given value is a valid internal link
5448 * - `target` is:
5449 * - When `type` is `null` or `'no-link'`: `false`
5450 * - When `type` is `'link-url'`: URL string corresponding to given value
5451 * - When `type` is `'link-title'`: Title object corresponding to given value
5452 */
5453 public function parseLinkParameter( $value ) {
5454 $chars = self::EXT_LINK_URL_CLASS;
5455 $addr = self::EXT_LINK_ADDR;
5456 $prots = $this->mUrlProtocols;
5457 $type = null;
5458 $target = false;
5459 if ( $value === '' ) {
5460 $type = 'no-link';
5461 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5462 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5463 $this->mOutput->addExternalLink( $value );
5464 $type = 'link-url';
5465 $target = $value;
5466 }
5467 } else {
5468 $linkTitle = Title::newFromText( $value );
5469 if ( $linkTitle ) {
5470 $this->mOutput->addLink( $linkTitle );
5471 $type = 'link-title';
5472 $target = $linkTitle;
5473 }
5474 }
5475 return [ $type, $target ];
5476 }
5477
5478 /**
5479 * @param string $caption
5480 * @param LinkHolderArray|bool $holders
5481 * @return mixed|string
5482 */
5483 protected function stripAltText( $caption, $holders ) {
5484 # Strip bad stuff out of the title (tooltip). We can't just use
5485 # replaceLinkHoldersText() here, because if this function is called
5486 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5487 if ( $holders ) {
5488 $tooltip = $holders->replaceText( $caption );
5489 } else {
5490 $tooltip = $this->replaceLinkHoldersText( $caption );
5491 }
5492
5493 # make sure there are no placeholders in thumbnail attributes
5494 # that are later expanded to html- so expand them now and
5495 # remove the tags
5496 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5497 $tooltip = Sanitizer::stripAllTags( $tooltip );
5498
5499 return $tooltip;
5500 }
5501
5502 /**
5503 * Set a flag in the output object indicating that the content is dynamic and
5504 * shouldn't be cached.
5505 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5506 */
5507 public function disableCache() {
5508 wfDebug( "Parser output marked as uncacheable.\n" );
5509 if ( !$this->mOutput ) {
5510 throw new MWException( __METHOD__ .
5511 " can only be called when actually parsing something" );
5512 }
5513 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5514 }
5515
5516 /**
5517 * Callback from the Sanitizer for expanding items found in HTML attribute
5518 * values, so they can be safely tested and escaped.
5519 *
5520 * @param string &$text
5521 * @param bool|PPFrame $frame
5522 * @return string
5523 */
5524 public function attributeStripCallback( &$text, $frame = false ) {
5525 $text = $this->replaceVariables( $text, $frame );
5526 $text = $this->mStripState->unstripBoth( $text );
5527 return $text;
5528 }
5529
5530 /**
5531 * Accessor
5532 *
5533 * @return array
5534 */
5535 public function getTags() {
5536 $this->firstCallInit();
5537 return array_merge(
5538 array_keys( $this->mTransparentTagHooks ),
5539 array_keys( $this->mTagHooks ),
5540 array_keys( $this->mFunctionTagHooks )
5541 );
5542 }
5543
5544 /**
5545 * @since 1.32
5546 * @return array
5547 */
5548 public function getFunctionSynonyms() {
5549 $this->firstCallInit();
5550 return $this->mFunctionSynonyms;
5551 }
5552
5553 /**
5554 * @since 1.32
5555 * @return string
5556 */
5557 public function getUrlProtocols() {
5558 return $this->mUrlProtocols;
5559 }
5560
5561 /**
5562 * Replace transparent tags in $text with the values given by the callbacks.
5563 *
5564 * Transparent tag hooks are like regular XML-style tag hooks, except they
5565 * operate late in the transformation sequence, on HTML instead of wikitext.
5566 *
5567 * @param string $text
5568 *
5569 * @return string
5570 */
5571 public function replaceTransparentTags( $text ) {
5572 $matches = [];
5573 $elements = array_keys( $this->mTransparentTagHooks );
5574 $text = self::extractTagsAndParams( $elements, $text, $matches );
5575 $replacements = [];
5576
5577 foreach ( $matches as $marker => $data ) {
5578 list( $element, $content, $params, $tag ) = $data;
5579 $tagName = strtolower( $element );
5580 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5581 $output = call_user_func_array(
5582 $this->mTransparentTagHooks[$tagName],
5583 [ $content, $params, $this ]
5584 );
5585 } else {
5586 $output = $tag;
5587 }
5588 $replacements[$marker] = $output;
5589 }
5590 return strtr( $text, $replacements );
5591 }
5592
5593 /**
5594 * Break wikitext input into sections, and either pull or replace
5595 * some particular section's text.
5596 *
5597 * External callers should use the getSection and replaceSection methods.
5598 *
5599 * @param string $text Page wikitext
5600 * @param string|int $sectionId A section identifier string of the form:
5601 * "<flag1> - <flag2> - ... - <section number>"
5602 *
5603 * Currently the only recognised flag is "T", which means the target section number
5604 * was derived during a template inclusion parse, in other words this is a template
5605 * section edit link. If no flags are given, it was an ordinary section edit link.
5606 * This flag is required to avoid a section numbering mismatch when a section is
5607 * enclosed by "<includeonly>" (T8563).
5608 *
5609 * The section number 0 pulls the text before the first heading; other numbers will
5610 * pull the given section along with its lower-level subsections. If the section is
5611 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5612 *
5613 * Section 0 is always considered to exist, even if it only contains the empty
5614 * string. If $text is the empty string and section 0 is replaced, $newText is
5615 * returned.
5616 *
5617 * @param string $mode One of "get" or "replace"
5618 * @param string $newText Replacement text for section data.
5619 * @return string For "get", the extracted section text.
5620 * for "replace", the whole page with the section replaced.
5621 */
5622 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5623 global $wgTitle; # not generally used but removes an ugly failure mode
5624
5625 $magicScopeVariable = $this->lock();
5626 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5627 $outText = '';
5628 $frame = $this->getPreprocessor()->newFrame();
5629
5630 # Process section extraction flags
5631 $flags = 0;
5632 $sectionParts = explode( '-', $sectionId );
5633 $sectionIndex = array_pop( $sectionParts );
5634 foreach ( $sectionParts as $part ) {
5635 if ( $part === 'T' ) {
5636 $flags |= self::PTD_FOR_INCLUSION;
5637 }
5638 }
5639
5640 # Check for empty input
5641 if ( strval( $text ) === '' ) {
5642 # Only sections 0 and T-0 exist in an empty document
5643 if ( $sectionIndex == 0 ) {
5644 if ( $mode === 'get' ) {
5645 return '';
5646 } else {
5647 return $newText;
5648 }
5649 } else {
5650 if ( $mode === 'get' ) {
5651 return $newText;
5652 } else {
5653 return $text;
5654 }
5655 }
5656 }
5657
5658 # Preprocess the text
5659 $root = $this->preprocessToDom( $text, $flags );
5660
5661 # <h> nodes indicate section breaks
5662 # They can only occur at the top level, so we can find them by iterating the root's children
5663 $node = $root->getFirstChild();
5664
5665 # Find the target section
5666 if ( $sectionIndex == 0 ) {
5667 # Section zero doesn't nest, level=big
5668 $targetLevel = 1000;
5669 } else {
5670 while ( $node ) {
5671 if ( $node->getName() === 'h' ) {
5672 $bits = $node->splitHeading();
5673 if ( $bits['i'] == $sectionIndex ) {
5674 $targetLevel = $bits['level'];
5675 break;
5676 }
5677 }
5678 if ( $mode === 'replace' ) {
5679 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5680 }
5681 $node = $node->getNextSibling();
5682 }
5683 }
5684
5685 if ( !$node ) {
5686 # Not found
5687 if ( $mode === 'get' ) {
5688 return $newText;
5689 } else {
5690 return $text;
5691 }
5692 }
5693
5694 # Find the end of the section, including nested sections
5695 do {
5696 if ( $node->getName() === 'h' ) {
5697 $bits = $node->splitHeading();
5698 $curLevel = $bits['level'];
5699 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5700 break;
5701 }
5702 }
5703 if ( $mode === 'get' ) {
5704 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5705 }
5706 $node = $node->getNextSibling();
5707 } while ( $node );
5708
5709 # Write out the remainder (in replace mode only)
5710 if ( $mode === 'replace' ) {
5711 # Output the replacement text
5712 # Add two newlines on -- trailing whitespace in $newText is conventionally
5713 # stripped by the editor, so we need both newlines to restore the paragraph gap
5714 # Only add trailing whitespace if there is newText
5715 if ( $newText != "" ) {
5716 $outText .= $newText . "\n\n";
5717 }
5718
5719 while ( $node ) {
5720 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5721 $node = $node->getNextSibling();
5722 }
5723 }
5724
5725 if ( is_string( $outText ) ) {
5726 # Re-insert stripped tags
5727 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5728 }
5729
5730 return $outText;
5731 }
5732
5733 /**
5734 * This function returns the text of a section, specified by a number ($section).
5735 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5736 * the first section before any such heading (section 0).
5737 *
5738 * If a section contains subsections, these are also returned.
5739 *
5740 * @param string $text Text to look in
5741 * @param string|int $sectionId Section identifier as a number or string
5742 * (e.g. 0, 1 or 'T-1').
5743 * @param string $defaultText Default to return if section is not found
5744 *
5745 * @return string Text of the requested section
5746 */
5747 public function getSection( $text, $sectionId, $defaultText = '' ) {
5748 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5749 }
5750
5751 /**
5752 * This function returns $oldtext after the content of the section
5753 * specified by $section has been replaced with $text. If the target
5754 * section does not exist, $oldtext is returned unchanged.
5755 *
5756 * @param string $oldText Former text of the article
5757 * @param string|int $sectionId Section identifier as a number or string
5758 * (e.g. 0, 1 or 'T-1').
5759 * @param string $newText Replacing text
5760 *
5761 * @return string Modified text
5762 */
5763 public function replaceSection( $oldText, $sectionId, $newText ) {
5764 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5765 }
5766
5767 /**
5768 * Get the ID of the revision we are parsing
5769 *
5770 * @return int|null
5771 */
5772 public function getRevisionId() {
5773 return $this->mRevisionId;
5774 }
5775
5776 /**
5777 * Get the revision object for $this->mRevisionId
5778 *
5779 * @return Revision|null Either a Revision object or null
5780 * @since 1.23 (public since 1.23)
5781 */
5782 public function getRevisionObject() {
5783 if ( !is_null( $this->mRevisionObject ) ) {
5784 return $this->mRevisionObject;
5785 }
5786
5787 // NOTE: try to get the RevisionObject even if mRevisionId is null.
5788 // This is useful when parsing revision that has not yet been saved.
5789 // However, if we get back a saved revision even though we are in
5790 // preview mode, we'll have to ignore it, see below.
5791 // NOTE: This callback may be used to inject an OLD revision that was
5792 // already loaded, so "current" is a bit of a misnomer. We can't just
5793 // skip it if mRevisionId is set.
5794 $rev = call_user_func(
5795 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5796 );
5797
5798 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
5799 // We are in preview mode (mRevisionId is null), and the current revision callback
5800 // returned an existing revision. Ignore it and return null, it's probably the page's
5801 // current revision, which is not what we want here. Note that we do want to call the
5802 // callback to allow the unsaved revision to be injected here, e.g. for
5803 // self-transclusion previews.
5804 return null;
5805 }
5806
5807 // If the parse is for a new revision, then the callback should have
5808 // already been set to force the object and should match mRevisionId.
5809 // If not, try to fetch by mRevisionId for sanity.
5810 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
5811 $rev = Revision::newFromId( $this->mRevisionId );
5812 }
5813
5814 $this->mRevisionObject = $rev;
5815
5816 return $this->mRevisionObject;
5817 }
5818
5819 /**
5820 * Get the timestamp associated with the current revision, adjusted for
5821 * the default server-local timestamp
5822 * @return string
5823 */
5824 public function getRevisionTimestamp() {
5825 if ( is_null( $this->mRevisionTimestamp ) ) {
5826 $revObject = $this->getRevisionObject();
5827 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5828
5829 # The cryptic '' timezone parameter tells to use the site-default
5830 # timezone offset instead of the user settings.
5831 # Since this value will be saved into the parser cache, served
5832 # to other users, and potentially even used inside links and such,
5833 # it needs to be consistent for all visitors.
5834 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
5835 }
5836 return $this->mRevisionTimestamp;
5837 }
5838
5839 /**
5840 * Get the name of the user that edited the last revision
5841 *
5842 * @return string User name
5843 */
5844 public function getRevisionUser() {
5845 if ( is_null( $this->mRevisionUser ) ) {
5846 $revObject = $this->getRevisionObject();
5847
5848 # if this template is subst: the revision id will be blank,
5849 # so just use the current user's name
5850 if ( $revObject ) {
5851 $this->mRevisionUser = $revObject->getUserText();
5852 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5853 $this->mRevisionUser = $this->getUser()->getName();
5854 }
5855 }
5856 return $this->mRevisionUser;
5857 }
5858
5859 /**
5860 * Get the size of the revision
5861 *
5862 * @return int|null Revision size
5863 */
5864 public function getRevisionSize() {
5865 if ( is_null( $this->mRevisionSize ) ) {
5866 $revObject = $this->getRevisionObject();
5867
5868 # if this variable is subst: the revision id will be blank,
5869 # so just use the parser input size, because the own substituation
5870 # will change the size.
5871 if ( $revObject ) {
5872 $this->mRevisionSize = $revObject->getSize();
5873 } else {
5874 $this->mRevisionSize = $this->mInputSize;
5875 }
5876 }
5877 return $this->mRevisionSize;
5878 }
5879
5880 /**
5881 * Mutator for $mDefaultSort
5882 *
5883 * @param string $sort New value
5884 */
5885 public function setDefaultSort( $sort ) {
5886 $this->mDefaultSort = $sort;
5887 $this->mOutput->setProperty( 'defaultsort', $sort );
5888 }
5889
5890 /**
5891 * Accessor for $mDefaultSort
5892 * Will use the empty string if none is set.
5893 *
5894 * This value is treated as a prefix, so the
5895 * empty string is equivalent to sorting by
5896 * page name.
5897 *
5898 * @return string
5899 */
5900 public function getDefaultSort() {
5901 if ( $this->mDefaultSort !== false ) {
5902 return $this->mDefaultSort;
5903 } else {
5904 return '';
5905 }
5906 }
5907
5908 /**
5909 * Accessor for $mDefaultSort
5910 * Unlike getDefaultSort(), will return false if none is set
5911 *
5912 * @return string|bool
5913 */
5914 public function getCustomDefaultSort() {
5915 return $this->mDefaultSort;
5916 }
5917
5918 private static function getSectionNameFromStrippedText( $text ) {
5919 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5920 $text = Sanitizer::decodeCharReferences( $text );
5921 $text = self::normalizeSectionName( $text );
5922 return $text;
5923 }
5924
5925 private static function makeAnchor( $sectionName ) {
5926 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5927 }
5928
5929 private function makeLegacyAnchor( $sectionName ) {
5930 $fragmentMode = $this->config->get( 'FragmentMode' );
5931 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
5932 // ForAttribute() and ForLink() are the same for legacy encoding
5933 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5934 } else {
5935 $id = Sanitizer::escapeIdForLink( $sectionName );
5936 }
5937
5938 return "#$id";
5939 }
5940
5941 /**
5942 * Try to guess the section anchor name based on a wikitext fragment
5943 * presumably extracted from a heading, for example "Header" from
5944 * "== Header ==".
5945 *
5946 * @param string $text
5947 * @return string Anchor (starting with '#')
5948 */
5949 public function guessSectionNameFromWikiText( $text ) {
5950 # Strip out wikitext links(they break the anchor)
5951 $text = $this->stripSectionName( $text );
5952 $sectionName = self::getSectionNameFromStrippedText( $text );
5953 return self::makeAnchor( $sectionName );
5954 }
5955
5956 /**
5957 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5958 * instead, if possible. For use in redirects, since various versions
5959 * of Microsoft browsers interpret Location: headers as something other
5960 * than UTF-8, resulting in breakage.
5961 *
5962 * @param string $text The section name
5963 * @return string Anchor (starting with '#')
5964 */
5965 public function guessLegacySectionNameFromWikiText( $text ) {
5966 # Strip out wikitext links(they break the anchor)
5967 $text = $this->stripSectionName( $text );
5968 $sectionName = self::getSectionNameFromStrippedText( $text );
5969 return $this->makeLegacyAnchor( $sectionName );
5970 }
5971
5972 /**
5973 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5974 * @param string $text Section name (plain text)
5975 * @return string Anchor (starting with '#')
5976 */
5977 public static function guessSectionNameFromStrippedText( $text ) {
5978 $sectionName = self::getSectionNameFromStrippedText( $text );
5979 return self::makeAnchor( $sectionName );
5980 }
5981
5982 /**
5983 * Apply the same normalization as code making links to this section would
5984 *
5985 * @param string $text
5986 * @return string
5987 */
5988 private static function normalizeSectionName( $text ) {
5989 # T90902: ensure the same normalization is applied for IDs as to links
5990 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5991 try {
5992
5993 $parts = $titleParser->splitTitleString( "#$text" );
5994 } catch ( MalformedTitleException $ex ) {
5995 return $text;
5996 }
5997 return $parts['fragment'];
5998 }
5999
6000 /**
6001 * Strips a text string of wikitext for use in a section anchor
6002 *
6003 * Accepts a text string and then removes all wikitext from the
6004 * string and leaves only the resultant text (i.e. the result of
6005 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6006 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6007 * to create valid section anchors by mimicing the output of the
6008 * parser when headings are parsed.
6009 *
6010 * @param string $text Text string to be stripped of wikitext
6011 * for use in a Section anchor
6012 * @return string Filtered text string
6013 */
6014 public function stripSectionName( $text ) {
6015 # Strip internal link markup
6016 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6017 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6018
6019 # Strip external link markup
6020 # @todo FIXME: Not tolerant to blank link text
6021 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6022 # on how many empty links there are on the page - need to figure that out.
6023 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6024
6025 # Parse wikitext quotes (italics & bold)
6026 $text = $this->doQuotes( $text );
6027
6028 # Strip HTML tags
6029 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6030 return $text;
6031 }
6032
6033 /**
6034 * strip/replaceVariables/unstrip for preprocessor regression testing
6035 *
6036 * @param string $text
6037 * @param Title $title
6038 * @param ParserOptions $options
6039 * @param int $outputType
6040 *
6041 * @return string
6042 */
6043 public function testSrvus( $text, Title $title, ParserOptions $options,
6044 $outputType = self::OT_HTML
6045 ) {
6046 $magicScopeVariable = $this->lock();
6047 $this->startParse( $title, $options, $outputType, true );
6048
6049 $text = $this->replaceVariables( $text );
6050 $text = $this->mStripState->unstripBoth( $text );
6051 $text = Sanitizer::removeHTMLtags( $text );
6052 return $text;
6053 }
6054
6055 /**
6056 * @param string $text
6057 * @param Title $title
6058 * @param ParserOptions $options
6059 * @return string
6060 */
6061 public function testPst( $text, Title $title, ParserOptions $options ) {
6062 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6063 }
6064
6065 /**
6066 * @param string $text
6067 * @param Title $title
6068 * @param ParserOptions $options
6069 * @return string
6070 */
6071 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6072 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6073 }
6074
6075 /**
6076 * Call a callback function on all regions of the given text that are not
6077 * inside strip markers, and replace those regions with the return value
6078 * of the callback. For example, with input:
6079 *
6080 * aaa<MARKER>bbb
6081 *
6082 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6083 * two strings will be replaced with the value returned by the callback in
6084 * each case.
6085 *
6086 * @param string $s
6087 * @param callable $callback
6088 *
6089 * @return string
6090 */
6091 public function markerSkipCallback( $s, $callback ) {
6092 $i = 0;
6093 $out = '';
6094 while ( $i < strlen( $s ) ) {
6095 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6096 if ( $markerStart === false ) {
6097 $out .= call_user_func( $callback, substr( $s, $i ) );
6098 break;
6099 } else {
6100 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6101 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6102 if ( $markerEnd === false ) {
6103 $out .= substr( $s, $markerStart );
6104 break;
6105 } else {
6106 $markerEnd += strlen( self::MARKER_SUFFIX );
6107 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6108 $i = $markerEnd;
6109 }
6110 }
6111 }
6112 return $out;
6113 }
6114
6115 /**
6116 * Remove any strip markers found in the given text.
6117 *
6118 * @param string $text
6119 * @return string
6120 */
6121 public function killMarkers( $text ) {
6122 return $this->mStripState->killMarkers( $text );
6123 }
6124
6125 /**
6126 * Save the parser state required to convert the given half-parsed text to
6127 * HTML. "Half-parsed" in this context means the output of
6128 * recursiveTagParse() or internalParse(). This output has strip markers
6129 * from replaceVariables (extensionSubstitution() etc.), and link
6130 * placeholders from replaceLinkHolders().
6131 *
6132 * Returns an array which can be serialized and stored persistently. This
6133 * array can later be loaded into another parser instance with
6134 * unserializeHalfParsedText(). The text can then be safely incorporated into
6135 * the return value of a parser hook.
6136 *
6137 * @deprecated since 1.31
6138 * @param string $text
6139 *
6140 * @return array
6141 */
6142 public function serializeHalfParsedText( $text ) {
6143 wfDeprecated( __METHOD__, '1.31' );
6144 $data = [
6145 'text' => $text,
6146 'version' => self::HALF_PARSED_VERSION,
6147 'stripState' => $this->mStripState->getSubState( $text ),
6148 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6149 ];
6150 return $data;
6151 }
6152
6153 /**
6154 * Load the parser state given in the $data array, which is assumed to
6155 * have been generated by serializeHalfParsedText(). The text contents is
6156 * extracted from the array, and its markers are transformed into markers
6157 * appropriate for the current Parser instance. This transformed text is
6158 * returned, and can be safely included in the return value of a parser
6159 * hook.
6160 *
6161 * If the $data array has been stored persistently, the caller should first
6162 * check whether it is still valid, by calling isValidHalfParsedText().
6163 *
6164 * @deprecated since 1.31
6165 * @param array $data Serialized data
6166 * @throws MWException
6167 * @return string
6168 */
6169 public function unserializeHalfParsedText( $data ) {
6170 wfDeprecated( __METHOD__, '1.31' );
6171 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6172 throw new MWException( __METHOD__ . ': invalid version' );
6173 }
6174
6175 # First, extract the strip state.
6176 $texts = [ $data['text'] ];
6177 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6178
6179 # Now renumber links
6180 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6181
6182 # Should be good to go.
6183 return $texts[0];
6184 }
6185
6186 /**
6187 * Returns true if the given array, presumed to be generated by
6188 * serializeHalfParsedText(), is compatible with the current version of the
6189 * parser.
6190 *
6191 * @deprecated since 1.31
6192 * @param array $data
6193 *
6194 * @return bool
6195 */
6196 public function isValidHalfParsedText( $data ) {
6197 wfDeprecated( __METHOD__, '1.31' );
6198 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6199 }
6200
6201 /**
6202 * Parsed a width param of imagelink like 300px or 200x300px
6203 *
6204 * @param string $value
6205 * @param bool $parseHeight
6206 *
6207 * @return array
6208 * @since 1.20
6209 */
6210 public static function parseWidthParam( $value, $parseHeight = true ) {
6211 $parsedWidthParam = [];
6212 if ( $value === '' ) {
6213 return $parsedWidthParam;
6214 }
6215 $m = [];
6216 # (T15500) In both cases (width/height and width only),
6217 # permit trailing "px" for backward compatibility.
6218 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6219 $width = intval( $m[1] );
6220 $height = intval( $m[2] );
6221 $parsedWidthParam['width'] = $width;
6222 $parsedWidthParam['height'] = $height;
6223 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6224 $width = intval( $value );
6225 $parsedWidthParam['width'] = $width;
6226 }
6227 return $parsedWidthParam;
6228 }
6229
6230 /**
6231 * Lock the current instance of the parser.
6232 *
6233 * This is meant to stop someone from calling the parser
6234 * recursively and messing up all the strip state.
6235 *
6236 * @throws MWException If parser is in a parse
6237 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6238 */
6239 protected function lock() {
6240 if ( $this->mInParse ) {
6241 throw new MWException( "Parser state cleared while parsing. "
6242 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6243 }
6244
6245 // Save the backtrace when locking, so that if some code tries locking again,
6246 // we can print the lock owner's backtrace for easier debugging
6247 $e = new Exception;
6248 $this->mInParse = $e->getTraceAsString();
6249
6250 $recursiveCheck = new ScopedCallback( function () {
6251 $this->mInParse = false;
6252 } );
6253
6254 return $recursiveCheck;
6255 }
6256
6257 /**
6258 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6259 *
6260 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6261 * or if there is more than one <p/> tag in the input HTML.
6262 *
6263 * @param string $html
6264 * @return string
6265 * @since 1.24
6266 */
6267 public static function stripOuterParagraph( $html ) {
6268 $m = [];
6269 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6270 if ( strpos( $m[1], '</p>' ) === false ) {
6271 $html = $m[1];
6272 }
6273 }
6274
6275 return $html;
6276 }
6277
6278 /**
6279 * Return this parser if it is not doing anything, otherwise
6280 * get a fresh parser. You can use this method by doing
6281 * $myParser = $wgParser->getFreshParser(), or more simply
6282 * $wgParser->getFreshParser()->parse( ... );
6283 * if you're unsure if $wgParser is safe to use.
6284 *
6285 * @since 1.24
6286 * @return Parser A parser object that is not parsing anything
6287 */
6288 public function getFreshParser() {
6289 if ( $this->mInParse ) {
6290 return $this->factory->create();
6291 } else {
6292 return $this;
6293 }
6294 }
6295
6296 /**
6297 * Set's up the PHP implementation of OOUI for use in this request
6298 * and instructs OutputPage to enable OOUI for itself.
6299 *
6300 * @since 1.26
6301 */
6302 public function enableOOUI() {
6303 OutputPage::setupOOUI();
6304 $this->mOutput->setEnableOOUI( true );
6305 }
6306 }
Wiklou, le Wiki du Wiklou