Remove unnecessary Parser::getConverterLanguage() indirection
[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 Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // phpcs:ignore Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102
103 # Regular expression for a non-newline space
104 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 /**
118 * @var string Prefix and suffix for temporary replacement strings
119 * for the multipass parser.
120 *
121 * \x7f should never appear in input as it's disallowed in XML.
122 * Using it at the front also gives us a little extra robustness
123 * since it shouldn't match when butted up against identifier-like
124 * string constructs.
125 *
126 * Must not consist of all title characters, or else it will change
127 * the behavior of <nowiki> in a link.
128 *
129 * Must have a character that needs escaping in attributes, otherwise
130 * someone could put a strip marker in an attribute, to get around
131 * escaping quote marks, and break out of the attribute. Thus we add
132 * `'".
133 */
134 const MARKER_SUFFIX = "-QINU`\"'\x7f";
135 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
136
137 # Markers used for wrapping the table of contents
138 const TOC_START = '<mw:toc>';
139 const TOC_END = '</mw:toc>';
140
141 /** @var int Assume that no output will later be saved this many seconds after parsing */
142 const MAX_TTS = 900;
143
144 # Persistent:
145 public $mTagHooks = [];
146 public $mTransparentTagHooks = [];
147 public $mFunctionHooks = [];
148 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
149 public $mFunctionTagHooks = [];
150 public $mStripList = [];
151 public $mDefaultStripList = [];
152 public $mVarCache = [];
153 public $mImageParams = [];
154 public $mImageParamsMagicArray = [];
155 public $mMarkerIndex = 0;
156 public $mFirstCall = true;
157
158 # Initialised by initialiseVariables()
159
160 /**
161 * @var MagicWordArray
162 */
163 public $mVariables;
164
165 /**
166 * @var MagicWordArray
167 */
168 public $mSubstWords;
169 # Initialised in constructor
170 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
171
172 # Initialized in getPreprocessor()
173 /** @var Preprocessor */
174 public $mPreprocessor;
175
176 # Cleared with clearState():
177 /**
178 * @var ParserOutput
179 */
180 public $mOutput;
181 public $mAutonumber;
182
183 /**
184 * @var StripState
185 */
186 public $mStripState;
187
188 public $mIncludeCount;
189 /**
190 * @var LinkHolderArray
191 */
192 public $mLinkHolders;
193
194 public $mLinkID;
195 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
196 public $mDefaultSort;
197 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
198 public $mExpensiveFunctionCount; # number of expensive parser function calls
199 public $mShowToc, $mForceTocPosition;
200
201 /**
202 * @var User
203 */
204 public $mUser; # User object; only used when doing pre-save transform
205
206 # Temporary
207 # These are variables reset at least once per parse regardless of $clearState
208
209 /**
210 * @var ParserOptions
211 */
212 public $mOptions;
213
214 /**
215 * @var Title
216 */
217 public $mTitle; # Title context, used for self-link rendering and similar things
218 public $mOutputType; # Output type, one of the OT_xxx constants
219 public $ot; # Shortcut alias, see setOutputType()
220 public $mRevisionObject; # The revision object of the specified revision ID
221 public $mRevisionId; # ID to display in {{REVISIONID}} tags
222 public $mRevisionTimestamp; # The timestamp of the specified revision ID
223 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
224 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
225 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
226 public $mInputSize = false; # For {{PAGESIZE}} on current page.
227
228 /**
229 * @var string Deprecated accessor for the strip marker prefix.
230 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
231 */
232 public $mUniqPrefix = self::MARKER_PREFIX;
233
234 /**
235 * @var array Array with the language name of each language link (i.e. the
236 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
237 * duplicate language links to the ParserOutput.
238 */
239 public $mLangLinkLanguages;
240
241 /**
242 * @var MapCacheLRU|null
243 * @since 1.24
244 *
245 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
246 */
247 public $currentRevisionCache;
248
249 /**
250 * @var bool|string Recursive call protection.
251 * This variable should be treated as if it were private.
252 */
253 public $mInParse = false;
254
255 /** @var SectionProfiler */
256 protected $mProfiler;
257
258 /**
259 * @var LinkRenderer
260 */
261 protected $mLinkRenderer;
262
263 /**
264 * @param array $conf
265 */
266 public function __construct( $conf = [] ) {
267 $this->mConf = $conf;
268 $this->mUrlProtocols = wfUrlProtocols();
269 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
270 self::EXT_LINK_ADDR .
271 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
272 if ( isset( $conf['preprocessorClass'] ) ) {
273 $this->mPreprocessorClass = $conf['preprocessorClass'];
274 } elseif ( defined( 'HPHP_VERSION' ) ) {
275 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
276 $this->mPreprocessorClass = Preprocessor_Hash::class;
277 } elseif ( extension_loaded( 'domxml' ) ) {
278 # PECL extension that conflicts with the core DOM extension (T15770)
279 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
280 $this->mPreprocessorClass = Preprocessor_Hash::class;
281 } elseif ( extension_loaded( 'dom' ) ) {
282 $this->mPreprocessorClass = Preprocessor_DOM::class;
283 } else {
284 $this->mPreprocessorClass = Preprocessor_Hash::class;
285 }
286 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
287 }
288
289 /**
290 * Reduce memory usage to reduce the impact of circular references
291 */
292 public function __destruct() {
293 if ( isset( $this->mLinkHolders ) ) {
294 unset( $this->mLinkHolders );
295 }
296 foreach ( $this as $name => $value ) {
297 unset( $this->$name );
298 }
299 }
300
301 /**
302 * Allow extensions to clean up when the parser is cloned
303 */
304 public function __clone() {
305 $this->mInParse = false;
306
307 // T58226: When you create a reference "to" an object field, that
308 // makes the object field itself be a reference too (until the other
309 // reference goes out of scope). When cloning, any field that's a
310 // reference is copied as a reference in the new object. Both of these
311 // are defined PHP5 behaviors, as inconvenient as it is for us when old
312 // hooks from PHP4 days are passing fields by reference.
313 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
314 // Make a non-reference copy of the field, then rebind the field to
315 // reference the new copy.
316 $tmp = $this->$k;
317 $this->$k =& $tmp;
318 unset( $tmp );
319 }
320
321 Hooks::run( 'ParserCloned', [ $this ] );
322 }
323
324 /**
325 * Do various kinds of initialisation on the first call of the parser
326 */
327 public function firstCallInit() {
328 if ( !$this->mFirstCall ) {
329 return;
330 }
331 $this->mFirstCall = false;
332
333 CoreParserFunctions::register( $this );
334 CoreTagHooks::register( $this );
335 $this->initialiseVariables();
336
337 // Avoid PHP 7.1 warning from passing $this by reference
338 $parser = $this;
339 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
340 }
341
342 /**
343 * Clear Parser state
344 *
345 * @private
346 */
347 public function clearState() {
348 if ( $this->mFirstCall ) {
349 $this->firstCallInit();
350 }
351 $this->mOutput = new ParserOutput;
352 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
353 $this->mAutonumber = 0;
354 $this->mIncludeCount = [];
355 $this->mLinkHolders = new LinkHolderArray( $this );
356 $this->mLinkID = 0;
357 $this->mRevisionObject = $this->mRevisionTimestamp =
358 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
359 $this->mVarCache = [];
360 $this->mUser = null;
361 $this->mLangLinkLanguages = [];
362 $this->currentRevisionCache = null;
363
364 $this->mStripState = new StripState( $this );
365
366 # Clear these on every parse, T6549
367 $this->mTplRedirCache = $this->mTplDomCache = [];
368
369 $this->mShowToc = true;
370 $this->mForceTocPosition = false;
371 $this->mIncludeSizes = [
372 'post-expand' => 0,
373 'arg' => 0,
374 ];
375 $this->mPPNodeCount = 0;
376 $this->mGeneratedPPNodeCount = 0;
377 $this->mHighestExpansionDepth = 0;
378 $this->mDefaultSort = false;
379 $this->mHeadings = [];
380 $this->mDoubleUnderscores = [];
381 $this->mExpensiveFunctionCount = 0;
382
383 # Fix cloning
384 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
385 $this->mPreprocessor = null;
386 }
387
388 $this->mProfiler = new SectionProfiler();
389
390 // Avoid PHP 7.1 warning from passing $this by reference
391 $parser = $this;
392 Hooks::run( 'ParserClearState', [ &$parser ] );
393 }
394
395 /**
396 * Convert wikitext to HTML
397 * Do not call this function recursively.
398 *
399 * @param string $text Text we want to parse
400 * @param Title $title
401 * @param ParserOptions $options
402 * @param bool $linestart
403 * @param bool $clearState
404 * @param int $revid Number to pass in {{REVISIONID}}
405 * @return ParserOutput A ParserOutput
406 */
407 public function parse(
408 $text, Title $title, ParserOptions $options,
409 $linestart = true, $clearState = true, $revid = null
410 ) {
411 if ( $clearState ) {
412 // We use U+007F DELETE to construct strip markers, so we have to make
413 // sure that this character does not occur in the input text.
414 $text = strtr( $text, "\x7f", "?" );
415 $magicScopeVariable = $this->lock();
416 }
417 // Strip U+0000 NULL (T159174)
418 $text = str_replace( "\000", '', $text );
419
420 $this->startParse( $title, $options, self::OT_HTML, $clearState );
421
422 $this->currentRevisionCache = null;
423 $this->mInputSize = strlen( $text );
424 if ( $this->mOptions->getEnableLimitReport() ) {
425 $this->mOutput->resetParseStartTime();
426 }
427
428 $oldRevisionId = $this->mRevisionId;
429 $oldRevisionObject = $this->mRevisionObject;
430 $oldRevisionTimestamp = $this->mRevisionTimestamp;
431 $oldRevisionUser = $this->mRevisionUser;
432 $oldRevisionSize = $this->mRevisionSize;
433 if ( $revid !== null ) {
434 $this->mRevisionId = $revid;
435 $this->mRevisionObject = null;
436 $this->mRevisionTimestamp = null;
437 $this->mRevisionUser = null;
438 $this->mRevisionSize = null;
439 }
440
441 // Avoid PHP 7.1 warning from passing $this by reference
442 $parser = $this;
443 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
444 # No more strip!
445 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
446 $text = $this->internalParse( $text );
447 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
448
449 $text = $this->internalParseHalfParsed( $text, true, $linestart );
450
451 /**
452 * A converted title will be provided in the output object if title and
453 * content conversion are enabled, the article text does not contain
454 * a conversion-suppressing double-underscore tag, and no
455 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
456 * automatic link conversion.
457 */
458 if ( !( $options->getDisableTitleConversion()
459 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
460 || isset( $this->mDoubleUnderscores['notitleconvert'] )
461 || $this->mOutput->getDisplayTitle() !== false )
462 ) {
463 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
464 if ( $convruletitle ) {
465 $this->mOutput->setTitleText( $convruletitle );
466 } else {
467 $titleText = $this->getTargetLanguage()->convertTitle( $title );
468 $this->mOutput->setTitleText( $titleText );
469 }
470 }
471
472 # Compute runtime adaptive expiry if set
473 $this->mOutput->finalizeAdaptiveCacheExpiry();
474
475 # Warn if too many heavyweight parser functions were used
476 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
477 $this->limitationWarn( 'expensive-parserfunction',
478 $this->mExpensiveFunctionCount,
479 $this->mOptions->getExpensiveParserFunctionLimit()
480 );
481 }
482
483 # Information on limits, for the benefit of users who try to skirt them
484 if ( $this->mOptions->getEnableLimitReport() ) {
485 $text .= $this->makeLimitReport();
486 }
487
488 # Wrap non-interface parser output in a <div> so it can be targeted
489 # with CSS (T37247)
490 $class = $this->mOptions->getWrapOutputClass();
491 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
492 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
493 }
494
495 $this->mOutput->setText( $text );
496
497 $this->mRevisionId = $oldRevisionId;
498 $this->mRevisionObject = $oldRevisionObject;
499 $this->mRevisionTimestamp = $oldRevisionTimestamp;
500 $this->mRevisionUser = $oldRevisionUser;
501 $this->mRevisionSize = $oldRevisionSize;
502 $this->mInputSize = false;
503 $this->currentRevisionCache = null;
504
505 return $this->mOutput;
506 }
507
508 /**
509 * Set the limit report data in the current ParserOutput, and return the
510 * limit report HTML comment.
511 *
512 * @return string
513 */
514 protected function makeLimitReport() {
515 global $wgShowHostnames;
516
517 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
518
519 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
520 if ( $cpuTime !== null ) {
521 $this->mOutput->setLimitReportData( 'limitreport-cputime',
522 sprintf( "%.3f", $cpuTime )
523 );
524 }
525
526 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
527 $this->mOutput->setLimitReportData( 'limitreport-walltime',
528 sprintf( "%.3f", $wallTime )
529 );
530
531 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
532 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
533 );
534 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
535 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
536 );
537 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
538 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
539 );
540 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
541 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
542 );
543 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
544 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
545 );
546 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
547 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
548 );
549
550 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
551 $this->mOutput->setLimitReportData( $key, $value );
552 }
553
554 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
555
556 $limitReport = "NewPP limit report\n";
557 if ( $wgShowHostnames ) {
558 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
559 }
560 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
561 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
562 $limitReport .= 'Dynamic content: ' .
563 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
564 "\n";
565
566 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
567 if ( Hooks::run( 'ParserLimitReportFormat',
568 [ $key, &$value, &$limitReport, false, false ]
569 ) ) {
570 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
571 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
572 ->inLanguage( 'en' )->useDatabase( false );
573 if ( !$valueMsg->exists() ) {
574 $valueMsg = new RawMessage( '$1' );
575 }
576 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
577 $valueMsg->params( $value );
578 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
579 }
580 }
581 }
582 // Since we're not really outputting HTML, decode the entities and
583 // then re-encode the things that need hiding inside HTML comments.
584 $limitReport = htmlspecialchars_decode( $limitReport );
585 // Run deprecated hook
586 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
587
588 // Sanitize for comment. Note '‐' in the replacement is U+2010,
589 // which looks much like the problematic '-'.
590 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
591 $text = "\n<!-- \n$limitReport-->\n";
592
593 // Add on template profiling data in human/machine readable way
594 $dataByFunc = $this->mProfiler->getFunctionStats();
595 uasort( $dataByFunc, function ( $a, $b ) {
596 return $b['real'] <=> $a['real']; // descending order
597 } );
598 $profileReport = [];
599 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
600 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
601 $item['%real'], $item['real'], $item['calls'],
602 htmlspecialchars( $item['name'] ) );
603 }
604 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
605 $text .= implode( "\n", $profileReport ) . "\n-->\n";
606
607 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
608
609 // Add other cache related metadata
610 if ( $wgShowHostnames ) {
611 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
612 }
613 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
614 $this->mOutput->getCacheTime() );
615 $this->mOutput->setLimitReportData( 'cachereport-ttl',
616 $this->mOutput->getCacheExpiry() );
617 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
618 $this->mOutput->hasDynamicContent() );
619
620 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
621 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
622 $this->mTitle->getPrefixedDBkey() );
623 }
624 return $text;
625 }
626
627 /**
628 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
629 * can be called from an extension tag hook.
630 *
631 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
632 * instead, which means that lists and links have not been fully parsed yet,
633 * and strip markers are still present.
634 *
635 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
636 *
637 * Use this function if you're a parser tag hook and you want to parse
638 * wikitext before or after applying additional transformations, and you
639 * intend to *return the result as hook output*, which will cause it to go
640 * through the rest of parsing process automatically.
641 *
642 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
643 * $text are not expanded
644 *
645 * @param string $text Text extension wants to have parsed
646 * @param bool|PPFrame $frame The frame to use for expanding any template variables
647 * @return string UNSAFE half-parsed HTML
648 */
649 public function recursiveTagParse( $text, $frame = false ) {
650 // Avoid PHP 7.1 warning from passing $this by reference
651 $parser = $this;
652 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
653 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
654 $text = $this->internalParse( $text, false, $frame );
655 return $text;
656 }
657
658 /**
659 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
660 * point can be called from an extension tag hook.
661 *
662 * The output of this function is fully-parsed HTML that is safe for output.
663 * If you're a parser tag hook, you might want to use recursiveTagParse()
664 * instead.
665 *
666 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
667 * $text are not expanded
668 *
669 * @since 1.25
670 *
671 * @param string $text Text extension wants to have parsed
672 * @param bool|PPFrame $frame The frame to use for expanding any template variables
673 * @return string Fully parsed HTML
674 */
675 public function recursiveTagParseFully( $text, $frame = false ) {
676 $text = $this->recursiveTagParse( $text, $frame );
677 $text = $this->internalParseHalfParsed( $text, false );
678 return $text;
679 }
680
681 /**
682 * Expand templates and variables in the text, producing valid, static wikitext.
683 * Also removes comments.
684 * Do not call this function recursively.
685 * @param string $text
686 * @param Title $title
687 * @param ParserOptions $options
688 * @param int|null $revid
689 * @param bool|PPFrame $frame
690 * @return mixed|string
691 */
692 public function preprocess( $text, Title $title = null,
693 ParserOptions $options, $revid = null, $frame = false
694 ) {
695 $magicScopeVariable = $this->lock();
696 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
697 if ( $revid !== null ) {
698 $this->mRevisionId = $revid;
699 }
700 // Avoid PHP 7.1 warning from passing $this by reference
701 $parser = $this;
702 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
703 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
704 $text = $this->replaceVariables( $text, $frame );
705 $text = $this->mStripState->unstripBoth( $text );
706 return $text;
707 }
708
709 /**
710 * Recursive parser entry point that can be called from an extension tag
711 * hook.
712 *
713 * @param string $text Text to be expanded
714 * @param bool|PPFrame $frame The frame to use for expanding any template variables
715 * @return string
716 * @since 1.19
717 */
718 public function recursivePreprocess( $text, $frame = false ) {
719 $text = $this->replaceVariables( $text, $frame );
720 $text = $this->mStripState->unstripBoth( $text );
721 return $text;
722 }
723
724 /**
725 * Process the wikitext for the "?preload=" feature. (T7210)
726 *
727 * "<noinclude>", "<includeonly>" etc. are parsed as for template
728 * transclusion, comments, templates, arguments, tags hooks and parser
729 * functions are untouched.
730 *
731 * @param string $text
732 * @param Title $title
733 * @param ParserOptions $options
734 * @param array $params
735 * @return string
736 */
737 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
738 $msg = new RawMessage( $text );
739 $text = $msg->params( $params )->plain();
740
741 # Parser (re)initialisation
742 $magicScopeVariable = $this->lock();
743 $this->startParse( $title, $options, self::OT_PLAIN, true );
744
745 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
746 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
747 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
748 $text = $this->mStripState->unstripBoth( $text );
749 return $text;
750 }
751
752 /**
753 * Set the current user.
754 * Should only be used when doing pre-save transform.
755 *
756 * @param User|null $user User object or null (to reset)
757 */
758 public function setUser( $user ) {
759 $this->mUser = $user;
760 }
761
762 /**
763 * Set the context title
764 *
765 * @param Title $t
766 */
767 public function setTitle( $t ) {
768 if ( !$t ) {
769 $t = Title::newFromText( 'NO TITLE' );
770 }
771
772 if ( $t->hasFragment() ) {
773 # Strip the fragment to avoid various odd effects
774 $this->mTitle = $t->createFragmentTarget( '' );
775 } else {
776 $this->mTitle = $t;
777 }
778 }
779
780 /**
781 * Accessor for the Title object
782 *
783 * @return Title
784 */
785 public function getTitle() {
786 return $this->mTitle;
787 }
788
789 /**
790 * Accessor/mutator for the Title object
791 *
792 * @param Title $x Title object or null to just get the current one
793 * @return Title
794 */
795 public function Title( $x = null ) {
796 return wfSetVar( $this->mTitle, $x );
797 }
798
799 /**
800 * Set the output type
801 *
802 * @param int $ot New value
803 */
804 public function setOutputType( $ot ) {
805 $this->mOutputType = $ot;
806 # Shortcut alias
807 $this->ot = [
808 'html' => $ot == self::OT_HTML,
809 'wiki' => $ot == self::OT_WIKI,
810 'pre' => $ot == self::OT_PREPROCESS,
811 'plain' => $ot == self::OT_PLAIN,
812 ];
813 }
814
815 /**
816 * Accessor/mutator for the output type
817 *
818 * @param int|null $x New value or null to just get the current one
819 * @return int
820 */
821 public function OutputType( $x = null ) {
822 return wfSetVar( $this->mOutputType, $x );
823 }
824
825 /**
826 * Get the ParserOutput object
827 *
828 * @return ParserOutput
829 */
830 public function getOutput() {
831 return $this->mOutput;
832 }
833
834 /**
835 * Get the ParserOptions object
836 *
837 * @return ParserOptions
838 */
839 public function getOptions() {
840 return $this->mOptions;
841 }
842
843 /**
844 * Accessor/mutator for the ParserOptions object
845 *
846 * @param ParserOptions $x New value or null to just get the current one
847 * @return ParserOptions Current ParserOptions object
848 */
849 public function Options( $x = null ) {
850 return wfSetVar( $this->mOptions, $x );
851 }
852
853 /**
854 * @return int
855 */
856 public function nextLinkID() {
857 return $this->mLinkID++;
858 }
859
860 /**
861 * @param int $id
862 */
863 public function setLinkID( $id ) {
864 $this->mLinkID = $id;
865 }
866
867 /**
868 * Get a language object for use in parser functions such as {{FORMATNUM:}}
869 * @return Language
870 */
871 public function getFunctionLang() {
872 return $this->getTargetLanguage();
873 }
874
875 /**
876 * Get the target language for the content being parsed. This is usually the
877 * language that the content is in.
878 *
879 * @since 1.19
880 *
881 * @throws MWException
882 * @return Language
883 */
884 public function getTargetLanguage() {
885 $target = $this->mOptions->getTargetLanguage();
886
887 if ( $target !== null ) {
888 return $target;
889 } elseif ( $this->mOptions->getInterfaceMessage() ) {
890 return $this->mOptions->getUserLangObj();
891 } elseif ( is_null( $this->mTitle ) ) {
892 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
893 }
894
895 return $this->mTitle->getPageLanguage();
896 }
897
898 /**
899 * Get the language object for language conversion
900 * @deprecated since 1.32, just use getTargetLanguage()
901 * @return Language|null
902 */
903 public function getConverterLanguage() {
904 return $this->getTargetLanguage();
905 }
906
907 /**
908 * Get a User object either from $this->mUser, if set, or from the
909 * ParserOptions object otherwise
910 *
911 * @return User
912 */
913 public function getUser() {
914 if ( !is_null( $this->mUser ) ) {
915 return $this->mUser;
916 }
917 return $this->mOptions->getUser();
918 }
919
920 /**
921 * Get a preprocessor object
922 *
923 * @return Preprocessor
924 */
925 public function getPreprocessor() {
926 if ( !isset( $this->mPreprocessor ) ) {
927 $class = $this->mPreprocessorClass;
928 $this->mPreprocessor = new $class( $this );
929 }
930 return $this->mPreprocessor;
931 }
932
933 /**
934 * Get a LinkRenderer instance to make links with
935 *
936 * @since 1.28
937 * @return LinkRenderer
938 */
939 public function getLinkRenderer() {
940 if ( !$this->mLinkRenderer ) {
941 $this->mLinkRenderer = MediaWikiServices::getInstance()
942 ->getLinkRendererFactory()->create();
943 $this->mLinkRenderer->setStubThreshold(
944 $this->getOptions()->getStubThreshold()
945 );
946 }
947
948 return $this->mLinkRenderer;
949 }
950
951 /**
952 * Replaces all occurrences of HTML-style comments and the given tags
953 * in the text with a random marker and returns the next text. The output
954 * parameter $matches will be an associative array filled with data in
955 * the form:
956 *
957 * @code
958 * 'UNIQ-xxxxx' => [
959 * 'element',
960 * 'tag content',
961 * [ 'param' => 'x' ],
962 * '<element param="x">tag content</element>' ]
963 * @endcode
964 *
965 * @param array $elements List of element names. Comments are always extracted.
966 * @param string $text Source text string.
967 * @param array &$matches Out parameter, Array: extracted tags
968 * @return string Stripped text
969 */
970 public static function extractTagsAndParams( $elements, $text, &$matches ) {
971 static $n = 1;
972 $stripped = '';
973 $matches = [];
974
975 $taglist = implode( '|', $elements );
976 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
977
978 while ( $text != '' ) {
979 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
980 $stripped .= $p[0];
981 if ( count( $p ) < 5 ) {
982 break;
983 }
984 if ( count( $p ) > 5 ) {
985 # comment
986 $element = $p[4];
987 $attributes = '';
988 $close = '';
989 $inside = $p[5];
990 } else {
991 # tag
992 $element = $p[1];
993 $attributes = $p[2];
994 $close = $p[3];
995 $inside = $p[4];
996 }
997
998 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
999 $stripped .= $marker;
1000
1001 if ( $close === '/>' ) {
1002 # Empty element tag, <tag />
1003 $content = null;
1004 $text = $inside;
1005 $tail = null;
1006 } else {
1007 if ( $element === '!--' ) {
1008 $end = '/(-->)/';
1009 } else {
1010 $end = "/(<\\/$element\\s*>)/i";
1011 }
1012 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1013 $content = $q[0];
1014 if ( count( $q ) < 3 ) {
1015 # No end tag -- let it run out to the end of the text.
1016 $tail = '';
1017 $text = '';
1018 } else {
1019 $tail = $q[1];
1020 $text = $q[2];
1021 }
1022 }
1023
1024 $matches[$marker] = [ $element,
1025 $content,
1026 Sanitizer::decodeTagAttributes( $attributes ),
1027 "<$element$attributes$close$content$tail" ];
1028 }
1029 return $stripped;
1030 }
1031
1032 /**
1033 * Get a list of strippable XML-like elements
1034 *
1035 * @return array
1036 */
1037 public function getStripList() {
1038 return $this->mStripList;
1039 }
1040
1041 /**
1042 * Add an item to the strip state
1043 * Returns the unique tag which must be inserted into the stripped text
1044 * The tag will be replaced with the original text in unstrip()
1045 *
1046 * @param string $text
1047 *
1048 * @return string
1049 */
1050 public function insertStripItem( $text ) {
1051 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1052 $this->mMarkerIndex++;
1053 $this->mStripState->addGeneral( $marker, $text );
1054 return $marker;
1055 }
1056
1057 /**
1058 * parse the wiki syntax used to render tables
1059 *
1060 * @private
1061 * @param string $text
1062 * @return string
1063 */
1064 public function doTableStuff( $text ) {
1065 $lines = StringUtils::explode( "\n", $text );
1066 $out = '';
1067 $td_history = []; # Is currently a td tag open?
1068 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1069 $tr_history = []; # Is currently a tr tag open?
1070 $tr_attributes = []; # history of tr attributes
1071 $has_opened_tr = []; # Did this table open a <tr> element?
1072 $indent_level = 0; # indent level of the table
1073
1074 foreach ( $lines as $outLine ) {
1075 $line = trim( $outLine );
1076
1077 if ( $line === '' ) { # empty line, go to next line
1078 $out .= $outLine . "\n";
1079 continue;
1080 }
1081
1082 $first_character = $line[0];
1083 $first_two = substr( $line, 0, 2 );
1084 $matches = [];
1085
1086 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1087 # First check if we are starting a new table
1088 $indent_level = strlen( $matches[1] );
1089
1090 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1091 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1092
1093 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1094 array_push( $td_history, false );
1095 array_push( $last_tag_history, '' );
1096 array_push( $tr_history, false );
1097 array_push( $tr_attributes, '' );
1098 array_push( $has_opened_tr, false );
1099 } elseif ( count( $td_history ) == 0 ) {
1100 # Don't do any of the following
1101 $out .= $outLine . "\n";
1102 continue;
1103 } elseif ( $first_two === '|}' ) {
1104 # We are ending a table
1105 $line = '</table>' . substr( $line, 2 );
1106 $last_tag = array_pop( $last_tag_history );
1107
1108 if ( !array_pop( $has_opened_tr ) ) {
1109 $line = "<tr><td></td></tr>{$line}";
1110 }
1111
1112 if ( array_pop( $tr_history ) ) {
1113 $line = "</tr>{$line}";
1114 }
1115
1116 if ( array_pop( $td_history ) ) {
1117 $line = "</{$last_tag}>{$line}";
1118 }
1119 array_pop( $tr_attributes );
1120 if ( $indent_level > 0 ) {
1121 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1122 } else {
1123 $outLine = $line;
1124 }
1125 } elseif ( $first_two === '|-' ) {
1126 # Now we have a table row
1127 $line = preg_replace( '#^\|-+#', '', $line );
1128
1129 # Whats after the tag is now only attributes
1130 $attributes = $this->mStripState->unstripBoth( $line );
1131 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1132 array_pop( $tr_attributes );
1133 array_push( $tr_attributes, $attributes );
1134
1135 $line = '';
1136 $last_tag = array_pop( $last_tag_history );
1137 array_pop( $has_opened_tr );
1138 array_push( $has_opened_tr, true );
1139
1140 if ( array_pop( $tr_history ) ) {
1141 $line = '</tr>';
1142 }
1143
1144 if ( array_pop( $td_history ) ) {
1145 $line = "</{$last_tag}>{$line}";
1146 }
1147
1148 $outLine = $line;
1149 array_push( $tr_history, false );
1150 array_push( $td_history, false );
1151 array_push( $last_tag_history, '' );
1152 } elseif ( $first_character === '|'
1153 || $first_character === '!'
1154 || $first_two === '|+'
1155 ) {
1156 # This might be cell elements, td, th or captions
1157 if ( $first_two === '|+' ) {
1158 $first_character = '+';
1159 $line = substr( $line, 2 );
1160 } else {
1161 $line = substr( $line, 1 );
1162 }
1163
1164 // Implies both are valid for table headings.
1165 if ( $first_character === '!' ) {
1166 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1167 }
1168
1169 # Split up multiple cells on the same line.
1170 # FIXME : This can result in improper nesting of tags processed
1171 # by earlier parser steps.
1172 $cells = explode( '||', $line );
1173
1174 $outLine = '';
1175
1176 # Loop through each table cell
1177 foreach ( $cells as $cell ) {
1178 $previous = '';
1179 if ( $first_character !== '+' ) {
1180 $tr_after = array_pop( $tr_attributes );
1181 if ( !array_pop( $tr_history ) ) {
1182 $previous = "<tr{$tr_after}>\n";
1183 }
1184 array_push( $tr_history, true );
1185 array_push( $tr_attributes, '' );
1186 array_pop( $has_opened_tr );
1187 array_push( $has_opened_tr, true );
1188 }
1189
1190 $last_tag = array_pop( $last_tag_history );
1191
1192 if ( array_pop( $td_history ) ) {
1193 $previous = "</{$last_tag}>\n{$previous}";
1194 }
1195
1196 if ( $first_character === '|' ) {
1197 $last_tag = 'td';
1198 } elseif ( $first_character === '!' ) {
1199 $last_tag = 'th';
1200 } elseif ( $first_character === '+' ) {
1201 $last_tag = 'caption';
1202 } else {
1203 $last_tag = '';
1204 }
1205
1206 array_push( $last_tag_history, $last_tag );
1207
1208 # A cell could contain both parameters and data
1209 $cell_data = explode( '|', $cell, 2 );
1210
1211 # T2553: Note that a '|' inside an invalid link should not
1212 # be mistaken as delimiting cell parameters
1213 # Bug T153140: Neither should language converter markup.
1214 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1215 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1216 } elseif ( count( $cell_data ) == 1 ) {
1217 // Whitespace in cells is trimmed
1218 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1219 } else {
1220 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1221 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1222 // Whitespace in cells is trimmed
1223 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1224 }
1225
1226 $outLine .= $cell;
1227 array_push( $td_history, true );
1228 }
1229 }
1230 $out .= $outLine . "\n";
1231 }
1232
1233 # Closing open td, tr && table
1234 while ( count( $td_history ) > 0 ) {
1235 if ( array_pop( $td_history ) ) {
1236 $out .= "</td>\n";
1237 }
1238 if ( array_pop( $tr_history ) ) {
1239 $out .= "</tr>\n";
1240 }
1241 if ( !array_pop( $has_opened_tr ) ) {
1242 $out .= "<tr><td></td></tr>\n";
1243 }
1244
1245 $out .= "</table>\n";
1246 }
1247
1248 # Remove trailing line-ending (b/c)
1249 if ( substr( $out, -1 ) === "\n" ) {
1250 $out = substr( $out, 0, -1 );
1251 }
1252
1253 # special case: don't return empty table
1254 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1255 $out = '';
1256 }
1257
1258 return $out;
1259 }
1260
1261 /**
1262 * Helper function for parse() that transforms wiki markup into half-parsed
1263 * HTML. Only called for $mOutputType == self::OT_HTML.
1264 *
1265 * @private
1266 *
1267 * @param string $text The text to parse
1268 * @param bool $isMain Whether this is being called from the main parse() function
1269 * @param PPFrame|bool $frame A pre-processor frame
1270 *
1271 * @return string
1272 */
1273 public function internalParse( $text, $isMain = true, $frame = false ) {
1274 $origText = $text;
1275
1276 // Avoid PHP 7.1 warning from passing $this by reference
1277 $parser = $this;
1278
1279 # Hook to suspend the parser in this state
1280 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1281 return $text;
1282 }
1283
1284 # if $frame is provided, then use $frame for replacing any variables
1285 if ( $frame ) {
1286 # use frame depth to infer how include/noinclude tags should be handled
1287 # depth=0 means this is the top-level document; otherwise it's an included document
1288 if ( !$frame->depth ) {
1289 $flag = 0;
1290 } else {
1291 $flag = self::PTD_FOR_INCLUSION;
1292 }
1293 $dom = $this->preprocessToDom( $text, $flag );
1294 $text = $frame->expand( $dom );
1295 } else {
1296 # if $frame is not provided, then use old-style replaceVariables
1297 $text = $this->replaceVariables( $text );
1298 }
1299
1300 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1301 $text = Sanitizer::removeHTMLtags(
1302 $text,
1303 [ $this, 'attributeStripCallback' ],
1304 false,
1305 array_keys( $this->mTransparentTagHooks ),
1306 [],
1307 [ $this, 'addTrackingCategory' ]
1308 );
1309 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1310
1311 # Tables need to come after variable replacement for things to work
1312 # properly; putting them before other transformations should keep
1313 # exciting things like link expansions from showing up in surprising
1314 # places.
1315 $text = $this->doTableStuff( $text );
1316
1317 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1318
1319 $text = $this->doDoubleUnderscore( $text );
1320
1321 $text = $this->doHeadings( $text );
1322 $text = $this->replaceInternalLinks( $text );
1323 $text = $this->doAllQuotes( $text );
1324 $text = $this->replaceExternalLinks( $text );
1325
1326 # replaceInternalLinks may sometimes leave behind
1327 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1328 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1329
1330 $text = $this->doMagicLinks( $text );
1331 $text = $this->formatHeadings( $text, $origText, $isMain );
1332
1333 return $text;
1334 }
1335
1336 /**
1337 * Helper function for parse() that transforms half-parsed HTML into fully
1338 * parsed HTML.
1339 *
1340 * @param string $text
1341 * @param bool $isMain
1342 * @param bool $linestart
1343 * @return string
1344 */
1345 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1346 $text = $this->mStripState->unstripGeneral( $text );
1347
1348 // Avoid PHP 7.1 warning from passing $this by reference
1349 $parser = $this;
1350
1351 if ( $isMain ) {
1352 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1353 }
1354
1355 # Clean up special characters, only run once, next-to-last before doBlockLevels
1356 $fixtags = [
1357 # French spaces, last one Guillemet-left
1358 # only if there is something before the space
1359 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1360 # french spaces, Guillemet-right
1361 '/(\\302\\253) /' => '\\1&#160;',
1362 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1363 ];
1364 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1365
1366 $text = $this->doBlockLevels( $text, $linestart );
1367
1368 $this->replaceLinkHolders( $text );
1369
1370 /**
1371 * The input doesn't get language converted if
1372 * a) It's disabled
1373 * b) Content isn't converted
1374 * c) It's a conversion table
1375 * d) it is an interface message (which is in the user language)
1376 */
1377 if ( !( $this->mOptions->getDisableContentConversion()
1378 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1379 ) {
1380 if ( !$this->mOptions->getInterfaceMessage() ) {
1381 # The position of the convert() call should not be changed. it
1382 # assumes that the links are all replaced and the only thing left
1383 # is the <nowiki> mark.
1384 $text = $this->getTargetLanguage()->convert( $text );
1385 }
1386 }
1387
1388 $text = $this->mStripState->unstripNoWiki( $text );
1389
1390 if ( $isMain ) {
1391 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1392 }
1393
1394 $text = $this->replaceTransparentTags( $text );
1395 $text = $this->mStripState->unstripGeneral( $text );
1396
1397 $text = Sanitizer::normalizeCharReferences( $text );
1398
1399 if ( MWTidy::isEnabled() ) {
1400 if ( $this->mOptions->getTidy() ) {
1401 $text = MWTidy::tidy( $text );
1402 }
1403 } else {
1404 # attempt to sanitize at least some nesting problems
1405 # (T4702 and quite a few others)
1406 $tidyregs = [
1407 # ''Something [http://www.cool.com cool''] -->
1408 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1409 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1410 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1411 # fix up an anchor inside another anchor, only
1412 # at least for a single single nested link (T5695)
1413 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1414 '\\1\\2</a>\\3</a>\\1\\4</a>',
1415 # fix div inside inline elements- doBlockLevels won't wrap a line which
1416 # contains a div, so fix it up here; replace
1417 # div with escaped text
1418 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1419 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1420 # remove empty italic or bold tag pairs, some
1421 # introduced by rules above
1422 '/<([bi])><\/\\1>/' => '',
1423 ];
1424
1425 $text = preg_replace(
1426 array_keys( $tidyregs ),
1427 array_values( $tidyregs ),
1428 $text );
1429 }
1430
1431 if ( $isMain ) {
1432 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1433 }
1434
1435 return $text;
1436 }
1437
1438 /**
1439 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1440 * magic external links.
1441 *
1442 * DML
1443 * @private
1444 *
1445 * @param string $text
1446 *
1447 * @return string
1448 */
1449 public function doMagicLinks( $text ) {
1450 $prots = wfUrlProtocolsWithoutProtRel();
1451 $urlChar = self::EXT_LINK_URL_CLASS;
1452 $addr = self::EXT_LINK_ADDR;
1453 $space = self::SPACE_NOT_NL; # non-newline space
1454 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1455 $spaces = "$space++"; # possessive match of 1 or more spaces
1456 $text = preg_replace_callback(
1457 '!(?: # Start cases
1458 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1459 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1460 (\b # m[3]: Free external links
1461 (?i:$prots)
1462 ($addr$urlChar*) # m[4]: Post-protocol path
1463 ) |
1464 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1465 ([0-9]+)\b |
1466 \bISBN $spaces ( # m[6]: ISBN, capture number
1467 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1468 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1469 [0-9Xx] # check digit
1470 )\b
1471 )!xu", [ $this, 'magicLinkCallback' ], $text );
1472 return $text;
1473 }
1474
1475 /**
1476 * @throws MWException
1477 * @param array $m
1478 * @return string HTML
1479 */
1480 public function magicLinkCallback( $m ) {
1481 if ( isset( $m[1] ) && $m[1] !== '' ) {
1482 # Skip anchor
1483 return $m[0];
1484 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1485 # Skip HTML element
1486 return $m[0];
1487 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1488 # Free external link
1489 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1490 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1491 # RFC or PMID
1492 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1493 if ( !$this->mOptions->getMagicRFCLinks() ) {
1494 return $m[0];
1495 }
1496 $keyword = 'RFC';
1497 $urlmsg = 'rfcurl';
1498 $cssClass = 'mw-magiclink-rfc';
1499 $trackingCat = 'magiclink-tracking-rfc';
1500 $id = $m[5];
1501 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1502 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1503 return $m[0];
1504 }
1505 $keyword = 'PMID';
1506 $urlmsg = 'pubmedurl';
1507 $cssClass = 'mw-magiclink-pmid';
1508 $trackingCat = 'magiclink-tracking-pmid';
1509 $id = $m[5];
1510 } else {
1511 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1512 substr( $m[0], 0, 20 ) . '"' );
1513 }
1514 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1515 $this->addTrackingCategory( $trackingCat );
1516 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1517 } elseif ( isset( $m[6] ) && $m[6] !== ''
1518 && $this->mOptions->getMagicISBNLinks()
1519 ) {
1520 # ISBN
1521 $isbn = $m[6];
1522 $space = self::SPACE_NOT_NL; # non-newline space
1523 $isbn = preg_replace( "/$space/", ' ', $isbn );
1524 $num = strtr( $isbn, [
1525 '-' => '',
1526 ' ' => '',
1527 'x' => 'X',
1528 ] );
1529 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1530 return $this->getLinkRenderer()->makeKnownLink(
1531 SpecialPage::getTitleFor( 'Booksources', $num ),
1532 "ISBN $isbn",
1533 [
1534 'class' => 'internal mw-magiclink-isbn',
1535 'title' => false // suppress title attribute
1536 ]
1537 );
1538 } else {
1539 return $m[0];
1540 }
1541 }
1542
1543 /**
1544 * Make a free external link, given a user-supplied URL
1545 *
1546 * @param string $url
1547 * @param int $numPostProto
1548 * The number of characters after the protocol.
1549 * @return string HTML
1550 * @private
1551 */
1552 public function makeFreeExternalLink( $url, $numPostProto ) {
1553 $trail = '';
1554
1555 # The characters '<' and '>' (which were escaped by
1556 # removeHTMLtags()) should not be included in
1557 # URLs, per RFC 2396.
1558 # Make &nbsp; terminate a URL as well (bug T84937)
1559 $m2 = [];
1560 if ( preg_match(
1561 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1562 $url,
1563 $m2,
1564 PREG_OFFSET_CAPTURE
1565 ) ) {
1566 $trail = substr( $url, $m2[0][1] ) . $trail;
1567 $url = substr( $url, 0, $m2[0][1] );
1568 }
1569
1570 # Move trailing punctuation to $trail
1571 $sep = ',;\.:!?';
1572 # If there is no left bracket, then consider right brackets fair game too
1573 if ( strpos( $url, '(' ) === false ) {
1574 $sep .= ')';
1575 }
1576
1577 $urlRev = strrev( $url );
1578 $numSepChars = strspn( $urlRev, $sep );
1579 # Don't break a trailing HTML entity by moving the ; into $trail
1580 # This is in hot code, so use substr_compare to avoid having to
1581 # create a new string object for the comparison
1582 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1583 # more optimization: instead of running preg_match with a $
1584 # anchor, which can be slow, do the match on the reversed
1585 # string starting at the desired offset.
1586 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1587 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1588 $numSepChars--;
1589 }
1590 }
1591 if ( $numSepChars ) {
1592 $trail = substr( $url, -$numSepChars ) . $trail;
1593 $url = substr( $url, 0, -$numSepChars );
1594 }
1595
1596 # Verify that we still have a real URL after trail removal, and
1597 # not just lone protocol
1598 if ( strlen( $trail ) >= $numPostProto ) {
1599 return $url . $trail;
1600 }
1601
1602 $url = Sanitizer::cleanUrl( $url );
1603
1604 # Is this an external image?
1605 $text = $this->maybeMakeExternalImage( $url );
1606 if ( $text === false ) {
1607 # Not an image, make a link
1608 $text = Linker::makeExternalLink( $url,
1609 $this->getTargetLanguage()->markNoConversion( $url, true ),
1610 true, 'free',
1611 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1612 # Register it in the output object...
1613 $this->mOutput->addExternalLink( $url );
1614 }
1615 return $text . $trail;
1616 }
1617
1618 /**
1619 * Parse headers and return html
1620 *
1621 * @private
1622 *
1623 * @param string $text
1624 *
1625 * @return string
1626 */
1627 public function doHeadings( $text ) {
1628 for ( $i = 6; $i >= 1; --$i ) {
1629 $h = str_repeat( '=', $i );
1630 // Trim non-newline whitespace from headings
1631 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1632 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1633 }
1634 return $text;
1635 }
1636
1637 /**
1638 * Replace single quotes with HTML markup
1639 * @private
1640 *
1641 * @param string $text
1642 *
1643 * @return string The altered text
1644 */
1645 public function doAllQuotes( $text ) {
1646 $outtext = '';
1647 $lines = StringUtils::explode( "\n", $text );
1648 foreach ( $lines as $line ) {
1649 $outtext .= $this->doQuotes( $line ) . "\n";
1650 }
1651 $outtext = substr( $outtext, 0, -1 );
1652 return $outtext;
1653 }
1654
1655 /**
1656 * Helper function for doAllQuotes()
1657 *
1658 * @param string $text
1659 *
1660 * @return string
1661 */
1662 public function doQuotes( $text ) {
1663 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1664 $countarr = count( $arr );
1665 if ( $countarr == 1 ) {
1666 return $text;
1667 }
1668
1669 // First, do some preliminary work. This may shift some apostrophes from
1670 // being mark-up to being text. It also counts the number of occurrences
1671 // of bold and italics mark-ups.
1672 $numbold = 0;
1673 $numitalics = 0;
1674 for ( $i = 1; $i < $countarr; $i += 2 ) {
1675 $thislen = strlen( $arr[$i] );
1676 // If there are ever four apostrophes, assume the first is supposed to
1677 // be text, and the remaining three constitute mark-up for bold text.
1678 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1679 if ( $thislen == 4 ) {
1680 $arr[$i - 1] .= "'";
1681 $arr[$i] = "'''";
1682 $thislen = 3;
1683 } elseif ( $thislen > 5 ) {
1684 // If there are more than 5 apostrophes in a row, assume they're all
1685 // text except for the last 5.
1686 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1687 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1688 $arr[$i] = "'''''";
1689 $thislen = 5;
1690 }
1691 // Count the number of occurrences of bold and italics mark-ups.
1692 if ( $thislen == 2 ) {
1693 $numitalics++;
1694 } elseif ( $thislen == 3 ) {
1695 $numbold++;
1696 } elseif ( $thislen == 5 ) {
1697 $numitalics++;
1698 $numbold++;
1699 }
1700 }
1701
1702 // If there is an odd number of both bold and italics, it is likely
1703 // that one of the bold ones was meant to be an apostrophe followed
1704 // by italics. Which one we cannot know for certain, but it is more
1705 // likely to be one that has a single-letter word before it.
1706 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1707 $firstsingleletterword = -1;
1708 $firstmultiletterword = -1;
1709 $firstspace = -1;
1710 for ( $i = 1; $i < $countarr; $i += 2 ) {
1711 if ( strlen( $arr[$i] ) == 3 ) {
1712 $x1 = substr( $arr[$i - 1], -1 );
1713 $x2 = substr( $arr[$i - 1], -2, 1 );
1714 if ( $x1 === ' ' ) {
1715 if ( $firstspace == -1 ) {
1716 $firstspace = $i;
1717 }
1718 } elseif ( $x2 === ' ' ) {
1719 $firstsingleletterword = $i;
1720 // if $firstsingleletterword is set, we don't
1721 // look at the other options, so we can bail early.
1722 break;
1723 } else {
1724 if ( $firstmultiletterword == -1 ) {
1725 $firstmultiletterword = $i;
1726 }
1727 }
1728 }
1729 }
1730
1731 // If there is a single-letter word, use it!
1732 if ( $firstsingleletterword > -1 ) {
1733 $arr[$firstsingleletterword] = "''";
1734 $arr[$firstsingleletterword - 1] .= "'";
1735 } elseif ( $firstmultiletterword > -1 ) {
1736 // If not, but there's a multi-letter word, use that one.
1737 $arr[$firstmultiletterword] = "''";
1738 $arr[$firstmultiletterword - 1] .= "'";
1739 } elseif ( $firstspace > -1 ) {
1740 // ... otherwise use the first one that has neither.
1741 // (notice that it is possible for all three to be -1 if, for example,
1742 // there is only one pentuple-apostrophe in the line)
1743 $arr[$firstspace] = "''";
1744 $arr[$firstspace - 1] .= "'";
1745 }
1746 }
1747
1748 // Now let's actually convert our apostrophic mush to HTML!
1749 $output = '';
1750 $buffer = '';
1751 $state = '';
1752 $i = 0;
1753 foreach ( $arr as $r ) {
1754 if ( ( $i % 2 ) == 0 ) {
1755 if ( $state === 'both' ) {
1756 $buffer .= $r;
1757 } else {
1758 $output .= $r;
1759 }
1760 } else {
1761 $thislen = strlen( $r );
1762 if ( $thislen == 2 ) {
1763 if ( $state === 'i' ) {
1764 $output .= '</i>';
1765 $state = '';
1766 } elseif ( $state === 'bi' ) {
1767 $output .= '</i>';
1768 $state = 'b';
1769 } elseif ( $state === 'ib' ) {
1770 $output .= '</b></i><b>';
1771 $state = 'b';
1772 } elseif ( $state === 'both' ) {
1773 $output .= '<b><i>' . $buffer . '</i>';
1774 $state = 'b';
1775 } else { // $state can be 'b' or ''
1776 $output .= '<i>';
1777 $state .= 'i';
1778 }
1779 } elseif ( $thislen == 3 ) {
1780 if ( $state === 'b' ) {
1781 $output .= '</b>';
1782 $state = '';
1783 } elseif ( $state === 'bi' ) {
1784 $output .= '</i></b><i>';
1785 $state = 'i';
1786 } elseif ( $state === 'ib' ) {
1787 $output .= '</b>';
1788 $state = 'i';
1789 } elseif ( $state === 'both' ) {
1790 $output .= '<i><b>' . $buffer . '</b>';
1791 $state = 'i';
1792 } else { // $state can be 'i' or ''
1793 $output .= '<b>';
1794 $state .= 'b';
1795 }
1796 } elseif ( $thislen == 5 ) {
1797 if ( $state === 'b' ) {
1798 $output .= '</b><i>';
1799 $state = 'i';
1800 } elseif ( $state === 'i' ) {
1801 $output .= '</i><b>';
1802 $state = 'b';
1803 } elseif ( $state === 'bi' ) {
1804 $output .= '</i></b>';
1805 $state = '';
1806 } elseif ( $state === 'ib' ) {
1807 $output .= '</b></i>';
1808 $state = '';
1809 } elseif ( $state === 'both' ) {
1810 $output .= '<i><b>' . $buffer . '</b></i>';
1811 $state = '';
1812 } else { // ($state == '')
1813 $buffer = '';
1814 $state = 'both';
1815 }
1816 }
1817 }
1818 $i++;
1819 }
1820 // Now close all remaining tags. Notice that the order is important.
1821 if ( $state === 'b' || $state === 'ib' ) {
1822 $output .= '</b>';
1823 }
1824 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1825 $output .= '</i>';
1826 }
1827 if ( $state === 'bi' ) {
1828 $output .= '</b>';
1829 }
1830 // There might be lonely ''''', so make sure we have a buffer
1831 if ( $state === 'both' && $buffer ) {
1832 $output .= '<b><i>' . $buffer . '</i></b>';
1833 }
1834 return $output;
1835 }
1836
1837 /**
1838 * Replace external links (REL)
1839 *
1840 * Note: this is all very hackish and the order of execution matters a lot.
1841 * Make sure to run tests/parser/parserTests.php if you change this code.
1842 *
1843 * @private
1844 *
1845 * @param string $text
1846 *
1847 * @throws MWException
1848 * @return string
1849 */
1850 public function replaceExternalLinks( $text ) {
1851 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1852 if ( $bits === false ) {
1853 throw new MWException( "PCRE needs to be compiled with "
1854 . "--enable-unicode-properties in order for MediaWiki to function" );
1855 }
1856 $s = array_shift( $bits );
1857
1858 $i = 0;
1859 while ( $i < count( $bits ) ) {
1860 $url = $bits[$i++];
1861 $i++; // protocol
1862 $text = $bits[$i++];
1863 $trail = $bits[$i++];
1864
1865 # The characters '<' and '>' (which were escaped by
1866 # removeHTMLtags()) should not be included in
1867 # URLs, per RFC 2396.
1868 $m2 = [];
1869 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1870 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1871 $url = substr( $url, 0, $m2[0][1] );
1872 }
1873
1874 # If the link text is an image URL, replace it with an <img> tag
1875 # This happened by accident in the original parser, but some people used it extensively
1876 $img = $this->maybeMakeExternalImage( $text );
1877 if ( $img !== false ) {
1878 $text = $img;
1879 }
1880
1881 $dtrail = '';
1882
1883 # Set linktype for CSS
1884 $linktype = 'text';
1885
1886 # No link text, e.g. [http://domain.tld/some.link]
1887 if ( $text == '' ) {
1888 # Autonumber
1889 $langObj = $this->getTargetLanguage();
1890 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1891 $linktype = 'autonumber';
1892 } else {
1893 # Have link text, e.g. [http://domain.tld/some.link text]s
1894 # Check for trail
1895 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1896 }
1897
1898 $text = $this->getTargetLanguage()->markNoConversion( $text );
1899
1900 $url = Sanitizer::cleanUrl( $url );
1901
1902 # Use the encoded URL
1903 # This means that users can paste URLs directly into the text
1904 # Funny characters like ö aren't valid in URLs anyway
1905 # This was changed in August 2004
1906 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1907 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1908
1909 # Register link in the output object.
1910 $this->mOutput->addExternalLink( $url );
1911 }
1912
1913 return $s;
1914 }
1915
1916 /**
1917 * Get the rel attribute for a particular external link.
1918 *
1919 * @since 1.21
1920 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1921 * nofollow if appropriate
1922 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1923 * @return string|null Rel attribute for $url
1924 */
1925 public static function getExternalLinkRel( $url = false, $title = null ) {
1926 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1927 $ns = $title ? $title->getNamespace() : false;
1928 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1929 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1930 ) {
1931 return 'nofollow';
1932 }
1933 return null;
1934 }
1935
1936 /**
1937 * Get an associative array of additional HTML attributes appropriate for a
1938 * particular external link. This currently may include rel => nofollow
1939 * (depending on configuration, namespace, and the URL's domain) and/or a
1940 * target attribute (depending on configuration).
1941 *
1942 * @param string $url URL to extract the domain from for rel =>
1943 * nofollow if appropriate
1944 * @return array Associative array of HTML attributes
1945 */
1946 public function getExternalLinkAttribs( $url ) {
1947 $attribs = [];
1948 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1949
1950 $target = $this->mOptions->getExternalLinkTarget();
1951 if ( $target ) {
1952 $attribs['target'] = $target;
1953 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1954 // T133507. New windows can navigate parent cross-origin.
1955 // Including noreferrer due to lacking browser
1956 // support of noopener. Eventually noreferrer should be removed.
1957 if ( $rel !== '' ) {
1958 $rel .= ' ';
1959 }
1960 $rel .= 'noreferrer noopener';
1961 }
1962 }
1963 $attribs['rel'] = $rel;
1964 return $attribs;
1965 }
1966
1967 /**
1968 * Replace unusual escape codes in a URL with their equivalent characters
1969 *
1970 * This generally follows the syntax defined in RFC 3986, with special
1971 * consideration for HTTP query strings.
1972 *
1973 * @param string $url
1974 * @return string
1975 */
1976 public static function normalizeLinkUrl( $url ) {
1977 # First, make sure unsafe characters are encoded
1978 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1979 function ( $m ) {
1980 return rawurlencode( $m[0] );
1981 },
1982 $url
1983 );
1984
1985 $ret = '';
1986 $end = strlen( $url );
1987
1988 # Fragment part - 'fragment'
1989 $start = strpos( $url, '#' );
1990 if ( $start !== false && $start < $end ) {
1991 $ret = self::normalizeUrlComponent(
1992 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1993 $end = $start;
1994 }
1995
1996 # Query part - 'query' minus &=+;
1997 $start = strpos( $url, '?' );
1998 if ( $start !== false && $start < $end ) {
1999 $ret = self::normalizeUrlComponent(
2000 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2001 $end = $start;
2002 }
2003
2004 # Scheme and path part - 'pchar'
2005 # (we assume no userinfo or encoded colons in the host)
2006 $ret = self::normalizeUrlComponent(
2007 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2008
2009 return $ret;
2010 }
2011
2012 private static function normalizeUrlComponent( $component, $unsafe ) {
2013 $callback = function ( $matches ) use ( $unsafe ) {
2014 $char = urldecode( $matches[0] );
2015 $ord = ord( $char );
2016 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2017 # Unescape it
2018 return $char;
2019 } else {
2020 # Leave it escaped, but use uppercase for a-f
2021 return strtoupper( $matches[0] );
2022 }
2023 };
2024 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2025 }
2026
2027 /**
2028 * make an image if it's allowed, either through the global
2029 * option, through the exception, or through the on-wiki whitelist
2030 *
2031 * @param string $url
2032 *
2033 * @return string
2034 */
2035 private function maybeMakeExternalImage( $url ) {
2036 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2037 $imagesexception = !empty( $imagesfrom );
2038 $text = false;
2039 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2040 if ( $imagesexception && is_array( $imagesfrom ) ) {
2041 $imagematch = false;
2042 foreach ( $imagesfrom as $match ) {
2043 if ( strpos( $url, $match ) === 0 ) {
2044 $imagematch = true;
2045 break;
2046 }
2047 }
2048 } elseif ( $imagesexception ) {
2049 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2050 } else {
2051 $imagematch = false;
2052 }
2053
2054 if ( $this->mOptions->getAllowExternalImages()
2055 || ( $imagesexception && $imagematch )
2056 ) {
2057 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2058 # Image found
2059 $text = Linker::makeExternalImage( $url );
2060 }
2061 }
2062 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2063 && preg_match( self::EXT_IMAGE_REGEX, $url )
2064 ) {
2065 $whitelist = explode(
2066 "\n",
2067 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2068 );
2069
2070 foreach ( $whitelist as $entry ) {
2071 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2072 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2073 continue;
2074 }
2075 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2076 # Image matches a whitelist entry
2077 $text = Linker::makeExternalImage( $url );
2078 break;
2079 }
2080 }
2081 }
2082 return $text;
2083 }
2084
2085 /**
2086 * Process [[ ]] wikilinks
2087 *
2088 * @param string $s
2089 *
2090 * @return string Processed text
2091 *
2092 * @private
2093 */
2094 public function replaceInternalLinks( $s ) {
2095 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2096 return $s;
2097 }
2098
2099 /**
2100 * Process [[ ]] wikilinks (RIL)
2101 * @param string &$s
2102 * @throws MWException
2103 * @return LinkHolderArray
2104 *
2105 * @private
2106 */
2107 public function replaceInternalLinks2( &$s ) {
2108 global $wgExtraInterlanguageLinkPrefixes;
2109
2110 static $tc = false, $e1, $e1_img;
2111 # the % is needed to support urlencoded titles as well
2112 if ( !$tc ) {
2113 $tc = Title::legalChars() . '#%';
2114 # Match a link having the form [[namespace:link|alternate]]trail
2115 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2116 # Match cases where there is no "]]", which might still be images
2117 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2118 }
2119
2120 $holders = new LinkHolderArray( $this );
2121
2122 # split the entire text string on occurrences of [[
2123 $a = StringUtils::explode( '[[', ' ' . $s );
2124 # get the first element (all text up to first [[), and remove the space we added
2125 $s = $a->current();
2126 $a->next();
2127 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2128 $s = substr( $s, 1 );
2129
2130 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2131 $e2 = null;
2132 if ( $useLinkPrefixExtension ) {
2133 # Match the end of a line for a word that's not followed by whitespace,
2134 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2135 global $wgContLang;
2136 $charset = $wgContLang->linkPrefixCharset();
2137 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2138 }
2139
2140 if ( is_null( $this->mTitle ) ) {
2141 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2142 }
2143 $nottalk = !$this->mTitle->isTalkPage();
2144
2145 if ( $useLinkPrefixExtension ) {
2146 $m = [];
2147 if ( preg_match( $e2, $s, $m ) ) {
2148 $first_prefix = $m[2];
2149 } else {
2150 $first_prefix = false;
2151 }
2152 } else {
2153 $prefix = '';
2154 }
2155
2156 $useSubpages = $this->areSubpagesAllowed();
2157
2158 # Loop for each link
2159 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2160 # Check for excessive memory usage
2161 if ( $holders->isBig() ) {
2162 # Too big
2163 # Do the existence check, replace the link holders and clear the array
2164 $holders->replace( $s );
2165 $holders->clear();
2166 }
2167
2168 if ( $useLinkPrefixExtension ) {
2169 if ( preg_match( $e2, $s, $m ) ) {
2170 $prefix = $m[2];
2171 $s = $m[1];
2172 } else {
2173 $prefix = '';
2174 }
2175 # first link
2176 if ( $first_prefix ) {
2177 $prefix = $first_prefix;
2178 $first_prefix = false;
2179 }
2180 }
2181
2182 $might_be_img = false;
2183
2184 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2185 $text = $m[2];
2186 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2187 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2188 # the real problem is with the $e1 regex
2189 # See T1500.
2190 # Still some problems for cases where the ] is meant to be outside punctuation,
2191 # and no image is in sight. See T4095.
2192 if ( $text !== ''
2193 && substr( $m[3], 0, 1 ) === ']'
2194 && strpos( $text, '[' ) !== false
2195 ) {
2196 $text .= ']'; # so that replaceExternalLinks($text) works later
2197 $m[3] = substr( $m[3], 1 );
2198 }
2199 # fix up urlencoded title texts
2200 if ( strpos( $m[1], '%' ) !== false ) {
2201 # Should anchors '#' also be rejected?
2202 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2203 }
2204 $trail = $m[3];
2205 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2206 # Invalid, but might be an image with a link in its caption
2207 $might_be_img = true;
2208 $text = $m[2];
2209 if ( strpos( $m[1], '%' ) !== false ) {
2210 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2211 }
2212 $trail = "";
2213 } else { # Invalid form; output directly
2214 $s .= $prefix . '[[' . $line;
2215 continue;
2216 }
2217
2218 $origLink = ltrim( $m[1], ' ' );
2219
2220 # Don't allow internal links to pages containing
2221 # PROTO: where PROTO is a valid URL protocol; these
2222 # should be external links.
2223 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2224 $s .= $prefix . '[[' . $line;
2225 continue;
2226 }
2227
2228 # Make subpage if necessary
2229 if ( $useSubpages ) {
2230 $link = $this->maybeDoSubpageLink( $origLink, $text );
2231 } else {
2232 $link = $origLink;
2233 }
2234
2235 // \x7f isn't a default legal title char, so most likely strip
2236 // markers will force us into the "invalid form" path above. But,
2237 // just in case, let's assert that xmlish tags aren't valid in
2238 // the title position.
2239 $unstrip = $this->mStripState->killMarkers( $link );
2240 $noMarkers = ( $unstrip === $link );
2241
2242 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2243 if ( $nt === null ) {
2244 $s .= $prefix . '[[' . $line;
2245 continue;
2246 }
2247
2248 $ns = $nt->getNamespace();
2249 $iw = $nt->getInterwiki();
2250
2251 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2252
2253 if ( $might_be_img ) { # if this is actually an invalid link
2254 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2255 $found = false;
2256 while ( true ) {
2257 # look at the next 'line' to see if we can close it there
2258 $a->next();
2259 $next_line = $a->current();
2260 if ( $next_line === false || $next_line === null ) {
2261 break;
2262 }
2263 $m = explode( ']]', $next_line, 3 );
2264 if ( count( $m ) == 3 ) {
2265 # the first ]] closes the inner link, the second the image
2266 $found = true;
2267 $text .= "[[{$m[0]}]]{$m[1]}";
2268 $trail = $m[2];
2269 break;
2270 } elseif ( count( $m ) == 2 ) {
2271 # if there's exactly one ]] that's fine, we'll keep looking
2272 $text .= "[[{$m[0]}]]{$m[1]}";
2273 } else {
2274 # if $next_line is invalid too, we need look no further
2275 $text .= '[[' . $next_line;
2276 break;
2277 }
2278 }
2279 if ( !$found ) {
2280 # we couldn't find the end of this imageLink, so output it raw
2281 # but don't ignore what might be perfectly normal links in the text we've examined
2282 $holders->merge( $this->replaceInternalLinks2( $text ) );
2283 $s .= "{$prefix}[[$link|$text";
2284 # note: no $trail, because without an end, there *is* no trail
2285 continue;
2286 }
2287 } else { # it's not an image, so output it raw
2288 $s .= "{$prefix}[[$link|$text";
2289 # note: no $trail, because without an end, there *is* no trail
2290 continue;
2291 }
2292 }
2293
2294 $wasblank = ( $text == '' );
2295 if ( $wasblank ) {
2296 $text = $link;
2297 if ( !$noforce ) {
2298 # Strip off leading ':'
2299 $text = substr( $text, 1 );
2300 }
2301 } else {
2302 # T6598 madness. Handle the quotes only if they come from the alternate part
2303 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2304 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2305 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2306 $text = $this->doQuotes( $text );
2307 }
2308
2309 # Link not escaped by : , create the various objects
2310 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2311 # Interwikis
2312 if (
2313 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2314 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2315 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2316 )
2317 ) {
2318 # T26502: filter duplicates
2319 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2320 $this->mLangLinkLanguages[$iw] = true;
2321 $this->mOutput->addLanguageLink( $nt->getFullText() );
2322 }
2323
2324 /**
2325 * Strip the whitespace interwiki links produce, see T10897
2326 */
2327 $s = rtrim( $s . $prefix ) . $trail; # T175416
2328 continue;
2329 }
2330
2331 if ( $ns == NS_FILE ) {
2332 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2333 if ( $wasblank ) {
2334 # if no parameters were passed, $text
2335 # becomes something like "File:Foo.png",
2336 # which we don't want to pass on to the
2337 # image generator
2338 $text = '';
2339 } else {
2340 # recursively parse links inside the image caption
2341 # actually, this will parse them in any other parameters, too,
2342 # but it might be hard to fix that, and it doesn't matter ATM
2343 $text = $this->replaceExternalLinks( $text );
2344 $holders->merge( $this->replaceInternalLinks2( $text ) );
2345 }
2346 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2347 $s .= $prefix . $this->armorLinks(
2348 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2349 continue;
2350 }
2351 } elseif ( $ns == NS_CATEGORY ) {
2352 /**
2353 * Strip the whitespace Category links produce, see T2087
2354 */
2355 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2356
2357 if ( $wasblank ) {
2358 $sortkey = $this->getDefaultSort();
2359 } else {
2360 $sortkey = $text;
2361 }
2362 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2363 $sortkey = str_replace( "\n", '', $sortkey );
2364 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2365 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2366
2367 continue;
2368 }
2369 }
2370
2371 # Self-link checking. For some languages, variants of the title are checked in
2372 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2373 # for linking to a different variant.
2374 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2375 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2376 continue;
2377 }
2378
2379 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2380 # @todo FIXME: Should do batch file existence checks, see comment below
2381 if ( $ns == NS_MEDIA ) {
2382 # Give extensions a chance to select the file revision for us
2383 $options = [];
2384 $descQuery = false;
2385 Hooks::run( 'BeforeParserFetchFileAndTitle',
2386 [ $this, $nt, &$options, &$descQuery ] );
2387 # Fetch and register the file (file title may be different via hooks)
2388 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2389 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2390 $s .= $prefix . $this->armorLinks(
2391 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2392 continue;
2393 }
2394
2395 # Some titles, such as valid special pages or files in foreign repos, should
2396 # be shown as bluelinks even though they're not included in the page table
2397 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2398 # batch file existence checks for NS_FILE and NS_MEDIA
2399 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2400 $this->mOutput->addLink( $nt );
2401 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2402 } else {
2403 # Links will be added to the output link list after checking
2404 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2405 }
2406 }
2407 return $holders;
2408 }
2409
2410 /**
2411 * Render a forced-blue link inline; protect against double expansion of
2412 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2413 * Since this little disaster has to split off the trail text to avoid
2414 * breaking URLs in the following text without breaking trails on the
2415 * wiki links, it's been made into a horrible function.
2416 *
2417 * @param Title $nt
2418 * @param string $text
2419 * @param string $trail
2420 * @param string $prefix
2421 * @return string HTML-wikitext mix oh yuck
2422 */
2423 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2424 list( $inside, $trail ) = Linker::splitTrail( $trail );
2425
2426 if ( $text == '' ) {
2427 $text = htmlspecialchars( $nt->getPrefixedText() );
2428 }
2429
2430 $link = $this->getLinkRenderer()->makeKnownLink(
2431 $nt, new HtmlArmor( "$prefix$text$inside" )
2432 );
2433
2434 return $this->armorLinks( $link ) . $trail;
2435 }
2436
2437 /**
2438 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2439 * going to go through further parsing steps before inline URL expansion.
2440 *
2441 * Not needed quite as much as it used to be since free links are a bit
2442 * more sensible these days. But bracketed links are still an issue.
2443 *
2444 * @param string $text More-or-less HTML
2445 * @return string Less-or-more HTML with NOPARSE bits
2446 */
2447 public function armorLinks( $text ) {
2448 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2449 self::MARKER_PREFIX . "NOPARSE$1", $text );
2450 }
2451
2452 /**
2453 * Return true if subpage links should be expanded on this page.
2454 * @return bool
2455 */
2456 public function areSubpagesAllowed() {
2457 # Some namespaces don't allow subpages
2458 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2459 }
2460
2461 /**
2462 * Handle link to subpage if necessary
2463 *
2464 * @param string $target The source of the link
2465 * @param string &$text The link text, modified as necessary
2466 * @return string The full name of the link
2467 * @private
2468 */
2469 public function maybeDoSubpageLink( $target, &$text ) {
2470 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2471 }
2472
2473 /**
2474 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2475 *
2476 * @param string $text
2477 * @param bool $linestart Whether or not this is at the start of a line.
2478 * @private
2479 * @return string The lists rendered as HTML
2480 */
2481 public function doBlockLevels( $text, $linestart ) {
2482 return BlockLevelPass::doBlockLevels( $text, $linestart );
2483 }
2484
2485 /**
2486 * Return value of a magic variable (like PAGENAME)
2487 *
2488 * @private
2489 *
2490 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2491 * @param bool|PPFrame $frame
2492 *
2493 * @throws MWException
2494 * @return string
2495 */
2496 public function getVariableValue( $index, $frame = false ) {
2497 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2498 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2499
2500 if ( is_null( $this->mTitle ) ) {
2501 // If no title set, bad things are going to happen
2502 // later. Title should always be set since this
2503 // should only be called in the middle of a parse
2504 // operation (but the unit-tests do funky stuff)
2505 throw new MWException( __METHOD__ . ' Should only be '
2506 . ' called while parsing (no title set)' );
2507 }
2508
2509 // Avoid PHP 7.1 warning from passing $this by reference
2510 $parser = $this;
2511
2512 /**
2513 * Some of these require message or data lookups and can be
2514 * expensive to check many times.
2515 */
2516 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2517 if ( isset( $this->mVarCache[$index] ) ) {
2518 return $this->mVarCache[$index];
2519 }
2520 }
2521
2522 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2523 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2524
2525 $pageLang = $this->getFunctionLang();
2526
2527 switch ( $index ) {
2528 case '!':
2529 $value = '|';
2530 break;
2531 case 'currentmonth':
2532 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2533 break;
2534 case 'currentmonth1':
2535 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2536 break;
2537 case 'currentmonthname':
2538 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2539 break;
2540 case 'currentmonthnamegen':
2541 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2542 break;
2543 case 'currentmonthabbrev':
2544 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2545 break;
2546 case 'currentday':
2547 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2548 break;
2549 case 'currentday2':
2550 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2551 break;
2552 case 'localmonth':
2553 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2554 break;
2555 case 'localmonth1':
2556 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2557 break;
2558 case 'localmonthname':
2559 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2560 break;
2561 case 'localmonthnamegen':
2562 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2563 break;
2564 case 'localmonthabbrev':
2565 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2566 break;
2567 case 'localday':
2568 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2569 break;
2570 case 'localday2':
2571 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2572 break;
2573 case 'pagename':
2574 $value = wfEscapeWikiText( $this->mTitle->getText() );
2575 break;
2576 case 'pagenamee':
2577 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2578 break;
2579 case 'fullpagename':
2580 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2581 break;
2582 case 'fullpagenamee':
2583 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2584 break;
2585 case 'subpagename':
2586 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2587 break;
2588 case 'subpagenamee':
2589 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2590 break;
2591 case 'rootpagename':
2592 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2593 break;
2594 case 'rootpagenamee':
2595 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2596 ' ',
2597 '_',
2598 $this->mTitle->getRootText()
2599 ) ) );
2600 break;
2601 case 'basepagename':
2602 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2603 break;
2604 case 'basepagenamee':
2605 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2606 ' ',
2607 '_',
2608 $this->mTitle->getBaseText()
2609 ) ) );
2610 break;
2611 case 'talkpagename':
2612 if ( $this->mTitle->canHaveTalkPage() ) {
2613 $talkPage = $this->mTitle->getTalkPage();
2614 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2615 } else {
2616 $value = '';
2617 }
2618 break;
2619 case 'talkpagenamee':
2620 if ( $this->mTitle->canHaveTalkPage() ) {
2621 $talkPage = $this->mTitle->getTalkPage();
2622 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2623 } else {
2624 $value = '';
2625 }
2626 break;
2627 case 'subjectpagename':
2628 $subjPage = $this->mTitle->getSubjectPage();
2629 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2630 break;
2631 case 'subjectpagenamee':
2632 $subjPage = $this->mTitle->getSubjectPage();
2633 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2634 break;
2635 case 'pageid': // requested in T25427
2636 $pageid = $this->getTitle()->getArticleID();
2637 if ( $pageid == 0 ) {
2638 # 0 means the page doesn't exist in the database,
2639 # which means the user is previewing a new page.
2640 # The vary-revision flag must be set, because the magic word
2641 # will have a different value once the page is saved.
2642 $this->mOutput->setFlag( 'vary-revision' );
2643 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2644 }
2645 $value = $pageid ?: null;
2646 break;
2647 case 'revisionid':
2648 # Let the edit saving system know we should parse the page
2649 # *after* a revision ID has been assigned.
2650 $this->mOutput->setFlag( 'vary-revision-id' );
2651 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2652 $value = $this->mRevisionId;
2653 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2654 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2655 $this->mOutput->setSpeculativeRevIdUsed( $value );
2656 }
2657 break;
2658 case 'revisionday':
2659 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2660 break;
2661 case 'revisionday2':
2662 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2663 break;
2664 case 'revisionmonth':
2665 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2666 break;
2667 case 'revisionmonth1':
2668 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2669 break;
2670 case 'revisionyear':
2671 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2672 break;
2673 case 'revisiontimestamp':
2674 # Let the edit saving system know we should parse the page
2675 # *after* a revision ID has been assigned. This is for null edits.
2676 $this->mOutput->setFlag( 'vary-revision' );
2677 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2678 $value = $this->getRevisionTimestamp();
2679 break;
2680 case 'revisionuser':
2681 # Let the edit saving system know we should parse the page
2682 # *after* a revision ID has been assigned for null edits.
2683 $this->mOutput->setFlag( 'vary-user' );
2684 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2685 $value = $this->getRevisionUser();
2686 break;
2687 case 'revisionsize':
2688 $value = $this->getRevisionSize();
2689 break;
2690 case 'namespace':
2691 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2692 break;
2693 case 'namespacee':
2694 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2695 break;
2696 case 'namespacenumber':
2697 $value = $this->mTitle->getNamespace();
2698 break;
2699 case 'talkspace':
2700 $value = $this->mTitle->canHaveTalkPage()
2701 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2702 : '';
2703 break;
2704 case 'talkspacee':
2705 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2706 break;
2707 case 'subjectspace':
2708 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2709 break;
2710 case 'subjectspacee':
2711 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2712 break;
2713 case 'currentdayname':
2714 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2715 break;
2716 case 'currentyear':
2717 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2718 break;
2719 case 'currenttime':
2720 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2721 break;
2722 case 'currenthour':
2723 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2724 break;
2725 case 'currentweek':
2726 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2727 # int to remove the padding
2728 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2729 break;
2730 case 'currentdow':
2731 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2732 break;
2733 case 'localdayname':
2734 $value = $pageLang->getWeekdayName(
2735 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2736 );
2737 break;
2738 case 'localyear':
2739 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2740 break;
2741 case 'localtime':
2742 $value = $pageLang->time(
2743 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2744 false,
2745 false
2746 );
2747 break;
2748 case 'localhour':
2749 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2750 break;
2751 case 'localweek':
2752 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2753 # int to remove the padding
2754 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2755 break;
2756 case 'localdow':
2757 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2758 break;
2759 case 'numberofarticles':
2760 $value = $pageLang->formatNum( SiteStats::articles() );
2761 break;
2762 case 'numberoffiles':
2763 $value = $pageLang->formatNum( SiteStats::images() );
2764 break;
2765 case 'numberofusers':
2766 $value = $pageLang->formatNum( SiteStats::users() );
2767 break;
2768 case 'numberofactiveusers':
2769 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2770 break;
2771 case 'numberofpages':
2772 $value = $pageLang->formatNum( SiteStats::pages() );
2773 break;
2774 case 'numberofadmins':
2775 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2776 break;
2777 case 'numberofedits':
2778 $value = $pageLang->formatNum( SiteStats::edits() );
2779 break;
2780 case 'currenttimestamp':
2781 $value = wfTimestamp( TS_MW, $ts );
2782 break;
2783 case 'localtimestamp':
2784 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2785 break;
2786 case 'currentversion':
2787 $value = SpecialVersion::getVersion();
2788 break;
2789 case 'articlepath':
2790 return $wgArticlePath;
2791 case 'sitename':
2792 return $wgSitename;
2793 case 'server':
2794 return $wgServer;
2795 case 'servername':
2796 return $wgServerName;
2797 case 'scriptpath':
2798 return $wgScriptPath;
2799 case 'stylepath':
2800 return $wgStylePath;
2801 case 'directionmark':
2802 return $pageLang->getDirMark();
2803 case 'contentlanguage':
2804 global $wgLanguageCode;
2805 return $wgLanguageCode;
2806 case 'pagelanguage':
2807 $value = $pageLang->getCode();
2808 break;
2809 case 'cascadingsources':
2810 $value = CoreParserFunctions::cascadingsources( $this );
2811 break;
2812 default:
2813 $ret = null;
2814 Hooks::run(
2815 'ParserGetVariableValueSwitch',
2816 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2817 );
2818
2819 return $ret;
2820 }
2821
2822 if ( $index ) {
2823 $this->mVarCache[$index] = $value;
2824 }
2825
2826 return $value;
2827 }
2828
2829 /**
2830 * @param int $start
2831 * @param int $len
2832 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2833 * @param string $variable Parser variable name
2834 * @return string
2835 */
2836 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2837 global $wgContLang;
2838
2839 # Get the timezone-adjusted timestamp to be used for this revision
2840 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2841 # Possibly set vary-revision if there is not yet an associated revision
2842 if ( !$this->getRevisionObject() ) {
2843 # Get the timezone-adjusted timestamp $mtts seconds in the future
2844 $resThen = substr(
2845 $wgContLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2846 $start,
2847 $len
2848 );
2849
2850 if ( $resNow !== $resThen ) {
2851 # Let the edit saving system know we should parse the page
2852 # *after* a revision ID has been assigned. This is for null edits.
2853 $this->mOutput->setFlag( 'vary-revision' );
2854 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2855 }
2856 }
2857
2858 return $resNow;
2859 }
2860
2861 /**
2862 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2863 *
2864 * @private
2865 */
2866 public function initialiseVariables() {
2867 $variableIDs = MagicWord::getVariableIDs();
2868 $substIDs = MagicWord::getSubstIDs();
2869
2870 $this->mVariables = new MagicWordArray( $variableIDs );
2871 $this->mSubstWords = new MagicWordArray( $substIDs );
2872 }
2873
2874 /**
2875 * Preprocess some wikitext and return the document tree.
2876 * This is the ghost of replace_variables().
2877 *
2878 * @param string $text The text to parse
2879 * @param int $flags Bitwise combination of:
2880 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2881 * included. Default is to assume a direct page view.
2882 *
2883 * The generated DOM tree must depend only on the input text and the flags.
2884 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2885 *
2886 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2887 * change in the DOM tree for a given text, must be passed through the section identifier
2888 * in the section edit link and thus back to extractSections().
2889 *
2890 * The output of this function is currently only cached in process memory, but a persistent
2891 * cache may be implemented at a later date which takes further advantage of these strict
2892 * dependency requirements.
2893 *
2894 * @return PPNode
2895 */
2896 public function preprocessToDom( $text, $flags = 0 ) {
2897 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2898 return $dom;
2899 }
2900
2901 /**
2902 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2903 *
2904 * @param string $s
2905 *
2906 * @return array
2907 */
2908 public static function splitWhitespace( $s ) {
2909 $ltrimmed = ltrim( $s );
2910 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2911 $trimmed = rtrim( $ltrimmed );
2912 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2913 if ( $diff > 0 ) {
2914 $w2 = substr( $ltrimmed, -$diff );
2915 } else {
2916 $w2 = '';
2917 }
2918 return [ $w1, $trimmed, $w2 ];
2919 }
2920
2921 /**
2922 * Replace magic variables, templates, and template arguments
2923 * with the appropriate text. Templates are substituted recursively,
2924 * taking care to avoid infinite loops.
2925 *
2926 * Note that the substitution depends on value of $mOutputType:
2927 * self::OT_WIKI: only {{subst:}} templates
2928 * self::OT_PREPROCESS: templates but not extension tags
2929 * self::OT_HTML: all templates and extension tags
2930 *
2931 * @param string $text The text to transform
2932 * @param bool|PPFrame $frame Object describing the arguments passed to the
2933 * template. Arguments may also be provided as an associative array, as
2934 * was the usual case before MW1.12. Providing arguments this way may be
2935 * useful for extensions wishing to perform variable replacement
2936 * explicitly.
2937 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2938 * double-brace expansion.
2939 * @return string
2940 */
2941 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2942 # Is there any text? Also, Prevent too big inclusions!
2943 $textSize = strlen( $text );
2944 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2945 return $text;
2946 }
2947
2948 if ( $frame === false ) {
2949 $frame = $this->getPreprocessor()->newFrame();
2950 } elseif ( !( $frame instanceof PPFrame ) ) {
2951 wfDebug( __METHOD__ . " called using plain parameters instead of "
2952 . "a PPFrame instance. Creating custom frame.\n" );
2953 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2954 }
2955
2956 $dom = $this->preprocessToDom( $text );
2957 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2958 $text = $frame->expand( $dom, $flags );
2959
2960 return $text;
2961 }
2962
2963 /**
2964 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2965 *
2966 * @param array $args
2967 *
2968 * @return array
2969 */
2970 public static function createAssocArgs( $args ) {
2971 $assocArgs = [];
2972 $index = 1;
2973 foreach ( $args as $arg ) {
2974 $eqpos = strpos( $arg, '=' );
2975 if ( $eqpos === false ) {
2976 $assocArgs[$index++] = $arg;
2977 } else {
2978 $name = trim( substr( $arg, 0, $eqpos ) );
2979 $value = trim( substr( $arg, $eqpos + 1 ) );
2980 if ( $value === false ) {
2981 $value = '';
2982 }
2983 if ( $name !== false ) {
2984 $assocArgs[$name] = $value;
2985 }
2986 }
2987 }
2988
2989 return $assocArgs;
2990 }
2991
2992 /**
2993 * Warn the user when a parser limitation is reached
2994 * Will warn at most once the user per limitation type
2995 *
2996 * The results are shown during preview and run through the Parser (See EditPage.php)
2997 *
2998 * @param string $limitationType Should be one of:
2999 * 'expensive-parserfunction' (corresponding messages:
3000 * 'expensive-parserfunction-warning',
3001 * 'expensive-parserfunction-category')
3002 * 'post-expand-template-argument' (corresponding messages:
3003 * 'post-expand-template-argument-warning',
3004 * 'post-expand-template-argument-category')
3005 * 'post-expand-template-inclusion' (corresponding messages:
3006 * 'post-expand-template-inclusion-warning',
3007 * 'post-expand-template-inclusion-category')
3008 * 'node-count-exceeded' (corresponding messages:
3009 * 'node-count-exceeded-warning',
3010 * 'node-count-exceeded-category')
3011 * 'expansion-depth-exceeded' (corresponding messages:
3012 * 'expansion-depth-exceeded-warning',
3013 * 'expansion-depth-exceeded-category')
3014 * @param string|int|null $current Current value
3015 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3016 * exceeded, provide the values (optional)
3017 */
3018 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3019 # does no harm if $current and $max are present but are unnecessary for the message
3020 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3021 # only during preview, and that would split the parser cache unnecessarily.
3022 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3023 ->text();
3024 $this->mOutput->addWarning( $warning );
3025 $this->addTrackingCategory( "$limitationType-category" );
3026 }
3027
3028 /**
3029 * Return the text of a template, after recursively
3030 * replacing any variables or templates within the template.
3031 *
3032 * @param array $piece The parts of the template
3033 * $piece['title']: the title, i.e. the part before the |
3034 * $piece['parts']: the parameter array
3035 * $piece['lineStart']: whether the brace was at the start of a line
3036 * @param PPFrame $frame The current frame, contains template arguments
3037 * @throws Exception
3038 * @return string The text of the template
3039 */
3040 public function braceSubstitution( $piece, $frame ) {
3041 // Flags
3042
3043 // $text has been filled
3044 $found = false;
3045 // wiki markup in $text should be escaped
3046 $nowiki = false;
3047 // $text is HTML, armour it against wikitext transformation
3048 $isHTML = false;
3049 // Force interwiki transclusion to be done in raw mode not rendered
3050 $forceRawInterwiki = false;
3051 // $text is a DOM node needing expansion in a child frame
3052 $isChildObj = false;
3053 // $text is a DOM node needing expansion in the current frame
3054 $isLocalObj = false;
3055
3056 # Title object, where $text came from
3057 $title = false;
3058
3059 # $part1 is the bit before the first |, and must contain only title characters.
3060 # Various prefixes will be stripped from it later.
3061 $titleWithSpaces = $frame->expand( $piece['title'] );
3062 $part1 = trim( $titleWithSpaces );
3063 $titleText = false;
3064
3065 # Original title text preserved for various purposes
3066 $originalTitle = $part1;
3067
3068 # $args is a list of argument nodes, starting from index 0, not including $part1
3069 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3070 # below won't work b/c this $args isn't an object
3071 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3072
3073 $profileSection = null; // profile templates
3074
3075 # SUBST
3076 if ( !$found ) {
3077 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3078
3079 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3080 # Decide whether to expand template or keep wikitext as-is.
3081 if ( $this->ot['wiki'] ) {
3082 if ( $substMatch === false ) {
3083 $literal = true; # literal when in PST with no prefix
3084 } else {
3085 $literal = false; # expand when in PST with subst: or safesubst:
3086 }
3087 } else {
3088 if ( $substMatch == 'subst' ) {
3089 $literal = true; # literal when not in PST with plain subst:
3090 } else {
3091 $literal = false; # expand when not in PST with safesubst: or no prefix
3092 }
3093 }
3094 if ( $literal ) {
3095 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3096 $isLocalObj = true;
3097 $found = true;
3098 }
3099 }
3100
3101 # Variables
3102 if ( !$found && $args->getLength() == 0 ) {
3103 $id = $this->mVariables->matchStartToEnd( $part1 );
3104 if ( $id !== false ) {
3105 $text = $this->getVariableValue( $id, $frame );
3106 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3107 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3108 }
3109 $found = true;
3110 }
3111 }
3112
3113 # MSG, MSGNW and RAW
3114 if ( !$found ) {
3115 # Check for MSGNW:
3116 $mwMsgnw = MagicWord::get( 'msgnw' );
3117 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3118 $nowiki = true;
3119 } else {
3120 # Remove obsolete MSG:
3121 $mwMsg = MagicWord::get( 'msg' );
3122 $mwMsg->matchStartAndRemove( $part1 );
3123 }
3124
3125 # Check for RAW:
3126 $mwRaw = MagicWord::get( 'raw' );
3127 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3128 $forceRawInterwiki = true;
3129 }
3130 }
3131
3132 # Parser functions
3133 if ( !$found ) {
3134 $colonPos = strpos( $part1, ':' );
3135 if ( $colonPos !== false ) {
3136 $func = substr( $part1, 0, $colonPos );
3137 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3138 $argsLength = $args->getLength();
3139 for ( $i = 0; $i < $argsLength; $i++ ) {
3140 $funcArgs[] = $args->item( $i );
3141 }
3142
3143 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3144
3145 // Extract any forwarded flags
3146 if ( isset( $result['title'] ) ) {
3147 $title = $result['title'];
3148 }
3149 if ( isset( $result['found'] ) ) {
3150 $found = $result['found'];
3151 }
3152 if ( array_key_exists( 'text', $result ) ) {
3153 // a string or null
3154 $text = $result['text'];
3155 }
3156 if ( isset( $result['nowiki'] ) ) {
3157 $nowiki = $result['nowiki'];
3158 }
3159 if ( isset( $result['isHTML'] ) ) {
3160 $isHTML = $result['isHTML'];
3161 }
3162 if ( isset( $result['forceRawInterwiki'] ) ) {
3163 $forceRawInterwiki = $result['forceRawInterwiki'];
3164 }
3165 if ( isset( $result['isChildObj'] ) ) {
3166 $isChildObj = $result['isChildObj'];
3167 }
3168 if ( isset( $result['isLocalObj'] ) ) {
3169 $isLocalObj = $result['isLocalObj'];
3170 }
3171 }
3172 }
3173
3174 # Finish mangling title and then check for loops.
3175 # Set $title to a Title object and $titleText to the PDBK
3176 if ( !$found ) {
3177 $ns = NS_TEMPLATE;
3178 # Split the title into page and subpage
3179 $subpage = '';
3180 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3181 if ( $part1 !== $relative ) {
3182 $part1 = $relative;
3183 $ns = $this->mTitle->getNamespace();
3184 }
3185 $title = Title::newFromText( $part1, $ns );
3186 if ( $title ) {
3187 $titleText = $title->getPrefixedText();
3188 # Check for language variants if the template is not found
3189 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3190 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3191 }
3192 # Do recursion depth check
3193 $limit = $this->mOptions->getMaxTemplateDepth();
3194 if ( $frame->depth >= $limit ) {
3195 $found = true;
3196 $text = '<span class="error">'
3197 . wfMessage( 'parser-template-recursion-depth-warning' )
3198 ->numParams( $limit )->inContentLanguage()->text()
3199 . '</span>';
3200 }
3201 }
3202 }
3203
3204 # Load from database
3205 if ( !$found && $title ) {
3206 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3207 if ( !$title->isExternal() ) {
3208 if ( $title->isSpecialPage()
3209 && $this->mOptions->getAllowSpecialInclusion()
3210 && $this->ot['html']
3211 ) {
3212 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3213 // Pass the template arguments as URL parameters.
3214 // "uselang" will have no effect since the Language object
3215 // is forced to the one defined in ParserOptions.
3216 $pageArgs = [];
3217 $argsLength = $args->getLength();
3218 for ( $i = 0; $i < $argsLength; $i++ ) {
3219 $bits = $args->item( $i )->splitArg();
3220 if ( strval( $bits['index'] ) === '' ) {
3221 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3222 $value = trim( $frame->expand( $bits['value'] ) );
3223 $pageArgs[$name] = $value;
3224 }
3225 }
3226
3227 // Create a new context to execute the special page
3228 $context = new RequestContext;
3229 $context->setTitle( $title );
3230 $context->setRequest( new FauxRequest( $pageArgs ) );
3231 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3232 $context->setUser( $this->getUser() );
3233 } else {
3234 // If this page is cached, then we better not be per user.
3235 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3236 }
3237 $context->setLanguage( $this->mOptions->getUserLangObj() );
3238 $ret = SpecialPageFactory::capturePath(
3239 $title, $context, $this->getLinkRenderer() );
3240 if ( $ret ) {
3241 $text = $context->getOutput()->getHTML();
3242 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3243 $found = true;
3244 $isHTML = true;
3245 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3246 $this->mOutput->updateRuntimeAdaptiveExpiry(
3247 $specialPage->maxIncludeCacheTime()
3248 );
3249 }
3250 }
3251 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3252 $found = false; # access denied
3253 wfDebug( __METHOD__ . ": template inclusion denied for " .
3254 $title->getPrefixedDBkey() . "\n" );
3255 } else {
3256 list( $text, $title ) = $this->getTemplateDom( $title );
3257 if ( $text !== false ) {
3258 $found = true;
3259 $isChildObj = true;
3260 }
3261 }
3262
3263 # If the title is valid but undisplayable, make a link to it
3264 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3265 $text = "[[:$titleText]]";
3266 $found = true;
3267 }
3268 } elseif ( $title->isTrans() ) {
3269 # Interwiki transclusion
3270 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3271 $text = $this->interwikiTransclude( $title, 'render' );
3272 $isHTML = true;
3273 } else {
3274 $text = $this->interwikiTransclude( $title, 'raw' );
3275 # Preprocess it like a template
3276 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3277 $isChildObj = true;
3278 }
3279 $found = true;
3280 }
3281
3282 # Do infinite loop check
3283 # This has to be done after redirect resolution to avoid infinite loops via redirects
3284 if ( !$frame->loopCheck( $title ) ) {
3285 $found = true;
3286 $text = '<span class="error">'
3287 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3288 . '</span>';
3289 $this->addTrackingCategory( 'template-loop-category' );
3290 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3291 wfEscapeWikiText( $titleText ) )->text() );
3292 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3293 }
3294 }
3295
3296 # If we haven't found text to substitute by now, we're done
3297 # Recover the source wikitext and return it
3298 if ( !$found ) {
3299 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3300 if ( $profileSection ) {
3301 $this->mProfiler->scopedProfileOut( $profileSection );
3302 }
3303 return [ 'object' => $text ];
3304 }
3305
3306 # Expand DOM-style return values in a child frame
3307 if ( $isChildObj ) {
3308 # Clean up argument array
3309 $newFrame = $frame->newChild( $args, $title );
3310
3311 if ( $nowiki ) {
3312 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3313 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3314 # Expansion is eligible for the empty-frame cache
3315 $text = $newFrame->cachedExpand( $titleText, $text );
3316 } else {
3317 # Uncached expansion
3318 $text = $newFrame->expand( $text );
3319 }
3320 }
3321 if ( $isLocalObj && $nowiki ) {
3322 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3323 $isLocalObj = false;
3324 }
3325
3326 if ( $profileSection ) {
3327 $this->mProfiler->scopedProfileOut( $profileSection );
3328 }
3329
3330 # Replace raw HTML by a placeholder
3331 if ( $isHTML ) {
3332 $text = $this->insertStripItem( $text );
3333 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3334 # Escape nowiki-style return values
3335 $text = wfEscapeWikiText( $text );
3336 } elseif ( is_string( $text )
3337 && !$piece['lineStart']
3338 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3339 ) {
3340 # T2529: if the template begins with a table or block-level
3341 # element, it should be treated as beginning a new line.
3342 # This behavior is somewhat controversial.
3343 $text = "\n" . $text;
3344 }
3345
3346 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3347 # Error, oversize inclusion
3348 if ( $titleText !== false ) {
3349 # Make a working, properly escaped link if possible (T25588)
3350 $text = "[[:$titleText]]";
3351 } else {
3352 # This will probably not be a working link, but at least it may
3353 # provide some hint of where the problem is
3354 preg_replace( '/^:/', '', $originalTitle );
3355 $text = "[[:$originalTitle]]";
3356 }
3357 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3358 . 'post-expand include size too large -->' );
3359 $this->limitationWarn( 'post-expand-template-inclusion' );
3360 }
3361
3362 if ( $isLocalObj ) {
3363 $ret = [ 'object' => $text ];
3364 } else {
3365 $ret = [ 'text' => $text ];
3366 }
3367
3368 return $ret;
3369 }
3370
3371 /**
3372 * Call a parser function and return an array with text and flags.
3373 *
3374 * The returned array will always contain a boolean 'found', indicating
3375 * whether the parser function was found or not. It may also contain the
3376 * following:
3377 * text: string|object, resulting wikitext or PP DOM object
3378 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3379 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3380 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3381 * nowiki: bool, wiki markup in $text should be escaped
3382 *
3383 * @since 1.21
3384 * @param PPFrame $frame The current frame, contains template arguments
3385 * @param string $function Function name
3386 * @param array $args Arguments to the function
3387 * @throws MWException
3388 * @return array
3389 */
3390 public function callParserFunction( $frame, $function, array $args = [] ) {
3391 global $wgContLang;
3392
3393 # Case sensitive functions
3394 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3395 $function = $this->mFunctionSynonyms[1][$function];
3396 } else {
3397 # Case insensitive functions
3398 $function = $wgContLang->lc( $function );
3399 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3400 $function = $this->mFunctionSynonyms[0][$function];
3401 } else {
3402 return [ 'found' => false ];
3403 }
3404 }
3405
3406 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3407
3408 // Avoid PHP 7.1 warning from passing $this by reference
3409 $parser = $this;
3410
3411 $allArgs = [ &$parser ];
3412 if ( $flags & self::SFH_OBJECT_ARGS ) {
3413 # Convert arguments to PPNodes and collect for appending to $allArgs
3414 $funcArgs = [];
3415 foreach ( $args as $k => $v ) {
3416 if ( $v instanceof PPNode || $k === 0 ) {
3417 $funcArgs[] = $v;
3418 } else {
3419 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3420 }
3421 }
3422
3423 # Add a frame parameter, and pass the arguments as an array
3424 $allArgs[] = $frame;
3425 $allArgs[] = $funcArgs;
3426 } else {
3427 # Convert arguments to plain text and append to $allArgs
3428 foreach ( $args as $k => $v ) {
3429 if ( $v instanceof PPNode ) {
3430 $allArgs[] = trim( $frame->expand( $v ) );
3431 } elseif ( is_int( $k ) && $k >= 0 ) {
3432 $allArgs[] = trim( $v );
3433 } else {
3434 $allArgs[] = trim( "$k=$v" );
3435 }
3436 }
3437 }
3438
3439 $result = $callback( ...$allArgs );
3440
3441 # The interface for function hooks allows them to return a wikitext
3442 # string or an array containing the string and any flags. This mungs
3443 # things around to match what this method should return.
3444 if ( !is_array( $result ) ) {
3445 $result = [
3446 'found' => true,
3447 'text' => $result,
3448 ];
3449 } else {
3450 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3451 $result['text'] = $result[0];
3452 }
3453 unset( $result[0] );
3454 $result += [
3455 'found' => true,
3456 ];
3457 }
3458
3459 $noparse = true;
3460 $preprocessFlags = 0;
3461 if ( isset( $result['noparse'] ) ) {
3462 $noparse = $result['noparse'];
3463 }
3464 if ( isset( $result['preprocessFlags'] ) ) {
3465 $preprocessFlags = $result['preprocessFlags'];
3466 }
3467
3468 if ( !$noparse ) {
3469 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3470 $result['isChildObj'] = true;
3471 }
3472
3473 return $result;
3474 }
3475
3476 /**
3477 * Get the semi-parsed DOM representation of a template with a given title,
3478 * and its redirect destination title. Cached.
3479 *
3480 * @param Title $title
3481 *
3482 * @return array
3483 */
3484 public function getTemplateDom( $title ) {
3485 $cacheTitle = $title;
3486 $titleText = $title->getPrefixedDBkey();
3487
3488 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3489 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3490 $title = Title::makeTitle( $ns, $dbk );
3491 $titleText = $title->getPrefixedDBkey();
3492 }
3493 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3494 return [ $this->mTplDomCache[$titleText], $title ];
3495 }
3496
3497 # Cache miss, go to the database
3498 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3499
3500 if ( $text === false ) {
3501 $this->mTplDomCache[$titleText] = false;
3502 return [ false, $title ];
3503 }
3504
3505 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3506 $this->mTplDomCache[$titleText] = $dom;
3507
3508 if ( !$title->equals( $cacheTitle ) ) {
3509 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3510 [ $title->getNamespace(), $title->getDBkey() ];
3511 }
3512
3513 return [ $dom, $title ];
3514 }
3515
3516 /**
3517 * Fetch the current revision of a given title. Note that the revision
3518 * (and even the title) may not exist in the database, so everything
3519 * contributing to the output of the parser should use this method
3520 * where possible, rather than getting the revisions themselves. This
3521 * method also caches its results, so using it benefits performance.
3522 *
3523 * @since 1.24
3524 * @param Title $title
3525 * @return Revision
3526 */
3527 public function fetchCurrentRevisionOfTitle( $title ) {
3528 $cacheKey = $title->getPrefixedDBkey();
3529 if ( !$this->currentRevisionCache ) {
3530 $this->currentRevisionCache = new MapCacheLRU( 100 );
3531 }
3532 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3533 $this->currentRevisionCache->set( $cacheKey,
3534 // Defaults to Parser::statelessFetchRevision()
3535 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3536 );
3537 }
3538 return $this->currentRevisionCache->get( $cacheKey );
3539 }
3540
3541 /**
3542 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3543 * without passing them on to it.
3544 *
3545 * @since 1.24
3546 * @param Title $title
3547 * @param Parser|bool $parser
3548 * @return Revision|bool False if missing
3549 */
3550 public static function statelessFetchRevision( Title $title, $parser = false ) {
3551 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3552
3553 return $rev;
3554 }
3555
3556 /**
3557 * Fetch the unparsed text of a template and register a reference to it.
3558 * @param Title $title
3559 * @return array ( string or false, Title )
3560 */
3561 public function fetchTemplateAndTitle( $title ) {
3562 // Defaults to Parser::statelessFetchTemplate()
3563 $templateCb = $this->mOptions->getTemplateCallback();
3564 $stuff = call_user_func( $templateCb, $title, $this );
3565 // We use U+007F DELETE to distinguish strip markers from regular text.
3566 $text = $stuff['text'];
3567 if ( is_string( $stuff['text'] ) ) {
3568 $text = strtr( $text, "\x7f", "?" );
3569 }
3570 $finalTitle = $stuff['finalTitle'] ?? $title;
3571 if ( isset( $stuff['deps'] ) ) {
3572 foreach ( $stuff['deps'] as $dep ) {
3573 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3574 if ( $dep['title']->equals( $this->getTitle() ) ) {
3575 // If we transclude ourselves, the final result
3576 // will change based on the new version of the page
3577 $this->mOutput->setFlag( 'vary-revision' );
3578 }
3579 }
3580 }
3581 return [ $text, $finalTitle ];
3582 }
3583
3584 /**
3585 * Fetch the unparsed text of a template and register a reference to it.
3586 * @param Title $title
3587 * @return string|bool
3588 */
3589 public function fetchTemplate( $title ) {
3590 return $this->fetchTemplateAndTitle( $title )[0];
3591 }
3592
3593 /**
3594 * Static function to get a template
3595 * Can be overridden via ParserOptions::setTemplateCallback().
3596 *
3597 * @param Title $title
3598 * @param bool|Parser $parser
3599 *
3600 * @return array
3601 */
3602 public static function statelessFetchTemplate( $title, $parser = false ) {
3603 $text = $skip = false;
3604 $finalTitle = $title;
3605 $deps = [];
3606
3607 # Loop to fetch the article, with up to 1 redirect
3608 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
3609 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3610 # Give extensions a chance to select the revision instead
3611 $id = false; # Assume current
3612 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3613 [ $parser, $title, &$skip, &$id ] );
3614
3615 if ( $skip ) {
3616 $text = false;
3617 $deps[] = [
3618 'title' => $title,
3619 'page_id' => $title->getArticleID(),
3620 'rev_id' => null
3621 ];
3622 break;
3623 }
3624 # Get the revision
3625 if ( $id ) {
3626 $rev = Revision::newFromId( $id );
3627 } elseif ( $parser ) {
3628 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3629 } else {
3630 $rev = Revision::newFromTitle( $title );
3631 }
3632 $rev_id = $rev ? $rev->getId() : 0;
3633 # If there is no current revision, there is no page
3634 if ( $id === false && !$rev ) {
3635 $linkCache = LinkCache::singleton();
3636 $linkCache->addBadLinkObj( $title );
3637 }
3638
3639 $deps[] = [
3640 'title' => $title,
3641 'page_id' => $title->getArticleID(),
3642 'rev_id' => $rev_id ];
3643 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3644 # We fetched a rev from a different title; register it too...
3645 $deps[] = [
3646 'title' => $rev->getTitle(),
3647 'page_id' => $rev->getPage(),
3648 'rev_id' => $rev_id ];
3649 }
3650
3651 if ( $rev ) {
3652 $content = $rev->getContent();
3653 $text = $content ? $content->getWikitextForTransclusion() : null;
3654
3655 Hooks::run( 'ParserFetchTemplate',
3656 [ $parser, $title, $rev, &$text, &$deps ] );
3657
3658 if ( $text === false || $text === null ) {
3659 $text = false;
3660 break;
3661 }
3662 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3663 global $wgContLang;
3664 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3665 if ( !$message->exists() ) {
3666 $text = false;
3667 break;
3668 }
3669 $content = $message->content();
3670 $text = $message->plain();
3671 } else {
3672 break;
3673 }
3674 if ( !$content ) {
3675 break;
3676 }
3677 # Redirect?
3678 $finalTitle = $title;
3679 $title = $content->getRedirectTarget();
3680 }
3681 return [
3682 'text' => $text,
3683 'finalTitle' => $finalTitle,
3684 'deps' => $deps ];
3685 }
3686
3687 /**
3688 * Fetch a file and its title and register a reference to it.
3689 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3690 * @param Title $title
3691 * @param array $options Array of options to RepoGroup::findFile
3692 * @return File|bool
3693 * @deprecated since 1.32, use fetchFileAndTitle instead
3694 */
3695 public function fetchFile( $title, $options = [] ) {
3696 wfDeprecated( __METHOD__, '1.32' );
3697 return $this->fetchFileAndTitle( $title, $options )[0];
3698 }
3699
3700 /**
3701 * Fetch a file and its title and register a reference to it.
3702 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3703 * @param Title $title
3704 * @param array $options Array of options to RepoGroup::findFile
3705 * @return array ( File or false, Title of file )
3706 */
3707 public function fetchFileAndTitle( $title, $options = [] ) {
3708 $file = $this->fetchFileNoRegister( $title, $options );
3709
3710 $time = $file ? $file->getTimestamp() : false;
3711 $sha1 = $file ? $file->getSha1() : false;
3712 # Register the file as a dependency...
3713 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3714 if ( $file && !$title->equals( $file->getTitle() ) ) {
3715 # Update fetched file title
3716 $title = $file->getTitle();
3717 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3718 }
3719 return [ $file, $title ];
3720 }
3721
3722 /**
3723 * Helper function for fetchFileAndTitle.
3724 *
3725 * Also useful if you need to fetch a file but not use it yet,
3726 * for example to get the file's handler.
3727 *
3728 * @param Title $title
3729 * @param array $options Array of options to RepoGroup::findFile
3730 * @return File|bool
3731 */
3732 protected function fetchFileNoRegister( $title, $options = [] ) {
3733 if ( isset( $options['broken'] ) ) {
3734 $file = false; // broken thumbnail forced by hook
3735 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3736 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3737 } else { // get by (name,timestamp)
3738 $file = wfFindFile( $title, $options );
3739 }
3740 return $file;
3741 }
3742
3743 /**
3744 * Transclude an interwiki link.
3745 *
3746 * @param Title $title
3747 * @param string $action
3748 *
3749 * @return string
3750 */
3751 public function interwikiTransclude( $title, $action ) {
3752 global $wgEnableScaryTranscluding;
3753
3754 if ( !$wgEnableScaryTranscluding ) {
3755 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3756 }
3757
3758 $url = $title->getFullURL( [ 'action' => $action ] );
3759
3760 if ( strlen( $url ) > 255 ) {
3761 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3762 }
3763 return $this->fetchScaryTemplateMaybeFromCache( $url );
3764 }
3765
3766 /**
3767 * @param string $url
3768 * @return mixed|string
3769 */
3770 public function fetchScaryTemplateMaybeFromCache( $url ) {
3771 global $wgTranscludeCacheExpiry;
3772 $dbr = wfGetDB( DB_REPLICA );
3773 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3774 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3775 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3776 if ( $obj ) {
3777 return $obj->tc_contents;
3778 }
3779
3780 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3781 $status = $req->execute(); // Status object
3782 if ( $status->isOK() ) {
3783 $text = $req->getContent();
3784 } elseif ( $req->getStatus() != 200 ) {
3785 // Though we failed to fetch the content, this status is useless.
3786 return wfMessage( 'scarytranscludefailed-httpstatus' )
3787 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3788 } else {
3789 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3790 }
3791
3792 $dbw = wfGetDB( DB_MASTER );
3793 $dbw->replace( 'transcache', [ 'tc_url' ], [
3794 'tc_url' => $url,
3795 'tc_time' => $dbw->timestamp( time() ),
3796 'tc_contents' => $text
3797 ] );
3798 return $text;
3799 }
3800
3801 /**
3802 * Triple brace replacement -- used for template arguments
3803 * @private
3804 *
3805 * @param array $piece
3806 * @param PPFrame $frame
3807 *
3808 * @return array
3809 */
3810 public function argSubstitution( $piece, $frame ) {
3811 $error = false;
3812 $parts = $piece['parts'];
3813 $nameWithSpaces = $frame->expand( $piece['title'] );
3814 $argName = trim( $nameWithSpaces );
3815 $object = false;
3816 $text = $frame->getArgument( $argName );
3817 if ( $text === false && $parts->getLength() > 0
3818 && ( $this->ot['html']
3819 || $this->ot['pre']
3820 || ( $this->ot['wiki'] && $frame->isTemplate() )
3821 )
3822 ) {
3823 # No match in frame, use the supplied default
3824 $object = $parts->item( 0 )->getChildren();
3825 }
3826 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3827 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3828 $this->limitationWarn( 'post-expand-template-argument' );
3829 }
3830
3831 if ( $text === false && $object === false ) {
3832 # No match anywhere
3833 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3834 }
3835 if ( $error !== false ) {
3836 $text .= $error;
3837 }
3838 if ( $object !== false ) {
3839 $ret = [ 'object' => $object ];
3840 } else {
3841 $ret = [ 'text' => $text ];
3842 }
3843
3844 return $ret;
3845 }
3846
3847 /**
3848 * Return the text to be used for a given extension tag.
3849 * This is the ghost of strip().
3850 *
3851 * @param array $params Associative array of parameters:
3852 * name PPNode for the tag name
3853 * attr PPNode for unparsed text where tag attributes are thought to be
3854 * attributes Optional associative array of parsed attributes
3855 * inner Contents of extension element
3856 * noClose Original text did not have a close tag
3857 * @param PPFrame $frame
3858 *
3859 * @throws MWException
3860 * @return string
3861 */
3862 public function extensionSubstitution( $params, $frame ) {
3863 static $errorStr = '<span class="error">';
3864 static $errorLen = 20;
3865
3866 $name = $frame->expand( $params['name'] );
3867 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3868 // Probably expansion depth or node count exceeded. Just punt the
3869 // error up.
3870 return $name;
3871 }
3872
3873 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3874 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3875 // See above
3876 return $attrText;
3877 }
3878
3879 // We can't safely check if the expansion for $content resulted in an
3880 // error, because the content could happen to be the error string
3881 // (T149622).
3882 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3883
3884 $marker = self::MARKER_PREFIX . "-$name-"
3885 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3886
3887 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3888 ( $this->ot['html'] || $this->ot['pre'] );
3889 if ( $isFunctionTag ) {
3890 $markerType = 'none';
3891 } else {
3892 $markerType = 'general';
3893 }
3894 if ( $this->ot['html'] || $isFunctionTag ) {
3895 $name = strtolower( $name );
3896 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3897 if ( isset( $params['attributes'] ) ) {
3898 $attributes = $attributes + $params['attributes'];
3899 }
3900
3901 if ( isset( $this->mTagHooks[$name] ) ) {
3902 $output = call_user_func_array( $this->mTagHooks[$name],
3903 [ $content, $attributes, $this, $frame ] );
3904 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3905 list( $callback, ) = $this->mFunctionTagHooks[$name];
3906
3907 // Avoid PHP 7.1 warning from passing $this by reference
3908 $parser = $this;
3909 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3910 } else {
3911 $output = '<span class="error">Invalid tag extension name: ' .
3912 htmlspecialchars( $name ) . '</span>';
3913 }
3914
3915 if ( is_array( $output ) ) {
3916 // Extract flags
3917 $flags = $output;
3918 $output = $flags[0];
3919 if ( isset( $flags['markerType'] ) ) {
3920 $markerType = $flags['markerType'];
3921 }
3922 }
3923 } else {
3924 if ( is_null( $attrText ) ) {
3925 $attrText = '';
3926 }
3927 if ( isset( $params['attributes'] ) ) {
3928 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3929 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3930 htmlspecialchars( $attrValue ) . '"';
3931 }
3932 }
3933 if ( $content === null ) {
3934 $output = "<$name$attrText/>";
3935 } else {
3936 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3937 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3938 // See above
3939 return $close;
3940 }
3941 $output = "<$name$attrText>$content$close";
3942 }
3943 }
3944
3945 if ( $markerType === 'none' ) {
3946 return $output;
3947 } elseif ( $markerType === 'nowiki' ) {
3948 $this->mStripState->addNoWiki( $marker, $output );
3949 } elseif ( $markerType === 'general' ) {
3950 $this->mStripState->addGeneral( $marker, $output );
3951 } else {
3952 throw new MWException( __METHOD__ . ': invalid marker type' );
3953 }
3954 return $marker;
3955 }
3956
3957 /**
3958 * Increment an include size counter
3959 *
3960 * @param string $type The type of expansion
3961 * @param int $size The size of the text
3962 * @return bool False if this inclusion would take it over the maximum, true otherwise
3963 */
3964 public function incrementIncludeSize( $type, $size ) {
3965 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3966 return false;
3967 } else {
3968 $this->mIncludeSizes[$type] += $size;
3969 return true;
3970 }
3971 }
3972
3973 /**
3974 * Increment the expensive function count
3975 *
3976 * @return bool False if the limit has been exceeded
3977 */
3978 public function incrementExpensiveFunctionCount() {
3979 $this->mExpensiveFunctionCount++;
3980 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3981 }
3982
3983 /**
3984 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3985 * Fills $this->mDoubleUnderscores, returns the modified text
3986 *
3987 * @param string $text
3988 *
3989 * @return string
3990 */
3991 public function doDoubleUnderscore( $text ) {
3992 # The position of __TOC__ needs to be recorded
3993 $mw = MagicWord::get( 'toc' );
3994 if ( $mw->match( $text ) ) {
3995 $this->mShowToc = true;
3996 $this->mForceTocPosition = true;
3997
3998 # Set a placeholder. At the end we'll fill it in with the TOC.
3999 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4000
4001 # Only keep the first one.
4002 $text = $mw->replace( '', $text );
4003 }
4004
4005 # Now match and remove the rest of them
4006 $mwa = MagicWord::getDoubleUnderscoreArray();
4007 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4008
4009 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4010 $this->mOutput->mNoGallery = true;
4011 }
4012 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4013 $this->mShowToc = false;
4014 }
4015 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4016 && $this->mTitle->getNamespace() == NS_CATEGORY
4017 ) {
4018 $this->addTrackingCategory( 'hidden-category-category' );
4019 }
4020 # (T10068) Allow control over whether robots index a page.
4021 # __INDEX__ always overrides __NOINDEX__, see T16899
4022 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4023 $this->mOutput->setIndexPolicy( 'noindex' );
4024 $this->addTrackingCategory( 'noindex-category' );
4025 }
4026 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4027 $this->mOutput->setIndexPolicy( 'index' );
4028 $this->addTrackingCategory( 'index-category' );
4029 }
4030
4031 # Cache all double underscores in the database
4032 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4033 $this->mOutput->setProperty( $key, '' );
4034 }
4035
4036 return $text;
4037 }
4038
4039 /**
4040 * @see ParserOutput::addTrackingCategory()
4041 * @param string $msg Message key
4042 * @return bool Whether the addition was successful
4043 */
4044 public function addTrackingCategory( $msg ) {
4045 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4046 }
4047
4048 /**
4049 * This function accomplishes several tasks:
4050 * 1) Auto-number headings if that option is enabled
4051 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4052 * 3) Add a Table of contents on the top for users who have enabled the option
4053 * 4) Auto-anchor headings
4054 *
4055 * It loops through all headlines, collects the necessary data, then splits up the
4056 * string and re-inserts the newly formatted headlines.
4057 *
4058 * @param string $text
4059 * @param string $origText Original, untouched wikitext
4060 * @param bool $isMain
4061 * @return mixed|string
4062 * @private
4063 */
4064 public function formatHeadings( $text, $origText, $isMain = true ) {
4065 global $wgMaxTocLevel;
4066
4067 # Inhibit editsection links if requested in the page
4068 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4069 $maybeShowEditLink = false;
4070 } else {
4071 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4072 }
4073
4074 # Get all headlines for numbering them and adding funky stuff like [edit]
4075 # links - this is for later, but we need the number of headlines right now
4076 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4077 # be trimmed here since whitespace in HTML headings is significant.
4078 $matches = [];
4079 $numMatches = preg_match_all(
4080 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4081 $text,
4082 $matches
4083 );
4084
4085 # if there are fewer than 4 headlines in the article, do not show TOC
4086 # unless it's been explicitly enabled.
4087 $enoughToc = $this->mShowToc &&
4088 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4089
4090 # Allow user to stipulate that a page should have a "new section"
4091 # link added via __NEWSECTIONLINK__
4092 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4093 $this->mOutput->setNewSection( true );
4094 }
4095
4096 # Allow user to remove the "new section"
4097 # link via __NONEWSECTIONLINK__
4098 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4099 $this->mOutput->hideNewSection( true );
4100 }
4101
4102 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4103 # override above conditions and always show TOC above first header
4104 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4105 $this->mShowToc = true;
4106 $enoughToc = true;
4107 }
4108
4109 # headline counter
4110 $headlineCount = 0;
4111 $numVisible = 0;
4112
4113 # Ugh .. the TOC should have neat indentation levels which can be
4114 # passed to the skin functions. These are determined here
4115 $toc = '';
4116 $full = '';
4117 $head = [];
4118 $sublevelCount = [];
4119 $levelCount = [];
4120 $level = 0;
4121 $prevlevel = 0;
4122 $toclevel = 0;
4123 $prevtoclevel = 0;
4124 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4125 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4126 $oldType = $this->mOutputType;
4127 $this->setOutputType( self::OT_WIKI );
4128 $frame = $this->getPreprocessor()->newFrame();
4129 $root = $this->preprocessToDom( $origText );
4130 $node = $root->getFirstChild();
4131 $byteOffset = 0;
4132 $tocraw = [];
4133 $refers = [];
4134
4135 $headlines = $numMatches !== false ? $matches[3] : [];
4136
4137 foreach ( $headlines as $headline ) {
4138 $isTemplate = false;
4139 $titleText = false;
4140 $sectionIndex = false;
4141 $numbering = '';
4142 $markerMatches = [];
4143 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4144 $serial = $markerMatches[1];
4145 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4146 $isTemplate = ( $titleText != $baseTitleText );
4147 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4148 }
4149
4150 if ( $toclevel ) {
4151 $prevlevel = $level;
4152 }
4153 $level = $matches[1][$headlineCount];
4154
4155 if ( $level > $prevlevel ) {
4156 # Increase TOC level
4157 $toclevel++;
4158 $sublevelCount[$toclevel] = 0;
4159 if ( $toclevel < $wgMaxTocLevel ) {
4160 $prevtoclevel = $toclevel;
4161 $toc .= Linker::tocIndent();
4162 $numVisible++;
4163 }
4164 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4165 # Decrease TOC level, find level to jump to
4166
4167 for ( $i = $toclevel; $i > 0; $i-- ) {
4168 if ( $levelCount[$i] == $level ) {
4169 # Found last matching level
4170 $toclevel = $i;
4171 break;
4172 } elseif ( $levelCount[$i] < $level ) {
4173 # Found first matching level below current level
4174 $toclevel = $i + 1;
4175 break;
4176 }
4177 }
4178 if ( $i == 0 ) {
4179 $toclevel = 1;
4180 }
4181 if ( $toclevel < $wgMaxTocLevel ) {
4182 if ( $prevtoclevel < $wgMaxTocLevel ) {
4183 # Unindent only if the previous toc level was shown :p
4184 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4185 $prevtoclevel = $toclevel;
4186 } else {
4187 $toc .= Linker::tocLineEnd();
4188 }
4189 }
4190 } else {
4191 # No change in level, end TOC line
4192 if ( $toclevel < $wgMaxTocLevel ) {
4193 $toc .= Linker::tocLineEnd();
4194 }
4195 }
4196
4197 $levelCount[$toclevel] = $level;
4198
4199 # count number of headlines for each level
4200 $sublevelCount[$toclevel]++;
4201 $dot = 0;
4202 for ( $i = 1; $i <= $toclevel; $i++ ) {
4203 if ( !empty( $sublevelCount[$i] ) ) {
4204 if ( $dot ) {
4205 $numbering .= '.';
4206 }
4207 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4208 $dot = 1;
4209 }
4210 }
4211
4212 # The safe header is a version of the header text safe to use for links
4213
4214 # Remove link placeholders by the link text.
4215 # <!--LINK number-->
4216 # turns into
4217 # link text with suffix
4218 # Do this before unstrip since link text can contain strip markers
4219 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4220
4221 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4222 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4223
4224 # Strip out HTML (first regex removes any tag not allowed)
4225 # Allowed tags are:
4226 # * <sup> and <sub> (T10393)
4227 # * <i> (T28375)
4228 # * <b> (r105284)
4229 # * <bdi> (T74884)
4230 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4231 # * <s> and <strike> (T35715)
4232 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4233 # to allow setting directionality in toc items.
4234 $tocline = preg_replace(
4235 [
4236 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4237 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4238 ],
4239 [ '', '<$1>' ],
4240 $safeHeadline
4241 );
4242
4243 # Strip '<span></span>', which is the result from the above if
4244 # <span id="foo"></span> is used to produce an additional anchor
4245 # for a section.
4246 $tocline = str_replace( '<span></span>', '', $tocline );
4247
4248 $tocline = trim( $tocline );
4249
4250 # For the anchor, strip out HTML-y stuff period
4251 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4252 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4253
4254 # Save headline for section edit hint before it's escaped
4255 $headlineHint = $safeHeadline;
4256
4257 # Decode HTML entities
4258 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4259
4260 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4261
4262 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4263 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4264 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4265 if ( $fallbackHeadline === $safeHeadline ) {
4266 # No reason to have both (in fact, we can't)
4267 $fallbackHeadline = false;
4268 }
4269
4270 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4271 # @todo FIXME: We may be changing them depending on the current locale.
4272 $arrayKey = strtolower( $safeHeadline );
4273 if ( $fallbackHeadline === false ) {
4274 $fallbackArrayKey = false;
4275 } else {
4276 $fallbackArrayKey = strtolower( $fallbackHeadline );
4277 }
4278
4279 # Create the anchor for linking from the TOC to the section
4280 $anchor = $safeHeadline;
4281 $fallbackAnchor = $fallbackHeadline;
4282 if ( isset( $refers[$arrayKey] ) ) {
4283 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4284 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4285 $anchor .= "_$i";
4286 $linkAnchor .= "_$i";
4287 $refers["${arrayKey}_$i"] = true;
4288 } else {
4289 $refers[$arrayKey] = true;
4290 }
4291 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4292 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4293 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4294 $fallbackAnchor .= "_$i";
4295 $refers["${fallbackArrayKey}_$i"] = true;
4296 } else {
4297 $refers[$fallbackArrayKey] = true;
4298 }
4299
4300 # Don't number the heading if it is the only one (looks silly)
4301 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4302 # the two are different if the line contains a link
4303 $headline = Html::element(
4304 'span',
4305 [ 'class' => 'mw-headline-number' ],
4306 $numbering
4307 ) . ' ' . $headline;
4308 }
4309
4310 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4311 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4312 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4313 }
4314
4315 # Add the section to the section tree
4316 # Find the DOM node for this header
4317 $noOffset = ( $isTemplate || $sectionIndex === false );
4318 while ( $node && !$noOffset ) {
4319 if ( $node->getName() === 'h' ) {
4320 $bits = $node->splitHeading();
4321 if ( $bits['i'] == $sectionIndex ) {
4322 break;
4323 }
4324 }
4325 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4326 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4327 $node = $node->getNextSibling();
4328 }
4329 $tocraw[] = [
4330 'toclevel' => $toclevel,
4331 'level' => $level,
4332 'line' => $tocline,
4333 'number' => $numbering,
4334 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4335 'fromtitle' => $titleText,
4336 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4337 'anchor' => $anchor,
4338 ];
4339
4340 # give headline the correct <h#> tag
4341 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4342 // Output edit section links as markers with styles that can be customized by skins
4343 if ( $isTemplate ) {
4344 # Put a T flag in the section identifier, to indicate to extractSections()
4345 # that sections inside <includeonly> should be counted.
4346 $editsectionPage = $titleText;
4347 $editsectionSection = "T-$sectionIndex";
4348 $editsectionContent = null;
4349 } else {
4350 $editsectionPage = $this->mTitle->getPrefixedText();
4351 $editsectionSection = $sectionIndex;
4352 $editsectionContent = $headlineHint;
4353 }
4354 // We use a bit of pesudo-xml for editsection markers. The
4355 // language converter is run later on. Using a UNIQ style marker
4356 // leads to the converter screwing up the tokens when it
4357 // converts stuff. And trying to insert strip tags fails too. At
4358 // this point all real inputted tags have already been escaped,
4359 // so we don't have to worry about a user trying to input one of
4360 // these markers directly. We use a page and section attribute
4361 // to stop the language converter from converting these
4362 // important bits of data, but put the headline hint inside a
4363 // content block because the language converter is supposed to
4364 // be able to convert that piece of data.
4365 // Gets replaced with html in ParserOutput::getText
4366 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4367 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4368 if ( $editsectionContent !== null ) {
4369 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4370 } else {
4371 $editlink .= '/>';
4372 }
4373 } else {
4374 $editlink = '';
4375 }
4376 $head[$headlineCount] = Linker::makeHeadline( $level,
4377 $matches['attrib'][$headlineCount], $anchor, $headline,
4378 $editlink, $fallbackAnchor );
4379
4380 $headlineCount++;
4381 }
4382
4383 $this->setOutputType( $oldType );
4384
4385 # Never ever show TOC if no headers
4386 if ( $numVisible < 1 ) {
4387 $enoughToc = false;
4388 }
4389
4390 if ( $enoughToc ) {
4391 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4392 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4393 }
4394 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4395 $this->mOutput->setTOCHTML( $toc );
4396 $toc = self::TOC_START . $toc . self::TOC_END;
4397 }
4398
4399 if ( $isMain ) {
4400 $this->mOutput->setSections( $tocraw );
4401 }
4402
4403 # split up and insert constructed headlines
4404 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4405 $i = 0;
4406
4407 // build an array of document sections
4408 $sections = [];
4409 foreach ( $blocks as $block ) {
4410 // $head is zero-based, sections aren't.
4411 if ( empty( $head[$i - 1] ) ) {
4412 $sections[$i] = $block;
4413 } else {
4414 $sections[$i] = $head[$i - 1] . $block;
4415 }
4416
4417 /**
4418 * Send a hook, one per section.
4419 * The idea here is to be able to make section-level DIVs, but to do so in a
4420 * lower-impact, more correct way than r50769
4421 *
4422 * $this : caller
4423 * $section : the section number
4424 * &$sectionContent : ref to the content of the section
4425 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4426 */
4427 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4428
4429 $i++;
4430 }
4431
4432 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4433 // append the TOC at the beginning
4434 // Top anchor now in skin
4435 $sections[0] = $sections[0] . $toc . "\n";
4436 }
4437
4438 $full .= implode( '', $sections );
4439
4440 if ( $this->mForceTocPosition ) {
4441 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4442 } else {
4443 return $full;
4444 }
4445 }
4446
4447 /**
4448 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4449 * conversion, substituting signatures, {{subst:}} templates, etc.
4450 *
4451 * @param string $text The text to transform
4452 * @param Title $title The Title object for the current article
4453 * @param User $user The User object describing the current user
4454 * @param ParserOptions $options Parsing options
4455 * @param bool $clearState Whether to clear the parser state first
4456 * @return string The altered wiki markup
4457 */
4458 public function preSaveTransform( $text, Title $title, User $user,
4459 ParserOptions $options, $clearState = true
4460 ) {
4461 if ( $clearState ) {
4462 $magicScopeVariable = $this->lock();
4463 }
4464 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4465 $this->setUser( $user );
4466
4467 // Strip U+0000 NULL (T159174)
4468 $text = str_replace( "\000", '', $text );
4469
4470 // We still normalize line endings for backwards-compatibility
4471 // with other code that just calls PST, but this should already
4472 // be handled in TextContent subclasses
4473 $text = TextContent::normalizeLineEndings( $text );
4474
4475 if ( $options->getPreSaveTransform() ) {
4476 $text = $this->pstPass2( $text, $user );
4477 }
4478 $text = $this->mStripState->unstripBoth( $text );
4479
4480 $this->setUser( null ); # Reset
4481
4482 return $text;
4483 }
4484
4485 /**
4486 * Pre-save transform helper function
4487 *
4488 * @param string $text
4489 * @param User $user
4490 *
4491 * @return string
4492 */
4493 private function pstPass2( $text, $user ) {
4494 global $wgContLang;
4495
4496 # Note: This is the timestamp saved as hardcoded wikitext to
4497 # the database, we use $wgContLang here in order to give
4498 # everyone the same signature and use the default one rather
4499 # than the one selected in each user's preferences.
4500 # (see also T14815)
4501 $ts = $this->mOptions->getTimestamp();
4502 $timestamp = MWTimestamp::getLocalInstance( $ts );
4503 $ts = $timestamp->format( 'YmdHis' );
4504 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4505
4506 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4507
4508 # Variable replacement
4509 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4510 $text = $this->replaceVariables( $text );
4511
4512 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4513 # which may corrupt this parser instance via its wfMessage()->text() call-
4514
4515 # Signatures
4516 if ( strpos( $text, '~~~' ) !== false ) {
4517 $sigText = $this->getUserSig( $user );
4518 $text = strtr( $text, [
4519 '~~~~~' => $d,
4520 '~~~~' => "$sigText $d",
4521 '~~~' => $sigText
4522 ] );
4523 # The main two signature forms used above are time-sensitive
4524 $this->mOutput->setFlag( 'user-signature' );
4525 }
4526
4527 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4528 $tc = '[' . Title::legalChars() . ']';
4529 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4530
4531 // [[ns:page (context)|]]
4532 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4533 // [[ns:page(context)|]] (double-width brackets, added in r40257)
4534 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?($tc+))\\|]]/";
4535 // [[ns:page (context), context|]] (using either single or double-width comma)
4536 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |,)$tc+|)\\|]]/";
4537 // [[|page]] (reverse pipe trick: add context from page title)
4538 $p2 = "/\[\[\\|($tc+)]]/";
4539
4540 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4541 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4542 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4543 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4544
4545 $t = $this->mTitle->getText();
4546 $m = [];
4547 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4548 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4549 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4550 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4551 } else {
4552 # if there's no context, don't bother duplicating the title
4553 $text = preg_replace( $p2, '[[\\1]]', $text );
4554 }
4555
4556 return $text;
4557 }
4558
4559 /**
4560 * Fetch the user's signature text, if any, and normalize to
4561 * validated, ready-to-insert wikitext.
4562 * If you have pre-fetched the nickname or the fancySig option, you can
4563 * specify them here to save a database query.
4564 * Do not reuse this parser instance after calling getUserSig(),
4565 * as it may have changed if it's the $wgParser.
4566 *
4567 * @param User &$user
4568 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4569 * @param bool|null $fancySig whether the nicknname is the complete signature
4570 * or null to use default value
4571 * @return string
4572 */
4573 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4574 global $wgMaxSigChars;
4575
4576 $username = $user->getName();
4577
4578 # If not given, retrieve from the user object.
4579 if ( $nickname === false ) {
4580 $nickname = $user->getOption( 'nickname' );
4581 }
4582
4583 if ( is_null( $fancySig ) ) {
4584 $fancySig = $user->getBoolOption( 'fancysig' );
4585 }
4586
4587 $nickname = $nickname == null ? $username : $nickname;
4588
4589 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4590 $nickname = $username;
4591 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4592 } elseif ( $fancySig !== false ) {
4593 # Sig. might contain markup; validate this
4594 if ( $this->validateSig( $nickname ) !== false ) {
4595 # Validated; clean up (if needed) and return it
4596 return $this->cleanSig( $nickname, true );
4597 } else {
4598 # Failed to validate; fall back to the default
4599 $nickname = $username;
4600 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4601 }
4602 }
4603
4604 # Make sure nickname doesnt get a sig in a sig
4605 $nickname = self::cleanSigInSig( $nickname );
4606
4607 # If we're still here, make it a link to the user page
4608 $userText = wfEscapeWikiText( $username );
4609 $nickText = wfEscapeWikiText( $nickname );
4610 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4611
4612 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4613 ->title( $this->getTitle() )->text();
4614 }
4615
4616 /**
4617 * Check that the user's signature contains no bad XML
4618 *
4619 * @param string $text
4620 * @return string|bool An expanded string, or false if invalid.
4621 */
4622 public function validateSig( $text ) {
4623 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4624 }
4625
4626 /**
4627 * Clean up signature text
4628 *
4629 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4630 * 2) Substitute all transclusions
4631 *
4632 * @param string $text
4633 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4634 * @return string Signature text
4635 */
4636 public function cleanSig( $text, $parsing = false ) {
4637 if ( !$parsing ) {
4638 global $wgTitle;
4639 $magicScopeVariable = $this->lock();
4640 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4641 }
4642
4643 # Option to disable this feature
4644 if ( !$this->mOptions->getCleanSignatures() ) {
4645 return $text;
4646 }
4647
4648 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4649 # => Move this logic to braceSubstitution()
4650 $substWord = MagicWord::get( 'subst' );
4651 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4652 $substText = '{{' . $substWord->getSynonym( 0 );
4653
4654 $text = preg_replace( $substRegex, $substText, $text );
4655 $text = self::cleanSigInSig( $text );
4656 $dom = $this->preprocessToDom( $text );
4657 $frame = $this->getPreprocessor()->newFrame();
4658 $text = $frame->expand( $dom );
4659
4660 if ( !$parsing ) {
4661 $text = $this->mStripState->unstripBoth( $text );
4662 }
4663
4664 return $text;
4665 }
4666
4667 /**
4668 * Strip 3, 4 or 5 tildes out of signatures.
4669 *
4670 * @param string $text
4671 * @return string Signature text with /~{3,5}/ removed
4672 */
4673 public static function cleanSigInSig( $text ) {
4674 $text = preg_replace( '/~{3,5}/', '', $text );
4675 return $text;
4676 }
4677
4678 /**
4679 * Set up some variables which are usually set up in parse()
4680 * so that an external function can call some class members with confidence
4681 *
4682 * @param Title|null $title
4683 * @param ParserOptions $options
4684 * @param int $outputType
4685 * @param bool $clearState
4686 */
4687 public function startExternalParse( Title $title = null, ParserOptions $options,
4688 $outputType, $clearState = true
4689 ) {
4690 $this->startParse( $title, $options, $outputType, $clearState );
4691 }
4692
4693 /**
4694 * @param Title|null $title
4695 * @param ParserOptions $options
4696 * @param int $outputType
4697 * @param bool $clearState
4698 */
4699 private function startParse( Title $title = null, ParserOptions $options,
4700 $outputType, $clearState = true
4701 ) {
4702 $this->setTitle( $title );
4703 $this->mOptions = $options;
4704 $this->setOutputType( $outputType );
4705 if ( $clearState ) {
4706 $this->clearState();
4707 }
4708 }
4709
4710 /**
4711 * Wrapper for preprocess()
4712 *
4713 * @param string $text The text to preprocess
4714 * @param ParserOptions $options
4715 * @param Title|null $title Title object or null to use $wgTitle
4716 * @return string
4717 */
4718 public function transformMsg( $text, $options, $title = null ) {
4719 static $executing = false;
4720
4721 # Guard against infinite recursion
4722 if ( $executing ) {
4723 return $text;
4724 }
4725 $executing = true;
4726
4727 if ( !$title ) {
4728 global $wgTitle;
4729 $title = $wgTitle;
4730 }
4731
4732 $text = $this->preprocess( $text, $title, $options );
4733
4734 $executing = false;
4735 return $text;
4736 }
4737
4738 /**
4739 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4740 * The callback should have the following form:
4741 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4742 *
4743 * Transform and return $text. Use $parser for any required context, e.g. use
4744 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4745 *
4746 * Hooks may return extended information by returning an array, of which the
4747 * first numbered element (index 0) must be the return string, and all other
4748 * entries are extracted into local variables within an internal function
4749 * in the Parser class.
4750 *
4751 * This interface (introduced r61913) appears to be undocumented, but
4752 * 'markerType' is used by some core tag hooks to override which strip
4753 * array their results are placed in. **Use great caution if attempting
4754 * this interface, as it is not documented and injudicious use could smash
4755 * private variables.**
4756 *
4757 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4758 * @param callable $callback The callback function (and object) to use for the tag
4759 * @throws MWException
4760 * @return callable|null The old value of the mTagHooks array associated with the hook
4761 */
4762 public function setHook( $tag, callable $callback ) {
4763 $tag = strtolower( $tag );
4764 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4765 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4766 }
4767 $oldVal = $this->mTagHooks[$tag] ?? null;
4768 $this->mTagHooks[$tag] = $callback;
4769 if ( !in_array( $tag, $this->mStripList ) ) {
4770 $this->mStripList[] = $tag;
4771 }
4772
4773 return $oldVal;
4774 }
4775
4776 /**
4777 * As setHook(), but letting the contents be parsed.
4778 *
4779 * Transparent tag hooks are like regular XML-style tag hooks, except they
4780 * operate late in the transformation sequence, on HTML instead of wikitext.
4781 *
4782 * This is probably obsoleted by things dealing with parser frames?
4783 * The only extension currently using it is geoserver.
4784 *
4785 * @since 1.10
4786 * @todo better document or deprecate this
4787 *
4788 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4789 * @param callable $callback The callback function (and object) to use for the tag
4790 * @throws MWException
4791 * @return callable|null The old value of the mTagHooks array associated with the hook
4792 */
4793 public function setTransparentTagHook( $tag, callable $callback ) {
4794 $tag = strtolower( $tag );
4795 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4796 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4797 }
4798 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4799 $this->mTransparentTagHooks[$tag] = $callback;
4800
4801 return $oldVal;
4802 }
4803
4804 /**
4805 * Remove all tag hooks
4806 */
4807 public function clearTagHooks() {
4808 $this->mTagHooks = [];
4809 $this->mFunctionTagHooks = [];
4810 $this->mStripList = $this->mDefaultStripList;
4811 }
4812
4813 /**
4814 * Create a function, e.g. {{sum:1|2|3}}
4815 * The callback function should have the form:
4816 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4817 *
4818 * Or with Parser::SFH_OBJECT_ARGS:
4819 * function myParserFunction( $parser, $frame, $args ) { ... }
4820 *
4821 * The callback may either return the text result of the function, or an array with the text
4822 * in element 0, and a number of flags in the other elements. The names of the flags are
4823 * specified in the keys. Valid flags are:
4824 * found The text returned is valid, stop processing the template. This
4825 * is on by default.
4826 * nowiki Wiki markup in the return value should be escaped
4827 * isHTML The returned text is HTML, armour it against wikitext transformation
4828 *
4829 * @param string $id The magic word ID
4830 * @param callable $callback The callback function (and object) to use
4831 * @param int $flags A combination of the following flags:
4832 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4833 *
4834 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4835 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4836 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4837 * the arguments, and to control the way they are expanded.
4838 *
4839 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4840 * arguments, for instance:
4841 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4842 *
4843 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4844 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4845 * working if/when this is changed.
4846 *
4847 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4848 * expansion.
4849 *
4850 * Please read the documentation in includes/parser/Preprocessor.php for more information
4851 * about the methods available in PPFrame and PPNode.
4852 *
4853 * @throws MWException
4854 * @return string|callable The old callback function for this name, if any
4855 */
4856 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4857 global $wgContLang;
4858
4859 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4860 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4861
4862 # Add to function cache
4863 $mw = MagicWord::get( $id );
4864 if ( !$mw ) {
4865 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4866 }
4867
4868 $synonyms = $mw->getSynonyms();
4869 $sensitive = intval( $mw->isCaseSensitive() );
4870
4871 foreach ( $synonyms as $syn ) {
4872 # Case
4873 if ( !$sensitive ) {
4874 $syn = $wgContLang->lc( $syn );
4875 }
4876 # Add leading hash
4877 if ( !( $flags & self::SFH_NO_HASH ) ) {
4878 $syn = '#' . $syn;
4879 }
4880 # Remove trailing colon
4881 if ( substr( $syn, -1, 1 ) === ':' ) {
4882 $syn = substr( $syn, 0, -1 );
4883 }
4884 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4885 }
4886 return $oldVal;
4887 }
4888
4889 /**
4890 * Get all registered function hook identifiers
4891 *
4892 * @return array
4893 */
4894 public function getFunctionHooks() {
4895 return array_keys( $this->mFunctionHooks );
4896 }
4897
4898 /**
4899 * Create a tag function, e.g. "<test>some stuff</test>".
4900 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4901 * Unlike parser functions, their content is not preprocessed.
4902 * @param string $tag
4903 * @param callable $callback
4904 * @param int $flags
4905 * @throws MWException
4906 * @return null
4907 */
4908 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4909 $tag = strtolower( $tag );
4910 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4911 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4912 }
4913 $old = $this->mFunctionTagHooks[$tag] ?? null;
4914 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4915
4916 if ( !in_array( $tag, $this->mStripList ) ) {
4917 $this->mStripList[] = $tag;
4918 }
4919
4920 return $old;
4921 }
4922
4923 /**
4924 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4925 * Placeholders created in Linker::link()
4926 *
4927 * @param string &$text
4928 * @param int $options
4929 */
4930 public function replaceLinkHolders( &$text, $options = 0 ) {
4931 $this->mLinkHolders->replace( $text );
4932 }
4933
4934 /**
4935 * Replace "<!--LINK-->" link placeholders with plain text of links
4936 * (not HTML-formatted).
4937 *
4938 * @param string $text
4939 * @return string
4940 */
4941 public function replaceLinkHoldersText( $text ) {
4942 return $this->mLinkHolders->replaceText( $text );
4943 }
4944
4945 /**
4946 * Renders an image gallery from a text with one line per image.
4947 * text labels may be given by using |-style alternative text. E.g.
4948 * Image:one.jpg|The number "1"
4949 * Image:tree.jpg|A tree
4950 * given as text will return the HTML of a gallery with two images,
4951 * labeled 'The number "1"' and
4952 * 'A tree'.
4953 *
4954 * @param string $text
4955 * @param array $params
4956 * @return string HTML
4957 */
4958 public function renderImageGallery( $text, $params ) {
4959 $mode = false;
4960 if ( isset( $params['mode'] ) ) {
4961 $mode = $params['mode'];
4962 }
4963
4964 try {
4965 $ig = ImageGalleryBase::factory( $mode );
4966 } catch ( Exception $e ) {
4967 // If invalid type set, fallback to default.
4968 $ig = ImageGalleryBase::factory( false );
4969 }
4970
4971 $ig->setContextTitle( $this->mTitle );
4972 $ig->setShowBytes( false );
4973 $ig->setShowDimensions( false );
4974 $ig->setShowFilename( false );
4975 $ig->setParser( $this );
4976 $ig->setHideBadImages();
4977 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4978
4979 if ( isset( $params['showfilename'] ) ) {
4980 $ig->setShowFilename( true );
4981 } else {
4982 $ig->setShowFilename( false );
4983 }
4984 if ( isset( $params['caption'] ) ) {
4985 $caption = $params['caption'];
4986 $caption = htmlspecialchars( $caption );
4987 $caption = $this->replaceInternalLinks( $caption );
4988 $ig->setCaptionHtml( $caption );
4989 }
4990 if ( isset( $params['perrow'] ) ) {
4991 $ig->setPerRow( $params['perrow'] );
4992 }
4993 if ( isset( $params['widths'] ) ) {
4994 $ig->setWidths( $params['widths'] );
4995 }
4996 if ( isset( $params['heights'] ) ) {
4997 $ig->setHeights( $params['heights'] );
4998 }
4999 $ig->setAdditionalOptions( $params );
5000
5001 // Avoid PHP 7.1 warning from passing $this by reference
5002 $parser = $this;
5003 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5004
5005 $lines = StringUtils::explode( "\n", $text );
5006 foreach ( $lines as $line ) {
5007 # match lines like these:
5008 # Image:someimage.jpg|This is some image
5009 $matches = [];
5010 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5011 # Skip empty lines
5012 if ( count( $matches ) == 0 ) {
5013 continue;
5014 }
5015
5016 if ( strpos( $matches[0], '%' ) !== false ) {
5017 $matches[1] = rawurldecode( $matches[1] );
5018 }
5019 $title = Title::newFromText( $matches[1], NS_FILE );
5020 if ( is_null( $title ) ) {
5021 # Bogus title. Ignore these so we don't bomb out later.
5022 continue;
5023 }
5024
5025 # We need to get what handler the file uses, to figure out parameters.
5026 # Note, a hook can overide the file name, and chose an entirely different
5027 # file (which potentially could be of a different type and have different handler).
5028 $options = [];
5029 $descQuery = false;
5030 Hooks::run( 'BeforeParserFetchFileAndTitle',
5031 [ $this, $title, &$options, &$descQuery ] );
5032 # Don't register it now, as TraditionalImageGallery does that later.
5033 $file = $this->fetchFileNoRegister( $title, $options );
5034 $handler = $file ? $file->getHandler() : false;
5035
5036 $paramMap = [
5037 'img_alt' => 'gallery-internal-alt',
5038 'img_link' => 'gallery-internal-link',
5039 ];
5040 if ( $handler ) {
5041 $paramMap = $paramMap + $handler->getParamMap();
5042 // We don't want people to specify per-image widths.
5043 // Additionally the width parameter would need special casing anyhow.
5044 unset( $paramMap['img_width'] );
5045 }
5046
5047 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5048
5049 $label = '';
5050 $alt = '';
5051 $link = '';
5052 $handlerOptions = [];
5053 if ( isset( $matches[3] ) ) {
5054 // look for an |alt= definition while trying not to break existing
5055 // captions with multiple pipes (|) in it, until a more sensible grammar
5056 // is defined for images in galleries
5057
5058 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5059 // splitting on '|' is a bit odd, and different from makeImage.
5060 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5061 // Protect LanguageConverter markup
5062 $parameterMatches = StringUtils::delimiterExplode(
5063 '-{', '}-', '|', $matches[3], true /* nested */
5064 );
5065
5066 foreach ( $parameterMatches as $parameterMatch ) {
5067 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5068 if ( $magicName ) {
5069 $paramName = $paramMap[$magicName];
5070
5071 switch ( $paramName ) {
5072 case 'gallery-internal-alt':
5073 $alt = $this->stripAltText( $match, false );
5074 break;
5075 case 'gallery-internal-link':
5076 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5077 $chars = self::EXT_LINK_URL_CLASS;
5078 $addr = self::EXT_LINK_ADDR;
5079 $prots = $this->mUrlProtocols;
5080 // check to see if link matches an absolute url, if not then it must be a wiki link.
5081 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5082 // Result of LanguageConverter::markNoConversion
5083 // invoked on an external link.
5084 $linkValue = substr( $linkValue, 4, -2 );
5085 }
5086 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5087 $link = $linkValue;
5088 $this->mOutput->addExternalLink( $link );
5089 } else {
5090 $localLinkTitle = Title::newFromText( $linkValue );
5091 if ( $localLinkTitle !== null ) {
5092 $this->mOutput->addLink( $localLinkTitle );
5093 $link = $localLinkTitle->getLinkURL();
5094 }
5095 }
5096 break;
5097 default:
5098 // Must be a handler specific parameter.
5099 if ( $handler->validateParam( $paramName, $match ) ) {
5100 $handlerOptions[$paramName] = $match;
5101 } else {
5102 // Guess not, consider it as caption.
5103 wfDebug( "$parameterMatch failed parameter validation\n" );
5104 $label = $parameterMatch;
5105 }
5106 }
5107
5108 } else {
5109 // Last pipe wins.
5110 $label = $parameterMatch;
5111 }
5112 }
5113 }
5114
5115 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5116 }
5117 $html = $ig->toHTML();
5118 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5119 return $html;
5120 }
5121
5122 /**
5123 * @param MediaHandler $handler
5124 * @return array
5125 */
5126 public function getImageParams( $handler ) {
5127 if ( $handler ) {
5128 $handlerClass = get_class( $handler );
5129 } else {
5130 $handlerClass = '';
5131 }
5132 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5133 # Initialise static lists
5134 static $internalParamNames = [
5135 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5136 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5137 'bottom', 'text-bottom' ],
5138 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5139 'upright', 'border', 'link', 'alt', 'class' ],
5140 ];
5141 static $internalParamMap;
5142 if ( !$internalParamMap ) {
5143 $internalParamMap = [];
5144 foreach ( $internalParamNames as $type => $names ) {
5145 foreach ( $names as $name ) {
5146 // For grep: img_left, img_right, img_center, img_none,
5147 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5148 // img_bottom, img_text_bottom,
5149 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5150 // img_border, img_link, img_alt, img_class
5151 $magicName = str_replace( '-', '_', "img_$name" );
5152 $internalParamMap[$magicName] = [ $type, $name ];
5153 }
5154 }
5155 }
5156
5157 # Add handler params
5158 $paramMap = $internalParamMap;
5159 if ( $handler ) {
5160 $handlerParamMap = $handler->getParamMap();
5161 foreach ( $handlerParamMap as $magic => $paramName ) {
5162 $paramMap[$magic] = [ 'handler', $paramName ];
5163 }
5164 }
5165 $this->mImageParams[$handlerClass] = $paramMap;
5166 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5167 }
5168 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5169 }
5170
5171 /**
5172 * Parse image options text and use it to make an image
5173 *
5174 * @param Title $title
5175 * @param string $options
5176 * @param LinkHolderArray|bool $holders
5177 * @return string HTML
5178 */
5179 public function makeImage( $title, $options, $holders = false ) {
5180 # Check if the options text is of the form "options|alt text"
5181 # Options are:
5182 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5183 # * left no resizing, just left align. label is used for alt= only
5184 # * right same, but right aligned
5185 # * none same, but not aligned
5186 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5187 # * center center the image
5188 # * frame Keep original image size, no magnify-button.
5189 # * framed Same as "frame"
5190 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5191 # * upright reduce width for upright images, rounded to full __0 px
5192 # * border draw a 1px border around the image
5193 # * alt Text for HTML alt attribute (defaults to empty)
5194 # * class Set a class for img node
5195 # * link Set the target of the image link. Can be external, interwiki, or local
5196 # vertical-align values (no % or length right now):
5197 # * baseline
5198 # * sub
5199 # * super
5200 # * top
5201 # * text-top
5202 # * middle
5203 # * bottom
5204 # * text-bottom
5205
5206 # Protect LanguageConverter markup when splitting into parts
5207 $parts = StringUtils::delimiterExplode(
5208 '-{', '}-', '|', $options, true /* allow nesting */
5209 );
5210
5211 # Give extensions a chance to select the file revision for us
5212 $options = [];
5213 $descQuery = false;
5214 Hooks::run( 'BeforeParserFetchFileAndTitle',
5215 [ $this, $title, &$options, &$descQuery ] );
5216 # Fetch and register the file (file title may be different via hooks)
5217 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5218
5219 # Get parameter map
5220 $handler = $file ? $file->getHandler() : false;
5221
5222 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5223
5224 if ( !$file ) {
5225 $this->addTrackingCategory( 'broken-file-category' );
5226 }
5227
5228 # Process the input parameters
5229 $caption = '';
5230 $params = [ 'frame' => [], 'handler' => [],
5231 'horizAlign' => [], 'vertAlign' => [] ];
5232 $seenformat = false;
5233 foreach ( $parts as $part ) {
5234 $part = trim( $part );
5235 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5236 $validated = false;
5237 if ( isset( $paramMap[$magicName] ) ) {
5238 list( $type, $paramName ) = $paramMap[$magicName];
5239
5240 # Special case; width and height come in one variable together
5241 if ( $type === 'handler' && $paramName === 'width' ) {
5242 $parsedWidthParam = self::parseWidthParam( $value );
5243 if ( isset( $parsedWidthParam['width'] ) ) {
5244 $width = $parsedWidthParam['width'];
5245 if ( $handler->validateParam( 'width', $width ) ) {
5246 $params[$type]['width'] = $width;
5247 $validated = true;
5248 }
5249 }
5250 if ( isset( $parsedWidthParam['height'] ) ) {
5251 $height = $parsedWidthParam['height'];
5252 if ( $handler->validateParam( 'height', $height ) ) {
5253 $params[$type]['height'] = $height;
5254 $validated = true;
5255 }
5256 }
5257 # else no validation -- T15436
5258 } else {
5259 if ( $type === 'handler' ) {
5260 # Validate handler parameter
5261 $validated = $handler->validateParam( $paramName, $value );
5262 } else {
5263 # Validate internal parameters
5264 switch ( $paramName ) {
5265 case 'manualthumb':
5266 case 'alt':
5267 case 'class':
5268 # @todo FIXME: Possibly check validity here for
5269 # manualthumb? downstream behavior seems odd with
5270 # missing manual thumbs.
5271 $validated = true;
5272 $value = $this->stripAltText( $value, $holders );
5273 break;
5274 case 'link':
5275 $chars = self::EXT_LINK_URL_CLASS;
5276 $addr = self::EXT_LINK_ADDR;
5277 $prots = $this->mUrlProtocols;
5278 if ( $value === '' ) {
5279 $paramName = 'no-link';
5280 $value = true;
5281 $validated = true;
5282 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5283 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5284 $paramName = 'link-url';
5285 $this->mOutput->addExternalLink( $value );
5286 if ( $this->mOptions->getExternalLinkTarget() ) {
5287 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5288 }
5289 $validated = true;
5290 }
5291 } else {
5292 $linkTitle = Title::newFromText( $value );
5293 if ( $linkTitle ) {
5294 $paramName = 'link-title';
5295 $value = $linkTitle;
5296 $this->mOutput->addLink( $linkTitle );
5297 $validated = true;
5298 }
5299 }
5300 break;
5301 case 'frameless':
5302 case 'framed':
5303 case 'thumbnail':
5304 // use first appearing option, discard others.
5305 $validated = !$seenformat;
5306 $seenformat = true;
5307 break;
5308 default:
5309 # Most other things appear to be empty or numeric...
5310 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5311 }
5312 }
5313
5314 if ( $validated ) {
5315 $params[$type][$paramName] = $value;
5316 }
5317 }
5318 }
5319 if ( !$validated ) {
5320 $caption = $part;
5321 }
5322 }
5323
5324 # Process alignment parameters
5325 if ( $params['horizAlign'] ) {
5326 $params['frame']['align'] = key( $params['horizAlign'] );
5327 }
5328 if ( $params['vertAlign'] ) {
5329 $params['frame']['valign'] = key( $params['vertAlign'] );
5330 }
5331
5332 $params['frame']['caption'] = $caption;
5333
5334 # Will the image be presented in a frame, with the caption below?
5335 $imageIsFramed = isset( $params['frame']['frame'] )
5336 || isset( $params['frame']['framed'] )
5337 || isset( $params['frame']['thumbnail'] )
5338 || isset( $params['frame']['manualthumb'] );
5339
5340 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5341 # came to also set the caption, ordinary text after the image -- which
5342 # makes no sense, because that just repeats the text multiple times in
5343 # screen readers. It *also* came to set the title attribute.
5344 # Now that we have an alt attribute, we should not set the alt text to
5345 # equal the caption: that's worse than useless, it just repeats the
5346 # text. This is the framed/thumbnail case. If there's no caption, we
5347 # use the unnamed parameter for alt text as well, just for the time be-
5348 # ing, if the unnamed param is set and the alt param is not.
5349 # For the future, we need to figure out if we want to tweak this more,
5350 # e.g., introducing a title= parameter for the title; ignoring the un-
5351 # named parameter entirely for images without a caption; adding an ex-
5352 # plicit caption= parameter and preserving the old magic unnamed para-
5353 # meter for BC; ...
5354 if ( $imageIsFramed ) { # Framed image
5355 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5356 # No caption or alt text, add the filename as the alt text so
5357 # that screen readers at least get some description of the image
5358 $params['frame']['alt'] = $title->getText();
5359 }
5360 # Do not set $params['frame']['title'] because tooltips don't make sense
5361 # for framed images
5362 } else { # Inline image
5363 if ( !isset( $params['frame']['alt'] ) ) {
5364 # No alt text, use the "caption" for the alt text
5365 if ( $caption !== '' ) {
5366 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5367 } else {
5368 # No caption, fall back to using the filename for the
5369 # alt text
5370 $params['frame']['alt'] = $title->getText();
5371 }
5372 }
5373 # Use the "caption" for the tooltip text
5374 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5375 }
5376
5377 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5378
5379 # Linker does the rest
5380 $time = $options['time'] ?? false;
5381 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5382 $time, $descQuery, $this->mOptions->getThumbSize() );
5383
5384 # Give the handler a chance to modify the parser object
5385 if ( $handler ) {
5386 $handler->parserTransformHook( $this, $file );
5387 }
5388
5389 return $ret;
5390 }
5391
5392 /**
5393 * @param string $caption
5394 * @param LinkHolderArray|bool $holders
5395 * @return mixed|string
5396 */
5397 protected function stripAltText( $caption, $holders ) {
5398 # Strip bad stuff out of the title (tooltip). We can't just use
5399 # replaceLinkHoldersText() here, because if this function is called
5400 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5401 if ( $holders ) {
5402 $tooltip = $holders->replaceText( $caption );
5403 } else {
5404 $tooltip = $this->replaceLinkHoldersText( $caption );
5405 }
5406
5407 # make sure there are no placeholders in thumbnail attributes
5408 # that are later expanded to html- so expand them now and
5409 # remove the tags
5410 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5411 $tooltip = Sanitizer::stripAllTags( $tooltip );
5412
5413 return $tooltip;
5414 }
5415
5416 /**
5417 * Set a flag in the output object indicating that the content is dynamic and
5418 * shouldn't be cached.
5419 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5420 */
5421 public function disableCache() {
5422 wfDebug( "Parser output marked as uncacheable.\n" );
5423 if ( !$this->mOutput ) {
5424 throw new MWException( __METHOD__ .
5425 " can only be called when actually parsing something" );
5426 }
5427 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5428 }
5429
5430 /**
5431 * Callback from the Sanitizer for expanding items found in HTML attribute
5432 * values, so they can be safely tested and escaped.
5433 *
5434 * @param string &$text
5435 * @param bool|PPFrame $frame
5436 * @return string
5437 */
5438 public function attributeStripCallback( &$text, $frame = false ) {
5439 $text = $this->replaceVariables( $text, $frame );
5440 $text = $this->mStripState->unstripBoth( $text );
5441 return $text;
5442 }
5443
5444 /**
5445 * Accessor
5446 *
5447 * @return array
5448 */
5449 public function getTags() {
5450 return array_merge(
5451 array_keys( $this->mTransparentTagHooks ),
5452 array_keys( $this->mTagHooks ),
5453 array_keys( $this->mFunctionTagHooks )
5454 );
5455 }
5456
5457 /**
5458 * Replace transparent tags in $text with the values given by the callbacks.
5459 *
5460 * Transparent tag hooks are like regular XML-style tag hooks, except they
5461 * operate late in the transformation sequence, on HTML instead of wikitext.
5462 *
5463 * @param string $text
5464 *
5465 * @return string
5466 */
5467 public function replaceTransparentTags( $text ) {
5468 $matches = [];
5469 $elements = array_keys( $this->mTransparentTagHooks );
5470 $text = self::extractTagsAndParams( $elements, $text, $matches );
5471 $replacements = [];
5472
5473 foreach ( $matches as $marker => $data ) {
5474 list( $element, $content, $params, $tag ) = $data;
5475 $tagName = strtolower( $element );
5476 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5477 $output = call_user_func_array(
5478 $this->mTransparentTagHooks[$tagName],
5479 [ $content, $params, $this ]
5480 );
5481 } else {
5482 $output = $tag;
5483 }
5484 $replacements[$marker] = $output;
5485 }
5486 return strtr( $text, $replacements );
5487 }
5488
5489 /**
5490 * Break wikitext input into sections, and either pull or replace
5491 * some particular section's text.
5492 *
5493 * External callers should use the getSection and replaceSection methods.
5494 *
5495 * @param string $text Page wikitext
5496 * @param string|int $sectionId A section identifier string of the form:
5497 * "<flag1> - <flag2> - ... - <section number>"
5498 *
5499 * Currently the only recognised flag is "T", which means the target section number
5500 * was derived during a template inclusion parse, in other words this is a template
5501 * section edit link. If no flags are given, it was an ordinary section edit link.
5502 * This flag is required to avoid a section numbering mismatch when a section is
5503 * enclosed by "<includeonly>" (T8563).
5504 *
5505 * The section number 0 pulls the text before the first heading; other numbers will
5506 * pull the given section along with its lower-level subsections. If the section is
5507 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5508 *
5509 * Section 0 is always considered to exist, even if it only contains the empty
5510 * string. If $text is the empty string and section 0 is replaced, $newText is
5511 * returned.
5512 *
5513 * @param string $mode One of "get" or "replace"
5514 * @param string $newText Replacement text for section data.
5515 * @return string For "get", the extracted section text.
5516 * for "replace", the whole page with the section replaced.
5517 */
5518 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5519 global $wgTitle; # not generally used but removes an ugly failure mode
5520
5521 $magicScopeVariable = $this->lock();
5522 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5523 $outText = '';
5524 $frame = $this->getPreprocessor()->newFrame();
5525
5526 # Process section extraction flags
5527 $flags = 0;
5528 $sectionParts = explode( '-', $sectionId );
5529 $sectionIndex = array_pop( $sectionParts );
5530 foreach ( $sectionParts as $part ) {
5531 if ( $part === 'T' ) {
5532 $flags |= self::PTD_FOR_INCLUSION;
5533 }
5534 }
5535
5536 # Check for empty input
5537 if ( strval( $text ) === '' ) {
5538 # Only sections 0 and T-0 exist in an empty document
5539 if ( $sectionIndex == 0 ) {
5540 if ( $mode === 'get' ) {
5541 return '';
5542 } else {
5543 return $newText;
5544 }
5545 } else {
5546 if ( $mode === 'get' ) {
5547 return $newText;
5548 } else {
5549 return $text;
5550 }
5551 }
5552 }
5553
5554 # Preprocess the text
5555 $root = $this->preprocessToDom( $text, $flags );
5556
5557 # <h> nodes indicate section breaks
5558 # They can only occur at the top level, so we can find them by iterating the root's children
5559 $node = $root->getFirstChild();
5560
5561 # Find the target section
5562 if ( $sectionIndex == 0 ) {
5563 # Section zero doesn't nest, level=big
5564 $targetLevel = 1000;
5565 } else {
5566 while ( $node ) {
5567 if ( $node->getName() === 'h' ) {
5568 $bits = $node->splitHeading();
5569 if ( $bits['i'] == $sectionIndex ) {
5570 $targetLevel = $bits['level'];
5571 break;
5572 }
5573 }
5574 if ( $mode === 'replace' ) {
5575 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5576 }
5577 $node = $node->getNextSibling();
5578 }
5579 }
5580
5581 if ( !$node ) {
5582 # Not found
5583 if ( $mode === 'get' ) {
5584 return $newText;
5585 } else {
5586 return $text;
5587 }
5588 }
5589
5590 # Find the end of the section, including nested sections
5591 do {
5592 if ( $node->getName() === 'h' ) {
5593 $bits = $node->splitHeading();
5594 $curLevel = $bits['level'];
5595 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5596 break;
5597 }
5598 }
5599 if ( $mode === 'get' ) {
5600 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5601 }
5602 $node = $node->getNextSibling();
5603 } while ( $node );
5604
5605 # Write out the remainder (in replace mode only)
5606 if ( $mode === 'replace' ) {
5607 # Output the replacement text
5608 # Add two newlines on -- trailing whitespace in $newText is conventionally
5609 # stripped by the editor, so we need both newlines to restore the paragraph gap
5610 # Only add trailing whitespace if there is newText
5611 if ( $newText != "" ) {
5612 $outText .= $newText . "\n\n";
5613 }
5614
5615 while ( $node ) {
5616 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5617 $node = $node->getNextSibling();
5618 }
5619 }
5620
5621 if ( is_string( $outText ) ) {
5622 # Re-insert stripped tags
5623 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5624 }
5625
5626 return $outText;
5627 }
5628
5629 /**
5630 * This function returns the text of a section, specified by a number ($section).
5631 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5632 * the first section before any such heading (section 0).
5633 *
5634 * If a section contains subsections, these are also returned.
5635 *
5636 * @param string $text Text to look in
5637 * @param string|int $sectionId Section identifier as a number or string
5638 * (e.g. 0, 1 or 'T-1').
5639 * @param string $defaultText Default to return if section is not found
5640 *
5641 * @return string Text of the requested section
5642 */
5643 public function getSection( $text, $sectionId, $defaultText = '' ) {
5644 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5645 }
5646
5647 /**
5648 * This function returns $oldtext after the content of the section
5649 * specified by $section has been replaced with $text. If the target
5650 * section does not exist, $oldtext is returned unchanged.
5651 *
5652 * @param string $oldText Former text of the article
5653 * @param string|int $sectionId Section identifier as a number or string
5654 * (e.g. 0, 1 or 'T-1').
5655 * @param string $newText Replacing text
5656 *
5657 * @return string Modified text
5658 */
5659 public function replaceSection( $oldText, $sectionId, $newText ) {
5660 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5661 }
5662
5663 /**
5664 * Get the ID of the revision we are parsing
5665 *
5666 * @return int|null
5667 */
5668 public function getRevisionId() {
5669 return $this->mRevisionId;
5670 }
5671
5672 /**
5673 * Get the revision object for $this->mRevisionId
5674 *
5675 * @return Revision|null Either a Revision object or null
5676 * @since 1.23 (public since 1.23)
5677 */
5678 public function getRevisionObject() {
5679 if ( !is_null( $this->mRevisionObject ) ) {
5680 return $this->mRevisionObject;
5681 }
5682 if ( is_null( $this->mRevisionId ) ) {
5683 return null;
5684 }
5685
5686 $rev = call_user_func(
5687 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5688 );
5689
5690 # If the parse is for a new revision, then the callback should have
5691 # already been set to force the object and should match mRevisionId.
5692 # If not, try to fetch by mRevisionId for sanity.
5693 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5694 $rev = Revision::newFromId( $this->mRevisionId );
5695 }
5696
5697 $this->mRevisionObject = $rev;
5698
5699 return $this->mRevisionObject;
5700 }
5701
5702 /**
5703 * Get the timestamp associated with the current revision, adjusted for
5704 * the default server-local timestamp
5705 * @return string
5706 */
5707 public function getRevisionTimestamp() {
5708 if ( is_null( $this->mRevisionTimestamp ) ) {
5709 global $wgContLang;
5710
5711 $revObject = $this->getRevisionObject();
5712 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5713
5714 # The cryptic '' timezone parameter tells to use the site-default
5715 # timezone offset instead of the user settings.
5716 # Since this value will be saved into the parser cache, served
5717 # to other users, and potentially even used inside links and such,
5718 # it needs to be consistent for all visitors.
5719 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5720
5721 }
5722 return $this->mRevisionTimestamp;
5723 }
5724
5725 /**
5726 * Get the name of the user that edited the last revision
5727 *
5728 * @return string User name
5729 */
5730 public function getRevisionUser() {
5731 if ( is_null( $this->mRevisionUser ) ) {
5732 $revObject = $this->getRevisionObject();
5733
5734 # if this template is subst: the revision id will be blank,
5735 # so just use the current user's name
5736 if ( $revObject ) {
5737 $this->mRevisionUser = $revObject->getUserText();
5738 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5739 $this->mRevisionUser = $this->getUser()->getName();
5740 }
5741 }
5742 return $this->mRevisionUser;
5743 }
5744
5745 /**
5746 * Get the size of the revision
5747 *
5748 * @return int|null Revision size
5749 */
5750 public function getRevisionSize() {
5751 if ( is_null( $this->mRevisionSize ) ) {
5752 $revObject = $this->getRevisionObject();
5753
5754 # if this variable is subst: the revision id will be blank,
5755 # so just use the parser input size, because the own substituation
5756 # will change the size.
5757 if ( $revObject ) {
5758 $this->mRevisionSize = $revObject->getSize();
5759 } else {
5760 $this->mRevisionSize = $this->mInputSize;
5761 }
5762 }
5763 return $this->mRevisionSize;
5764 }
5765
5766 /**
5767 * Mutator for $mDefaultSort
5768 *
5769 * @param string $sort New value
5770 */
5771 public function setDefaultSort( $sort ) {
5772 $this->mDefaultSort = $sort;
5773 $this->mOutput->setProperty( 'defaultsort', $sort );
5774 }
5775
5776 /**
5777 * Accessor for $mDefaultSort
5778 * Will use the empty string if none is set.
5779 *
5780 * This value is treated as a prefix, so the
5781 * empty string is equivalent to sorting by
5782 * page name.
5783 *
5784 * @return string
5785 */
5786 public function getDefaultSort() {
5787 if ( $this->mDefaultSort !== false ) {
5788 return $this->mDefaultSort;
5789 } else {
5790 return '';
5791 }
5792 }
5793
5794 /**
5795 * Accessor for $mDefaultSort
5796 * Unlike getDefaultSort(), will return false if none is set
5797 *
5798 * @return string|bool
5799 */
5800 public function getCustomDefaultSort() {
5801 return $this->mDefaultSort;
5802 }
5803
5804 private static function getSectionNameFromStrippedText( $text ) {
5805 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5806 $text = Sanitizer::decodeCharReferences( $text );
5807 $text = self::normalizeSectionName( $text );
5808 return $text;
5809 }
5810
5811 private static function makeAnchor( $sectionName ) {
5812 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5813 }
5814
5815 private static function makeLegacyAnchor( $sectionName ) {
5816 global $wgFragmentMode;
5817 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5818 // ForAttribute() and ForLink() are the same for legacy encoding
5819 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5820 } else {
5821 $id = Sanitizer::escapeIdForLink( $sectionName );
5822 }
5823
5824 return "#$id";
5825 }
5826
5827 /**
5828 * Try to guess the section anchor name based on a wikitext fragment
5829 * presumably extracted from a heading, for example "Header" from
5830 * "== Header ==".
5831 *
5832 * @param string $text
5833 * @return string Anchor (starting with '#')
5834 */
5835 public function guessSectionNameFromWikiText( $text ) {
5836 # Strip out wikitext links(they break the anchor)
5837 $text = $this->stripSectionName( $text );
5838 $sectionName = self::getSectionNameFromStrippedText( $text );
5839 return self::makeAnchor( $sectionName );
5840 }
5841
5842 /**
5843 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5844 * instead, if possible. For use in redirects, since various versions
5845 * of Microsoft browsers interpret Location: headers as something other
5846 * than UTF-8, resulting in breakage.
5847 *
5848 * @param string $text The section name
5849 * @return string Anchor (starting with '#')
5850 */
5851 public function guessLegacySectionNameFromWikiText( $text ) {
5852 # Strip out wikitext links(they break the anchor)
5853 $text = $this->stripSectionName( $text );
5854 $sectionName = self::getSectionNameFromStrippedText( $text );
5855 return self::makeLegacyAnchor( $sectionName );
5856 }
5857
5858 /**
5859 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5860 * @param string $text Section name (plain text)
5861 * @return string Anchor (starting with '#')
5862 */
5863 public static function guessSectionNameFromStrippedText( $text ) {
5864 $sectionName = self::getSectionNameFromStrippedText( $text );
5865 return self::makeAnchor( $sectionName );
5866 }
5867
5868 /**
5869 * Apply the same normalization as code making links to this section would
5870 *
5871 * @param string $text
5872 * @return string
5873 */
5874 private static function normalizeSectionName( $text ) {
5875 # T90902: ensure the same normalization is applied for IDs as to links
5876 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5877 try {
5878
5879 $parts = $titleParser->splitTitleString( "#$text" );
5880 } catch ( MalformedTitleException $ex ) {
5881 return $text;
5882 }
5883 return $parts['fragment'];
5884 }
5885
5886 /**
5887 * Strips a text string of wikitext for use in a section anchor
5888 *
5889 * Accepts a text string and then removes all wikitext from the
5890 * string and leaves only the resultant text (i.e. the result of
5891 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5892 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5893 * to create valid section anchors by mimicing the output of the
5894 * parser when headings are parsed.
5895 *
5896 * @param string $text Text string to be stripped of wikitext
5897 * for use in a Section anchor
5898 * @return string Filtered text string
5899 */
5900 public function stripSectionName( $text ) {
5901 # Strip internal link markup
5902 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5903 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5904
5905 # Strip external link markup
5906 # @todo FIXME: Not tolerant to blank link text
5907 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5908 # on how many empty links there are on the page - need to figure that out.
5909 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5910
5911 # Parse wikitext quotes (italics & bold)
5912 $text = $this->doQuotes( $text );
5913
5914 # Strip HTML tags
5915 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5916 return $text;
5917 }
5918
5919 /**
5920 * strip/replaceVariables/unstrip for preprocessor regression testing
5921 *
5922 * @param string $text
5923 * @param Title $title
5924 * @param ParserOptions $options
5925 * @param int $outputType
5926 *
5927 * @return string
5928 */
5929 public function testSrvus( $text, Title $title, ParserOptions $options,
5930 $outputType = self::OT_HTML
5931 ) {
5932 $magicScopeVariable = $this->lock();
5933 $this->startParse( $title, $options, $outputType, true );
5934
5935 $text = $this->replaceVariables( $text );
5936 $text = $this->mStripState->unstripBoth( $text );
5937 $text = Sanitizer::removeHTMLtags( $text );
5938 return $text;
5939 }
5940
5941 /**
5942 * @param string $text
5943 * @param Title $title
5944 * @param ParserOptions $options
5945 * @return string
5946 */
5947 public function testPst( $text, Title $title, ParserOptions $options ) {
5948 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5949 }
5950
5951 /**
5952 * @param string $text
5953 * @param Title $title
5954 * @param ParserOptions $options
5955 * @return string
5956 */
5957 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5958 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5959 }
5960
5961 /**
5962 * Call a callback function on all regions of the given text that are not
5963 * inside strip markers, and replace those regions with the return value
5964 * of the callback. For example, with input:
5965 *
5966 * aaa<MARKER>bbb
5967 *
5968 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5969 * two strings will be replaced with the value returned by the callback in
5970 * each case.
5971 *
5972 * @param string $s
5973 * @param callable $callback
5974 *
5975 * @return string
5976 */
5977 public function markerSkipCallback( $s, $callback ) {
5978 $i = 0;
5979 $out = '';
5980 while ( $i < strlen( $s ) ) {
5981 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5982 if ( $markerStart === false ) {
5983 $out .= call_user_func( $callback, substr( $s, $i ) );
5984 break;
5985 } else {
5986 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5987 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5988 if ( $markerEnd === false ) {
5989 $out .= substr( $s, $markerStart );
5990 break;
5991 } else {
5992 $markerEnd += strlen( self::MARKER_SUFFIX );
5993 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5994 $i = $markerEnd;
5995 }
5996 }
5997 }
5998 return $out;
5999 }
6000
6001 /**
6002 * Remove any strip markers found in the given text.
6003 *
6004 * @param string $text
6005 * @return string
6006 */
6007 public function killMarkers( $text ) {
6008 return $this->mStripState->killMarkers( $text );
6009 }
6010
6011 /**
6012 * Save the parser state required to convert the given half-parsed text to
6013 * HTML. "Half-parsed" in this context means the output of
6014 * recursiveTagParse() or internalParse(). This output has strip markers
6015 * from replaceVariables (extensionSubstitution() etc.), and link
6016 * placeholders from replaceLinkHolders().
6017 *
6018 * Returns an array which can be serialized and stored persistently. This
6019 * array can later be loaded into another parser instance with
6020 * unserializeHalfParsedText(). The text can then be safely incorporated into
6021 * the return value of a parser hook.
6022 *
6023 * @deprecated since 1.31
6024 * @param string $text
6025 *
6026 * @return array
6027 */
6028 public function serializeHalfParsedText( $text ) {
6029 wfDeprecated( __METHOD__, '1.31' );
6030 $data = [
6031 'text' => $text,
6032 'version' => self::HALF_PARSED_VERSION,
6033 'stripState' => $this->mStripState->getSubState( $text ),
6034 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6035 ];
6036 return $data;
6037 }
6038
6039 /**
6040 * Load the parser state given in the $data array, which is assumed to
6041 * have been generated by serializeHalfParsedText(). The text contents is
6042 * extracted from the array, and its markers are transformed into markers
6043 * appropriate for the current Parser instance. This transformed text is
6044 * returned, and can be safely included in the return value of a parser
6045 * hook.
6046 *
6047 * If the $data array has been stored persistently, the caller should first
6048 * check whether it is still valid, by calling isValidHalfParsedText().
6049 *
6050 * @deprecated since 1.31
6051 * @param array $data Serialized data
6052 * @throws MWException
6053 * @return string
6054 */
6055 public function unserializeHalfParsedText( $data ) {
6056 wfDeprecated( __METHOD__, '1.31' );
6057 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6058 throw new MWException( __METHOD__ . ': invalid version' );
6059 }
6060
6061 # First, extract the strip state.
6062 $texts = [ $data['text'] ];
6063 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6064
6065 # Now renumber links
6066 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6067
6068 # Should be good to go.
6069 return $texts[0];
6070 }
6071
6072 /**
6073 * Returns true if the given array, presumed to be generated by
6074 * serializeHalfParsedText(), is compatible with the current version of the
6075 * parser.
6076 *
6077 * @deprecated since 1.31
6078 * @param array $data
6079 *
6080 * @return bool
6081 */
6082 public function isValidHalfParsedText( $data ) {
6083 wfDeprecated( __METHOD__, '1.31' );
6084 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6085 }
6086
6087 /**
6088 * Parsed a width param of imagelink like 300px or 200x300px
6089 *
6090 * @param string $value
6091 * @param bool $parseHeight
6092 *
6093 * @return array
6094 * @since 1.20
6095 */
6096 public static function parseWidthParam( $value, $parseHeight = true ) {
6097 $parsedWidthParam = [];
6098 if ( $value === '' ) {
6099 return $parsedWidthParam;
6100 }
6101 $m = [];
6102 # (T15500) In both cases (width/height and width only),
6103 # permit trailing "px" for backward compatibility.
6104 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6105 $width = intval( $m[1] );
6106 $height = intval( $m[2] );
6107 $parsedWidthParam['width'] = $width;
6108 $parsedWidthParam['height'] = $height;
6109 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6110 $width = intval( $value );
6111 $parsedWidthParam['width'] = $width;
6112 }
6113 return $parsedWidthParam;
6114 }
6115
6116 /**
6117 * Lock the current instance of the parser.
6118 *
6119 * This is meant to stop someone from calling the parser
6120 * recursively and messing up all the strip state.
6121 *
6122 * @throws MWException If parser is in a parse
6123 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6124 */
6125 protected function lock() {
6126 if ( $this->mInParse ) {
6127 throw new MWException( "Parser state cleared while parsing. "
6128 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6129 }
6130
6131 // Save the backtrace when locking, so that if some code tries locking again,
6132 // we can print the lock owner's backtrace for easier debugging
6133 $e = new Exception;
6134 $this->mInParse = $e->getTraceAsString();
6135
6136 $recursiveCheck = new ScopedCallback( function () {
6137 $this->mInParse = false;
6138 } );
6139
6140 return $recursiveCheck;
6141 }
6142
6143 /**
6144 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6145 *
6146 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6147 * or if there is more than one <p/> tag in the input HTML.
6148 *
6149 * @param string $html
6150 * @return string
6151 * @since 1.24
6152 */
6153 public static function stripOuterParagraph( $html ) {
6154 $m = [];
6155 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6156 if ( strpos( $m[1], '</p>' ) === false ) {
6157 $html = $m[1];
6158 }
6159 }
6160
6161 return $html;
6162 }
6163
6164 /**
6165 * Return this parser if it is not doing anything, otherwise
6166 * get a fresh parser. You can use this method by doing
6167 * $myParser = $wgParser->getFreshParser(), or more simply
6168 * $wgParser->getFreshParser()->parse( ... );
6169 * if you're unsure if $wgParser is safe to use.
6170 *
6171 * @since 1.24
6172 * @return Parser A parser object that is not parsing anything
6173 */
6174 public function getFreshParser() {
6175 global $wgParserConf;
6176 if ( $this->mInParse ) {
6177 return new $wgParserConf['class']( $wgParserConf );
6178 } else {
6179 return $this;
6180 }
6181 }
6182
6183 /**
6184 * Set's up the PHP implementation of OOUI for use in this request
6185 * and instructs OutputPage to enable OOUI for itself.
6186 *
6187 * @since 1.26
6188 */
6189 public function enableOOUI() {
6190 OutputPage::setupOOUI();
6191 $this->mOutput->setEnableOOUI( true );
6192 }
6193 }