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